Response to Comment on Barbour and Feig. Metformin for Gestational Diabetes Mellitus: Progeny, Perspective, and a Personalized Approach. Diabetes Care 2019;42:396–399

We appreciate Dr. Rowan’s thoughtful comments (1) on our article (2), as her comments make a number of important points. As she notes, there are conflicting mouse data with respect to the effect of metformin. Furthermore, it is difficult to extrapolate mouse studies to humans given that there are marked differences in placentation and oxidative/proliferative pathways, the early postnatal period may have stronger developmental effects than later pregnancy, and mice are typically born with only 2–3% fat compared with 6–14% fat in humans. However, in one of the mouse studies Dr. Rowan cited, prenatal metformin did lead to increased weight gain, mesenteric fat, and liver weight after a high-fat diet, and males showed glucose intolerance (3).

Dr. Rowan makes a fundamental point that an intrauterine environment of maternal nutrient excess (likely both hyperglycemia and hyperlipidemia) may potentially render differences in offspring effects from metformin compared with less nutrient exposure. Hanem et al. (4) recently reported the 5- to 10-year follow-up of the PregMet randomized controlled trial of 144 children (55% of eligible children) from mothers with polycystic ovary syndrome randomized to metformin versus placebo (without gestational diabetes mellitus [GDM]). Although not different at birth, metformin-exposed children (mean 7.5 years) demonstrated statistically higher BMI z-scores, waist-to-height ratio z-scores, and waist circumference z-scores and borderline significantly higher body fat (P = 0.05), especially when mothers were obese.

Given the potential of fetal nutrient restriction from metformin that when followed by postnatal nutrient excess could affect childhood growth, we agree that the maternal metabolic phenotype should also be considered. However, maternal obesity alone also results in 10% higher glucoses and 30–40% higher triglycerides compared with those of normal-weight mothers (5), similar to some GDM women, supporting significant metabolic heterogeneity in obesity alone versus GDM with/without obesity. Also, mothers with obesity, GDM, type 2 diabetes, and polycystic ovary syndrome are all at higher risk of placental insufficiency, which could further restrict nutrition. Appropriately, metformin was stopped in the Metformin in Gestational Diabetes (MiG) trial for evidence of placental insufficiency by fetal growth or development of preeclampsia. One GDM randomized trial of metformin versus insulin showed a statistically significant decrease in head and chest circumference and ponderal index at birth in metformin-exposed offspring (6), suggesting that despite a GDM intrauterine environment, nutrient restriction may affect fetal growth unrelated to fat. In the MiG Auckland 9-year-old children, Dr. Rowan utilized MRI, which gives an estimate of visceral/abdominal fat superior to that of DEXA, and showed increased abdominal fat volume and visceral fat at borderline significance (both P = 0.51) in metformin-exposed offspring (7).

We are in agreement that when/if deciding to use metformin, not only should specific maternal glycemic profiles, circumstances, and risks be considered but the fetal nutrient environment should also be appraised. Any concerns about potential nutrient insufficiency (caloric restriction, insufficient weight gain, normal BMI, placental insufficiency, hypertension) should dissuade the use of metformin. Metabolic studies in which the offspring from nonhuman primate mothers (who share our placenta) are exposed in utero to metformin could markedly increase our understanding at a cellular level in the pancreas, liver, and mitochondria. Certainly, long-term human studies remain critical to discern any childhood effects when metformin is given to mothers with different metabolic phenotypes and nutrient exposures.

• © 2019 by the American Diabetes Association.