Screening and application of early pregnancy urine metabolism marker for gestational diabetes mellitus

Abstract

The invention relates to screening and application of early pregnancy urine metabolism markers of gestational diabetes mellitus, and belongs to the field of metabonomics and human health research. On the basis of prospective research design, eight early pregnancy urine metabolites significantly associated with the gestational diabetes mellitus occurrence risk are screened out, and valine, 3-hydroxyisobutyric acid and creatine levels are positively associated with the gestational diabetes mellitus occurrence risk, and valine, 3-hydroxyisobutyric acid and creatine levels are positively associated with the gestational diabetes mellitus occurrence risk. The levels of pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indol sulfate and p-hydroxyhippuric acid are in negative correlation with the occurrence risk of gestational diabetes mellitus; and the eight metabolites can significantly improve the prediction efficiency of a traditional early prediction model of gestational diabetes mellitus. The urine metabolism marker can be used as a detection target for early prediction of gestational diabetes mellitus, or can be used as a target for research and development of related drugs for prevention and treatment of gestational diabetes mellitus.

Description

Technical Field

[0001] The present invention belongs to the field of metabolomics and human health research, and more specifically, relates to a series of early pregnancy urine metabolic markers associated with the occurrence of gestational diabetes, and especially to the screening and application of early pregnancy urine metabolic markers for gestational diabetes. Background Art

[0002] Gestational diabetes mellitus (GDM) is one of the most common complications during pregnancy, affecting the short-term and long-term health of pregnant women and their offspring (Johns, Denison et al. 2018). According to statistics, the standardized prevalence of gestational diabetes worldwide in 2021 was approximately 14.0% (Wang, Li et al. 2022). A systematic review published in 2019 showed that the overall prevalence of gestational diabetes in my country was as high as 14.8% (Gao, Sun et al. 2019). Although early prevention and detection of gestational diabetes are crucial to reducing its disease burden, its etiology and pathogenesis are still unclear, and research on early predictive factors is insufficient.

[0003] The development of gestational diabetes is accompanied by metabolic abnormalities such as pancreatic β-cell dysfunction and increased insulin resistance (McIntyre, Catalano et al. 2019), which is bound to affect its downstream metabolic processes. Metabolomics is a powerful tool for studying the metabolic response of the body to pathophysiological stimuli (Nicholson, Lindon et al. 1999) and has been applied to the study of gestational diabetes (Xie, Li et al. 2023). However, most of the current metabolomics studies of gestational diabetes are based on blood samples (Xie et al. 2023), and the collection of blood samples is somewhat invasive and requires the participation of professional samplers. In contrast, the collection of urine samples is non-invasive, convenient, and can be performed by the subjects themselves. Although urine metabolomics has been used in gestational diabetes research, most studies use urine samples collected at the time of disease diagnosis or after diagnosis (Mao, Chen et al. 2017), which limits its application in obtaining metabolic information related to the risk of gestational diabetes and early diagnosis. Although a few studies have reported changes in the urine metabolome of patients with gestational diabetes in early pregnancy (Law, Han et al. 2017; Zhao, Zheng et al. 2022), the sample size was small (less than 65 cases), and the association and potential mechanism between the urine metabolome in early pregnancy and the risk of gestational diabetes were not evaluated. Although metabolomics has also been used in early prediction studies of gestational diabetes, most of them are based on blood samples and have not been compared with traditional prediction models (Mennickent, Rodriguez et al. 2022). Conducting a urine metabolomics study of gestational diabetes in early pregnancy in a large population of pregnant women in my country can provide high-quality local research evidence for early non-invasive prediction, prevention and intervention studies of gestational diabetes, and provide a reference for its pathogenesis research. Summary of the invention

[0004] The purpose of the present invention is to screen urine metabolic markers in early pregnancy for gestational diabetes, and finally screen out 8 urine metabolites that are significantly associated with gestational diabetes, and establish a prediction model for gestational diabetes based on this. The present invention can provide guidance for early prevention of gestational diabetes, risk prediction, early identification of high-risk pregnant women, early intervention and drug treatment. In addition, the present invention can also provide a reference for the development of new drugs or nutritional supplements related to the pathogenesis of gestational diabetes, prevention and treatment of gestational diabetes.

[0005] According to a first aspect of the present invention, there is provided a use of early pregnancy urine metabolic markers for gestational diabetes in the preparation of a kit; the kit is used to predict the risk of gestational diabetes in mid-pregnancy in pregnant women in early pregnancy;

[0006] The metabolic marker is at least one of valine, 3-hydroxyisobutyric acid, creatine, pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid;

[0007] The concentrations of valine, 3-hydroxyisobutyric acid and creatine in the urine of the gestational diabetes patients in early pregnancy were higher than those of the non-patient group;

[0008] The concentrations of pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid in the early pregnancy urine of the gestational diabetes mellitus group were lower than those of the non-patient group.

[0009] Preferably, the valine, 3-hydroxyisobutyric acid and creatine are each independently positively correlated with the risk of gestational diabetes.

[0010] Preferably, the pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid are each independently negatively correlated with the risk of gestational diabetes.

