Skip to main content
NCBI home page
Journal of Diabetes Investigation logoLink to Journal of Diabetes Investigation
. 2023 Nov 27;15(3):346–354. doi: 10.1111/jdi.14115

Body composition and risk of gestational diabetes mellitus: A univariable and multivariable Mendelian randomization study

Yuan Gao 1,2, Min Zhou 1,✉, Xuan Xu 1, Jun Yang Ma 3,4, Man Fen Qin 1,2
PMCID: PMC10906022  PMID: 38013660

ABSTRACT

Objective

In this study, we employed a Mendelian randomization (MR) approach to investigate the independent causal associations of six body composition traits with gestational diabetes mellitus (GDM).

Methods

Genome‐wide significant levels (P < 10 × 5−8) of single nucleotide polymorphisms associated with body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage were used as instrumental variables. Data on GDM were obtained from the FinnGen Consortium, and both univariable and multivariable Mendelian randomization were performed. We utilized five different analytical methods including inverse‐variance weighted (IVW), MR Egger, weighted median, simple mode, and weighted mode to assess the robustness of the results.

Results

With univariable Mendelian randomization, the risk of GDM increased per 1‐standard deviation (SD) increase in weight (ORweight = 1.297, P = 3.06 × 10−5), whole body fat mass (ORwhole body fat mass = 1.408, P = 1.32 × 10−6), and the risk of GDM increased per percent increase in body fat percentage (ORbody fat percentage = 1.661, P = 1.01 × 10−8). Total protein had a protective effect on the risk of GDM (ORtotal protein = 0.880, P = 0.048). However, there was no significant causal association between increases in body water mass and whole body fat‐free mass per SD and the risk of GDM. Causal associations between weight, whole body fat mass, body fat percentage, and total protein with GDM were reduced to null in multivariable Mendelian randomization.

Conclusion

The present study furnishes genetic evidence to elucidate the causal relationship between body composition traits and GDM. Additionally, further studies are imperative to establish a causal connection between body composition traits and gestational diabetes mellitus.

Keywords: Body composition, Gestational diabetes mellitus, Mendelian randomization

The potential causal relationship between body composition traits and gestational diabetes mellitus (GDM) reminds clinical doctors and nurses to pay attention to the early screening of pregnant women in order to reduce the risk of GDM and other complications. Additionally, further studies are imperative to establish a causal connection between body composition traits and GDM.

graphic file with name JDI-15-346-g004.jpg

INTRODUCTION

Gestational diabetes mellitus (GDM) is defined as carbohydrate intolerance first diagnosed during pregnancy 1 . In recent years, the prevalence of GDM has been increasing. According to 2021 survey data from the International Diabetes Federation (IDF), 21.1 million (16.7%) of live births to women in 2021 had some form of hyperglycemia in pregnancy. Of these, 80.3% were due to GDM 2 . GDM not only increases adverse perinatal outcomes for pregnant women, such as premature rupture of membranes, infection, and preterm delivery but also increases the risk of neonatal hypoglycemia, neonatal hyperbilirubinemia, macrosomia, and fetal intrauterine death 3 , 4 , 5 , 6 , 7 , 8 . At present, the etiology of GDM is controversial, and early identification of high‐risk factors for GDM is of practical significance.

Body composition traits include body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage, etc. In addition, body composition traits have been shown to be causal to many diseases, such as atrial fibrillation, atrial flutter, varicose veins, deep venous thrombosis, and pulmonary embolism 9 . Retrospective studies have shown that body composition traits are independently related to the occurrence of GDM, the findings indicate that every unit rise in the waist‐to‐hip ratio amplifies the risk of GDM by a factor of 4.562, whereas each 1 kg gain in gestational weight gain (GWG) augments the risk by a factor of 4.08, additionally, body water mass, total protein, whole body fat‐free mass, and basal metabolic rate demonstrate a protective effect against the development of GDM 10 . However, due to correlations between body composition traits, it is difficult to identify the causal relationship between body composition traits and the risk of GDM through data from observational studies. In addition, the results of observational studies may be biased due to confounding factors, and reverse causality may affect the scientific conclusions of clinical studies 11 . Randomized controlled trials are the best protocol to verify the causal relationship between exposure factors and outcomes, but ethical as well as other factors make it difficult to conduct randomized controlled trials in etiologic studies. In this study, Mendelian randomization (MR) can provide support for the identification of causality.

