Lifestyle intervention modifies the effect of the MC4R genotype on changes in insulin resistance among women with prior gestational diabetes: Tianjin Gestational Diabetes Mellitus Prevention Program

ABSTRACT

Background

A history of gestational diabetes mellitus (GDM) has been related to an elevated risk of type 2 diabetes. The melanocortin-4 receptor (MC4R) genotype has been related to glycemic changes in women with prior GDM.

Objective

The objective of this study was to analyze whether lifestyle intervention modified the association between the MC4R genotype and changes in insulin sensitivity among women with prior GDM.

Methods

We genotyped MC4R rs6567160 and measured glucose and insulin in fasting plasma samples at baseline and during the first 2 follow-up visits in 1128 women with prior GDM. They were randomly assigned to either a 4-y lifestyle intervention involving both diet and physical activity or a control group from a randomized clinical trial, the Tianjin Gestational Diabetes Mellitus Prevention Program. We analyzed the interaction between the MC4R genotype and lifestyle intervention on changes in insulin resistance.

Results

From baseline to 1.28 y, the MC4R genotype was related to changes in fasting insulin, HOMA-IR, and homeostasis model assessment of β cell function (HOMA-B) in the intervention group. Each risk allele (C) of rs6567160 was associated with a 0.08-unit greater decrease in log(insulin), log(HOMA-IR), and log(HOMA-B) (P = 0.02, 0.04, and 0.04, respectively), whereas in the control group, each C allele tended to be associated with a greater increase in HOMA-IR (P = 0.09). We found significant interactions between the MC4R genotype and lifestyle intervention on 1.28-y changes in fasting insulin and HOMA-IR (P = 0.006 and 0.008, respectively), and such interaction remained significant when we analyzed the trajectory of changes in insulin and HOMA-IR from baseline to 2.55 y (both P = 0.03).

Conclusions

The exploratory results from the first 2 follow-up visits indicate that women with prior GDM carrying a diabetes-increasing MC4R genotype (CC or TC) may obtain better improvement than the TT genotype in insulin resistance through lifestyle intervention. This trial was registered at clinicaltrials.gov as NCT01554358.

Introduction

Gestational diabetes mellitus (GDM), one of the most common complications of pregnancy, is a risk factor for type 2 diabetes (T2D) (1, 2). The prevalence of GDM ranges from 9.3% to 25.5% worldwide (3), and the prevalence has been continuously increasing (2, 4, 5). Lifestyle intervention has been widely accepted as a major approach to prevent and delay the progression to diabetes among women with prior GDM (6–8).

In our previous analysis, we found that a genetic variant near the melanocortin-4 receptor gene (MC4R) was related to postpartum changes in glycated hemoglobin (HbA1c) and 2-h oral-glucose-tolerance test (OGTT) glucose among women with prior GDM (9). The MC4R genotype has also been related to obesity (10, 11), insulin resistance (12), hyperinsulinemia (13), and T2D (14–16) in previous studies. Emerging evidence, including that from our group, has consistently shown that lifestyle interventions may modify the genetic effects on glycemic changes among women without a history of GDM and men (17–21). However, women with prior GDM have at least a 7-fold increased risk of developing T2D compared with women who had a normoglycemic pregnancy (1). Thus, it is also important to investigate whether lifestyle interventions modify the associations between genetic factor and longitudinal glycemic changes among women with prior GDM in order to develop better personalized intervention for women with high T2D risk.

In this study, we aimed to examine whether lifestyle intervention modified the genetic associations of the MC4R variant with 1.28- and 2.55-y improvement of insulin sensitivity in a randomized clinical trial, the Tianjin Gestational Diabetes Mellitus Prevention Program.

