Unsaturated plasma phospholipids are consistently lower in the patients diagnosed with gestational diabetes mellitus throughout pregnancy: A longitudinal metabolomics study of Chinese pregnant women part 1
Introduction
Over the last few decades, the increasing prevalence of hyperglycemia in pregnancy has become a major challenge in maternal foetal medicine [1]. Gestational diabetes mellitus (GDM) is hyperglycemia that is first detected in pregnancy [2]. It occurs as insulin resistance or hyperglycemia, which is reflected in high blood glucose measures [3]. For the majority of cases, the condition is asymptomatic and is only detected through routine screening [4]. Risk factors for the development of GDM include advanced maternal age, ethnicity, obesity, and family history of type 2 diabetes [5]. Cellular mechanisms of the condition are largely unknown. Elucidation of the pathogenesis of GDM would not only provide opportunities to improve therapeutic intervention for women diagnosed with GDM but may also provide deeper insight into the evolution of other glycaemic disorders [6].
It has long been suggested that hyperglycemia in GDM (along with other glycaemic disorders) is accompanied by alterations in fasting, postprandial, and integrated 24-h plasma concentrations of amino acids and lipids [7]. The progressive increase in reproductive hormones secreted by the foetal-placental unit (e.g., estrogen, progesterone, cortisol, Human placental lactogen, and placental insulinase), along with adipocytokines released by adipose tissue (e.g., leptin, adiponectin, tumour necrosis factor-α, interleukin-6, resistin, visfatin and apelin) in the maternal circulation have been shown to be associated with the insulin desensitization in normal pregnancies, and potentially pregnancies complicated by GDM [8]. Several novel protein markers of GDM have been also proposed (reviewed in [9]), including several apolipoproteins involved in lipid metabolism and transportation [10], and transthyretin–retinol binding protein–retinol complex associated with glucose metabolism and insulin resistance [11]. Notably, these changes in the serum proteome were observed as early as the first trimester of pregnancy.
Plasma and serum is the most commonly used biofluid for clinical chemistry diagnoses due to the fact that the blood profile contains a large number of various classes of important biomolecules. A number of metabolomics studies of GDM have investigated maternal blood using various analytical approaches, including LC-MS [12], [13], [14], GC-MS [15], [16], and NMR [17], [18]. However, results of these studies are not complementing one another (especially when comparing results across different analytical platforms). Thus, there has been no consensus among metabolomics researchers as to the pathogenesis or the pathophysiology of GDM. Furthermore, none of these studies have investigated the longitudinal changes of the maternal blood metabolome as pregnancy and GDM progress - notwithstanding longitudinal studies would be extremely powerful for studying the subtle changes of the metabolome induced by GDM. This might be because data analysis of a longitudinal data set presents an analytical challenge relative to a simple case-control study design.
Previous longitudinal metabolomic studies have relied on classical multivariate statistical methods, such as principal component analysis (PCA) and partial least-squares discriminant analysis (PLS-DA) [19], [20], [21]. Not only are these models unable to comprehend the complexity of a longitudinal data set, they do not take into account the intrinsic characteristics of the data set - that is repeated measurement of the same subjects in a number of time points. Statistical models distinguishing the ‘between’ and ‘within’ factors, or multilevel approaches have been developed for longitudinal and time-course studies. Methods include ANOVA-simultaneous component analysis (ASCA) [22], [23], dynamic probabilistic principal components analysis (DPPCA) [24], multilevel PCA (mPCA), and multilevel PLS-DA (mPLS-DA) [25], [26]. These methods allow visualisation of the time-related patterns. Furthermore, a range of solutions has been proposed to model the complex curves of longitudinal trajectories [27]. Methods include linear model, linear mixed models [24], [28], and empirical Bayesian statistics [29].
