Gestational diabetes mellitus (GDM) is often diagnosed in the second or third trimester of pregnancy, which was not clearly overt before gestation(Reference Wu, Tang and Zeng1). GDM affects approximately 16·9 % of pregnancies globally(Reference Guariguata, Linnenkamp and Beagley2) with its prevalence ranging from 5 to 24 % in China(3). GDM can lead to adverse perinatal outcomes including macrosomia, caesarean delivery and shoulder dystocia(Reference Li, Zhong and Li4,Reference Farrar, Simmonds and Bryant5) . In addition, GDM has been linked to long-term adverse outcomes in both mothers (e.g. higher incident type 2 diabetes later in life) and their offspring (e.g. increasing rate of adiposity and diabetes in teenagerhood or young adulthood)(Reference Li, Li and Chavarro6–Reference Fetita, Sobngwi and Serradas9). Therefore, identifying modifiable risk factors for GDM prevention is imperative. Despite the unclear pathogenesis of GDM, a growing body of evidence indicates that healthy diet and lifestyle factors could benefit the establishment of prevention strategies for GDM(Reference Li, Zhong and Li4,Reference Li, Li and Chavarro6,Reference Zhang and Ning10,Reference Zhang, Tobias and Chavarro11) .
Choline, an essential nutrient, serves as a precursor to synthesise the neurotransmitter acetylcholine(Reference Zhu, Liu and He12). It also contributes to the component of cell membrane and lipoprotein cholesterol(Reference Zhu, Liu and He12). In addition, choline can also irreversibly convert to betaine, which donates its one-carbon group for remethylating homocysteine to methionine and regulates methylation modification(Reference Zhu, Liu and He12). Both choline and betaine can be synthesised endogenously, but their de novo synthesis may not be sufficient(Reference Zhu, Liu and He12,Reference Golzarand, Bahadoran and Mirmiran13) . Thus, exogenous sources of choline and betaine by diet are essential for optimal human health(Reference Zhu, Liu and He12,Reference Golzarand, Bahadoran and Mirmiran13) .
Previous experimental studies demonstrated that choline or betaine supplementation improved glucose intolerance and insulin resistance in murine models(Reference Hammoud, Pannia and Kubant14–Reference Szkudelska, Chan and Okulicz16). However, human studies on choline or betaine status and GDM were scanty, and results remained inconsistent. Previous prospective cohort studies reported that circulating betaine was inversely associated with GDM risk in Chinese women(Reference Huo, Li and Cao17,Reference Gong, Du and Li18) , whereas no significant difference in blood choline or betaine was observed between GDM and non-GDM Mexican women(Reference Fernandez-Osornio, Gomez-Diaz and Mondragon-Gonzalez19) or Canadian women(Reference Barzilay, Moon and Plumptre20). In addition, both intake and plasma levels of choline, but not betaine, were observed higher in the US pregnant women with GDM than those with non-GDM(Reference Kadam, Dalloul and Hausser21). Moreover, it is unclear whether higher intakes of these two nutrients during pregnancy are related to lower GDM risk. Therefore, the current study aims to investigate the correlation between maternal intake of choline or betaine during pregnancy and incident GDM among a hospital-based Chinese cohort.
Materials and methods
Study design and population
This hospital-based case–control study included pregnant women who were recruited at 24–28 gestational weeks when they attended a universal screening for GDM at the Seventh People’s Hospital affiliated to the Shanghai University of Chinese Medicine from December 2019 to May 2021. They were followed up through the postpartum period. This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the Medical Ethics Committee of the Shanghai Seventh People’s Hospital, Shanghai, China. Written informed consent was obtained from all study participants. This study was registered at www.chictr.org.cn (ChiCTR1900027764).
To estimate the sample size, we assumed that the type I error rate of α was 0·05, the power of test was 80 % (β = 20 %) and a case-to-control ratio was 1:2. Based on the previous similar study conducted among the pregnant women in Shanghai(Reference Zhu, Liu and He12), the proportion of choline intake in the lowest quintile in the control group was estimated as 24·8%, and considering the expected proportion of that in the case group equal to 37·8 % based on the expert opinion, the minimal required sample size was estimated as 159 in the case group and 318 in the control group. In addition, considering the response rate was 90 %, we assumed the required sample size was 530. Open Source Statistics for Public Health (OpenEpi) version 3.01 was adopted to determine the sample size(22).