Methods

Study participants

The Tianjin Gestational Diabetes Mellitus Prevention Program is a 4-y randomized clinical trial that is being carried out at the Tianjin Women's and Children's Health Center in Tianjin, China. The details of the study, including design and methods, have been described previously (7, 22). Briefly, all women aged 20–49 y at baseline and diagnosed with GDM between 2005 and 2009 were eligible for the study. GDM was diagnosed according to WHO criteria (23). Pregnant women at 26–30 weeks of gestation participated in a 1-h, 50-g OGTT at Tianjin Women and Children's Health Center, and they were invited to undergo a 2-h, 75-g glucose OGTT if they had a glucose concentration ≥7.8 mmol/L. Women who underwent the 2-h, 75-g glucose OGTT with the results confirming either diabetes (fasting glucose ≥7 mmol/L or 2-h glucose ≥11.1 mmol/L) or impaired glucose tolerance (2-h glucose ≥7.8 and <11.1 mmol/L) were diagnosed as having GDM. Major exclusion criteria were 1) diagnosis of diabetes at the screening visit, 2) taking medicines known to alter OGTTs, 3) the presence of any chronic diseases that could seriously reduce the life expectancy or the ability to participate in the trial, and 4) currently pregnant or planning to become pregnant in the next 2 y.

After participants completed the baseline survey between August 2009 and July 2011, 1180 eligible women with prior GDM were randomly assigned (1:1) to either a 4-y lifestyle intervention or a control group (with usual care) including 4 follow-up visits. To date, the first (December 2010 to September 2013) and second (March 2012 to February 2016) follow-up visits have been completed, and the data from these visits are analyzed in this study (Supplemental Figure 1). The mean follow-up time for the first visit was 1.28 y (range: 0.27–3.70 y), and that for the second visit was 2.55 y (range: 1.20–6.00 y). This study provides an interim analysis of a predetermined 4-y intervention study. For the intervention group, the following 5 goals were included: 1) among women with BMI (in kg/m²) ≥24, reducing initial body weight by 5–10% by a reduction of at least 10% of total calories in their normal meals, and among women with BMI <24, no weight loss was requested; 2) <30% of energy derived from total fat intake; 3) 55–65% of energy derived from carbohydrate intake; 4) 20–30 g fiber intake per day; and 5) performing moderate or vigorous exercise for at least 30 min daily. Major elements of the intervention included 6 face-to-face sessions with study dietitians and 2 telephone calls in the first year, and 2 additional sessions and 4 telephone calls in each subsequent year. A dietitian provided participants advice on modifying their diets, including 1) appropriate energy intake; 2) inclusion of appropriate amounts of eggs, low-fat milk, fish, and lean meat and reduction in animal fat and fatty meat; 3) avoidance/reduction of refined carbohydrates and simple sugars; and 4) inclusion of more foods rich in fiber, such as whole grains, wheat flour with standard grade, brown rice, corn/corn starch, fruits, and vegetables. The control group was given general oral and written information on the awareness of diabetes, dietary modification, and increasing physical activity at subsequent annual visits, but no specific individualized programs were offered. Ethics approval was granted by the Human Subjects Committee of the Tianjin Women's and Children's Health Center, and all participants provided written informed consent.

In this study, we analyzed data from 1128 participants with single nucleotide polymorphism (SNP) data available at baseline. Data on insulin and insulin resistance were available for 1128 participants at baseline, 893 participants at the first visit (1.28 y), and 745 participants at the second visit (2.55 y).

Measurements

Information on sociodemographic characteristics, family history of diabetes, alcohol consumption, smoking habits, dietary habits (a self-administered FFQ to measure the frequency and quantity of intake of 33 major food groups and beverages during the past year) (24), and physical activity (the frequency and duration of 5 domains of physical activity: occupational, commuting, leisure time, household, and sedentary activities) (25) was obtained by questionnaire at the baseline survey. Changes in lifestyle were measured using a questionnaire on changes in major dietary and physical activity habits at each visit. A 3-d, 24-h food record and a self-administered FFQ were completed at each visit. The performance of 3-d, 24-h food records (24), the FFQ (24), and the questionnaire assessing physical activity (25) were validated in the China National Nutrition and Health Survey in 2002.

