Women With Gestational Diabetes Mellitus Randomized to a Higher–Complex Carbohydrate/Low-Fat Diet Manifest Lower Adipose Tissue Insulin Resistance, Inflammation, Glucose, and Free Fatty Acids: A Pilot Study

Diet therapy is the first line of treatment in women with gestational diabetes mellitus (GDM). Yet, the optimal diet in GDM is debated. The American College of Obstetricians and Gynecologists (1) and the Endocrine Society (2) support a low-carbohydrate diet, whereas the American Diabetes Association has withdrawn a specific diet recommendation. A diet lower in simple carbohydrates blunts postprandial glucose, which is associated with large-for-gestational-age infants (3). However, this approach often necessitates an increase in dietary fat given that protein intake is relatively constant (3). Higher dietary fat in humans and animals has been shown to promote insulin resistance (IR) (3). There is concern that increasing maternal IR could shunt more nutrients to the fetus and increase fetal fat accretion (3). Animal and nonhuman primate models support an intrauterine influence of dietary fat in promoting offspring adiposity, hepatic steatosis, and metabolic syndrome (4). In humans, maternal triglycerides (TGs) and free fatty acids (FFAs) are also strong predictors of excess fetal fat accretion (5,6). Thus, a low-carbohydrate, higher-fat (LC/CONV) diet in obese women with GDM with preexisting IR may have unintended consequences on infant health.

We set out to elucidate the effects of prepared diets on glucose and lipid profiles, adipose tissue (AT) IR, and infant adiposity in GDM. This highly controlled pilot began with a short-term crossover design. Women were randomized to either a conventional LC/CONV diet or a higher–complex carbohydrate, lower-fat (CHOICE, choosing healthy options in carbohydrate energy) diet at the time of GDM diagnosis. They remained on the second diet through delivery. We have previously published that in the randomized crossover trial, both diets resulted in controlled glycemia (7). We hypothesized in this pilot study that maternal IR (HOMA-IR, AT) and offspring adiposity would be improved after continuing the CHOICE diet versus LC/CONV for 6–7 weeks until delivery.

This study was approved by the Colorado Multiple and the Kaiser Permanente Colorado institutional review boards. As previously described (7), GDM was diagnosed at 24–28 weeks using Carpenter and Coustan criteria (1). Women were aged 20–36 years and had BMI 26–39 kg/m², had fasting blood glucose (FBG) <110 mg/dL without comorbidities, and were treated with diet alone.

Twelve women started diet intervention (all meals provided) in gestational weeks 30–31 and continued until delivery (7). A fasting (10-h) blood sample was collected at baseline (31 weeks), blood and AT were collected after 6–7 weeks of diet (37 weeks), and newborns were studied 2 weeks after birth. Women self-monitored fasting and 2-h postprandial glucose (targets <95 and <120 mg/dL, respectively) throughout pregnancy.

Thorough information describing the diet protocol has been previously published (7). Diets were eucaloric and macronutrient distributions were as follows: LC/CONV, 40% carbohydrate/45% fat/15% protein; CHOICE, 60% carbohydrate/25% fat/15% protein, as previously described (7). Both diets were matched for fat (35% saturated fatty acids/45% monounsaturated fatty acids/20% polyunsaturated fatty acids) and simple sugars (≤18% of kilocalories), and daily kilocalories were partitioned as breakfast 25%/lunch 25%/dinner 30%/snacks 20%. Both diets contained mainly low– to moderate–glycemic index foods. As previously described (7), we define “complex carbohydrate” as “polysaccharides and starches primarily derived from grains, vegetables, and fruits that tend to attenuate a sharp postprandial rise in plasma glucose.” Fiber content was similar (LC/CONV ∼23.5 g/day; CHOICE ∼29.3 g/day). Menus were tailored to food preferences, and all meals were prepared by the Clinical Translational Research Center (CTRC) Nutrition Services. Women met with investigators when they picked up their meals (every 72 h).

Subcutaneous AT (10-h fasted) was biopsied (n = 11; one subject declined) from the upper gluteal/flank region using our established protocol (8,9). Isoproterenol-stimulated AT lipolysis ± insulin was measured as described by Engfeldt et al. (10). Gene expression (mRNA) for markers of inflammation and lipogenesis associated with IR in AT was measured by quantitative RT-PCR (11). In brief, RNA was isolated from AT samples using RNeasy (Qiagen, Valencia, CA), reverse transcribed (Bio-Rad, Hercules, CA), and run on an iQ5 instrument (Bio-Rad). mRNA was normalized to the mean of reference genes RPL13a, UBC, and TFRC.

Lipids, ApoB, ApoA1, and Lp(a) were measured using VAP cholesterol testing (Atherotech, Birmingham, AL) and OxLDL by ELISA (Mercodia, Uppsala, Sweden) as previously reported (7).

Infant weight, length, and adiposity (% fat by PEAPOD; Cosmed Inc., Concord, CA) were measured by the same investigators 2 weeks after birth.

This pilot and feasibility study was an extension of the published randomized crossover study (7) performed to estimate power for a larger randomized clinical trial. Reported outcomes were prespecified, but not powered, end points. Equality of variance was confirmed using the Levene test. Paired Student t tests assessed within-group changes; independent Student t tests assessed between-group differences and change from baseline. Correlation analyses were Pearson R (SPSS V22.0; IBM, Armonk, NY).

