Is Energy Balance in Pregnancy Involved in the Etiology of Gestational Diabetes in Women with Obesity?

Lifestyle intervention with dietary counseling and increased physical activity is the first-line strategy for prevention of gestational diabetes mellitus (GDM). This approach is based on epidemiological data suggesting that the etiology of GDM is similar to type 2 diabetes, with common risk factors including an unhealthy, hypercaloric diet and a sedentary lifestyle (Agha-Jaffar et al., 2016). Preliminary studies in obese pregnant mothers at risk for GDM indeed supported the notion that lifestyle modification may reduce GDM incidence through restriction of pregnancy weight gain (e.g., Thornton et al., 2009). Based on such preliminary findings, well-powered randomized controlled lifestyle intervention trials such as LIMIT (Dodd et al., 2014b), UPBEAT (Poston et al., 2015), RADIEL (Koivusalo et al., 2016), DALI (Simmons et al., 2017), and LifeMoms (Peaceman et al., 2018), which collectively involved more than 5,000 patients, were more recently undertaken to assess the efficacy of lifestyle interventions to reduce GDM incidence. With the exception of RADIEL (Koivusalo et al., 2016), all other trials were ineffective at lowering GDM incidence. This is surprising, since these trials achieved acceptable levels of patient adherence (Koivusalo et al., 2016; Poston et al., 2015) and successful alterations of lifestyle factors targeted by the interventions—such as reduced dietary intake (Dodd et al., 2014a; Poston et al., 2015), improved diet quality (Dodd et al., 2014a; Koivusalo et al., 2016; Poston et al., 2015; Simmons et al., 2017), and increased physical activity (Dodd et al., 2014b; Koivusalo et al., 2016; Poston et al., 2015)—which indeed translated to lower rates of gestational weight gain (Koivusalo et al., 2016; Peaceman et al., 2018; Poston et al., 2015; Simmons et al., 2017) in all but one trial (Dodd et al., 2014b) without lowering GDM incidence.

The inability of lifestyle modification to prevent GDM despite successful restriction of weight gain raises the essential question of whether energy imbalance (energy intake > energy expenditure) is involved in the development or prevention of GDM. To date, information regarding energy balance is based on subjective data (e.g., self-reported diet and physical activity), which is prone to recall bias, rather than objective data (e.g., energy intake and expenditure by stable isotopes) derived from state-of-the art methods in human nutrition research. Consequently, empirical evidence to indicate that a more positive energy balance early in pregnancy favors the development of GDM is lacking, and yet this is the chief scientific premise supporting many lifestyle modification programs for GDM prevention (Dodd et al., 2014b; Koivusalo et al., 2016; Peaceman et al., 2018; Simmons et al., 2017).

In a prospective, observational energy balance study in 62 pregnant women with obesity (BMI > 30 kg/m²) we used state-of-the-art methodology to simultaneously measure energy intake and energy expenditure over ~12 weeks starting between 13 and 16 weeks of gestation and concluding between 24 and 27 weeks. Unlike previous studies based on dietary self-report, we assessed energy intake across the 12-week period using the energy intake-balance method (see Supplementary Methods). This method calculates energy intake from the sum of total daily energy expenditure determined from doubly labeled water (1.25 g of 10% enriched H₂¹⁸O and 0.10 g of 99.9% enriched ²H₂O per kg body weight) and the change in body energy stores (771 kcal/kg fat-free mass and 9,500 kcal/kg fat mass).

Nine (15%) of the 62 obese women developed GDM (Figure S1A; Table S1A), and thus we compared components of energy balance between women who developed GDM and those who maintained normal glucose tolerance (n = 53). Over the ~12 week observation period prior to GDM diagnosis (second trimester of pregnancy), energy intake in women who developed GDM was 2,744 ± 92 kcal/day and did not differ from energy intake in the controls (2,606 ± 54 kcal/day; p = 0.31; Figure S1B). Like energy intake, the caloric expenditure over 7 days was comparable between the GDM and controls (GDM: 2,855 ± 148; control: 2,631 ± 51 kcal/day; p = 0.11; Figure S1B), and consequently, energy balance did not differ between groups (GDM: +16 ± 110; control: +84 ± 44 kcal/day; p = 0.56). The macronutrient composition of the diet was derived from a validated food photography method (see Supplementary Methods) and for GDM women comprised 42% ± 4% carbohydrate, 41% ± 3% fat, and 18% ± 2% protein (Healthy Eating Index: 52 ± 2.7), which was also not different from the control group (Figures S1B and S1C). Using indirect calorimetry in a respiratory chamber (see Supplementary Methods), we found that energy expenditure measured during sleep was also comparable between women that developed GDM and those who did not (Figure S1B). Physical activity assessed with calorimetry and accelerometry over 7 days was also similar between groups (Table S1A).

In agreement with a similar energy balance in the two groups, weight gain across the ~12-week period as well as total body fat accumulation (Figure S1B) and fat distribution in the abdomen as measured by magnetic resonance imaging were not different between women who developed GDM and those who did not (Figure S1B). Notably, in an attempt to account for potentially relevant covariates, we replicated these results in our entire cohort (n = 62) in a sensitivity analysis that included only the 9 women with GDM and 9 controls that were pair-matched for age, race, parity, BMI, and fetal sex (Table S1B; Figures S1D and S1E). Overall, this study in obese women shows that an energy imbalance throughout the second trimester did not contribute to the development of GDM.

