Gestational diabetes mellitus (GDM) affects one in five pregnancies according to the International Association of the Diabetes and Pregnancy Study Groups (IADPSG) criteria [1
]. This could rise to above 40% in certain ethnic groups such as Middle Eastern populations [2
]. Women with previous GDM are seven times more likely to develop type 2 diabetes (T2DM) compared with women with normoglycemic pregnancies [3
]. Weight gain after childbirth is a significant risk factor for T2DM development after GDM; the incidence of T2DM increases incrementally with weight gain [4
]. Due to the risks associated with obesity in the development of T2DM, weight loss is one of the intervention goals in diabetes prevention programs for the general population and for women after GDM [6
The development of T2DM in high-risk groups can be prevented or delayed with lifestyle interventions. Numerous real-world diabetes prevention programs in the general population have demonstrated a significant reduction in the incidence of diabetes through diet and physical activity modifications [9
]. Both the Finnish Diabetes Prevention Study and United States Diabetes Prevention Program (DPP) reported a reduction diabetes incidence of up to 58% with approximately 5 kg weight loss in their lifestyle intervention arms [6
]. Younger populations such as those targeting women after GDM report much smaller weight loss, averaging between 1 to 3 kg, with correspondingly smaller effects on the reduction of diabetes incidence (relative risk 0.75, 95% confidence interval (CI) 0.55–1.03) [12
]. Greater weight regain among younger participants (45 years and below) has also been reported in a 10-year follow up study of the US DPP study [13
]. The challenges with weight loss and weight loss maintenance for younger women in diabetes prevention programs are consistent with longitudinal studies in the general population, which showed that women between 18 to 40 years are at higher risk of weight gain compared with older women [14
]. The secular trend of weight gain in this age group may have attenuated the weight loss effect of diabetes prevention programs in younger women with histories of GDM. In addition, a sub group analysis of the US DPP comparing women who previously had GDM on average 12 years before with women who had not had GDM, the women who previously had GDM were much less likely to respond to lifestyle modification [16
The reasons for weight gain in women of childbearing age are complex. Being married, having children and starting work are risk factors for weight gain in younger women [17
]. These changes often result in increased sedentary behaviours, decreased physical activity and a worsening of dietary habits [17
]. Among postpartum women, a lack of time, tiredness, financial constraints and changes in priorities after childbirth further contribute to the barriers to a healthy lifestyle [19
]. Interventions may not be able to address many of these barriers. While higher intervention intensity may increase weight loss, the presence of contextual barriers to lifestyle modification in younger women has meant that higher intensity interventions are met by low participation or retention rates [21
]. Thus, interventions with lower intensity and a smaller weight loss goal may be more feasible for this group. However, a paucity of data on the metabolic effect of lower-level weight loss or weight gain prevention in women after GDM remains.
Owing in part to the lower weight loss achieved in diabetes prevention programs for women after GDM, these programs also tend to produce smaller, inconsistent and often insignificant improvements in glycemic markers such as glycated hemoglobin (HbA1C) or fasting plasma glucose [12
]. As insulin resistance and hyperinsulinemia precedes hyperglycemia in the natural history of T2DM development, measuring changes in markers of insulin resistance may more appropriate and relevant than changes in glycemic markers when measuring progress in diabetes prevention in younger cohorts [24
]. Recently, triglyceride-to-high-density lipoprotein cholesterol ratio (triglyceride:HDL) has been suggested to be marker of insulin resistance in some population groups [25
]. Changes in triglyceride:HDL was not previously investigated in diabetes prevention programs.
Considering the effect of postpartum weight gain on the development of T2DM [4
], prevention of weight gain may be efficacious in preventing diabetes in women after GDM. There is little direct evidence available to confirm the benefit of weight gain prevention on the cardiometabolic risk of women after GDM. Consistently lower-level weight loss achieved in diabetes prevention programs in this group suggests that greater weight loss may not be feasible at a large scale in this population. This necessitates a greater understanding on the metabolic benefits of small weight changes in this group within the context of diabetes prevention. A stratified approach to investigate the effect of lower-level weight changes on cardiometabolic effects has not been previously investigated in women after GDM. Therefore, this study aims to compare the cardiometabolic effects of weight loss, weight maintenance and weight gain following a diabetes prevention program in women after GDM.
