Genetic Risk Score of Hemoglobin A_1c interact with Postpartum Weight Reduction on Glycemic Changes: Gene-Weight Interaction Analysis

Abstract

Aim:

We aimed to investigate the interaction between genetic risk score of HbA_1c and weight changes during and after pregnancy(postpartum weight reduction and gestational weight gain) on long-term glycemic changes in the largest cohort of women with preceding history of gestational diabetes mellitus (GDM).

Methods:

This is a retrospective cohort using the baseline data from the Tianjin Gestational Diabetes Mellitus Prevention Program. A genetic risk score was established by combining 10 HbA_1c related single nucleotide polymorphisms (SNPs), which were identified by genome-wide association studies (GWAS). General linear regression models were applied to evaluate the interaction between HbA_1c genetic risk score and weight changes during and after pregnancy(postpartum weight reduction and gestational weight gain) on glycemic changes.

Results:

A total of 1156 women with preceding history of GDM were included in this respective cohort study. Statistical differences in pre-pregnancy weight, pre-delivery weight and postpartum weight were evidenced across different groups of postpartum weight reduction. After adjusted for covariates, statistical significance in changes of HbA_1c % was only observed in postpartum weight reduction<5 kg/yr group (p= 0.002), and significant interaction between HbA_1c genetic risk score and postpartum weight reduction on long-term changes of HbA_1c was evidenced (p-interaction=0.01). In the category of postpartum weight reduction≥8kg/y group, women with a lower genetic risk score of HbA_1c had a greater decrease of HbA_1c.

Conclusions:

A genetic risk score of HbA_1c interact with postpartum weight reduction on long-term changes of HbA_1c among women with preceding history of GDM.

INTRODUCTION

During pregnancy, women will gain about 20 percent or more of body weight ¹. Abnormal metabolism of glucose are greatly affected by weight changes during and after pregnancy ^2,3. Compelling evidence indicates that low postpartum weight reduction is associated with aberrant glucose metabolism and an increased risk of type 2 diabetes in later life ^1,4–7, women with greater postpartum weight retention and preceding history of gestational diabetes mellitus (GDM) are more likely to involve in this metabolic disease^8–11.

As a common biomarker, hemoglobin A_1c (HbA_1c) is regarded as a stable biomarker for hyperglycemia. Genetic susceptibility to higher HbA_1c levels had been verified by genome-wide association studies(GWAS) ^12,13. However, the interactions between genetic variations of HbA_1c and weight changes during and after pregnancy (postpartum weight reduction and gestational weight gain) on long-term changes of glycemic traits remain unclear. We hypothesized that weight changes during and after pregnancy (postpartum weight reduction and gestational weight gain) might modify the genetic variations of HbA_1c on glycemic changes.

Taken together, we investigated the interaction between genetic risk score of HbA_1c and weight changes during and after pregnancy (postpartum weight reduction and gestational weight gain) on long-term changes of glycemic traits in the largest cohort of women with preceding history of GDM.

2. MATERIALS AND METHODS

2.1. Study samples

Our study is a retrospective cohort using the baseline data from the Tianjin Gestational Diabetes Mellitus Prevention Program¹⁴. n accordance with the criteria of World Health Organization) at 26–30 gestational weeks(2005–2009) were recruitedWomen with pregnancy, resided in the 6 central urban districts of Tianjin and with diagnosis of GDM(i to engage in a postpartum survey(2009–2011). Those with chronic diseases, diabetes after delivery, usage of medicine to change the glucose values, pregnant and going to be pregnant in the next two years were excluded. A group of 4644 women with preceding history of GDM (postpartum at 1–5 years) were recruited from the Tianjin Gestational Diabetes Mellitus Prevention Study^14–17.

A total of 1263 women with preceding history of GDM finished the questionnaire. Glucose levels, the prevalence of impaired glucose tolerance and diabetes at 26–30 gestational weeks didn’t differ between women who returned the questionnaire and those who did not. The 3-day 24-hour food records were obtained by a dietitian. The final analysis was restricted to a group of 1156 participants with available genotype data. The proposal of this respective cohort study was approved by the Human Subjects Committee of Tianjin Women’s and Children’s Health Center. Informed consent were acquired for all of the women with preceding history of GDM.