In line with previous studies, the primary risk factors for GDM development, such as excess adiposity and insulin resistance, were evident early in pregnancy (Agha-Jaffar et al., 2016; Wexler et al., 2018). Women in this study that developed GDM tended to be heavier (+10.5 ± 5.8 kg; p = 0.08), had more fat mass (+7.3 ± 4.1 kg; p = 0.08) and significantly more visceral adipose tissue (+3.3 ± 0.9% of total abdominal fat; p = 0.001; Figure S1B). Furthermore, women who developed GDM reported a greater prevalence of diabetes in first degree relatives (p = 0.004), significantly higher fasting glucose (p < 0.001, Figure S1B) and HbA1c (p = 0.04), and hence greater prevalence of prediabetes (p = 0.001). Among the women who developed GDM, the prevalence of prediabetes (HbA1c ≥ 5.7%) prior to 15 weeks was 4-fold higher (67%, n = 6/9) compared to those without GDM (17%, n = 9/53). Together, our data reinforce the recommendation for earlier screening of dysglycemia in all women (Wexler et al., 2018) since obesity itself does not necessarily reflect an “at-risk” phenotype to trigger a universal prescription for lifestyle modification in clinical practice.

Screening for dysglycemia has been shown to reveal heterogeneity in glucose tolerance and identification of at least three distinct phenotypes of GDM (Powe et al., 2016). The notion that not all GDM is the same and the differential impact of lifestyle interventions on GDM incidence raise the question of whether prevention approaches may be more effective when tailored to individual patients (Wexler et al., 2018). To this end, in RADIEL—the only successful intervention — patients were notably less insulin resistant in early pregnancy as compared to patients in DALI (HOMA-IR: 1.8 versus 2.6, not reported in others). Furthermore, RADIEL included significantly more women with a history of GDM as compared to trials that were ineffective (~25% versus < 5% in others). A recent review assessing lifestyle intervention trials for GDM prevention (Agha-Jaffar et al., 2016) concluded that the multiplicity of differences in study designs, including participant characteristics and GDM screening methods alongside variations in intervention modalities and adherence, precludes any robust understanding of efficacy (or lack thereof) between trials.

Although this prospective cohort study was conducted throughout the period of pregnancy when the majority of lifestyle interventions targeting GDM prevention have been implemented (Dodd et al., 2014b; Koivusalo et al., 2016; Poston et al., 2015; Simmons et al., 2017), weight (and fat) gain prior to the second trimester probably contribute to the development of GDM (MacDonald et al., 2017). However, such observations are inconsistent for women with obesity, and studies are weakened by use of self-reported weight at conception.

We acknowledge that our cohort is small in comparison to those in intervention trials (Dodd et al., 2014b; Koivusalo et al., 2016; Peaceman et al., 2018; Simmons et al., 2017), but the scope of this study is analogous to those in the energy balance field with similar designs and methodologies. A post-hoc analysis of our data showed that 288 kcal per day (> 80% power) was the minimal detectable difference in energy intake that could be detected in 62 patients with 15% incidence of GDM. A study that measured energy intake, albeit by self-report, has demonstrated larger differences in energy intake between patient groups but with no difference in intake in women with GDM (Vesco et al., 2014). Nevertheless, given the few cases of GDM in this cohort, we were unable to conduct a multivariate analysis to account for other classical determinants of GDM such as family history, maternal race, insulin resistance, and weight at conception. Thus, while our study shows that a more positive energy balance does not explain GDM development in women with obesity, we cannot dismiss the hypothesis that an energy imbalance influences the development of GDM in distinct phenotypes or earlier in pregnancy. Of note, none of the women in this cohort reported a history of GDM, which is among the most determining risk factors of GDM development.

GDM leads to significant deleterious outcomes for the mother and her child that extend well beyond pregnancy (reviewed in Agha-Jaffar et al., 2016). Women with a history of GDM are seven times more likely to develop type 2 diabetes, and children exposed to GDM in utero have increased incidence of type 2 diabetes and a higher lifetime risk of overweight and obesity. Thus, carefully conducted energy balance studies are needed to understand the contribution of diet, physical activity, and weight gain to the etiology of GDM and to identify patients most likely to benefit from lifestyle intervention approaches. Our data suggest that energy balance in pregnancy may not determine the development of GDM. It does not argue the potential benefits of a lifestyle characterized by physical activity and a “healthy diet” in pregnancy, but it does underscore the necessity of better understanding the pathogenesis of GDM in women with obesity and developing approaches that are tailored for individuals with different metabolic phenotypes. This study thereby warrants duplication in other cohorts, such as women with a history of GDM, who are at highest risk for GDM development and those with differing metabolic phenotypes identified early in pregnancy. If energy intake and energy expenditure are not the major causes of the development of GDM in the second trimester, the field then can turn its attention to other potential mechanisms—e.g., pre-pregnancy or first-trimester maladaptation—and develop evidence-based dietary, physical activity, and pharmacological prevention strategies accordingly.