Obgyn.ubc.ca

T h e n e w e ng l a n d j o u r na l o f m e dic i n e Metformin versus Insulin for the Treatment Janet A. Rowan, M.B., Ch.B., William M. Hague, M.D., Wanzhen Gao, Ph.D., Malcolm R. Battin, M.B., Ch.B., and M. Peter Moore, M.B., Ch.B., Background
Metformin is a logical treatment for women with gestational diabetes mellitus, but From National Women’s Health, Auck-
land City Hospital, Auckland (J.A.R.); the randomized trials to assess the efficacy and safety of its use for this condition are National Institute of Public Health and Mental Health Research, Faculty of Health and Environmental Sciences, Auckland University of Technology, Auckland (W.G.); the Department of Pediatrics, University We randomly assigned 751 women with gestational diabetes mellitus at 20 to 33 weeks of Auckland, Auckland City Hospital, of gestation to open treatment with metformin (with supplemental insulin if re- Auckland (M.R.B.); and the Diabetes Cen- tre, Christchurch Hospital, Christchurch quired) or insulin. The primary outcome was a composite of neonatal hypoglycemia, (M.P.M.) — all in New Zealand; and the respiratory distress, need for phototherapy, birth trauma, 5-minute Apgar score less Department of Obstetrics, Women’s and than 7, or prematurity. The trial was designed to rule out a 33% increase (from 30% Children’s Hospital, University of Ade- to 40%) in this composite outcome in infants of women treated with metformin as dress correspondence to Dr. Rowan at compared with those treated with insulin. Secondary outcomes included neonatal National Women’s Health, SMO Room, anthropometric measurements, maternal glycemic control, maternal hypertensive 9th Floor Support Bldg., Auckland City complications, postpartum glucose tolerance, and acceptability of treatment.
Gestational Diabetes (MiG) Trial are list- Of the 363 women assigned to metformin, 92.6% continued to receive metformin ed in the Appendix.
until delivery and 46.3% received supplemental insulin. The rate of the primary com- posite outcome was 32.0% in the group assigned to metformin and 32.2% in the N Engl J Med 2008;358:2003-15.
Copyright 2008 Massachusetts Medical Society. insulin group (relative risk, 0.99; 95% confidence interval, 0.80 to 1.23). More women in the metformin group than in the insulin group stated that they would choose to receive their assigned treatment again (76.6% vs. 27.2%, P<0.001). The rates of other secondary outcomes did not differ significantly between the groups. There were no serious adverse events associated with the use of metformin.
Conclusions
In women with gestational diabetes mellitus, metformin (alone or with supplemental
insulin) is not associated with increased perinatal complications as compared with insulin. The women preferred metformin to insulin treatment. (Australian New Zealand Clinical Trials Registry number, 12605000311651.) n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e Gestational diabetes is a complica- Study Subjects
tion in about 5% of pregnancies, is in- Women were eligible for inclusion if they were be- creasing in prevalence, and is associated tween 18 and 45 years of age, had received a diag- with complications to the pregnancy and a long- nosis of gestational diabetes mellitus according to term risk of diabetes in both mother and off- the criteria of the Australasian Diabetes in Preg- spring.1-5 Intervention to change lifestyle and, if nancy Society (ADIPS),25 were pregnant with a sin- maternal hyperglycemia persists, treatment with gle fetus between 20 and 33 weeks of gestation, additional insulin have been shown to improve met the hospital’s usual criteria for starting insulin perinatal outcomes.6,7 Women who begin insulin treatment, and, after lifestyle intervention consist- therapy require education to ensure the safe ad- ing of advice about diet and exercise, had more than ministration of insulin. Use of insulin is also as- one capillary blood glucose measurement above sociated with hypoglycemia and weight gain. The 5.4 mmol per liter (97.2 mg per deciliter) after an use of safe and effective oral agents may offer ad- overnight fast or more than one 2-hour postpran- dial blood glucose measurement above 6.7 mmol Oral metformin is a logical option for women per liter (120.6 mg per deciliter). The exclusion with gestational diabetes mellitus. It improves in- criteria were a prepregnancy diagnosis of diabe- sulin sensitivity, probably by activating AMP ki- tes, a contraindication to metformin, a fetal anom- nase, and is not associated with weight gain or aly, gestational hypertension, preeclampsia, fetal hypoglycemia.8,9 Reported outcomes of its use dur- growth restriction, and ruptured membranes.
ing pregnancy have been favorable10-19 except for Randomization was performed with a block one small, retrospective cohort study20 that showed size of four and was stratified according to site increased rates of perinatal loss and preeclampsia and gestational age (from 20 weeks to 27 weeks as compared with insulin treatment. Metformin 6 days, or from 28 weeks to 33 weeks 6 days). All crosses the placenta and could affect fetal phys- sites agreed to aim for the capillary glucose levels iology directly.21 Its use in pregnancy remains recommended by the ADIPS25 (level after an over- controversial; to our knowledge, only two small, night fast, <5.5 mmol per liter [99 mg per deci- randomized trials addressing this question have liter]; 2-hour postprandial level, <7.0 mmol per liter [126 mg per deciliter]), although several sites We designed the Metformin in Gestational Dia- aimed for lower levels.