Eligible women for this study were those (1) who were aged 18–50 years at enrolment; (2) who were at 24–28 gestational weeks when visiting for GDM screening; (3) who were able to read, understand and sign informed consent; (4) who were either a Shanghai native or had been a resident in Shanghai for at least 5 years; and (5) who were Han Chinese with non-smoking status. The recruitment was conducted during the daytime (from 07.00 to 17.00). When one GDM woman was recruited in the case group, two healthy pregnant women were selected in the control group for each GDM case based on age (±3 years) and gestational age when the blood sample was collected (±1 week).
We originally recruited 570 pregnant women during their clinical examination at 24–28 weeks of gestation. Participants were excluded if (1) they were diagnosed with type 1 or type 2 diabetes prior to the present pregnancy (n 5); (2) they were taking drugs that affect choline and betaine metabolism or glucose metabolism (n 7); (3) implausible energy intake (≤ 600 or ≥ 4500 kcal) (n 4); and (4) they had severe heart, brain, liver or kidney diseases, gestational thyroid disease, placenta previa, any infectious diseases, any autoimmune diseases or any history of chemotherapy (n 11). In total, 543 participants (168 GDM cases and 375 controls) were enrolled in the present study.
Ascertainment of gestational diabetes mellitus
After overnight fasting for ≥ 8 h, all enrolled pregnant women first had fasting blood drawn and then underwent a 75-g oral glucose tolerance test, which was followed by blood drawn at 1 h and 2 h post-oral glucose tolerance test. GDM was diagnosed according to the recommendation by the International Association of Diabetes and Pregnancy Study Groups if at least one of the following criteria was met: fasting plasma glucose ≥ 5·1 mmol/l, 1-h plasma glucose ≥ 10·0 mmol/l, or 2-h plasma glucose ≥ 8·5 mmol/l. The case group included women diagnosed with GDM, and the control group included healthy women with a complication-free pregnancy(Reference Metzger, Gabbe and Persson23).
Dietary assessment
Dietary intake data during pregnancy (24–28 weeks of gestation) were collected using a validated semi-quantitative FFQ, which included 120 most common food items in the Chinese diet(Reference Zhu, Liu and He12). The dietary assessment using this semi-quantitative FFQ was administered during face-to-face interviews at the same clinical examination (24–28 weeks of gestation) by the trained registered dietitians. Serving size of food items was reported by participants using a picture book, which displayed common Chinese food items with various kitchen utensils portion sizes. Then, the consumed serving size for each food item was converted to the amount (g) adopting the conversion coefficients for each item. Food intake frequencies were recorded in times per d, week, month or never. Supplement consumption was also recorded.
Participants’ common macro- and micro-nutrient consumption was computed via entering the amounts into Nutrition Star Nutrient Calculator Software (Shanghai Zhending Health Technology Co. Ltd). This software was developed according to the Chinese Food Composition Table(24). In addition, choline or betaine intake was calculated separately by a food composition database that was established according to the Chinese Food Composition Table(Reference Yang25) and the US Department of Agriculture Food Composition Databases(24,Reference Zeisel, Mar and Howe26) .
Assessment of covariates
Data about participants’ social demographics and lifestyles were collected in person by trained interviewers at enrolment using a questionnaire. Pre-pregnancy BMI (kg/m2) was calculated by dividing pre-gestational weight (kg) by the square of height (m). The BMI was classified based on the Chinese criteria for adults: underweight BMI (< 18·5 kg/m2), normal weight BMI (18·5–23·9 kg/m2), overweight BMI (24·0–27·9 kg/m2) and obese BMI (≥ 28·0 kg/m2)(27). Regarding physical activity, the International Physical Activity Questionnaire(Reference Wang, Wei and Zhang28,Reference Craig, Marshall and Sjöström29) was applied to estimate participants’ occupational activities and leisure time during pregnancy. The metabolic equivalent (min/week) was computed by using data on frequency and length for each activity. Different activity levels were determined based on the instruction of maternal physical activity assessment during pregnancy(Reference Wang, Wei and Zhang28–Reference Harizopoulou, Kritikos and Papanikolaou30). All the participants were followed till they gave birth. In addition, medical records were reviewed to obtain clinical information (e.g. health history, blood biochemical index examination, gravidity and parity history, delivery mode, pregnancy outcome and complications).