The Consolidated Standards of Reporting Trials (CONSORT) diagram describing the flow of participants is as shown in Figure 1
. From a total of 573 participants randomized, 434 completed the study at 12-months. The main results of the RCT were previously published [8
]. The intervention resulted in a small but statistically significant weight loss compared to control (−0.23 kg, (95% CI −0.89, 0.43) vs. +0.72 kg (95% CI 0.09, 1.35); group-by-treatment interaction p
= 0.04). No other intervention effect was found on the other outcomes. This secondary analysis included participants from the intervention arm who completed the study (n
= 206). Baseline characteristics by weight change categories are shown in Table 1
. There were no statistically significant differences in the baseline characteristics between the groups. The most prevalent age categories across all groups were 30- to 34-year-olds and 35- to 39-year-olds. Most women reported being married and undertaking home duties when asked their marital status and work situation respectively. Most participants were overweight or obese at baseline in all groups. Attainment of a Bachelor degree was the most prevalent highest level of education reported across all groups. Most participants were within 1 year postpartum across all groups.
Changes in cardiometabolic measures over time within each group is shown in Table 2
. Significant improvements in weight, waist and body mass index (BMI) were seen for the weight loss group (Table 2
). They also showed significant improvements in glycemic control from HbA1c, and in total cholesterol, triglyceride and triglyceride:HDL ratio but not fasting plasma glucose or HDL. In the weight stable group which maintained weight within 2 kg, a small but significant decrease in waist was observed while weight and BMI remained unchanged. This was accompanied by a significant worsening of fasting plasma glucose although HbA1c was not significantly changed in this group. This group also had significant improvements in total cholesterol and LDL cholesterol but not in triglyceride. Within the weight gain group a significant deterioration in weight, BMI and waist was seen. A significant improvement in total cholesterol was seen for this group but there was also a significant deterioration in triglyceride and triglyceride:HDL ratio. HDL cholesterol decreased significantly over time in all groups.
Between-group comparisons showed significant differences in all anthropometric measures between weight loss, weight maintenance and weight gain groups (Table 2
). The weight loss group had significantly smaller increase in fasting plasma glucose and significantly greater decrease in HbA1c compared with the weight gain group. The weight loss group had significantly greater decrease in total cholesterol and LDL cholesterol compared with the other two groups. There appeared to be a dose-response relationship between weight change and triglyceride with significant differences between weight loss, weight stability and weight gain groups. In terms of insulin resistance, the weight gain group had significant greater increase in triglyceride:HDL ratio, a sometimes used indicator of insulin resistance, compared with the other two groups. No time or group effect was seen in systolic and diastolic blood pressure. Similar results for the between-group comparisons were obtained after corrected for baseline weight.
We explored the effect of weight change on anthropometric and cardiometabolic outcomes in postpartum women with a history of gestational diabetes at 1 year following a diabetes prevention program. We report that weight loss of more than 2 kg was associated with significant improvements in glycemic control, total and LDL cholesterol, triglyceride and triglyceride:HDL ratio. We found weight gain of more than 2 kg significantly worsened triglyceride and triglyceride:HDL ratio.
Diabetes prevention programs for women after gestational diabetes typically reports much smaller weight loss, averaging between 1 to 3 kg compared with 2.5 kg or more seen in “real world” diabetes prevention programs in the general population, or 5 kg in controlled trials [6
]. The clinical significance of such small degees/amounts of weight loss on glycemic control and other metabolic outcomes is unclear. Previous studies in postpartum women after gestational diabetes with small weight loss reported no significant change in lipids [33
] and fasting plasma glucose [34
]. Similarly, the primary analysis of the current trial did not find a significant change to the intervention participants as a group [8
]. When we stratified the group according to weight loss achieved, we found that weight loss as little as over 2 kg is associated with improvements in HbA1c, fasting plasma glucose, triglyceride:HDL, LDL cholesterol, total cholesterol and triglyceride compared with those who gained weight. This provides evidence that the small average weight loss seen in this group could still be beneficial in improving the metabolic risks.
Women of reproductive age are at high risk of weight gain [15
]. Adulthood weight gain (from 18 years onwards) is associated with increased risk of chronic diseases [35
]. There was little direct evidence demonstrating that weight gain increases metabolic risk in the postpartum period. The current study suggest that weight gain as little as 2 kg or more worsens triglyceride and triglyceride:HDL. triglyceride:HDL has been found to be a predictor of insulin resistance in certain populations including individuals of Aboriginal, Chinese, and European origin [25
]. This is consistent with the known relationship between weight gain and increasing insulin resistance [36
]. Considering that triglyceride:HDL was increased in weight gain but similar between weight loss and weight maintenance groups in the current analysis, this finding suggests that prevention of weight gain could prevent the progression of insulin resistance, and should be one of the diabetes prevention goals in postpartum women with history of gestational diabetes.