betes Trial to rule out a 33% increase in a compos- The women obtained the prescribed medica- ite of perinatal complications in infants of women tions from their local pharmacies. Metformin treated with metformin as compared with insulin. (Metomin in New Zealand [Pacific Pharmaceuti- Our hypotheses were that perinatal outcomes cals] and Diaformin in Australia [Alphapharm and would be similar for both treatments, that wom- other nonspecified manufacturers]) was started en would consider metformin a more acceptable at a dose of 500 mg once or twice daily with treatment than insulin, and that metformin would food and increased, typically over a period of 1 to improve markers of insulin sensitivity in the 2 weeks, to meet glycemic targets up to a maxi- mum daily dose of 2500 mg. If the targets were not achieved with metformin alone, insulin was added. Metformin was stopped if maternal con- traindications (such as liver or renal impairment or Study Design
sepsis) or fetal growth restriction developed. In- We conducted this randomized, open-label trial sulin was prescribed according to usual practice.
comparing metformin with insulin treatment in 10 New Zealand and Australian urban obstetrical Data Collection
hospitals. The ethics review boards at all sites ap- Demographic and clinical data were recorded at proved the study, and the participants gave writ- enrollment. Blood samples were obtained after an ten informed consent. Details of the study design overnight fast to assess baseline glycemia and to have been published elsewhere.24 Dr. Rowan vouch- ensure that the results of renal- and liver-function es for the integrity and completeness of the data. tests did not preclude the use of metformin. Glu- n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. 751 Women with gestational diabetes mellitus who were pregnant with a single fetus at 20–33 wk of gestation and required medication underwent randomization Figure 1. Enrollment of Subjects.
1 of 1 processing within 90 minutes. After centrifuga- SIZE ncentration was measured
Combo le laboratory with the use of a Roche Di- those of the Diabetes Control and Complications agnostics Elec AUTHOR, PLEASE NOTE: sys 2010 automated analyzer and an
Trial. The women performed blood glucose mea- Elecsys 1010/2010 ins Figure has been redrawn and type has been reset.
Please check carefully.
surements at home with the use of a MediSense equal volume of 25% polyethylene glycol-6000 was 2-hour postprandial results were downloaded and before analysis. The coefficients of variation of transcribed into the database. At 36 to 37 weeks insulin concentration were 1.8% at 1050 pmol per of gestation, venous plasma glucose was again liter (175 mU per liter), 2.0% at 330 pmol per liter measured after an overnight fast. At delivery, com- (55 mU per liter), and 4.8% at 38 pmol per liter plications of pregnancy, the indication for deliv- (6 mU per liter).
ery, the mode of delivery, and neonatal complica- Study Outcomes
After the umbilical cord had been clamped, The primary outcome was a composite of neona- cord blood was collected in tubes containing EDTA tal complications, with components chosen to re- and in plain tubes and sent for processing within flect important adverse effects of fetal exposure to n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e maternal hyperglycemia that might be modified ing until consecutive glucose values of 2.6 mmol by treatment and directly influenced by the pas- per liter (46.8 mg per deciliter) or greater were sage of metformin across the placenta.24 The com- achieved. Readings below 2.6 mmol per liter and ponents of the composite outcome were neonatal below 1.6 mmol per liter (28.8 mg per deciliter) hypoglycemia (two or more neonatal glucose val- were documented, as was treatment for hypogly- ues <2.6 mmol per liter [46.8 mg per deciliter]), cemia.
respiratory distress (need for at least 4 hours of re- Maternal hypertensive complications were di- spiratory support with supplemental oxygen, con- agnosed according to Australasian guidelines.26 tinuous positive airway pressure, or intermittent Birth-weight percentiles were calculated with the positive-pressure ventilation during the first 24 use of a customized calculator27 that adjusts for hours after delivery), need for phototherapy, birth sex and gestational age of the infant, as well as trauma (injury to the baby at delivery, documented maternal height, weight in early pregnancy, ethnic as mild if bruises or abrasions were present at birth group, and parity. Neonatal anthropometric mea- but resolved before 6 weeks post partum; more se- surements, including crown–heel length, crown– rious injuries were also recorded), 5-minute Apgar rump length, head circumference, chest circum- score below 7, or premature birth (<37 weeks of ference, abdominal circumference, mid-upper-arm gestation). The neonates were monitored for hy- circumference, triceps skin-fold thickness, and poglycemia by measuring blood glucose levels subscapular skin-fold thickness, were obtained within 2 hours after birth and before each feed- within 48 hours after birth by trained personnel.24 Table 1. Maternal Characteristics at Baseline.*
Metformin Group
Insulin Group
Characteristic
≥3 Pregnancy terminations or miscarriages n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. Table 1. (Continued.)
Metformin Group
Insulin Group
Characteristic
Results of preintervention 75-g oral glucose-tolerance Plasma glucose level after an overnight fast Plasma glucose level at enrollment after an overnight fast * Plus–minus values are means ±SD.
† Body-mass index is the weight in kilograms divided by the square of the height in meters.
‡ The body-mass index values in early pregnancy (i.e., before 20 weeks of gestation) are based on 281 women in the metformin group and 304 in the insulin group.
§ Race and ethnic group were self-reported.
¶ The Polynesian subjects included 137 Pacific Islanders and 19 Maori.