Mendelian randomization is a natural experimental method to infer the causal relationship between exposure factors and outcome based on genetic variation. It is considered similar to a randomized controlled experiment, using genetic variation as a natural random assignment, not only eliminates confounding factors and avoids the effect of reverse causality but also improves external validity by using large‐scale genomic data and makes the results of a study more relevant 12 . Therefore, the purpose of this study was to evaluate the independent causal relationship between six different body composition traits and the risk of GDM via univariable and multivariable Mendelian randomization using UK Biobank and summary data from the FinnGen Consortium.

MATERIALS AND METHODS

Study design

In the two‐sample Mendelian randomization, genetic variation tools were used to assess a causal association with body composition traits, and body composition was also considered a risk factor for gestational diabetes mellitus. Mendelian randomization is based on three assumptions: (1) the variation is associated with the exposure; (2) the variation is not associated with the outcome via a confounding pathway; (3) the variant does not affect the outcome directly, only possibly indirectly via the exposure 13 . Univariable Mendelian randomization allows genetic variation to estimate a direct causal relationship between each exposure and outcome in a single analytical model, therefore, we first explored the causal relationship between body composition traits and gestational diabetes mellitus by univariate Mendelian randomization analysis, and after performing sensitivity analysis, we validated our results by heterogeneity and horizontal pleiotropy. Multivariable Mendelian randomization is an extension of univariable Mendelian randomization. Genetic variation is allowed to be associated with more than one exposure factor, as long as it is not related to confounding factors of exposure or outcome or directly affects the outcome 14 . Subsequently, we used multivariate Mendelian randomization to evaluate the direct impact of each body composition trait on the occurrence of gestational diabetes mellitus 15 .

Data sources for body composition

The data sources for exposure factors were shown in Table 1. To quantify the effect of body composition on gestational diabetes mellitus, we used six body composition traits: body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage. Body water mass (N = 454,888), total protein (N = 431,772), whole body fat‐free mass (N = 454,850), weight (N = 461,632), whole body fat mass (N = 454,137), and body fat percentage (N = 454,633) from the UK Biobank in the MRC Integrated Epidemiological Unit (MRC‐IEU) database 16 were used as summary data to determine the genetic variance tools for Mendelian randomization analysis. More information about exposure factors can be accessed in the complete database of the UK Biobank. The genome‐wide association study (GWAS) pipeline of the MRC‐IEU UK biological sample bank has been optimized to quickly, and efficiently standardize the complete 500,000 estimated genetic datasets from the UK biological sample bank. The Bayesian Linear Mixed Model (BOLT‐LMM) uses a linear mixed model (LMM) to explain kinship and demographic stratification, and the population is limited to individuals of European origin after standard exclusion.

Table 1.

Characteristics of data sources

Traits Consortium Data source Year Sample size No. SNP (total) Population
Exposures
Body water mass MRC‐IEU UK Biobank 2018 454,888 9,851,867 European
Total protein MRC‐IEU UK Biobank 2018 431,772 13,585,298 European
Whole body fat‐free mass MRC‐IEU UK Biobank 2018 454,850 9,851,867 European
Weight MRC‐IEU UK Biobank 2018 461,632 9,851,867 European
Whole body fat mass MRC‐IEU UK Biobank 2018 454,137 9,851,867 European
Body fat percentage MRC‐IEU UK Biobank 2018 454,633 9,851,867 European
Outcome
Gestational diabetes mellitus FinnGen 2022 190,879 20,160,256 European
Open in a new tab

MRC‐IEU, the MRC IEU Open GWAS data infrastructure; SNP, single‐nucleotide polymorphism.

Genetic instrument selection

As an independent instrument, we set a genome‐wide significance threshold of P < 5 × 10−8 to select single‐nucleotide polymorphisms (SNPs) related to exposure factors and used the clumping procedure with a cutoff of R 2 = 0.001, kb > 10,000 to remove linkage disequilibrium (LD). Likewise, we used the PhenoScanner tool to exclude any other phenotype‐associated SNPs that affected the risk of gestational diabetes mellitus independent of the selected six body composition traits 17 . Palindromic SNPs (referring to SNPs with A/T or G/C alleles) were excluded, and to avoid potential statistical deviations in the original GWAS data, SNPs with a minor allele frequency (MAF) < 0.01 were also excluded because they usually carried a low confidence level. In addition, MR Pleiotropy RESidual Sum and Outlier (MR‐PRESSO) was used to remove potential outliers before each analysis. LDlink (https://ldlink.nci.nih.gov/) was used to find a proxy SNP when the SNP was not available for the outcome. We also calculated R 2 to represent the variance of exposure explained by instrument variables and the F statistic F = [R 2 (N − K − 1)/K (1 − R 2)] to measure the strength of the instrument. Instrument variables with F < 10 were excluded and are often referred to as ‘weak instrument variables’.