Fasting peripheral venous EDTA blood samples were collected at baseline and first and second visits. Glucose was measured by an automatic analyzer (TBA-120FR; Toshiba), and insulin was determined by electrochemiluminescent immunoassay (ADVIA Centaur CP; Bayer). HbA1c was measured using an automatic glycohemoglobin analyzer (ADAMS A1c HA-8160; Arkray). HOMA models were used to estimate insulin resistance (HOMA-IR) and β-cell function (HOMA-B), which were calculated by Equations 1 and 2, respectively:

(1)

(2)

Body weight, waist and hip circumferences, body fat (SC-240 Body Composition Analyzer; Tanita), systolic blood pressure (SBP), and diastolic blood pressure (DBP) were measured using the standardized protocol at the baseline, first, and second visits. Height was measured at baseline.

Genotyping

DNA was extracted from the buffy coat fraction of centrifuged blood with the QIAmp Blood Kit (Qiagen). SNP rs6567160 in the MC4R gene was genotyped (as TT, TC, or CC) by quantitative real-time TaqMan PCR (Applied Biosystems). The success rate of genotyping was >98%. For quality control, 10% of the samples were re-genotyped, with >99% concordance (26).

Statistical analysis

The primary endpoints in this study were changes in fasting insulin, HOMA-IR, HOMA-B, fasting glucose, and HbA1c. Concentrations of insulin, HOMA-IR, and HOMA-B were log transformed before analysis to improve the normality of distributions. General linear models for continuous variables and chi-square test for categorical variables were used to compare characteristics according to the rs6567160 genotype at baseline. Additive genetic models were used in the analyses. In our exploratory analysis, for gene–lifestyle interaction, we examined the genotype, lifestyle intervention or not, and genotype-by-lifestyle interaction as independent predictors of changes in primary outcomes adjusted for age, baseline BMI, weight change, baseline value for the respective outcome trait, income class, alcohol consumption, smoking, family history of diabetes and SBP at baseline, and SBP changes in general linear models. Second, generalized linear mixed models were used to test interactions between gene and lifestyle intervention on the 2.55-y trajectory of changes in primary outcomes, with genotype, lifestyle intervention or not, genotype-by-lifestyle interaction, age, baseline BMI, weight change, baseline value for the respective outcome trait, income class, alcohol consumption, smoking, and family history of diabetes as fixed effects and intervention time within individuals as a random effect. All reported P values are nominal and 2-sided, and P < 0.05 was considered significant. All statistical analyses were performed with SAS version 9.4 (SAS Institute).

Results

Characteristics of the study population according to the MC4R genotype

The baseline characteristics of the total participants are presented in Supplemental Table 1. The mean age of the participants was 32.4 ± 3.5 y, and the mean BMI was 24.0 ± 3.8 kg/m². The baseline characteristics of the participants according to the MC4R rs6567160 genotype (TT, TC, or CC) are presented in Table 1. The risk allele (C) frequency was 0.22 in the study population. Mean age and intervention group distribution were similar across the MC4R genotypes (all P > 0.5). After adjustment for age, the C allele was significantly associated with higher height, weight, BMI, waist circumference, body fat, and HbA1c (all P ≤ 0.04), whereas no significant difference in fasting glucose, fasting insulin, HOMA-IR, or HOMA-B was observed across the MC4R genotypes (all P > 0.1). In addition, participants carrying the C allele tended to have higher SBP and DBP and more total energy and fat intake but less carbohydrate intake (all P ≤ 0.08). No other differences in baseline characteristics were observed across the MC4R genotypes.

TABLE 1.