‖ Nulliparous women had had no previous pregnancy beyond 20 weeks.
** P = 0.03; all other differences were not significant (P≥0.05).
†† A family history was recorded if the condition was present in a first-degree or second-degree relative.
‡‡ In women who did not have a 75-g oral glucose-tolerance test, the investigator verified the presence of gestational di- abetes mellitus by the presence of a high glucose level in a random sample or a sample taken 1 hour after a 50-g glu- cose load, and lifestyle intervention and capillary glucose monitoring were initiated before the women were consid- ered for enrollment in the trial. The values for fasting plasma glucose are based on 321 women in the metformin group and 330 in the insulin group, the values for 2-hour postprandial plasma glucose are based on 320 women in the metformin group and 319 in the insulin group, and the values for fasting plasma glucose at enrollment are based on 324 women in the metformin group and 294 in the insulin group. To convert values for glucose to millimoles per §§ The values are based on 345 women in the metformin group and 319 in the insulin group.
A questionnaire was administered to the mothers at discharge from the hospital. In cases in which in the first postpartum week to assess acceptabil- fetal pH was measured in blood from the scalp or umbilical cord, the lowest value was recorded. Adverse events were reported to the data and The secondary outcome measures were maternal safety monitoring committee. Side effects of med- and neonatal body composition, maternal glyce- ication and complications of pregnancy were docu- mic control, maternal hypertensive complications, mented at clinic visits, and the investigators were maternal glucose tolerance at 6 to 8 weeks post informed of hospitalizations. Congenital anom- partum, and acceptability of treatment.
alies and events that were fatal, life-threatening, associated with serious disability or incapacity, re- Statistical Analysis
quired prolonged hospitalization (apart from hos- The pretrial estimate of the frequency of the pri- pitalization related to expected pregnancy events), mary outcome was 30% (on the basis of local data or required a major intervention to prevent an- from women who had been treated with insulin). other serious outcome were classified as serious The anticipated rates for each component were 14% adverse events. Other measures of neonatal com- for hypoglycemia, 5% for respiratory distress, 5% plications were admission to a level 2 or level 3 for phototherapy, 1.5% for birth trauma, less than neonatal intensive care unit, duration of stay in 1% for Apgar scores below 7, and 15% for preterm the neonatal intensive care unit, and diagnosis delivery. The infants could meet one or more of n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e Table 2. Primary Outcome and Additional Neonatal Complications.*
Metformin
Insulin Group
Relative Risk
Primary composite outcome
Recurrent blood glucose level <46.8 mg/dl† Additional neonatal complications
Admission to level 2 or 3 neonatal intensive care unit
>24-Hr stay in neonatal intensive care unit * Treatment included supplemental feeding for 129 infants (35.5%) in the metformin group and 145 (39.2%) in the insu- lin group (P = 0.31), nasogastric feeding for 9 infants (2.5%) in the metformin group and 14 (3.8%) in the insulin group (P = 0.31), and intravenous dextrose for 25 infants (6.9%) in the metformin group and 22 (5.9%) in the insulin group (P = 0.60). Induction of labor was performed in 196 women (54.0%) in the metformin group and 208 (56.2%) in the in- sulin group (P = 0.55), cesarean section in 131 women (36.1%) in the metformin group and 142 (38.4%) in the insulin group (P = 0.52), and emergency cesarean section in 55 women (15.2%) in the metformin group and 63 (17.0%) in the insulin group (P = 0.49). Shoulder dystocia occurred in 6 deliveries (1.7%) in the metformin group and 11 (3.0%) in the † To convert values for glucose to millimoles per liter, divide by 18.
‡ Nine infants in each group were treated with continuous positive airway pressure. Five infants in the insulin group re- quired intermittent positive-pressure ventilation (one of these infants subsequently received a diagnosis of tetralogy of § Moderate or severe birth trauma included Erb’s palsy, which resolved by 6 weeks of age in one infant, and a severe bra- ¶ All these infants had 5-minute Apgar scores of 6. In the metformin group, one infant had facial bruising and one pre- term infant was admitted to the neonatal intensive care unit to establish feeding. Delivery was complicated by shoulder dystocia in one infant in the insulin group, who was admitted to the neonatal intensive care unit for observation.
‖ The values are based on 94 infants in the metformin group and 101 in the insulin group.
the criteria. The primary aim of the study was to The data were summarized as frequencies or rule out a clinically significant increase (from 30% percentages for categorical variables and as means to 40%) in the metformin group. Two-tailed cal- and standard deviations or medians and interquar- culations were used to rule out a significant dif- tile ranges for continuous variables, depending on ference in either direction. For 80% power and a the distribution. Differences between the treat- 5% significance level, 375 subjects were required ment groups were compared by the chi-square or Fisher’s exact test for categorical variables and a n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. two-sample t-test or Mann–Whitney test for con- metformin was 2500 mg (interquartile range, 1750 tinuous variables. In addition to comparisons be- to 2500). For women who required supplemental tween the randomly assigned groups, data from insulin, the median maximum daily dose of in- women in the metformin group who were treated sulin was 42 units (interquartile range, 22 to 81), with metformin alone and those who had sup- which was lower than the maximum daily dose in plemental insulin were analyzed separately, with those assigned to insulin (50 units; interquartile recognition that these were not randomized sub- range, 30 to 90) (P = 0.002). Supplemental insulin groups. Analyses were performed with SAS soft- was started at a median of 20.4 days (interquar- ware, version 9.1 for Windows. Relative risks are tile range, 12.4 to 27.5) after random assignment reported with 95% confidence intervals.