Fasting venous blood was collected with and without EDTA-coated tubes separately from participants during their visit for routine GDM screening. After centrifugation, fasting plasma glucose and serum lipid profile (e.g. TAG, total cholesterol, LDL-cholesterol and HDL-cholesterol) were measured by using a HITACHI 7600-DDP/7600–020 automatic biochemical analyser (Hitachi, Ltd) with the respective commercial assay kits (Biosino Bio-Technology and Science, Inc.). In addition, serum folate and vitamin B12 were measured by using the respective commercial assay kit (Abbott Ireland Diagnostic Division, Ireland) according to the manufacturer’s instructions. All the above measurements were conducted by professional staff in a standardised clinical laboratory at the Seventh People’s Hospital in Shanghai. The CV for measurement of TAG, total cholesterol, LDL-cholesterol, HDL-cholesterol, folate and vitamin B12 was 1·2 , 1·7 , 1·5 , 1·1 , 2·9 and 2·3 %, respectively.
Statistical analysis
Categorical variables between cases and controls were summarised as numbers (%) and differences between groups were analysed by the χ 2 test or Fisher’s exact test when cell numbers were less than five. Continuous variables were summarised as mean (s d) or median (IQR) and differences between groups were analysed by Student’s t test, Mann–Whitney U test or Kruskal–Wallis H-test as appropriate.
Dietary intake of choline or betaine was adjusted for total energy intake by using the residual method(Reference Willett, Howe and Kushi31). Energy-adjusted choline or betaine ingestion was categorised into quartiles based on distributions among controls with the lowest quartile as the reference group. Unconditional logistic regression models were adopted to examine the OR and corresponding 95 % CI of GDM by quartiles of choline or betaine intake. Potential confounders were selected and adjusted in the logistic regression model, which included age, pre-pregnancy BMI, parity, TAG, LDL-cholesterol/HDL-cholesterol and serum folate as appropriate. To test whether the correlations appear linear trend, we employed the median value for each quartile and treated it as a continuous variable in the unconditional logistic regression model.
Additionally, stratified analyses were conducted by using multivariable unconditional logistic regression models to determine whether the aforementioned correlations will be modified by age, pre-pregnancy BMI, parity, TAG and HDL-cholesterol to estimate the multiplicative interactions via including each interaction item. All statistical tests were conducted with R (version 4.1.2). P values presented were 2-tailed and P values less than 0·05 were considered as significant.
Results
Characteristics of the participants
Table 1 summarises the social-demographic and clinical characteristics of 543 participants (168 GDM cases and 375 controls). Compared with the healthy control subjects, women with GDM tended to be older, have higher pre-pregnancy BMI, have more frequencies of gravidity and parity, have higher fasting serum TAG and lower HDL-cholesterol and have higher fasting serum folate.
Table 1. Sociodemographic and clinical characteristics among cases and controls among Chinese women in Shanghai, China * (Numbers and percentages; median values and interquartile ranges; mean values and standard deviations)
GDM, gestational diabetes mellitus; MET, metabolic equivalent.
* Age at enrolment, height and physical activity level were assessed with Student’s t test. Total energy intake data were assessed with the Mann–Whitney U test. Pre-pregnancy BMI, occupation and education data were assessed with the χ 2 test of independence.
† Data are mean (sd) or n (%).
‡ Data are median (interquartile range).
Association between dietary intake of choline or betaine and gestational diabetes mellitus risk
As shown in Table 2, the energy-adjusted dietary choline consumption was not correlated with GDM risk, after adjusting for age, pre-pregnancy BMI, parity, TAG, LDL-cholesterol/HDL-cholesterol and serum folate. The multivariate OR of GDM comparing the pregnant women in the highest quartile of choline consumption to those in the lowest quartile was 0·77 (95 % CI 0·41, 1·43). Similarly, the energy-adjusted dietary betaine intake demonstrated a null association with incident GDM after controlling for the aforementioned confounders (OR for highest v. lowest quartile: 0·80; 95 % CI 0·42, 152) (Table 3).
Table 2. OR and 95 % CI of gestational diabetes according to quartiles of energy-adjusted daily choline intake *
Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile.
* Unconditional logistic regression was used to estimate OR and 95 % CI.
† Adjusted for age, pre-pregnancy BMI, parity, serum folate, TAG and LDL-cholesterol/HDL-cholesterol.
Table 3. OR and 95 % CI of gestational diabetes according to quartiles of energy-adjusted daily betaine intake *
Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile.
* Unconditional logistic regression was used to estimate OR and 95 % CI.
† Adjusted for age, pre-pregnancy BMI, parity, serum folate, TAG and LDL-cholesterol/HDL-cholesterol.