Outcome source

The summary data associated with selected single‐nucleotide polymorphisms and gestational diabetes mellitus come from the FinnGen Consortium. The FinnGen research project is an ongoing Finnish consortium study that combines genotypic data from the Finnish Biobank with digital health records from the Finnish Health Registry since 2017. The data analysis is based on FinnGen release 8 (http://r8.finngen.fi) and includes 11,279 women of European origin with GDM 18 . The data analyzed in this study are based on published aggregate statistics, which have been approved by the Ethics Committee for human experiments and can be found according to the website cited. This study does not involve additional ethical proof.

Statistical analyses

Five different Mendelian randomization methods (inverse‐variance weighted [IVW], MR Egger, weighted median, simple mode, weighted mode) were used in this study to address heterogeneity and pleiotropy effects. IVW was the primary method of analysis method and assumed that all genetic variants were valid instrumental variables and provided estimates with the highest power 19 . The MR Egger method can evaluate whether genetic variation demonstrates pleiotropy in results with an average difference of zero (horizontal pleiotropy) and provides a consistent estimation of causal effects under a weaker hypothesis‐InSIDE (instrument strength independent of direct effects), but it has low estimation accuracy and exaggerates the probability of type I error 20 . The weighted median approach complements the MREgger approach, which differs from IVW in that the estimate is consistent even if up to 50% of the information comes from invalid instrumental variables 21 . Although the simple mode and weighted mode are not as powerful as IVW, the simple mode can monitor the robustness of gene polymorphism, and the weighted mode is more sensitive to mode selection bandwidth 22 , 23 .

The global test in the MR‐PRESSO method is used to test the existence of horizontal pleiotropy, and the MR‐PRESSO outlier test is used to eliminate abnormal SNPs and to estimate the corrected result. The MR‐PRESSO distortion test is also used to test whether there is a difference before and after correction 24 . In addition, the MR Egger intercept test and leave‐one‐out test were used to further analyze horizontal pleiotropy. Heterogeneity was tested by Cochran's Q test and I 2. The I 2 statistic was calculated to evaluate the heterogeneity between SNPs. Scatter plots, funnel plots, forest plots, and leave‐one‐out plots were used to assess significant genetic and phenotypic correlations between SNPs and body composition traits, and the multivariable inverse variance weighted method was used to compare the effects of body composition traits on the risk of developing gestational diabetes mellitus. The use of OR values to express causality is based on the increase in standard deviation (SD) of body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage. In multivariable Mendelian randomization, P < 0.008 was considered statistically significant, and in univariable Mendelian randomization, P values between 0.008 and 0.05 were considered statistically significant. The above analyses were performed with TwoSampleMR (version 0.5.6) 25 , Mendelian Randomization (version 0.7.0) 26 , MVMR (version 0.3) 27 , and MR‐PRESSO (version 1.0) 28 in statistical Software R (version 4.1.3).

RESULTS

Univariable MR analysis

The study design and the characteristics of exposure factors (body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage) are shown in Figure 1 and Table 1. For body water mass, total protein, whole body fat‐free mass, weight, whole body fat mass, and body fat percentage, 416, 137, 414, 345, 262, and 238 genome‐wide independent SNPs were identified as univariable Mendelian randomization tool variables from the results of the UK Biobank consortium in IEU OpenGWAS. We used 496 instrumental variables in a multivariable Mendelian randomization analysis of total protein, body weight, whole body fat mass, and body fat percentage. The F‐statistics for these exposure factors were 35.15, 24.34, 36.01, 39.4, 39.60, and 38.61, respectively, all of which were greater than 10 and were strong instrumental variables (Table S1).

Figure 1.

Figure 1

Open in a new tab

Overview of study design. There are three key assumptions for Mendelian randomization (MR). Assumption 1: the genetic variants selected as instrumental variables should be robustly associated with the body composition traits. Assumption 2: the used genetic instruments are not associated with the outcome via a confounding pathway. Assumption 3: the genetic instruments of an exposure do not affect the gestational diabetes mellitus directly, only possibly indirectly via the exposure. SNP, single‐nucleotide polymorphism.