Two interim analyses were conducted and re- viewed by the data and safety monitoring commit- Study Outcomes
tee. P values were adjusted by the Peto–Haybittle The primary outcome — a composite of neonatal method. The investigators were to be informed if complications — did not differ significantly be- a difference in the primary outcome between treat- tween groups (32.0% in the metformin group and ment groups of three or more standard deviations 32.2% in the insulin group, P = 0.95) (Table 2). Of the outcomes included in the composite, severe hypoglycemia (glucose level <1.6 mmol per li- ter) was less common in the metformin group (P = 0.008), but preterm birth (before 37 weeks of Study Subjects
gestation) was more common in the metformin Enrollment of women for the study was complet- group (P = 0.04) (Table 2). There was one delivery ed between October 2002 and November 2006 before 32 weeks of gestation in each treatment (Fig. 1). Of the 751 women enrolled, 6 underwent group. In the insulin group, delivery was due to randomization twice as a result of investigator an intrauterine fetal death at 29 weeks of gesta- errors; the errors were immediately recognized, tion, the only perinatal loss in the study. This and the first randomized treatment was assigned. mother had been recruited at 20 weeks of gesta- Data after randomization were not available for tion with a glycated hemoglobin value of 10.2%; 10 women in the metformin group and 8 in the at postmortem examination, the fetus had the insulin group, who therefore could not be includ- Budd–Chiari syndrome. In an analysis according ed in the analysis. The analyses included the re- to indication for preterm birth, the frequency of maining 363 women in the metformin group and iatrogenic preterm births was similar in both treat- the 370 in the insulin group and were performed ment groups, but there was a trend toward more according to the intention-to-treat principle.
spontaneous preterm births (spontaneous labor or Supplemental insulin was required in 168 wom- preterm ruptured membranes) in the metformin en (46.3%) in the metformin group. Metformin group (Table 2). There was a statistically signifi- treatment was stopped in 27 women (7.4%) before cant but clinically small difference in the mean delivery (Fig. 1). Treatment was stopped in 11 of gestational age at delivery between the metformin these women in accordance with the trial protocol group (38.3 weeks) and the insulin group (38.5 (9 women had obstetrical complications, 1 had weeks, P = 0.02).
sepsis, and 1 had worsening abnormal liver- function test results); treatment was stopped in Secondary Outcomes and Additional
7 women (1.9%) because of gastrointestinal side Measures of Neonatal Complications
effects; 5 women chose to stop metformin; and The secondary outcomes are summarized in Ta- 4 women were advised to stop by other health ble 3. There were no significant differences be- professionals who were not involved in the trial. tween the groups in neonatal anthropometric mea- Metformin doses were reduced because of gastro- sures or measurements of umbilical-cord serum intestinal side effects in 32 women (8.8%); all but insulin concentrations. Although the overall mean 1 of these women were able to maintain a dose of maternal 2-hour postprandial glucose levels were slightly lower in the metformin group, these val- The baseline characteristics of the two groups ues did not differ significantly between the groups were similar (Table 1). The median daily dose of in the 2 weeks before delivery (Table 3), a finding n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e Table 3. Secondary Outcomes and Additional Measurements.*
Metformin
Insulin Group
Variable
Neonatal
Birth weight <10th percentile — no. (%) Birth weight >90th percentile — no. (%) Subscapular skin-fold thickness — mm‖ Umbilical-cord serum insulin concentration — pmol/liter†† Maternal‡‡
Glycemic control from randomization until delivery Capillary glucose level after an overnight fast — mg/dl 2-Hr postprandial capillary glucose level — mg/dl Glycemic control at 1 week after randomization Capillary glucose level after an overnight fast — mg/dl Postprandial capillary glucose level — mg/dl Glycemic control during the last 2 wk before delivery Capillary glucose level after an overnight fast — mg/dl 2-Hr postprandial capillary glucose — mg/dl Glycated hemoglobin at wk 36–37 — %§§ Plasma glucose level at wk 36–37 after an overnight fast — mg/dl¶¶ Capillary glucose level 12 hr before delivery — mg/dl‖‖ Hypertensive complications — no. (%)*** Results of 75-g oral glucose-tolerance test at 6 to 8 wk post partum††† 2-Hr postprandial plasma glucose level — mg/dl Impaired glucose level after an overnight fast — no./total no. (%)‡‡‡ Impaired glucose tolerance — no./total no. (%)§§§ n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. Table 3. (Continued.)