Tianjin gestational diabetes mellitus screening program

The Tianjin Gestational Diabetes Mellitus Prevention Program is ongoing and 2-arm, 4-year randomized controlled trial, which is conducted by Tianjin Women’s and Children’s Health Centre, Tianjin, China. Compared to the usual-care group, six face-to-face sessions with study dietitians in the first year, and two extra sessions and two telephone calls in second year were performed in the intervention group. The detailed information about the study design has been described in elsewhere¹⁴. The registered number is NCT01554358 at ClinicalTrials.gov. Among the 1263 women with preceding history of GDM, who finished the questionnaire, a total of 1180 women were randomly assigned (1:1) to receive a 4-year lifestyle intervention or standard care¹⁴. Our current analysis is retrospective, the outcomes are changes in glycemic traits from pregnancy to baseline of the intervention trial. Therefore, intervention would not affect our analyses.

Variable definition

All participants completed a baseline self-administered questionnaire. Pre-pregnancy weight, postpartum weight, and weight gain in pregnancy were self-reported. Baseline weight was measured by trained physicians¹⁴. Body weight and height were measured using the standardized protocol according to the WHO MONICA project¹⁸. Weight was measured to the nearest 0.1 kg and height(without shoes) to the nearest 0.1 cm.

The postpartum weight retention (weight change between pre-pregnancy and 1–5 y postpartum) was classified as gestational weight gain and postpartum weight reduction. The annual post-partum weight reduction was calculated by dividing the difference between postpartum weight and predelivery weight (pre-pregnancy weight plus gestational weight gain) by follow-up years. Gestational weight gain was categorized as inadequate, adequate, and excessive according to 2009 Institute of Medicine guideline¹⁹. For the range of pre-pregnancy BMI of <18.5 kg/m², 18.5–24.9 kg/m², 25.0–29.9 kg/m², BMI of ≥30 kg/m², weight in the range of 12.5 to 18 kg, 11.5 to 16 kg, 7 to 11.5 kg, and 5 to 9 kg were defined as adequate gestational weight gain²⁰. Otherwise under or higher than the cut-off values were taken as inadequate or excessive, separately²⁰. Fasting glucose, 2-hour glucose and HbA_1c were measured according to the standard protocols (with analytical machines TBA-120FR and ADAMS A_1c HA-8160; Arkray, Japan, respectively). The changes in glycemic traits were computed as the corresponding values at postpartum minus the values at pre-gnancy(GDM diagnosis).

2.2. Calculation of HbA_1c genetic risk score

DNA was extracted from the buffy coat fraction of centrifuged blood using a QIAamp Blood Maxi Kit (Qiagen, Chatsworth, CA, USA). SNPs were genotyped by aquantitative real-time TaqMan PCR(Applied Biosystems) with a success rate ≥ 98%. The HbA_1c genetic risk score was established by aggregating the HbA_1c-increasing alleles according to the HbA_1c susceptibility SNPs discovered by GWAS (Table S1) ^12,13, and mutual independence were observed for all HbA_1c SNPs. Among them, rs1800562 and rs855791 were not in accordance with Hardy-Weinberg equilibrium in our data, thus rs266717 and rs12597579 were used as proxies, respectively(r²=1). GWAS provided the β-coefficients (effect sizes) for the 10 HbA_1c susceptibility SNPs ^12,13 (Table S1). A genetic risk score was established by summing the 10 effect alleles, weighted by the β-coefficients. We actually transformed the β-coefficients to indicate a higher HbA_1c. The HbA_1c genetic risk score was divided into tertiles (tertile 1 for the lowest, tertile 2 in the middle, and tertile 3 for the highest).

2.3. Statistical analysis

We did a retrospective analysis using the baseline data from Tianjin Gestational Diabetes Mellitus Prevention Program, the final analysis was restricted to a group of 1156 participants with available genotype data. The baseline characteristics across different groups of annual postpartum weight reduction (<5, 5–8, and ≥8 kg/y) and gestational weight gain(inadequate, adequate, excessive) were estimated by chi-square tests and general linear model(GLM), as appropriate. The association of HbA_1C genetic risk score and changes in glycemic traits by annual weight reduction groups was tested by GLM. An interaction between the HbA_1C genetic risk score and annual postpartum weight reduction on changes in glycemic traits was tested by including an interaction term in the models. The long-term effect of weight changes on the glycemic traits stratified by tertiles of HbA_1c genetic risk score were also estimated by GLM. In addition, sensitivity analyses were performed to estimate the effect of follow-up years on the observed association. SAS 9.4 was used to perform all of the analysis with two-sided significance at 0.05 (SAS Institute, Cary, NC).