Baseline characteristics of participants according to the MC4R genotype¹

	MC4R rs6567160 genotype
Characteristics	TT (n = 674)	TC (n = 405)	CC (n = 49)	P
Age, y	32.4 ± 3.5	32.3 ± 3.4	32.4 ± 3.4	0.54
Height, cm	159.9 ± 5.4	160.7 ± 5.2	161.5 ± 5.6	0.006
Weight, kg	60.8 ± 10.1	62.5 ± 10.8	64.1 ± 11.5	0.003
BMI, kg/m2	23.8 ± 3.7	24.2 ± 3.9	24.6 ± 4.3	0.04
Waist circumference, cm	79.5 ± 9.0	80.8 ± 9.5	81.5 ± 10.1	0.02
Waist-to-hip ratio	0.82 ± 0.06	0.82 ± 0.05	0.82 ± 0.07	0.39
Body fat, %	32.5 ± 5.6	33.2 ± 5.8	33.6 ± 6.3	0.04
Fasting glucose, mmol/L	5.2 ± 0.5	5.2 ± 0.5	5.3 ± 0.6	0.70
Fasting insulin, μU/mL	6.7 (4.7, 9.7)	6.6 (4.8, 10.2)	7.8 (5.1, 12.4)	0.13
HOMA-IR	1.6 (1.1, 2.3)	1.5 (1.1, 2.4)	1.9 (1.1, 2.9)	0.14
HOMA-B	81.6 (56.8, 121.2)	86.6 (56.9, 126.1)	102.9 (57.5, 130.7)	0.21
HbA1c, %	5.5 ± 0.5	5.6 ± 0.7	5.7 ± 0.9	0.01
SBP, mm Hg	106.6 ± 11.2	107.4 ± 12	109.4 ± 11.4	0.06
DBP, mm Hg	73.2 ± 9.2	74 ± 9.6	75.1 ± 9.2	0.07
Intervention group, n (%)	330 (49.0)	204 (50.4)	25 (51.0)	0.88
Dietary intake per day
Energy, kcal	1709 ± 435	1765 ± 467	1741 ± 537	0.08
Protein,% of energy	16.0 ± 2.7	16.1 ± 2.7	16.0 ± 2.3	0.57
Fat, % of energy	32.5 ± 6.5	33.3 ± 6.1	32.9 ± 4.8	0.08
Carbohydrate, % of energy	51.5 ± 7.5	50.6 ± 6.8	51.1 ± 5.5	0.08
Physical activity ≥30 min/d2	114 (16.9)	84 (20.7)	5 (10.2)	0.10
Current drinker, n (%)	154 (22.8)	85 (21.0)	8 (16.3)	0.49
Current smoker, n (%)	15 (2.2)	11 (2.7)	1 (2.0)	0.87
Family history of diabetes, n (%)	228 (33.6)	133 (32.8)	21 (42.9)	0.38

¹Values are means ± SDs, median (25th, 75th), or n (%). Insulin, HOMA-IR, and HOMA-B were log transformed before analysis. P values were calculated by using the chi-square test for categorical variables and general linear regression test for continuous variables after adjusting for age (note that age itself was not adjusted for age). HOMA-B, homeostasis model assessment of β cell function; MC4R, melanocortin-4 receptor. CC, TC, and TT are 3 genotypes of MC4R.

²Physical activity means time spent on leisure time and commuting time physical activity per day.

During the 2.55-y intervention, dietary intake and physical activity were assessed (Supplemental Table 2). No significant differences in nutrient intake and physical activity were observed at 1.28 and 2.55 y across the MC4R genotypes (all P > 0.05).

Main associations between the MC4R genotype and longitudinal glycemic changes among women with prior GDM

The main associations between the MC4R genotype and glycemic changes in both intervention and control groups are presented in Table 2. After 1.28-y follow-up, we found each C allele was associated with a 0.08-unit greater decrease in log(insulin), log(HOMA-IR), and log(HOMA-B) (P = 0.02, 0.04, and 0.04, respectively) in the intervention group after adjustment for age, baseline value for the respective outcome, baseline BMI, concurrent weight change, income class, alcohol consumption, smoking, and family history of diabetes at baseline, whereas participants with the C allele had an increasing trend in HOMA-IR in the control group (P = 0.09). After 2.55-y follow-up, we found no significant associations between the MC4R genotype and glycemic changes in the intervention group. In the control group, we found that each C allele was associated with a 0.09-unit greater increase in 2.55-y changes in fasting glucose (P = 0.04). No other main associations were observed in glycemic changes in the intervention and control groups.