Univariable Mendelian randomized primary analysis IVW results showed that increases in weight, whole body fat mass, and body fat percentage per SD or % were associated with an increased risk of developing gestational diabetes mellitus (ORweight = 1.297, 95% CI = 1.147–1.466, P = 3.07 × 10−5; ORwhole body fat mass = 1.408, 95% CI = 1.225–1.617, P = 1.33 × 10−6; ORbody fat percentage = 1.661, 95% CI = 1.396–1.976, P = 1.01 × 10−8). The increase in total protein per SD had a protective effect on the risk of gestational diabetes mellitus (ORtotal protein = 0.869, 95% CI = 0.765–0.988, P = 0.032). However, the increase in body water mass and whole body fat‐free mass per SD had no significant correlation with the risk of GDM (ORbody water mass = 1.041, 95% CI = 0.894–1.212, P = 0.597; ORwhole body fat‐free mass = 0.986, 95% CI = 0.857–1.135, P = 0.854) (Figure 2). The estimates of different statistical methods are consistent with the results of the main IVW method (Table S2). In addition, we performed sensitivity analyses using MR‐PRESSO, MR Egger, Cochran's Q test, and leave‐one‐out tests. At the time of the MR‐PRESSO test, there were some pleiotropic outlier SNPs in the instrumental variables of body water mass, weight, and whole body fat mass, and they were associated with the risk of developing GDM (P < 0.05). However, after excluding and correcting for these pleiotropic outlier SNPs, the results were in the same direction as the previous univariable analysis, supporting the robustness of the previous results (Table 2). The results of the MR Egger intercept test showed that the variable P Egger intercept was greater than 0.05 for all variables except total protein, which did not have horizontal pleiotropy (Table 2). The leave‐one‐out results showed that the causal estimates were robust after stepwise removal of individual SNPs (Figure S4). Heterogeneity was calculated using Cochran's Q test and the formula I 2 = [Q − (K − 1)]/Q to calculate I 2 to assess the magnitude of heterogeneity. The results showed low heterogeneity between 15% and 28% 29 . Finally, scatter plots, funnel plots, and forest plots did not reveal significantly abnormal or pleiotropic SNPs (Figures S1–S3).

Figure 2.

Figure 2

Open in a new tab

Association of six body composition traits with risk of gestational diabetes mellitus in five analysis methods.

Table 2.

Sensitivity analysis results

Exposures Outcome MR Egger Q‐statistics I 2 MR PRESSO
MR Egger intercept Value (Egger intercept) Global test P‐value Outlier‐corrected OR (95% CI) Outlier‐corrected P‐value
Body water mass −0.00006 0.977 549.26 0.244 0.001 0.950 (0.825, 1.094) 0.48
Total protein Gestational diabetes mellitus 0.009 0.008 184.21 0.261 0.002 NA NA
Whole body fat‐free mass 0.002 0.299 549.41 0.248 0.001 NA NA
Weight 0.0004 0.856 461.38 0.254 0.001 1.310 (1.162, 1.478) 1.051 × 10−5
Whole body fat mass 0.003 0.317 365.56 0.286 0.001 1.421 (1.238, 1.630) 5.550 × 10−7
Body fat percentage 0.002 0.591 279.59 0.152 0.001 NA NA
Open in a new tab

The I 2 statistic was calculated to evaluate the heterogeneity between SNPs.

CI, confidence interval; MR, Mendelian randomization; MR‐PRESSO, Mendelian randomization Pleiotropy RESidual Sum and Outlier; OR, odds ratio.

Multivariable MR analysis

The results of multivariable Mendelian randomization analysis are shown in Figure 3. In the multivariable Mendelian randomization with mutual adjustment of total protein, weight, whole body fat mass, and body fat percentage, the causal estimates of total protein, weight, whole body fat mass, and body fat percentage on the risk of gestational diabetes mellitus were significantly reduced, and no causal correlations were found (ORtotal protein = 0.984, 95% CI = 0.891–1.050, P = 0.423; ORweight = 0.454, 95% CI = 0.085–2.413, P = 0.354; ORwhole body fat mass = 1.964, 95% CI = 0.569–6.778, P = 0.362; ORbody fat percentage = 1.340, 95% CI = 0.029–3.870, P = 0.385) (Table S3).

Figure 3.

Figure 3

Open in a new tab

Associations of total protein, weight, whole body fat mass and body fat percentage with gestational diabetes mellitus in multivariable inverse‐variance weighted model based on European population.

DISCUSSION

By using a large amount of GWAS data and Mendelian randomization, our univariable Mendelian randomized results confirmed the association between total protein, weight, whole body fat mass, and body fat percentage and the risk of gestational diabetes mellitus, but there was no association between body water mass or whole body fat‐free mass and gestational diabetes mellitus. However, after adjusting for total protein, weight, whole body fat mass, and body fat percentage in multivariable Mendelian randomization, these correlations are no longer significant, which may indicate that there is no direct causal relationship between total protein, weight, body fat mass and body fat percentage and the risk of gestational diabetes mellitus.