3. RESULTS

Characteristics across different groups of postpartum weight reduction and gestational weight gain are summarized in Table 1 and Table S2, respectively. Statistical differences in pre-pregnancy weight, pre-delivery weight and postpartum weight were evidenced across different groups of postpartum weight reduction. Others characteristics and HbA_1c genetic risk score were not significant across different groups (Table 1).

Table 1.

Characteristics of women with prior gestational diabetes mellitus by annual weight reduction

Variables	Weight reduction
	<5 kg/year	5 to 8 kg/year	≥8 kg/year	P value
Number	428	396	332
Age (years)	32.46 ± 3.45	32.79 ± 3.76	31.89 ± 3.56	0.0548
Follow-up (years)	2.71 ± 0.91	2.42 ± 0.80	1.94 ± 0.78	0.1029
Pre-pregnancy BMI (kg/m2)	23.27 ± 3.47	22.95 ± 2.83	22.61 ± 2.96	0.0051
Weight (kg)
Pre-pregnancy	59.74 ± 9.52	58.46 ± 8.43	58.56 ± 8.00	0.0486
Pre-delivery	75.94 ± 11.18	72.83 ± 9.43	78.53 ± 10.39	0.0122
Postpartum	64.49 ± 10.50	64.16 ± 8.39	71.14 ± 9.94	<0.0001
Fasting glucose (mmol/L)
At GDM diagnosis	5.34 ± 0.81	5.23 ± 0.71	5.31 ± 0.83	0.4442
Postpartum	5.13 ± 0.76	5.66 ± 0.58	5.36 ± 0.95	0.4144
2-h glucose (mmol/L)
At GDM diagnosis	9.19 ± 1.27	8.99 ± 1.16	9.06 ± 1.33	0.0824
Postpartum	6.05 ± 1.03	8.30 ± 2.12	3.90 ± 1.69	0.1982
HbA1C (%)
At GDM diagnosis	5.84 ± 0.64	5.73 ± 0.63	5.78 ± 0.59	0.1479
Postpartum	5.61 ± 0.75	5.69 ± 0.67	5.69 ± 0.78	0.1040
Family history of diabetes	140 (32.71%)	117 (29.54%)	122 (36.75%)	0.7282
Current Smokers	10 (2.33%)	0 (0.00%)	10 (3.01%)	0.2654
Alcohol drinker	99 (23.13%)	79 (19.95%)	95 (28.61%)	0.4069
Leisure time physical activity
0 min/d	120 (28.04%)	320 (80.80%)	246 (74.10%)	0.4294
1 to 30 min/d	140 (32.71%)	68 (17.17%)	79 (23.79%)
≥30 min/d	168 (39.25%)	8 (1.49%)	7 (12.11%)
Sitting time (h/d)	3.19 ± 2.15	3.51 ± 2.26	3.19 ± 1.80	0.7509
Total energy intake (kal/d)	1679.27 ± 435.85	1713.65 ± 438.90	1638.34 ± 445.58	0.2889
HbA1C GRS	10.96 ± 2.07	11.11 ± 1.85	11.07 ± 1.97	0.3602

The distribution of the HbA_1c genetic risk score is illustrated in Figure S1. There were no statistical differences for the HbA_1c genetic risk score with reference levels of glycemic traits, even adjusted for covariates. The association between HbA_1c genetic risk score with changes in glycemic traits stratified by annual postpartum weight reduction and gestational weight gain are summarized in Table 2 and Table S3, respectively. After adjusted for covariates, statistical significance in changes of HbA_1c was only observed in postpartum weight reduction<5 kg/yr group (p= 0.002) (Table 2). Significant interactions were evidenced for the association between HbA_1c genetic risk score and postpartum weight reduction on changes of HbA_1c (p-interaction=0.01). In the category of postpartum weight reduction≥8 kg/y, women with a lower genetic predisposition to high HbA_1c might be more susceptible to the improvement of HbA_1c (P=0.036). Along with the accumulation of HbA_1c genetic risk score, elevated HbA_1c was evidenced in the largest postpartum weight loss group (Figure 1C). In the group of postpartum weight reduction<5 kg/yr, HbA_1c levels were decreased across tertiles of HbA_1c genetic risk score, whereas, HbA_1c levels were increased in the group of postpartum weight reduction≥8 kg/y (Figure 1C). We also estimated the tertiles of HbA_1c genetic risk score divided by weight reduction (kg/yr) on changes of HbA_1c (Table S5). In the category of postpartum weight reduction ≥8kg/y, women with a lower genetic predisposition to high HbA_1c showed more[Double check]improvement of HbA_1c (p=0.036) (Table S5).We didn’t observe any significant interactions on variations of fasting glucose and 2-hour glucose, as well as in different groups of gestational weight gain (Table S2).