TABLE 2.

Effects of the MC4R rs6567160 genotype on 1.28- and 2.55-y changes in glycemic traits among women with prior GDM¹

	Change from baseline to 1.28 y (n = 893)							Change from baseline to 2.55 y (n = 745)
	Intervention group (n = 442)			Control group (n = 451)				Intervention group (n = 367)			Control group (n = 378)
Outcomes	β	SE	P	β	SE	P	P-interaction	β	SE	P	β	SE	P	P- interaction
Model 1
Fasting insulin, μU/mL	−0.08	0.04	0.03	0.06	0.04	0.11	0.005	−0.02	0.04	0.64	0.02	0.04	0.60	0.45
HOMA-IR	−0.08	0.04	0.06	0.07	0.04	0.09	0.008	−0.02	0.05	0.62	0.04	0.04	0.41	0.33
HOMA-B	−0.08	0.04	0.04	0.02	0.04	0.55	0.06	−0.0009	0.04	0.98	−0.01	0.04	0.76	0.91
Fasting glucose, mmol/L	0.03	0.04	0.45	0.03	0.04	0.40	0.88	−0.02	0.05	0.74	0.10	0.05	0.04	0.11
HbA1c, %	0.02	0.04	0.70	−0.01	0.04	0.70	0.44	−0.04	0.03	0.25	0.02	0.03	0.47	0.13
Model 2
Fasting insulin, μU/mL	−0.08	0.03	0.02	0.05	0.03	0.15	0.006	−0.04	0.04	0.36	0.01	0.04	0.74	0.38
HOMA-IR	−0.08	0.04	0.04	0.06	0.04	0.12	0.009	−0.04	0.04	0.32	0.03	0.04	0.49	0.24
HOMA-B	−0.08	0.04	0.03	0.01	0.04	0.80	0.09	−0.005	0.04	0.90	−0.03	0.04	0.47	0.76
Fasting glucose, mmol/L	0.03	0.04	0.44	0.03	0.04	0.48	0.99	−0.03	0.05	0.51	0.09	0.04	0.04	0.08
HbA1c, %	0.02	0.04	0.65	−0.02	0.04	0.65	0.40	−0.04	0.03	0.16	0.01	0.03	0.61	0.16
Model 3
Fasting insulin, μU/mL	−0.08	0.04	0.02	0.06	0.04	0.11	0.006	−0.04	0.04	0.29	0.02	0.04	0.64	0.38
HOMA-IR	−0.08	0.04	0.04	0.06	0.04	0.09	0.008	−0.05	0.04	0.25	0.03	0.04	0.41	0.24
HOMA-B	−0.08	0.04	0.04	0.02	0.04	0.72	0.10	−0.01	0.04	0.81	−0.02	0.04	0.55	0.78
Fasting glucose, mmol/L	0.03	0.04	0.40	0.03	0.04	0.40	0.92	−0.03	0.05	0.49	0.09	0.04	0.04	0.08
HbA1c, %	0.02	0.04	0.69	−0.01	0.04	0.76	0.43	−0.04	0.03	0.17	0.02	0.03	0.56	0.15

¹ P values for model 1 were adjusted for age and baseline values of the respective outcomes; P values for model 2 were further adjusted for baseline BMI and concurrent weight change; P values for model 3 were further adjusted for monthly income (<$5,000, $5,000 to $7,999, and ≥$8,000), smoking (no, past, or current), drinking (no, past, or current), and family history of diabetes (no or yes) at baseline. Insulin, HOMA-B, and HOMA-IR were log transformed before analysis; β represents changes in outcomes per additional copy of the rs6567160 C allele. GDM, gestational diabetes mellitus; HbA1c, glycated hemoglobin; HOMA-B, homeostasis model assessment of β cell function; MC4R, melanocortin-4 receptor.