The incidence of gestational diabetes mellitus is increasing year by year, and previous studies have demonstrated the important role of variants in susceptibility genes such as glucokinase (GCK) and TCF7L2 genes in the development of gestational diabetes mellitus 30 . Similarly, the interaction of genetic, ethnic, and environmental factors affects the progress of GDM, although our research is focused on the role of body composition traits in the development of gestational diabetes mellitus 31 . Our univariate Mendelian randomization findings demonstrate no statistically significant association between body water mass increase and the incidence of gestational diabetes (OR: 1.041, 95% CI = 0.894–1.212), this is inconsistent with the results of Li et al. who investigated the protective effect of body water mass against gestational diabetes mellitus, and there may be several reasons for this discrepancy. First, the data in the present study were derived from a European population, whereas the subjects in Li et al.'s study were derived from China, and there were significant dietary differences between the subjects in the two studies, which may have affected their body composition traits (including body water mass). Second, the data collected in this study were from subjects throughout their pregnancy, whereas the body composition traits data collected in the study by Li et al. were from early pregnancy (the first 14 weeks of pregnancy), and the nutritional status and body composition status of pregnant women varied more as the weeks of pregnancy increased. Therefore, further investigation is needed regarding the association between body water mass and the risk of gestational diabetes mellitus.

Protein is essential to the growth and development of the mother and fetus during pregnancy and is an important part of a healthy human diet. The insulin‐promoting properties of dietary proteins enhance the ability of peripheral tissues to remove glucose from the blood, thereby reducing blood glucose levels and affecting blood glucose patterns in the body 32 . The results of a cohort study by Bao et al. 33 showed that excessive intake of animal protein, especially red meat, leads to protein energy overload in pregnant women, and a higher risk of gestational diabetes mellitus, but insufficient protein content in the body leads to serious adverse pregnancy outcomes, which affects maternal and fetal health. Whereas the results of univariate Mendelian randomization in the present study confirmed total protein as a protective factor for the development of gestational diabetes mellitus, the results of multivariate Mendelianization similarly showed that total protein was negatively associated with the risk of developing gestational diabetes mellitus, but lacked statistical significance. The results of a prospective cohort study involving 11 clinical centers showed a negative correlation between whole body fat‐free mass and GDM 34 . In contrast, the results of our univariate Mendelian randomization suggest that whole body fat‐free mass is not first significantly associated with an increased risk of gestational diabetes mellitus. The reason for the negative correlation between whole body fat‐free mass and the risk of GDM derived by Xu et al. may be related to the fact that total energy expenditure in the body depends on the level of basal metabolic rate, of which defatted body weight is a determinant. An increase in whole body fat‐free mass is not only closely related to endogenous glucose output but also positively related to the consumption of glucose in the body, which is beneficial to the control of blood glucose. This may explain the negative correlation between whole body fat‐free mass and gestational diabetes mellitus 10 .

Among the many known risk factors for gestational diabetes mellitus, weight is a relatively controllable factor. Research has demonstrated that women with normal weight, class I obesity, and class II obesity, who experience weight gain beyond the expected range during the first trimester of pregnancy, are at an elevated risk of developing gestational diabetes mellitus. Furthermore, among normal weight pregnant women, a single standard weight gain difference in the first trimester greater than its predicted value was linked to a 23% increase in the risk of GDM 35 . In addition, some scholars have used latent trajectory models to further verify the risk of gestational diabetes mellitus in pregnant women with excessive weight gain and a high risk of macrosomia and cesarean section 36 . However, the relationship between body composition traits and the risk of GDM differs in pregnant women with similar weights. Other researchers have proposed that BMI is the best index to evaluate the risk of gestational diabetes mellitus, but BMI cannot provide information about obesity and the distribution of body fat 37 .

Whole body fat mass is the sum of visceral adipose tissue, subcutaneous adipose tissue, and muscle adipose tissue, which can reflect the level of serum adiponectin and insulin resistance 30 . Studies have shown that visceral adipose tissue leads to insulin resistance in the liver and impairs islet function through high lipolysis activity and the release of free fatty acids into the portal circulation. In addition, the accumulation of subcutaneous adipose tissue leads to an increase in leptin and the secretion of tumor necrosis factor‐α (TNF‐α), which reduces insulin sensitivity 38 , 39 . Therefore, an increase in whole body fat mass increases insulin resistance and leads to the impairment of islet function, which leads to an increased risk of gestational diabetes mellitus; this scenario is consistent with the results of previous studies and our univariable Mendelian randomization 40 . Furthermore, the results of multivariate Mendelian randomization analyses indicate that there exists no significant correlation between body fat mass and the risk of developing GDM. Given these findings, further research is necessary to elucidate the underlying mechanisms that link body fat mass and the development of gestational diabetes mellitus.