We report in the current analysis that improvements in total and LDL cholesterol were seen not only in the weight loss group but also in the weight maintenance and weight gain groups at 1 year after randomization to the lifestyle intervention. This suggests that improvements in metabolic health such as lipid outcomes could occur independent of weight loss, presumably due to improvements in diet quality resulting from the lifestyle intervention. The benefit of diet quality on reducing cancer, cardiovascular and all-cause mortality independent of weight change were also previously observed in the general population [37
]. We also observed greater improvements of total and LDL cholesterol in the weight loss group (about 10% improvement) compared with the other groups, as consistent with previous evidence on the protective effect of weight loss on lipid metabolism in young women [21
]. This is particularly relevant to women with histories of GDM as a recent meta-analysis found that GDM increases the risk of cardiovascular event by 2.3-fold in the first decade postpartum [38
There was a decrease in HDL across all weight change groups without significant group differences. A recent meta-analysis (20 studies, n
= 2016) on low-fat compared with high-fat diet on cardiometabolic outcomes in people without metabolic disturbance found that low-fat diet was associated with a decrease in HDL-cholesterol (WMD: −2.57 mg/dL (−0.07 mmol/L); 95% CI −3.85, −1.28; p
< 0.001) [39
]. As two of the five goals of MAGDA were on reducing total and saturated fat intake, this may have contributed to the observed change in HDL. We have also previously observed a reduction in HDL-cholesterol following a low-fat diet in women of reproductive age [21
]. Weight loss should increase HDL-cholesterol, although the level of weight loss required for this is usually not seen in postpartum women or women of reproductive age [40
]. Higher physical activity levels could also increase HDL-cholesterol [41
] but the current group did not have a significant change in physical activity during the intervention [8
We found that fasting plasma glucose worsened slightly over time in all weight change groups despite receiving lifestyle intervention, without reaching the diagnostic threshold for impaired fasting glucose. This is consistent with the increased risk of type 2 diabetes development associated with the history of gestational diabetes [42
]. In particular, there was a significantly greater increase in fasting plasma glucose in the weight gain group compared with that in the weight loss group. This provides further evidence supporting the increased risk of type 2 diabetes with adulthood weight gain as previously reported in the Nurses’ Health Study [35
]. A small but significant reduction in HbA1c was seen in the group with weight loss of 2 kg or more. HbA1c has been found to have lower sensitivity and higher specificity in detecting type 2 diabetes in at-risk populations compared to fasting plasma glucose [43
]. The current findings of significantly increased fasting plasma glucose in weight gain group compared to weight loss group, and significantly decreased HbA1c in weight loss group compared to weight gain group, suggest that modest weight loss in the postpartum period could slow the progression of type 2 diabetes in this group. Longer term follow-up is required to determine if this would result in reduced incidence of diabetes. This is consistent with a previous study reporting that postpartum weight changes correspond to the incidence of type 2 diabetes among women with history of gestational diabetes [44
]. This current finding on the benefit of small weight loss on glycemic improvement provide further support for modest weight loss as treatment goals in this group.
Women of reproductive age face unique barriers to lifestyle modification in terms of time, motivation, cost and social support, which may explain the high risk of weight gain in this group [15
]. All postpartum women face additional lifestyle change barriers due to infant needs and shifts in priority [19
]. Considering these significant barriers to lifestyle modification in the postpartum stage, lower weight loss or prevention of weight gain may be a more realistic and useful goal for this group. Achieving lower level weight loss or even weight stability actually constitutes a considerable achievement in light of the population trend of weight gain at this lifestage [15
]. Our findings provide evidence that even a small amount of weight loss or stability in the postpartum period carries significant health benefit in women with history of GDM.
The strengths of this randomized trial include the length of follow-up and rigorous data collection methods. Limitations include this study being a secondary analysis, which is cross-sectional in nature, and as such we cannot derive causal relationships between the variables described. Also the delivery format of the intervention was resource intensive yet only achieved a low level of engagement from participants, which makes it unattractive to funders of health services.