Interactions between the MC4R genotype and lifestyle interventions on 1.28-y changes in insulin resistance and other glycemic traits

In this study, we found that lifestyle intervention significantly modified the associations between the MC4R genotype and 1.28-y changes in fasting insulin (P_{model 3} = 0.006) and HOMA-IR (P_{model 3} = 0.008) after adjustment for age and baseline value for the respective outcome trait in model 1; with additional adjustment for baseline BMI and weight change at 1.28 y in model 2; and with further adjustment for income class, alcohol consumption, smoking, and family history of diabetes at baseline in model 3 (Figure 1, Table 2). The interaction still remained significant after adjusting for baseline SBP and concurrent SBP changes (Supplemental Table 3). In all 4 models, the directions of genetic associations were consistent. Each C allele was associated with greater decreases in both fasting insulin and HOMA-IR in the intervention group. Opposite-directional associations were observed in the control group, in which each C allele tended to be associated with a greater increase in insulin and HOMA-IR. No significant interactions were observed between lifestyle intervention and the MC4R genotype on changes in fasting glucose, HOMA-B, or HbA1c (all P > 0.05). We also did not observe significant interactions between lifestyle intervention and the MC4R genotype on changes in BMI after adjusting for age, BMI, income class, alcohol consumption, smoking, and family history of diabetes at baseline (data not shown).

FIGURE 1.

Effects of the MC4R and lifestyle intervention on 1.28- and 2.55-y changes in insulin and HOMA-IR in women with prior GDM. Values are means ± SEs, adjusted for age, baseline BMI, weight change at 1.28 y, baseline values of the respective outcomes, monthly income (<$5000, $5000 to $8000, and ≥$8000), smoking (no, past, or current), drinking (no, past, or current), and family history of diabetes (no or yes) at baseline. Insulin and HOMA-IR were log transformed before analysis. At 1.28 y, for the intervention group: TT, n = 260; TC, n = 162; CC, n = 20; for the control group: TT, n = 268; TC, n = 162; CC, n = 21. At 2.55 y, for the intervention group: TT, n = 215; TC, n = 137; CC, n = 15; for the control group: TT, n = 222; TC, n = 137; CC, n = 19. (A) Changes in insulin at 1.28 y, (B) changes in HOMA-IR at 1.28 y, (C) changes in insulin at 2.55 y, and (D) changes in HOMA-IR at 2.55 y. GDM, gestational diabetes mellitus; MC4R, melanocortin-4 receptor. CC, TC, and TT are 3 genotypes of MC4R.

Interactions between the MC4R genotype and lifestyle interventions on 2.55-y trajectory changes in insulin and insulin resistance

The 2.55-y longitudinal changes in fasting insulin and insulin resistance are shown in Figure 2. From 1.28 y to 2.55 y, the effects of lifestyle intervention were attenuated so that all participants regained the levels of fasting insulin and HOMA-IR, with participants carrying the CC allele regaining the least. In the control group, the levels of the two outcomes continued to increase in all participants, and those with the C allele had a greater increase in the levels of the two outcomes. However, the interactions between the MC4R genotype and lifestyle intervention on the 2.55-y trajectory changes in fasting insulin and HOMA-IR remained significant by using generalized linear mixed models with repeated measures from baseline to 2.55 y (both P for interaction = 0.03). Participants with the C allele had a decreasing trend in fasting insulin (P = 0.07) and HOMA-IR (P = 0.10) in the intervention group, whereas participants with the C allele had an increasing trend in HOMA-IR (P = 0.10) from baseline to 2.55 y.

FIGURE 2.