Metformin
Insulin Group
Variable
Loss from enrollment to postpartum visit ¶¶¶ Gain from early pregnancy to enrollment‖‖‖ Gain from enrollment to 36 or 37 wk of gestation**** The values are based on 294 subjects in the metformin group and 297 in the insulin group. The values are based on 299 subjects in the metformin group and 301 in the insulin group. The values are based on 298 subjects in the metformin group and 300 in the insulin group. The values are based on 292 subjects in the metformin group and 289 in the insulin group. The values are based on 291 subjects in the metformin group and 289 in the insulin group. ** The ponderal index is calculated as 100 times the birth weight in grams divided by the cube of the crown–heel †† There were 140 infants in the metformin group and 154 in the insulin group. To convert values for insulin to micro- grams per liter, divide by 172.2. The P value was calculated by a t-test for the logarithm of umbilical-cord serum in- ‡‡ To convert values for glucose to millimoles per liter, divide by 18.
The values are based on 280 subjects in the metformin group and 308 in the insulin group. ¶¶ The values are based on 279 subjects in the metformin group and 300 in the insulin group.
‖‖ The values are based on 311 subjects in the metformin group and 327 in the insulin group.
*** The diagnosis of hypertensive complications was made according to the Australasian Hypertension in Pregnancy ††† Results are for 270 women in the metformin group and 282 in the insulin group, except for body-mass index (the weight in kilograms divided by the square of the height in meters), for which the results are for 301 women in the metformin group and 304 in the insulin group. ‡‡‡ Impaired glucose level after an overnight fast is defined as a plasma glucose level of 6.1 to 6.9 mmol per liter (110 §§§ Impaired glucose tolerance is defined as a 2-hour postprandial plasma glucose level of 7.8 to 11.0 mmol per liter ¶¶¶ The values are based on 301 subjects in the metformin group and 303 in the insulin group.
‖‖‖ The values are based on 281 subjects in the metformin group and 302 in the insulin group.
**** The values are based on 329 subjects in the metformin group and 300 in the insulin group. The weight gain from enrollment to 36 or 37 weeks of gestation in women taking metformin and supplemental insulin was 0.8±3.2 kg, as compared with 2.0±3.3 kg in those taking insulin alone (P<0.001).
suggesting that glucose targets were reached soon- in the insulin group said they would choose insu- er in the metformin group. The rates of maternal lin again (P<0.001). More women in the metformin hypertensive complications did not differ signifi- group than in the insulin group said that taking medication was the easiest part of treatment At 6 to 8 weeks post partum, 552 women (59.0% vs. 35.3%, P<0.001), and fewer women in (75.3%) underwent a 75-g oral glucose-tolerance the metformin group than in the insulin group test. In 62 of 270 women (23.0%) in the metformin said that taking medication was the hardest part group and 58 of 282 (20.6%) in the insulin group, of treatment (10.5% vs. 27.2%, P<0.001).
glucose tolerance was impaired or diagnostic of Serious adverse events affecting either the diabetes. Women in the metformin group had mother or the child are summarized in Table 5. greater weight loss between the time of enroll- There were no significant differences between the ment and the postpartum visit and less weight treatment groups, and none of the serious adverse gain between the time of enrollment and 36 events were considered by the data and safety weeks of gestation than did women in the insulin monitoring committee to be treatment-related.
The results of the postpartum questionnaire Metformin Treatment and Supplemental
assessing acceptability of treatment are shown in insulin
Table 4. Among women treated with metformin, As compared with women who were treated with 76.6% said they would choose metformin in a metformin alone, women requiring supplemental subsequent pregnancy, whereas 27.2% of women insulin had a higher BMI and had higher baseline n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e Table 4. Results of Questionnaire on Acceptability of Treatment.*
Metformin Group
Insulin Group
Question
How often did you forget to take your medication?† Which medication would you choose in another pregnancy? In another pregnancy, if you were told you were likely to need insulin injections to control the sugar levels but could try metformin first, what would you prefer? Start with metformin and add insulin if needed Which part of your diabetes treatment was the easiest? Which part of your diabetes treatment was the hardest? * Percentages may not sum to 100 because of rounding.
† One woman in the metformin group did not respond to this question.
glucose levels (see the Supplementary Appendix, groups, but severe hypoglycemia (<1.6 mmol of available with the full text of this article at www. glucose per liter [28.8 mg per deciliter]) occurred nejm.org). The rates of the primary outcome did significantly less often in infants of women tak- not differ between women treated with metformin ing metformin. We did not see a reduction in the alone and those treated with supplemental insu- insulin concentration in umbilical-cord serum in lin (29.7% and 34.5%, respectively; relative risk, the metformin group, but this finding should be 0.86; 95% confidence interval [CI], 0.64 to 1.16; interpreted with caution, since results were avail- able for less than half of the participants.
Prematurity was included as part of the com- posite outcome. The rationale was that if metfor- min had any unanticipated adverse effect on fetal We found no significant increase in a composite growth or well-being, there would be more iatro- measure of neonatal complications among women genic preterm births. The frequency of preterm with gestational diabetes mellitus who were ran- birth was higher in the metformin group than in domly assigned to metformin as compared with the insulin group, but the difference was associ- those who were assigned to insulin. Rates of neo- ated with a greater frequency of spontaneous natal hypoglycemia, one of the components of the (rather than iatrogenic) preterm births that could composite end point, were similar in the two be due to chance or to an unrecognized effect of n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. Table 5. Adverse Events.