Moreover, stratified analyses showed that the null association between dietary consumption of choline or betaine and risk of GDM was not significantly different according to the subgroups of age (≥ 35 years, < 35 years), HDL-cholesterol (< 2·28, ≥ 2·28), TAG (≥ 3·10, < 3·10), parity (≥ 1, < 1) and pre-pregnancy BMI (> 28, 24–27·9, 18·5–23·9, < 18·5) (P for interaction = 0·005–0·946) (Figs. 1 and 2). However, there was a significant interaction between betaine intake and parity on the risk of GDM (P for interaction = 0·01) (Fig. 2). Among those with no parity history, the multivariate OR of GDM comparing the women in the highest quartile of betaine consumption with those in the lowest quartile was 0·25 (95 % CI 0·06, 0·81) after adjusting the covariates (Fig. 2).
Fig. 1. Interaction between the energy-adjusted daily choline intake and risk factors on gestational diabetes mellitus risk. Unconditional logistic regression was used to produce OR and 95 % CI. The model was adjusted for the following cofactors except for the stratification-related factors: age, pre-pregnancy BMI, parity, serum folate, TAG and LDL-cholesterol/HDL-cholesterol. Abbreviations: Q1, first quartile; Q4, fourth quartile.
Fig. 2. Interaction between the energy-adjusted daily betaine intake and risk factors on gestational diabetes mellitus risk. Unconditional logistic regression was used to produce OR and 95 % CI. The model was adjusted for the following cofactors except for the stratification-related factors: age, pre-pregnancy BMI, parity, serum folate, TAG and LDL-cholesterol/HDL-cholesterol. Abbreviations: Q1, first quartile; Q4, fourth quartile.
Discussion
To the best of our knowledge, this hospital-based case–control study is the first to explore the association of dietary choline or betaine intake with the risk of GDM among Chinese pregnant women. We found a null correlation between choline or betaine consumption during pregnancy and GDM risk among the overall participants. However, it is interesting to note that higher betaine intake was related to lower GDM risk among the women who had no parity history before the present pregnancy, and the correlation remained fairly consistent after adjusting multivariable confounders.
It was reported that type 2 diabetes mellitus (T2DM) and GDM shared many common risk factors and pathogenesis(Reference Vounzoulaki, Khunti and Abner7,Reference Zhang and Ning10,Reference Huo, Li and Cao17,Reference Gong, Du and Li18,Reference Greenberg, Jiang and Tinker32–Reference Garcia, Osté and Bennett38) . Although previous epidemiology studies investigating the association between intake of choline and risk of GDM were scanty, there were several studies examining the associations between dietary choline intake and incident diabetes. However, the results remain contradictory. In the prospective Women’s Health Initiative cohort, higher consumption of choline was significantly associated with augmented risk of diabetes among 46 263 postmenopausal women for 13·3 years of follow-up(Reference Greenberg, Jiang and Tinker32). Similarly, in the Atherosclerosis Risk in Communities study, choline ingestion was positively associated with T2DM risk among 7392 US middle-aged women over a median follow-up of 9 years(Reference Dibaba, Johnson and Kucharska-Newton33). A similar unfavourable relationship between choline consumption and diabetes was also observed in a cross-sectional study among 8621 US adult population aged 20 years or older(Reference Zhou, Li and Li34). However, a prospective study comprising 2332 males aged 42–60 years in eastern Finland during a mean 19·3-year follow-up reported a negative association between choline consumption and incident T2DM(Reference Virtanen, Tuomainen and Voutilainen35). Moreover, another prospective cohort study manifested that dietary choline intake was not associated with T2DM risk with a median 9-year follow-up among 13 440 US participants aged 57–61 years(Reference Dibaba, Johnson and Kucharska-Newton33). The discrepant results in the above investigations may be partly explained by different ethnicities, diverse frequency of dietary intake measurements and different follow-up periods. In the current study, we observed a null correlation between choline consumption during pregnancy and GDM risk, which warrants more studies to confirm it.