The effects of MC4R genotype on changes in insulin and HOMA-IR in response to lifestyle intervention from baseline to 2.55 y. GLMMs were used to test potential interactions between SNP rs6567160 and lifestyle interventions on the trajectory of changes in insulin and HOMA-IR from baseline to 2.55 y. P = 0.03 for both interactions. Values are means ± SEs after adjustment for age, baseline BMI, weight change, baseline values for respective phenotypes, monthly income (<$5000, $5000 to $8000, and ≥$8000), smoking (no, past, or current), drinking (no, past, or current), and family history of diabetes (no or yes). GLMMs were also used to calculate regression coefficients (β) of per additional copy of the rs6567160 C allele on changes in outcomes in intervention and control groups separately. Insulin and HOMA-IR were log transformed before analysis. At 1.28 y, for the intervention group: TT, n = 260; TC, n = 162; CC, n = 20; for the control group: TT, n = 268; TC, n = 162; CC, n = 21. At 2.55 years, for the intervention group: TT, n = 215; TC, n = 137; CC, n = 15; for the control group: TT, n = 222; TC, n = 137; CC, n = 19. (A) Changes in insulin in the lifestyle intervention group, (B) changes in insulin in the control group, (C) changes in HOMA-IR in the lifestyle intervention group, and (D) changes in HOMA-IR in the control group. GLMM, generalized linear mixed model; MC4R, melanocortin-4 receptor; SNP, single nucleotide polymorphism. CC, TC, and TT are 3 genotypes of MC4R.

Discussion

In this study, we found that the MC4R genotype was significantly related to 1.28-y changes in fasting insulin, HOMA-IR, and HOMA-B in the intervention group and 2.55-y changes in fasting glucose in the control group among women with prior GDM in the Tianjin Gestational Diabetes Mellitus Prevention Program. In addition, we observed significant interactions between the MC4R genotype and lifestyle intervention on 1.28-y changes in fasting insulin and HOMA-IR, and such gene–intervention interactions remained significant at 2.55 y. Participants with the obesity- and diabetes-increasing C allele showed greater improvement of insulin resistance in response to lifestyle intervention compared with those without the allele, and the associations were independent of concurrent weight change.

Our findings support the modification effects of lifestyle factors on the genetic associations with metabolic outcomes among high-risk populations. In the Tianjin Gestational Diabetes Mellitus Prevention Program, lifestyle intervention included 1) reducing total calorie intake for women with BMI ≥24, 2) reducing fat intake, 3) controlling carbohydrate intake, 4) increasing fiber intake, and 5) increasing physical activity. Both dietary factors and physical activity interventions may interact with genetic variants on insulin resistance and diabetes risk (20, 27–32). Among the nondiabetic, overweight persons in the diabetes prevention program, each copy of the minor allele of MC4R rs17066829 corresponded to a lower diabetes risk in the lifestyle intervention group, whereas no such association was observed in the placebo group (21). However, women with prior GDM have a higher risk of progression to T2D compared with non-GDM women (1), so studies testing whether such interaction is applicable to women with prior GDM are needed in order to develop more targeted and effective diabetes prevention. The observed modifications of lifestyle intervention on the association between the MC4R genotype and changes in insulin resistance in our study were in line with previously reported MC4R–diet interactions. In a case–control study, carriers of the risk allele of MC4R rs17782313 had a higher T2D risk with low adherence to the Mediterranean diet, whereas those who had high adherence to the Mediterranean diet tended to have a lower T2D risk (33).

We found that the genetic effects of MC4R were opposite between the lifestyle intervention and control groups with regard to improvement in insulin resistance. Previous studies investigating gene–environment interactions also observed that some genetic variants were associated with an increased risk of pathology in negative environments but greater than average resilience in enriched environments (33–35). This could be partly explained by the differential susceptibility hypothesis (36, 37), in which the risk alleles are conceptualized as “plasticity alleles,” making individuals more susceptible to environmental influences—for better and for worse.