Table 2.

Association between HbA_1C genetic risk score and changes in glycemic traits by annual weight reduction

Changes in glycemic traits	<5 kg/year		5 to 8 kg/year		≥8 kg/year		P for interaction
	β (SE)	P value	β (SE)	P value	β (SE)	P value
Fasting glucose, mmol/L
Age-adjusted	0.163 (0.372)	0.661	−0.707 (0.778)	0.365	−0.242 (0.316)	0.445	0.371
Multivariable-adjusted†	0.182 (0.371)	0.624	−0.508 (0.298)	0.091	−0.219 (0.320)	0.493	0.428
2-h glucose, mmol/L
Age-adjusted	0.274 (0.824)	0.739	0.039 (0.093)	0.966	0.094 (0.073)	0.899	0.712
Multivariable-adjusted†	0.192 (0.795)	0.809	−0.122 (0.089)	0.174	−0.364 (0.720)	0.613	0.965
HbA1C, %
Age-adjusted	−0.933 (0.322)	0.003	−0.226 (0.329)	0.493	−0.141 (0.103)	0.171	0.025
Multivariable-adjusted†	−0.972 (0.323)	0.002	−0.125 (0.363)	0.729	−0.176 (0.105)	0.095	0.019

^†Covariates included glycemic traits at GDM diagnosis, age, duration of follow-up, BMI at pre-pregnancy, kal/d for energy intake, time for sitting, smoking, drinking, physical activity and family history of diabetes.

Figure 1.

Changes in glycemic traits according to weight reduction and HbA_1c genetic risk score.

The HbA_1c genetic risk score were divided into tertiles (tertile 1 for the lowest, tertile 2 in the middle, and tertile 3 for the highest). From tertile 1 to tertile 3, the β values for every 1-unit postpartum weight reduction with changes of fasting glucose were 0.182 (P = 0.624), −0.508 (P =0.091), and −0.219 (P = 0.493) mmol/L, respectively (Figure 2A). The β values for every 1-unit postpartum weight reduction with changes of 2-hour glucose were 0.192 (P = 0.809), −0.122 (P =0.174), and −0.364 (P =0.613) mmol/L, respectively (Figure 2B). And the β values for every 1-unit postpartum weight reduction with changes of HbA_1c were −0.972% (P = 0.002), −0.124% (P = 0.729) and −0.176% (P = 0.095), respectively (Figure 2C).

Figure 2.

Predicted differences in glycemic traits stratified by HbA_1c genetic risk score.

A sensitivity analyses by limiting the study to participants with delivery within the first 1.5 years was also performed(sample size equal 207). No statistical significant interactions between HbA_1c genetic risk score and postpartum weight reduction/gestational weight gain on any glycemic changes were observed (Table 3 and Table S4, respectively).

Table 3.

Association between HbA_1C genetic risk score and changes in glycemic traits by weight reduction within 1.5 years

Changes in glycemic trait	<5 kg/year (N=129)		5 to 8 kg/year (N=44)		≥8 kg/year (N=34)		P for interaction
	β (SE)	P value	β (SE)	P value	β (SE)	P value
Fasting glucose, mmol/L
Age-adjusted	0.024 (0.489)	0.960	−0.249 (0.446)	0.585	−0.857 (0.864)	0.328	0.604
Multivariable-adjusted†	0.130 (0.498)	0.793	−0.130 (1.214)	0.915	−1.290 (1.046)	0.090	0.723
2-h glucose, mmol/L
Age-adjusted	0.137 (1.135)	0.903	1.363 (0.611)	0.185	4.859 (1.714)	0.700	0.088
Multivariable-adjusted†	0.200 (1.102)	0.856	−1.283 (1.549)	0.454	−6.149 (1.912)	0.343	0.097
HbA1C, %
Age-adjusted	−0.032 (0.376)	0.931	−0.656 (0.572)	0.273	−0.879 (0.625)	0.168	0.332
Multivariable-adjusted†	−0.019 (0.379)	0.958	−0.607 (1.216)	0.643	−1.191 (0.667)	0.084	0.287