Subject characteristics did not differ statistically between maternal groups (Table 1). However, in the CHOICE group, four women were Caucasian and two were Asian. In LC/CONV, two were Caucasian, three were Hispanic, and one was Asian. Physical activity was not different between groups (7). Women attended all clinic visits and picked up all prepared meals for the duration of the protocol.

There were no effects of either diet on cholesterol or lipoproteins (Table 1). Fasting TGs increased within both groups (P < 0.01, LC/CONV; P < 0.05, CHOICE). Both FBG (P < 0.05) and FFAs (P = 0.06) decreased on the CHOICE diet, whereas FBG increased on LC/CONV (P < 0.05) (Table 1).

Insulin suppression of isoproterenol-stimulated lipolysis was higher on CHOICE versus LC/CONV (56 vs. 31%, respectively, P < 0.01) (Table 1). AT gene expression of proinflammatory biomarkers interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), toll-like receptor-4 (TLR-4), cluster of differentiation molecule 11B (CD11B), CD11C, CD1D, and glycoprotein receptor 43 (GPR43) was lower in AT from CHOICE versus LC/CONV subjects (all P < 0.01) (Table 1).

Maternal fasting insulin (r = 0.697, P = 0.01) and HOMA-IR (r = 0.731, P < 0.01) at 37 weeks’ gestation were correlated with infant adiposity.

This highly controlled pilot intervention demonstrated that women with GDM randomized to CHOICE compared with LC/CONV diet have greater AT insulin sensitivity (i.e., better suppression of isoproterenol-stimulated lipolysis) and strikingly lower proinflammatory gene expression. FBG was also reduced after CHOICE compared with an increase after LC/CONV, both of which limited simple carbohydrates. Fasting FFAs decreased on both eucaloric diets, but with a greater decrement on the CHOICE diet (33% compared with 19%). Among all women with GDM, fasting insulin and HOMA-IR measured at 37 weeks’ gestation were strong correlates of increased infant adiposity. Taken together, our data suggest that the CHOICE diet, which is characterized by including more nutritious complex carbohydrates, may improve AT IR, resulting in a lower FBG and FFA exposure to the fetus. Therefore, AT insulin action may be an important target for diet therapy in GDM.

We previously demonstrated that glycemia was well controlled in these women with GDM when they consumed both diets for 3 days each in a randomized crossover trial. Importantly, postprandial FFAs were lower on CHOICE compared with the LC/CONV diet (7). Fasting glycemia was the strongest predictor for large-for-gestational-age infants in the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) trial (12), and in this pilot, prolonged exposure to the CHOICE diet decreased FBG. Mounting evidence supports that maternal TGs and FFAs also contribute to excess fetal fat accretion (5,6). In the second and third trimester, systemic IR increases to ensure adequate nutrient flux to the fetus, but insulin clamp studies demonstrate that women with GDM already enter pregnancy with chronic IR (13) and/or insufficient β-cell reserve. We have shown worsened skeletal muscle IR in GDM (14), and these data are the first to support an effect of diet on AT IR. Lowering dietary fat may improve insulin signaling through suppressing inflammation, resulting in less circulating FFAs, which we observed in subjects on the CHOICE diet (3). The reduced pattern of AT inflammatory and cytokine gene expression with CHOICE further supports this assertion. A reduction in the short-chain fatty acid receptor GPR43 suggests a potential favorable influence on the microbiota on the CHOICE diet (15).

Limitations to our study include the following: 1) the small sample size, 2) lack of a baseline (31 weeks) AT biopsy, and 3) lack of a systemic measure of IR (beyond HOMA-IR estimates). It is a further limitation that three of six women in the LC/CONV group were Hispanic, as this could have influenced the outcomes. However, the well-matched women, highly controlled randomized design in which all meals were provided, and exclusion of the confounding effects of glucose-lowering medications make these data provocative (3). Based on these pilot data, we are conducting a larger randomized clinical trial powered to detect a between-group difference in infant adiposity following these two diets.

Our pilot data raise questions about current low-carbohydrate dietary recommendations for women with GDM. If maternal IR is a key regulator of nutrient exposure to the fetal-placental unit, and it may be attenuated by reducing dietary fat, then potential long-term programming effects and excessive fetal growth could be modifiable by diet therapy in GDM.

Acknowledgments. The authors acknowledge the support of Dr. Robert H. Eckel, Dr. Bryan R. Haugen (University of Colorado [CU] Department of Medicine), Dr. Jill Davies (CU Department of Obstetrics and Gynecology), Dr. William R. Hay (CU Department of Pediatrics), and the study participants. The authors acknowledge Dr. Janine Higgins (CU CTRC Bionutrition Department), who made the dietary manipulation possible.

Funding. The authors acknowledge funding support from the National Institutes of Health (NIH) (R21 DK 088324), the CU Diabetes and Endocrinology Research Center, the NIH/National Center for Advancing Translational Sciences Colorado Clinical and Translational Research Institute (UL1 TR 000154), the Colorado Nutrition and Obesity Research Center (NIH P30 DK 048520-15), and the Colorado Program for Nutrition and Healthy Development.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.L.H. and L.A.B. researched data and wrote, reviewed, and edited the manuscript. R.E.V., M.A.A., M.S.R., R.M.R., B.A.d.l.H., M.H., C.C.-L., R.C.J., and J.E.F. researched data, reviewed and edited the manuscript, and contributed to discussion. W.T.D. and L.J.D. researched data and reviewed and edited the manuscript. T.L.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.