Metformin Group
Insulin Group
Serious maternal adverse events
Pelvic arthropathy requiring opiate analgesia Serious fetal or neonatal adverse events
Neonatal infection requiring hospitalization§ Important maternal adverse events‖
Gastrointestinal events resulting in dose limiting Gastrointestinal events resulting in treatment cessation * In the metformin group, one mother underwent surgery for cholecystectomy and one for placement of a right ureteric stent; in the insulin group, one mother had an appendectomy for appendicitis complicated by abscess.
† In the metformin group, there was one mother each with antepartum severe superficial thrombophlebitis, antepartum granulomatous mastitis, postpartum iatrogenic pulmonary edema, and postpartum anaphylactic reaction to anesthetic with stridor. In the insulin group, there were two mothers with severe preeclampsia (requiring prolonged hospitaliza- tion or care in an obstetrical intensive care unit) and one mother each with postpartum hemorrhage requiring transfu- ‡ In the metformin group, two infants had small ventricular septal defects, three had hypospadias, one had hydro- nephrosis, two had talipes equinovarus, two had a dislocatable hip, and one had a bifid thumb. In the insulin group, two infants had ventricular septal defects, one had coarctation of the aorta, one had tetralogy of Fallot, one had an atri- al septal defect with a minor cleft palate, one had choanal atresia, one had a ureterocele, one had hydronephrosis, one had hemifacial microsomia, three had talipes equinovarus, three had preauricular skin tags, and three had a dislocat- § In the metformin group, one infant each had gastroenteritis, viral pneumonia, renal sepsis, and sepsis of unknown source. In the insulin group, one infant each had viral meningitis and cytomegalovirus infection.
¶ In the metformin group, one infant each had neonatal abstinence syndrome due to maternal use of opiates, stridor due to a short aryepiglottic fold, sensorineural hearing loss (probably due to a connexin mutation), and inguinal hernia re- quiring repair. In the insulin group, one infant each had trisomy 13 and hypoxic encephalopathy with Erb’s palsy.
‖ Important adverse events were those leading to dose reduction or cessation of treatment or those occurring with a fre- metformin on the labor process. The increased mean gestational age at delivery was clinically in- rate of preterm birth was not associated with significant. In a previous cohort of women with higher rates of other complications, probably be- gestational diabetes mellitus who were treated cause the difference between the two groups in with either insulin or glyburide, the rates of pre- n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. T h e n e w e ng l a n d j o u r na l o f m e dic i n e term delivery were 13% for the insulin group and ments). Nonetheless, the 95% confidence inter- 12% for the glyburide group, but the causes of val for the relative risk of the composite measure of neonatal complications suggests that an in- In our study, 46.3% of women taking metfor- crease in risk of more than 23% with metformin min required supplemental insulin. This propor- is implausible. In addition, a post hoc analysis us- tion is likely to vary in other populations, depend- ing a noninferiority design and a proposed mar- ing on patient characteristics and target levels of gin of 1.33 (33% change in complications) sup- ports the conclusion that metformin is not inferior How do the effects of metformin in women to insulin (relative risk, 0.99; 97.5% CI, 0.78 to with gestational diabetes mellitus compare with 1.26). Finally, the composite outcome included those reported for glyburide? In a randomized outcomes of differing clinical significance.32 How- trial comparing glyburide and insulin in 404 ever, data pertaining to the individual compo- women with gestational diabetes mellitus, glyce- nents of the composite measure serve to inform mic control and pregnancy outcomes were simi- clinicians about several relevant effects of met- lar between groups (although the trial was un- formin.24 derpowered to address neonatal complications).29 Clinicians may remain circumspect about us- Subsequent experience with glyburide shows that ing metformin until follow-up data for offspring approximately 20% of women change to insu- are available. The offspring from this trial are lin.30,31 To our knowledge, there are no pub- being assessed at 2 years of age. Data at 18 months lished trials comparing metformin with glybu- of age from 126 infants of women with polycys- ride, and comparisons among available data are tic ovarian syndrome who were treated with met- limited by differences in study populations and formin have provided preliminary reassurance of a lack of effect on growth and on motor and Although our study was not designed to com- social development.33 pare the outcome of combined treatment with that In conclusion, our findings suggest that met- of either treatment alone, the rate of neonatal com- formin, alone or with supplemental insulin, is an plications did not differ significantly between effective and safe treatment option for women women who required supplemental insulin and with gestational diabetes mellitus who meet the those who received metformin alone. Moreover, usual criteria for starting insulin, and that met- women receiving combined treatment required formin is more acceptable to women with gesta- less insulin and gained less weight than those tional diabetes mellitus than is insulin. Further follow-up data are needed to establish long-term Strengths of this trial are that it took place safety.
within routine clinical practice and included the Supported by grants from the Auckland Medical Research spectrum of women with a diagnosis of gesta- Foundation, the National Women’s Evelyn Bond Charitable tional diabetes mellitus. A weakness is that treat- Trust, the Health Research Council of New Zealand, and the ment was open-label, since blinding was not National Health and Medical Research Council of Australia.