For metabolomic analysis of plasma or sera (especially by mass spectrometry), the chemical complexity of the blood presents another analytical challenge [30]. Blood plasma and sera contain high concentrations of protein that must be removed during extraction processes. An excessive amount of organic solvent is most commonly used to precipitate the proteins from plasma and serum. One problem associated with organic solvent extraction is that the phospholipid content of plasma and serum can be significantly greater than many other metabolites. High concentrations of phospholipids in the organic solvent-extracted sample disturbs subsequent LC-MS analysis, by masking the detection of metabolites present in lower concentrations (ion suppression) and degrading liquid chromatography columns (irreversible binding to reverse-phase stationary phase). Therefore, many early LC-MS-based metabolomic studies of plasma and serum, and even those recently reported, were conducted with compromised methodologies. A combination of protein removal and phospholipid separation is the most effective means to improve metabolite coverage and maintain a stable analytical system essential for the holistic analysis [31]. Novel blood-derived sample preparation methods, such as turbulent flow chromatography, solid-supported liquid-liquid extraction, and hybrid protein precipitation/phospholipid removal plates have been developed and made commercially available [32], [33], [34].
We report the results of our longitudinal study of the maternal plasma metabolome in normal pregnancies and pregnancies complicated by GDM. This study aimed to examine whether GDM progressed as pregnancy progressed, or whether it is a pre-existing condition that has manifested during pregnancy (i.e., changes of the plasma metabolome could be detected before the onset of GDM). This study used MS-based metabolomics to advance our knowledge of GDM. Both hydrophilic interaction liquid chromatography (HILIC) and reversed-phase liquid chromatography (RPLC) were employed. To enhance our capabilities of untargeted metabolomics analysis of maternal plasma by UPLC-MS, we employed a recently introduced protein precipitation/phospholipid separation technique. To take into account the repeated measurements on the same subject at different stages of pregnancy, we further employed innovative multilevel multivariate statistics [35], [36] and multivariate empirical Bayes analysis (MEBA) [37] to identify the subtle changes of the plasma metabolome induced by GDM.
Section snippets
Study design
This observational longitudinal study was approved by the ethics committee of the First Affiliated Hospital of Chongqing Medical University (University Hospital). Samples were collected from each participant after obtaining an informed consent.
Participant screening and entry criteria
Participants were chosen from pregnant women who attended their first prenatal visit at the outpatient unit of the obstetric department of the University Hospital, during their first 10–14 weeks of gestation, in 2013. Planned delivery at the University
Demographic and clinical characteristics of the longitudinal cohort
Of the 61 participants, 34 had normal glucose tolerance (controls), and 27 met the diagnostic criteria for GDM. Selected clinical characteristics of the participants are presented in Table 1. All participants were in their normal childbearing age, ranged from 19 to 35 y. The GDM group had a higher average age when compared to the control group, and significantly higher BMI in their first and second trimesters. A total of 17 GDM cases and 18 controls had a history of pregnancy, and 16 GDM cases
Discussion
The mechanisms leading to the development of GDM have not been fully elucidated but are probably related to an exacerbation of the β cell dysfunction in women genetically predisposed to β cell alterations [6], [50]. It has been recognised that GDM tends to be milder in women with a normal β cell response, and they are at relatively low risk for developing long-term diabetes [51]. Regarding the β cell dysfunction, several mechanisms could be involved in this process. Obesity is one of the known
Conclusion
The application of metabolomics in diabetes research has provided the scientific community with new insights into the pathogenesis of diabetes. Studies on GDM are increasingly employing metabolomics. In this study, we devised innovative solutions to enhance our ability to analyse maternal plasma and complex longitudinal data sets. Our results demonstrated that a number of polyunsaturated or chemically modified phospholipids were significantly lower in the plasma of women diagnosed with GDM
Acknowledgement
The authors thank Kim-Anh Lê Cao (mixOmics) and Jianguo Xia (MetaboAnalyst) for excellent implementation of the innovative statistical methods in their platform and answering our questions. The authors would also like to thank Xuyang Chen and Thibaut Delplancke for assisting sample preparation. The authors would also like to thank Jamie de Seymour for proofreading and comments.
References (61)
- et al.
Global estimates of the prevalence of hyperglycemia in pregnancy
Diabetes Res. Clin. Pract.