^†Covariates included glycemic traits at GDM diagnosis, age, duration of follow-up, BMI at pre-pregnancy, kal/d for energy intake, time for sitting, smoking, drinking, physical activity and family history of diabetes.

4. CONCLUSION

To our knowledge, significant interaction between genetic risk score of HbA_1c and postpartum weight reduction on changes of HbA_1c was evidenced in currently the largest cohort of women with preceding history of GDM. Our data demonstrated that across different categories of postpartum weight reduction, the genetic risk score of HbA_1c differed significantly on changes of HbA_1c. Interestingly, In the group of postpartum weight reduction<5 kg/yr, HbA_1c levels were decreased across tertiles of HbA_1c genetic risk score, whereas HbA_1c levels were increased in the group of postpartum weight reduction≥8 kg/y. Our data indicated that postpartum weight reduction might modify the genetic risk score of HbA_1c on long-term HbA_1c changes, the “vulnerability genes” of HbA_1c may act as “plasticity genes”²¹, with some women more susceptible to postpartum weight reduction, others not. In the category of postpartum weight reduction≥8 kg/y, women with a lower genetic predisposition to high HbA_1c might be more susceptible to the improvement of HbA_1c (P=0.036).

Women with greater postpartum weight retention are more susceptible to insulin resistance ²². Improvement in glucose metabolism in women with GDM was observed after moderate postpartum weight loss⁹. Low postpartum weight reduction contributes significantly to the long-term risk of type 2 diabetes in later life, especially in groups of women with preceding history of GDM ^1,4–7. Therefore, postpartum weight reduction is a great concern in this population ^23,24.

To the best of our knowledge, our findings demonstrated that genetic predisposition to high HbA_1c interacted significantly with postpartum weight reduction on postpartum changes of HbA_1c. This study brings novel perspectives to understand the role of postpartum body weight management in glycemic improvement under the genetic background. Furthermore, we created the HbA_1c genetic risk score using widely accepted methods to sum up the genetic variations weighted by their effect sizes ^25,26. Individual genetic variants explain a very small fraction of the overall variation, and a combination of a higher number of included variants with weak to moderate effect sizes into a genetic risk score can explain a larger proportion of heritability ^27,28.

Two large GWAS analyzed the common genetic determinants of HbA_1c levels with larger sample sizes and independent replications, the identified SNPs reached genome-wide significance levels (p<5×10⁻⁸), thus providing more reliable estimates^12,29. Currently, the construction of GRS use the weighted published effect size from GWAS identified SNPs that achieved genome-wide significance^30,31. However, the SNPs in our study didn’t reach the stringent levels, therefore, we used the weighted β-coefficients from the published two large GWAS^12,29, and the results were similar when we used the un-weighted GRS scores. Besides, genetic architectures for HbA_1c and fasting glucose were partially overlapped, GWAS of fasting glucose suggested that SNPs near three loci (G6PC2, MTNR1B, and GCK) were also associated with HbA_1c levels¹³. Therefore, we not only investigated the interaction between HbA_1c genetic risk score and postpartum weight reduction on HbA_1c changes, but also on changes of fasting glucose and 2-h glucose.

The results should be mentioned cautiously, our findings may not be generalized to other populations, therefore different studies are required to verify our findings. The sample size of women in the sensitivity analysis was comparatively small, thus the interaction effect was not significant in this sub-group of analysis. The genetic susceptibility of HbA_1c SNPs were primarily originated from European ancestry^12,29, which may not be representative of the genetic susceptibility of Chinese women with GDM.

Taken together, we found postpartum weight reduction interacted with genetic predisposition to HbA_1c on long-term changes of HbA_1c. Contradictory to women with a higher genetic risk score, women with a lower genetic risk score demonstrated a more distinct association between postpartum weight reduction and HbA_1c improvement. In women genetically predisposed to high HbA_1c, postpartum weight management is of great concern.