Dr. Moore reports receiving speaking fees from Sanofi-Aven- considered practical or ethical. A methodologic tis. No other potential conflict of interest relevant to this article limitation is that we used a superiority design to was reported. assess whether insulin was superior to metformin We thank all the clinic teams that provided daily support to the researchers and all the women with gestational diabetes mellitus and have accepted rather than proved the null hy- who carefully considered the study and agreed to take part.
pothesis (that there is no difference between treat- Appendix
The following investigators participated in the Metformin in Gestational Diabetes Trial: steering committee members in Auckland, New Zealand — J. Rowan, T. Cundy, R. Elder, M. Battin; investigators in Adelaide, Australia, providing input to the steering committee — B. Hague, R. Haslam; site coordinators — A. Siegers (overall coordinator), M. Cropper, J. Rafferty, S. Coat (Australia national coordinator), C. Parker, A. Barry, G. Smith, T. Clarke, M. Cram, S. Hendon; data and safety monitoring committees, Green Lane (Auckland) and Flinders (Adelaide) Clinical Trial Centers — J. Harding, L. McCowan, R. Cutfield, W. Gao; umbilical-cord blood assays — P. Skidmore, EndoLab Christchurch Hospital; study sites and principal investigators (with numbers of subjects in parentheses): National Women’s Hospital, Auckland, New Zealand — J. Rowan (282); Middlemore Hospital, Auckland, New Zealand — J. Rowan, J. Griffiths (114); Women’s and Children’s Hospital, Adelaide, Australia — W. Hague (181); King Edward Memorial Hospital, Perth, Australia — D. Graham, B. Walters (56); Royal Women’s Hospital, Brisbane, Australia — K. Lust (29); Mater Misericordae Hospital, Brisbane, Austra- lia — H.D. McIntyre (6); Christchurch Women’s Hospital, Christchurch, New Zealand — P. Moore (29); Royal Women’s Hospital, Melbourne, Australia — J. Oats, P. Wein (18); Wellington Hospital, Wellington, New Zealand — C. Eagleton (11); Blacktown Hospital, n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. References
1. Ben-Haroush A, Yogev Y, Hod M. Epi-
14. Gilbert C, Valois M, Koren G. Preg-
demiology of gestational diabetes mellitus nancy outcome after first-trimester expo- betes [MiG] trial). Diabetes Care 2007;30: and its association with Type 2 diabetes. sure to metformin: a meta-analysis. Fertil Suppl:S214-S219.
25. Hoffman L, Nolan C, Wilson JD, Oats
2. Dabelea D, Hanson RL, Lindsay RS, et 15. Glueck CJ, Bornovali S, Pranikoff J, JJ, Simmons D. Gestational diabetes mel-
al. Intrauterine exposure to diabetes con- Goldenberg N, Dharashivkar S, Wang P. litus — management guidelines. Med J veys risks for type 2 diabetes and obesity: Metformin, pre-eclampsia, and pregnan- Aust 1998;169:93-7.
a study of discordant sibships. Diabetes cy outcomes in women with polycystic 26. Brown MA, Hague WM, Higgins J, et
ovary syndrome. Diabet Med 2004;21:829- al. The detection, investigation and man- 3. Ferrara A, Kahn HS, Quesenberry CP, 36.
Riley C, Hedderson MM. An increase in the 16. Glueck CJ, Phillips H, Cameron D, full consensus statement. Aust N Z J Ob-
incidence of gestational diabetes mellitus: Sieve-Smith L, Wang P. Continuing met- stet Gynaecol 2000;40:139-55.
Northern California, 1991-2000. Obstet formin throughout pregnancy in women 27. McCowan L, Stewart AW, Francis A,
Gynecol 2004;103:526-33. [Erratum, Obstet with polycystic ovary syndrome appears Gardosi J. A customised birthweight cen- to safely reduce first-trimester spontane- tile calculator developed for a New Zea- 4. Silverman BL, Metzger BE, Cho NH, ous abortion: a pilot study. Fertil Steril land population. Aust N Z J Obstet Gyne-
Loeb CA. Impaired glucose tolerance in 2001;75:46-52.
adolescent offspring of diabetic mothers: 17. Glueck CJ, Wang P, Goldenberg N, 28. Jacobson GF, Ramos GA, Ching JY,
relationship to fetal hyperinsulinism. Di- Sieve-Smith L. Pregnancy outcomes among Kirby RS, Ferrara A, Field R. Comparison women with polycystic ovary syndrome of glyburide and insulin for the manage- 5. Schaefer-Graf UM, Pawliczak J, Pas-
treated with metformin. Hum Reprod 2002; ment of gestational diabetes in a large sow D, et al. Birth weight and parental 17:2858-64.
BMI predict overweight in children from 18. Gutzin SJ, Kozer E, Magee LA, Feig DS, Gynecol 2005;193:118-24.
mothers with gestational diabetes. Diabe- Koren G. The safety of oral hypoglycemic 29. Langer O, Conway DL, Berkus MD,
agents in the first trimester of pregnancy: Xenakis EM, Gonzales O. A comparison 6. Crowther CA, Hiller JE, Moss JR, a meta-analysis. Can J Clin Pharmacol of glyburide and insulin in women with
McPhee AJ, Jeffries WS, Robinson JS. Ef- 2003;10(4):179-83.
fect of treatment of gestational diabetes 19. Hughes RCE, Rowan JA. Pregnancy in Med 2000;343:1134-8.
mellitus on pregnancy outcomes. N Engl women with Type 2 diabetes: who takes 30. Conway DL, Gonzales O, Skiver D.