(2014) - et al.
Proteomic-driven biomarker discovery in gestational diabetes mellitus: a review
J. Proteome
(2015) - et al.
Metabolic fingerprint of gestational diabetes mellitus
J. Proteome
(2014) - et al.
Association between serum bile acid profiles and gestational diabetes mellitus: a targeted metabolomics study
Clin. Chim. Acta
(2016) - et al.
Recent developments in sample-pretreatment techniques for mass spectrometry-based metabolomics
TrAC Trends Anal. Chem.
(2014) - et al.
Mevalonate pathway: a review of clinical and therapeutical implications
Clin. Biochem.
(2007) - et al.
Maternal plasma phospholipid polyunsaturated fatty acids in pregnancy with and without gestational diabetes mellitus: relations with maternal factors
Am. J. Clin. Nutr.
(1999) - et al.
A prospective study of prepregnancy dietary fat intake and risk gestational diabetes
Am. J. Clin. Nutr.
(2012) Diagnosis and classification of diabetes mellitus
Diabetes Care
(2010)- et al.
Pathogenesis of gestational diabetes mellitus
Clinically useful estimates of insulin sensitivity during pregnancy: validation studies in women with normal glucose tolerance and gestational diabetes mellitus
Diabetes Care
Standards of Medical Care in Diabetes—2013
Diabetes Care
What is gestational diabetes?
Diabetes Care
Effects of gestational diabetes on diurnal profiles of plasma glucose, lipids, and individual amino acids
Diabetes Care
Gestational diabetes mellitus and macrosomia: a literature review
Ann. Nutr. Metab.
First-trimester multimarker prediction of gestational diabetes mellitus using targeted mass spectrometry
Diabetologia
First trimester concentrations of the TTR-RBP4-retinol complex components as early markers of insulin-treated gestational diabetes mellitus
Clin. Chem. Lab. Med.
Comprehensive analysis of serum metabolites in gestational diabetes mellitus by UPLC/Q-TOF-MS
Anal. Bioanal. Chem.
Metabolomics reveals broad-scale metabolic perturbations in hyperglycemic mothers during pregnancy
Diabetes Care
Maternal early pregnancy serum metabolites and risk of gestational diabetes mellitus
J. Clin. Endocrinol. Metab.
Prediction of gestational diabetes through NMR metabolomics of maternal blood
J. Proteome Res.
Metabolic biomarkers of prenatal disorders: an exploratory NMR metabonomics study of second trimester maternal urine and blood plasma
J. Proteome Res.
Metabolic changes in urine during and after pregnancy in a large, multiethnic population-based cohort study of gestational diabetes
PLoS One
Pregnancy-induced metabolic phenotype variations in maternal plasma
J. Proteome Res.
The longitudinal cerebrospinal fluid metabolomic profile of amyotrophic lateral sclerosis
Amyotroph. Lateral Scler. Frontotemporal Degener.
ANOVA-simultaneous component analysis (ASCA): a new tool for analyzing designed metabolomics data
Bioinformatics
Discovering gene expression patterns in time course microarray experiments by ANOVA–SCA
Bioinformatics
A dynamic probabilistic principal components model for the analysis of longitudinal metabolomics data
J. R. Stat. Soc.: Ser. C: Appl. Stat.
Multilevel data analysis of a crossover designed human nutritional intervention study
J. Proteome Res.
Multivariate paired data analysis: multilevel PLSDA versus OPLSDA
Metabolomics
Cited by (32)
In utero metabolomic signatures of refined grain intake and risk of gestational diabetes: A metabolome-wide association study
2023, American Journal of Clinical NutritionMetabolomics in gestational diabetes mellitus: A review
2023, Clinica Chimica ActaPlasma lipidomics in early pregnancy and risk of gestational diabetes mellitus: a prospective nested case-control study in Chinese women
2021, American Journal of Clinical NutritionF <inf>2</inf> -isoprostanes and fatty acids profile in early pregnancy complicated by pre-existing diabetes
2018, Prostaglandins Leukotrienes and Essential Fatty Acids