Previous epidemiology studies exploring the association between betaine status and incident diabetes were also limited, and results remained inconsistent. Several prospective cohort studies manifested that betaine consumption was not associated with incident diabetes among postmenopausal US women(Reference Greenberg, Jiang and Tinker32) and in the US middle-aged adults(Reference Dibaba, Johnson and Kucharska-Newton33). However, other prospective cohort studies reported an inverse association of blood betaine levels with the risk of T2DM among Chinese adults aged 40–75 years(Reference Lu, Huang and Li36), Norwegian adults aged 54–70 years(Reference Svingen, Schartum-Hansen and Pedersen37) and a Dutch cohort of predominantly middle-aged (52·6 (sd 11·5) years) Caucasians(Reference Garcia, Osté and Bennett38). Favourable associations of dietary betaine intakes or serum betaine levels with insulin resistance were also found in a cross-sectional study among a general Canadian population(Reference Gao, Wang and Sun39,Reference Gao, Randell and Tian40) . Moreover, a case–control study that was nested within a prospective cohort of Chinese pregnant women, including 243 GDM cases and 243 controls, manifested that higher serum betaine in early pregnancy was associated with lower GDM risk(Reference Huo, Li and Cao17). Likewise, another prospective cohort study reported an inverse association between blood betaine concentration in the second trimester and GDM risk(Reference Gong, Du and Li18). Although we observed the null correlation between dietary betaine consumption during pregnancy and GDM risk among all participants, differences by parity were present. We found that increasing betaine intake was correlated with a lower rate of GDM among the participants with no parity history prior to the present pregnancy. Reasons for these inconsistent findings of epidemiology studies could be attributed to a variety of dietary sources, diversity in participants’ ethnicities and different study designs as well as covariates adopted in the adjustment. Our findings warrant to be confirmed by utilising the biomarkers of dietary betaine intakes.
The specific mechanisms of the role of choline or betaine in the pathophysiology of diabetes are still unclear, although several proposed mechanisms were investigated. Previous studies manifested that both choline and betaine supplementation could enhance cellular insulin sensitivity and improve insulin resistance by suppressing oxidative stress and ameliorating inflammation(Reference Mehta, Singh and Arora41–Reference Kim, Kim and Lee44). In addition, betaine administration was evidenced to improve gluconeogenesis and glycogen synthesis by enhancing hepatic insulin receptor substrate 1 phosphorylation(Reference Kathirvel, Morgan and Nandgiri45). Betaine could also improve insulin resistance and maintain glucose homeostasis via elevating fibroblast growth factor 21 levels in blood and liver(Reference Ejaz, Martinez-Guino and Goldfine46) and suppressing the Forkhead box O1 that binds to thioredoxin-interacting proteins(Reference Kim, Kim and Lee44). Moreover, betaine could participate in the one-carbon unit metabolism via conversion from choline as well as dietary intake and donate the methyl group for modifying the methylation status of genetic loci signals related to insulin signalling(Reference Zhao, Yang and Hu47), which may affect diabetes development. Nevertheless, the above plausible mechanisms warrant further investigation in human studies.
The present study’s strengths include a comprehensive dietary assessment of choline and betaine intakes employing a validated semi-quantitative FFQ, detailed data collection on multiple risk factors as well as potential covariates and using stratified analyses across several GDM risk factors to examine the potentially differential influence of dietary choline and betaine consumptions.
There are also several limitations in the present study. First, because of the nature of the case–control study design, we were unable to infer whether low betaine intake causes incident GDM. Second, residual or unmeasured confounding (such as genetic factors, biomarkers of one-carbon metabolism profile) could not be completely ruled out, although we adjusted for multiple significant potential confounders. Third, although dietary assessment was conducted by a validated semi-quantitative FFQ, the potential recall bias may exist. Furthermore, the participants are Chinese pregnant women living in Shanghai. Owing to genetic variation encoding the key enzymes among diverse populations, various dietary patterns, different living environments and diverse food cultures, the generalisability of our findings may not be applied to other women.
Conclusions
Although dietary intake of choline during the first and second trimester of pregnancy was not significantly correlated with incident GDM among the Chinese pregnant women, the increasing betaine intake may be related to a lower incident GDM among the Chinese women with no prior parity history. For future research, we suggest conducting studies with a double-blinded, placebo-controlled, randomised clinical trial design to explore the causality and elucidate the underlying mechanisms.
Acknowledgements
The authors acknowledge all the participants and the personnel of this study.
This work was jointly funded by the university-level project of Shanghai University of Traditional Chinese Medicine (2019LK042), the National Natural Science Foundation of China (81501338), the 2021 Research Enhancement Program, the 2022 Multidisciplinary Internal Research Grant and THRC/CHERR Faculty Fellowship Funding at Texas State University. The funders had no role in the study design, implementation, analysis or decision to publish or reparation of the manuscript.
J.J., Y.H.L., K.F.Y. and J.Z. contributed to the conception and design of the work. K.L., L.X., J.J., Y.H.L., K.F.Y., H.W., L.L. and J.Z. contributed to the acquisition and collection of data. K.L., L.X., K.P.W., H.W., L.L. and X.X.S. contributed to the analysis and interpretation of data. K.L. and J.Z. contributed to the original draft preparation. K.P.W., Y.H.L., K.F.Y., C.M.J., J.J. and J.Z. revised the draft. All authors have read and agreed to the published version of the manuscript.
The authors have no financial or personal conflicts of interest to declare.
Open access