The body fat percentage is the percentage of whole body fat mass in the weight, and it is considered normal for the percentage of female body fat to be <30% and for male body fat to be <25% 41 . Inflammation caused by excessive accumulation of adipose tissue and an increase in cytokines influences the development of diabetes. According to a recent prospective longitudinal cohort study by Michelle et al., an increase in body fat percentage during the initial 2 months of pregnancy is significantly associated with an elevated risk of gestational diabetes mellitus. Furthermore, the study highlights the potential utility of body fat percentage as a powerful predictor of GDM, underscoring the importance of monitoring maternal body composition during early pregnancy as a means of identifying women at risk for this condition. These findings have important implications for clinical practice, as they suggest that early intervention to manage body fat percentage may help to mitigate the risk of gestational diabetes mellitus and its associated maternal and fetal complications 42 . At present, the pathophysiological mechanisms between body composition traits, obesity and insulin resistance, insulin sensitivity, and gestational diabetes mellitus still require investigation.

As far as we know, this is the first study to use aggregate GWAS data to investigate the causal relationship between body composition traits and gestational diabetes mellitus. Secondly, the Mendelian randomization method we used is less susceptible to confounding factors and reverse causality than observational studies, and we used sensitivity analysis in addition to the main analysis to ensure the robustness of the results and to minimize the impact of pleiotropic SNPs on the results. In addition, we used exposure data from UK Biobank and outcome data from FinnGen, both of which had a large sample size that enhanced statistical power. Of course, this study had some limitations. First, our subjects were of European origin, but body composition traits and fat distribution are affected by race, which explains significant correlations between body composition traits and gestational diabetes mellitus in Asian populations in previous studies but a significant decline in our study. Therefore, the generalizability of our results is limited. Second, we cannot completely rule out SNPs pleiotropy hidden in the study, which may lead to deviations in causality assessment; but, when we used MREgger and MR‐PRESSO for sensitivity analysis, we did not observe significant multiplicity or heterogeneity.

CONCLUSIONS

The present study furnishes genetic evidence to elucidate the causal relationship between body composition traits and gestational diabetes mellitus. Univariable Mendelian randomization results corroborate the association between gestational diabetes mellitus risk and total protein, weight, whole body fat mass, and body fat percentage, whereas no such evidence is observed for body water mass or whole body fat‐free mass. However, multivariable Mendelian randomization analysis reveals that the aforementioned associations become null after adjusting for total protein, weight, whole body fat mass, and body fat percentage. The potential causal relationship between body composition traits and gestational diabetes mellitus reminds clinical doctors and nurses to pay attention to the early screening of pregnant women in order to reduce the risk of GDM and other complications. Additionally, further studies are imperative to establish a causal connection between body composition traits and gestational diabetes mellitus.

FUNDING

This study was supported by the Jiangxi provincial science and technology department Key R&D program (20202BBGL73075).

DISCLOSURE

The authors declare no conflict of interest.

Approval of the research protocol: N/A.

Informed consent: N/A.

Registry and the registration no. of the study/trial: N/A.

Animal studies: N/A.

Supporting information

Table S1 | Genetic instruments selected for body composition traits in UK Biobank

Table S2 Associations of body composition traits with gestational diabetes mellitus in univariable inverse‐variance weighted model based on European population

Table S3 Associations of total protein, weight, whole body fat mass, and body fat percentage with gestational diabetes mellitus in multivariable inverse‐variance weighted model based on European population

Figure S1 Comparison of the causal effects of body composition traits on gestational diabetes using scatter plots from different Mendelian randomization methods.

Figure S2 Forest plot of MR effect size using MR‐Egger and IVW methods for causal associations between body composition traits and gestational diabetes.

Figure S3 Funnel plot of causal associations between body composition traits and gestational diabetes.

Figure S4 Leave‐one‐out plot to assess if a single variant is driving the association between body composition traits and gestational diabetes mellitus.

JDI-15-346-s001.docx (1.5MB, docx)

DATA AVAILABILITY STATEMENT

GWAS dataset are available at IEU OpenGWAS project (https://gwas.mrcieu.ac.uk/). Data on GDM were obtained from the FinnGen Consortium (http://r8.finngen.fi).