In our study, the interaction between the MC4R genotype and lifestyle intervention on improvement in insulin and HOMA-IR remained significant after adjustment for concurrent weight change, suggesting such interaction might not act via the pro-obesity effect of MC4R. Several previous observational studies also reported that the association between the MC4R genotype and risk of T2D was independent of adiposity (38, 39). In a recent genome-wide association study, SNP rs6567160 was also found to be associated with SBP (40), so we further adjusted for baseline SBP and concurrent SBP changes to minimize the bias, and the interaction remained significant. The potential mechanism remains unclear, but there are several lines of evidence. The MC4R is a key gene in the control of food intake and energy expenditure (41), and its inactivation would cause hyperphagia, hyperinsulinemia, and hyperglycemia (42). Animal studies have found that dietary fat can differentially affect the expression of MC4R in different mouse lines (43), and MC4R expression can be increased by food restriction and exercise (44). Calorie and fat intake and physical activity were included in our intervention measures.

Our analyses were performed in one of the largest, long-term randomized clinical trials to date on T2D prevention through lifestyle modifications in women with prior GDM. Comprehensive collection of clinical measures enabled us to control for potential confounding in our analyses. Nevertheless, several potential limitations must be addressed. Instead of the gold standard (the hyperinsulinemic–euglycemic clamp and the hyperglycemic clamp), we used the homeostatic model to assess insulin resistance; however, this model can also yield valuable data when primary input data are robust (45). Second, our study was an exploratory analysis, and all the participants in the study were Chinese women with prior GDM, aged 24–49 y; further investigations in populations with other races, ethnicities, or age groups are warranted to validate and generalize our findings. However, women aged <25 y are at low risk for GDM (46). For women with prior GDM, the cumulative incidence of T2D increased markedly in the first 5 y after delivery and reached a plateau after 10 y (47). One intervention study indicated that for women aged >60 y with prior GDM, intervention was no more effective than placebo (8). Thus, the age group of women with prior GDM in our study is among the target population for lifestyle intervention to prevent T2D. Third, although the original study design includes a 4-y intervention period, currently, only the first and second follow-up outcomes are available, so we analyzed only the 2.55-y changes in glycemic traits as an interim analysis. However, the diminished adherence of an intervention trial usually occurs between 6 mo and 2 y (48–50), so the best intervention effect might be observed during this time frame.

In conclusion, our study indicates that women with prior GDM with a diabetes-increasing MC4R genotype (CC or TC) may obtain better improvement than the TT genotype in insulin resistance through lifestyle intervention—that is, increasing physical activity and following healthy dietary habits. Our findings provide evidence for the importance of lifestyle interventions in preventing diabetes especially among women at high risk.

Notes

This work was supported by grants from the European Foundation for the Study of Diabetes/Chinese Diabetes Society/Lilly Program for Collaborative Research between China and Europe. GH was partly supported by grants from the National Institute of General Medical Sciences (U54GM104940) and the National Institute of Diabetes and Digestive and Kidney Diseases (R01DK100790). LQ was supported by grants from the National Heart, Lung, and Blood Institute (HL071981, HL034594, HL126024), the National Institute of Diabetes and Digestive and Kidney Diseases (DK115679, DK091718, DK100383, DK078616), and the Fogarty International Center (TW010790). YC is a recipient of a scholarship under the China Scholarship Council to pursue her studies in the United States (201706240060).

Data described in the manuscript, code book, and analytic code will not be made available because the Tianjin Gestational Diabetes Mellitus Prevention Program is a collaboration with other institutions and the authors do not have the authority to make data public.

Supplemental Figure 1 and Supplemental Tables 1–3 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.

Abbreviations used: DBP, diastolic blood pressure; GDM, gestational diabetes mellitus; HbA1c, glycated hemoglobin; HOMA-B, homeostasis model assessment of β cell function; MC4R, melanocortin-4 receptor; OGTT, oral-glucose-tolerance test; SBP, systolic blood pressure; SNP, single nucleotide polymorphism; T2D, type 2 diabetes.