metformin and what is the outcome? Dia- Use of glyburide for the treatment of ges- 7. Langer O, Yogev Y, Most O, Xenakis bet Med 2006;23:318-22.
tational diabetes: the San Antonio experi- EMJ. Gestational diabetes: the conse- 20. Hellmuth E, Damm P, Molsted-Peder- ence. J Matern Fetal Neonatal Med 2004;
quences of not treating. Am J Obstet Gy- son L. Oral hypoglycaemic agents in 118 15:51-5.
diabetic pregnancies. Diabet Med 2000;17: 31. Rochon M, Rand L, Roth L, Gaddipati
8. Kirpichnikov D, McFarlane SI, Sowers 507-11.
JR. Metformin: an update. Ann Intern Med 21. Charles B, Norris R, Xiao X, Hague W. tational diabetes: risk factors predictive
Population pharmacokinetics of metfor- of failure and associated pregnancy out- 9. Hawthorne G. Metformin use and di-
min in late pregnancy. Ther Drug Monit comes. Am J Obstet Gynecol 2006;195: abetic pregnancy — has its time come? 2006;28:67-72.
22. Vanky E, Salvesen KA, Heimstad R, 32. Ross S. Composite outcomes in ran-
10. Coetzee EJ, Jackson WP. Pregnancy in Fougner KJ, Romundstad P, Carlsen SM. domized trials: arguments for and against.
established non-insulin-dependent dia- Metformin reduces pregnancy complica- Am J Obstet Gynecol 2007;196(2):119. betics. S Afr Med J 1980;58:795-802.
tions without affecting androgen levels in e1-6.
11. Idem. Oral hypoglycaemics in the first pregnant polycystic ovary syndrome wom-
33. Glueck CJ, Goldenberg N, Pranikoff J,
trimester and fetal outcome. S Afr Med J en: results of a randomised study. Hum Loftspring M, Sieve L, Wang P. Height, 12. Idem. Metformin in management of 23. Hague WM, Davoren PM, Oliver J, during the first 18 months of life in 126
pregnant insulin-independent diabetics. Rowan D. Contraindications to use of infants born to 109 mothers with polycys- metformin: metformin may be useful in tic ovary syndrome who conceived on and 13. Idem. The management of non-insu-
gestational diabetes. BMJ 2003;326:762.
lin-dependent diabetes during pregnancy. 24. Rowan JA, MiG Investigators. A trial Hum Reprod 2004;19:1323-30.
Diabetes Res Clin Pract 1985-1986;1(5): in progress: gestational diabetes: treat- Copyright 2008 Massachusetts Medical Society. n engl j med 358;19 www.nejm.org may 8, 2008 Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved. CORRECTION
Metformin versus Insulin for the Treatment of
Gestational Diabetes
Metformin versus Insulin for the Treatment of Gestational Diabetes .
In the abstract (page 2003), under Results, the parenthetical informa- tion should read ``(relative risk, 0.99; 95% confidence interval, 0.80 to 1.23).´´ In the introduction (page 2004), the last sentence of the sec- ond paragraph should read as follows: ``Its use in pregnancy remains controversial; to our knowledge, only two small, randomized trials ad- dressing this question have been reported to date.´´ In Table 2 (page 2008), the data listed under Relative Risk (95% CI) are incorrect; they should read as follows: for Primary composite outcome, 0.99 (0.80–1.23); Recurrent blood glucose level <46.8 mg/dl, 0.81 (0.59– 1.12); Any blood glucose level <28.8 mg/dl, 0.41 (0.21–0.78); Res- piratory distress, 0.76 (0.37–1.59); Phototherapy, 0.95 (0.59–1.55); Birth trauma, 0.96 (0.49–1.87); 5-Min Apgar score, 3.06 (0.32–29.26); Preterm birth (<37 wk of gestation), 1.60 (1.02–2.52); Iatrogenic (in- dicated), 1.41 (0.70–2.84); Spontaneous, 1.77 (0.95–3.28); Admis- sion to level 2 or 3 neonatal intensive care unit, 0.89 (0.66–1.19); and>24-Hr stay in neonatal intensive care unit, 1.04 (0.71–1.53). In thefirst footnote below Table 2, the sentence that begins with ``Induction of labor´´ should say that ``cesarean section was performed in 131 women (36.1%) in the metformin group and 142 (38.4%) in the insulin group´´ (rather than nonemergency cesarean section). In the Discus- sion section, in the third pararaph from the end of the text (page 2014), the sentence beginning with ``Nonetheless´´ should read as follows: ``Nonetheless, the 95% confidence interval for the relative risk of the composite measure of neonatal complications suggests that an in- crease in risk of more than 23% with metformin is implausible´´ (rather than ``an increase in risk of more than 10%´´). In the same paragraph, the parenthetical information should read ``(relative risk, 0.99; 97.5% confidence interval, 0.78 to 1.26).´´ The article has been corrected on the Journal’s Web site at www.nejm.org.
Downloaded from www.nejm.org at UNIVERSITY OF BRITISH COLUMBIA on March 1, 2009 . Copyright 2008 Massachusetts Medical Society. All rights reserved.

Source: http://www.obgyn.ubc.ca/Education/Residency/JournalClub/docs/Burrows_001.pdf