REFERENCES

  • 1. Dłuski DF, Wolińska E, Skrzypczak M. Epigenetic changes in gestational diabetes mellitus. Int J Mol Sci 2021; 22: 7649. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Magliano DJ, Boyko EJ, Committee IDFDAtes . IDF Diabetes Atlas, 10th edn. Brussels: International Diabetes Federation, 2021. [Google Scholar]
  • 3. Mack LR, Tomich PG. Gestational diabetes: diagnosis, classification, and clinical care. Obstet Gynecol Clin North Am 2017; 44: 207–217. [DOI] [PubMed] [Google Scholar]
  • 4. Sacks DA, Black MH, Li X, et al. Adverse pregnancy outcomes using the International Association of the Diabetes and Pregnancy Study Groups criteria: glycemic thresholds and associated risks. Obstet Gynecol 2015; 126: 67–73. [DOI] [PubMed] [Google Scholar]
  • 5. Ye W, Luo C, Huang J, et al. Gestational diabetes mellitus and adverse pregnancy outcomes: systematic review and meta‐analysis. BMJ 2022; 377: e067946. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Tabrizi R, Asemi Z, Lankarani KB, et al. Gestational diabetes mellitus in association with macrosomia in Iran: a meta‐analysis. J Diabetes Metab Disord 2019; 18: 41–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Kouhkan A, Najafi L, Malek M, et al. Gestational diabetes mellitus: major risk factors and pregnancy‐related outcomes: a cohort study. Int J Reprod Biomed 2021; 19: 827–836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Li YX, Long DL, Liu J, et al. Gestational diabetes mellitus in women increased the risk of neonatal infection via inflammation and autophagy in the placenta. Medicine (Baltimore) 2020; 99: e22152. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Zeng H, Lin C, Wang S, et al. Genetically predicted body composition in relation to cardiometabolic traits: a Mendelian randomization study. Eur J Epidemiol 2021; 36: 1157–1168. [DOI] [PubMed] [Google Scholar]
  • 10. Li X, Xu D, Zhang L, et al. Correlation of body composition in early pregnancy on gestational diabetes mellitus under different body weights before pregnancy. Front Endocrinol 2022; 13: 916883. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet 2002; 359: 248–252. [DOI] [PubMed] [Google Scholar]
  • 12. Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol 2004; 33: 30–42. [DOI] [PubMed] [Google Scholar]
  • 13. Didelez V, Sheehan N. Mendelian randomization as an instrumental variable approach to causal inference. Stat Methods Med Res 2007; 16: 309–330. [DOI] [PubMed] [Google Scholar]
  • 14. Lu Z, Jiao Y, Li J. Higher genetically predicted triglycerides, LDL, and HDL increase the vitamin D deficiency: a Mendelian randomization study. Front Nutr 2022; 9: 862942. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Sanderson E, Davey Smith G, Windmeijer F, et al. An examination of multivariable mendelian randomization in the single‐sample and two‐sample summary data settings. Int J Epidemiol 2019; 48: 713–727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Elsworth BL, Lyon MS, Alexander T, et al. The MRC IEU OpenGWAS Data Infrastructure. New York, NY: Cold Spring Harbor Laboratory, 2020. [Google Scholar]
  • 17. Kamat MA, Blackshaw JA, Robin Y, et al. PhenoScanner V2: an expanded tool for searching human genotype–phenotype associations. Bioinformatics 2019; 35: 4851–4853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Kurki MI, Karjalainen J, Palta P, et al. FinnGen provides genetic insights from a well‐phenotyped isolated population. Nature 2023; 613(7944): 508–518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Xu J, Zhang S, Tian Y, et al. Genetic causal association between iron status and osteoarthritis: a two‐sample Mendelian randomization. Nutrients 2022; 14: 3683. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Burgess S, Thompson SG. Interpreting findings from mendelian randomization using the MR‐egger method. Eur J Epidemiol 2017; 32: 377–389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Bowden J, Smith GD, Haycock PC, et al. Consistent estimation in mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 2016; 40: 304–314. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Hu J, Song J, Chen Z, et al. Reverse causal relationship between periodontitis and shortened telomere length: bidirectional two‐sample mendelian random analysis. Front Immunol 2022; 13: 1057602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Hartwig FP, Smith GD, Bowden J. Robust inference in summary data mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol 2017; 46: 1985–1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Verbanck M, Chen C‐Y, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nature Genetics 2018; 50: 1196. [DOI] [PubMed] [Google Scholar]
  • 25. Hemani G, Zheng J, Elsworth B, et al. The MR‐base platform supports systematic causal inference across the human phenome. Elife 2018; 7: e34408. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Yavorska OO, Burgess S. MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. Int J Epidemiol 2017; 46: 1734–1739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Rasooly D, Peloso GM. Two‐sample multivariable mendelian randomization analysis using R. Curr Protoc 2021; 1: e335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Verbanck M, Chen CY, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Bowden J, Del Greco FM, Minelli C, et al. Assessing the suitability of summary data for two‐sample Mendelian randomization analyses using MR‐egger regression: the role of the I‐2 statistic. Int J Epidemiol 2016; 45: 1961–1974. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Freathy RM, Hayes MG, Urbanek M, et al. Hyperglycemia and adverse pregnancy outcome (HAPO) study: common genetic variants in GCK and TCF7L2 are associated with fasting and postchallenge glucose levels in pregnancy and with the new consensus definition of gestational diabetes mellitus from the International Association of Diabetes and Pregnancy Study Groups. Diabetes 2010; 59: 2682–2689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Baz B, Riveline JP, Gautier JF. Gestational diabetes mellitus: definition, aetiological and clinical aspects. Eur J Endocrinol 2016; 174: R43–R51. [DOI] [PubMed] [Google Scholar]
  • 32. Rietman A, Schwarz J, Tome D, et al. High dietary protein intake, reducing or eliciting insulin resistance? Eur J Clin Nutr 2014; 68: 973–979. [DOI] [PubMed] [Google Scholar]
  • 33. Bao W, Bowers K, Tobias DK, et al. Prepregnancy dietary protein intake, major dietary protein sources, and the risk of gestational diabetes mellitus a prospective cohort study. Diabetes Care 2013; 36: 2001–2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Xu Q, Gao ZY, Li LM, et al. The Association of Maternal Body Composition and Dietary Intake with the risk of gestational diabetes mellitus during the second trimester in a cohort of Chinese pregnant women. Biomed Environ Sci 2016; 29: 1–11. [DOI] [PubMed] [Google Scholar]
  • 35. Buck CO, Shabanova V, Taylor SN. Growth patterns among late preterm infants of mothers with diabetes. J Matern Fetal Neonatal Med 2022; 35: 10116–10123. [DOI] [PubMed] [Google Scholar]
  • 36. Peng YZ, Han N, Su T, et al. Gestational weight gain and the risk of gestational diabetes mellitus: a latent class trajectory analysis using birth cohort data. Diabetes Res Clin Pract 2021; 182: 109130. [DOI] [PubMed] [Google Scholar]
  • 37. Sommer C, Morkrid K, Jenum AK, et al. Weight gain, total fat gain and regional fat gain during pregnancy and the association with gestational diabetes: a population‐based cohort study. Int J Obes (Lond) 2014; 38: 76–81. [DOI] [PubMed] [Google Scholar]
  • 38. Patel P, Abate N. Body fat distribution and insulin resistance. Nutrients 2013; 5: 2019–2027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Koszowska AU, Nowak J, Dittfeld A, et al. Obesity, adipose tissue function and the role of vitamin D. Cent Eur J Immunol 2014; 39: 260–264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40. Liu YP, Liu J, Gao YJ, et al. The body composition in early pregnancy is associated with the risk of development of gestational diabetes mellitus late during the second trimester. Diabetes Metab Syndr Obes 2020; 13: 2367–2374. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41. Liu F, Pang D, Hua L, et al. Analysis of body composition of residents with different stroke risk classes in the community. Nurs Res 2017; 31: 4642–4645. [Google Scholar]
  • 42. Bai M, Susic D, O'Sullivan AJ, et al. Reproducibility of bioelectrical impedance analysis in pregnancy and the association of body composition with the risk of gestational diabetes: a substudy of MUMS cohort. J Obes 2020; 2020: 3128767. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1 | Genetic instruments selected for body composition traits in UK Biobank

Table S2 Associations of body composition traits with gestational diabetes mellitus in univariable inverse‐variance weighted model based on European population

Table S3 Associations of total protein, weight, whole body fat mass, and body fat percentage with gestational diabetes mellitus in multivariable inverse‐variance weighted model based on European population

Figure S1 Comparison of the causal effects of body composition traits on gestational diabetes using scatter plots from different Mendelian randomization methods.

Figure S2 Forest plot of MR effect size using MR‐Egger and IVW methods for causal associations between body composition traits and gestational diabetes.

Figure S3 Funnel plot of causal associations between body composition traits and gestational diabetes.

Figure S4 Leave‐one‐out plot to assess if a single variant is driving the association between body composition traits and gestational diabetes mellitus.

JDI-15-346-s001.docx (1.5MB, docx)

Data Availability Statement

GWAS dataset are available at IEU OpenGWAS project (https://gwas.mrcieu.ac.uk/). Data on GDM were obtained from the FinnGen Consortium (http://r8.finngen.fi).

Articles from Journal of Diabetes Investigation are provided here courtesy of Wiley

RESOURCES