[0011] According to another aspect of the present invention, there is provided the use of early pregnancy urine metabolic markers of gestational diabetes in the preparation of drugs related to the prevention and / or treatment of gestational diabetes;

[0012] The metabolic marker is at least one of valine, 3-hydroxyisobutyric acid, creatine, pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid;

[0013] The concentrations of valine, 3-hydroxyisobutyric acid and creatine in the urine of the gestational diabetes patients in early pregnancy were higher than those of the non-patient group;

[0014] The concentrations of pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid in the early pregnancy urine of the gestational diabetes mellitus group were lower than those of the non-patient group.

[0015] Preferably, the drug is an amino acid metabolism drug, a nucleotide metabolism drug or an intestinal flora metabolism drug.

[0016] According to another aspect of the present invention, a system for screening urine metabolic markers in early pregnancy for gestational diabetes is provided, comprising:

[0017] Metabolome detection module: The metabolome detection module is used for the collection and processing of urine metabolite spectra, as well as the identification and quantification of metabolites;

[0018] Metabolite concentration pre-processing module: The metabolite concentration pre-processing module is used to normalize and standardize the concentration of each metabolite;

[0019] Differential metabolite screening and analysis module: The differential metabolite screening and analysis module is used to screen through univariate conditional logistic regression, multivariate conditional logistic regression and false discovery rate correction methods to obtain metabolites with different concentrations between the gestational diabetes group and the non-gestational diabetes group and metabolites that are significantly associated with the risk of gestational diabetes.

[0020] Preferably, it also includes:

[0021] Gestational diabetes risk prediction module: The gestational diabetes risk prediction module is used to construct a new gestational diabetes prediction model by adding early pregnancy urine metabolites related to the risk of gestational diabetes on the basis of traditional risk factors for gestational diabetes. The predictive efficacy of the established prediction model is evaluated by the area under the curve of the receiver operating characteristic curve, the net reclassification improvement index and the comprehensive discrimination improvement index.

[0022] In general, the above technical solution conceived by the present invention has the following technical advantages compared with the prior art:

[0023] (1) The present invention is based on a large birth cohort in China and conducted a prospective nested case-control study. By performing metabolomics analysis on the early pregnancy urine of pregnant women with gestational diabetes and pregnant women without gestational diabetes, 8 urine metabolite markers were screened out, a corresponding prediction model was established, and the prediction ability of the model was comprehensively evaluated. It was found that these urine metabolite markers can significantly improve the predictive efficiency of the traditional early prediction model for gestational diabetes. The present invention screened out a series of early pregnancy urine metabolite markers that are significantly associated with the risk of gestational diabetes. The study included 395 cases and 790 controls, and used age, gestational age and sampling time for 1:2 matching. The sample size was sufficient and the statistical power was high. After statistical analysis methods such as univariate conditional logistic regression, multivariate conditional logistic regression and multiple correction, a series of early pregnancy urine metabolite markers were screened out; after incorporating these urine metabolite markers, the predictive efficiency of the traditional prediction model for gestational diabetes was significantly improved. The screening steps of the present invention are more rigorous, the screening criteria are more stringent, and the results of the obtained urine metabolite markers are more robust. The urine metabolic markers screened out by the present invention can be used as detection targets for the preparation of kits to predict the risk of gestational diabetes or identify pregnant women at high risk of gestational diabetes in early pregnancy; they can also be used as targets for the development of drugs and nutritional supplements related to the prevention and treatment of gestational diabetes.

[0024] (2) The present invention uses nuclear magnetic resonance spectroscopy technology to detect the urine metabolome. This detection method has simple sample preparation, non-invasive detection, high reproducibility, and can accurately quantify metabolites in urine without bias.

[0025] (3) Urine is one of the most commonly used biological samples in metabolomics research. It carries rich information about endogenous and exogenous metabolites and can reflect the metabolic changes of the body under different pathophysiological states. Urine is also one of the most common and easily obtained biological samples in clinical practice. Compared with blood samples: the collection of urine samples is non-invasive, has higher subject compliance, has lower professional requirements for sampling personnel, and is easier to implement dynamic collection. In addition, since urine routine is a must-check item for pregnant women during early pregnancy check-ups, the present invention will not increase the number of examinations and burden on pregnant women.

[0026] (4) The gestational diabetes metabolic markers based on the early pregnancy urine metabolome proposed in the present invention can significantly improve the predictive ability of traditional predictive factors and indicate the metabolic pathways closely related to the occurrence of gestational diabetes. Therefore, this technology has high value for early prevention, early prediction, identification of high-risk pregnant women, selection of therapeutic targets, early intervention and evaluation of the effects of intervention measures of gestational diabetes. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 The epidemiological design, detection method and statistical method derived from the present invention are used to screen urine metabolic markers that are significantly associated with gestational diabetes, and based on this, a prediction model for early pregnancy diseases is established.

[0028] Figure 2 The urine sample of the present invention 1 H NMR spectrum, which shows the metabolite spectrum of a representative pregnant woman's early pregnancy urine sample, including metabolites in multiple metabolic pathways.

[0029] Figure 3 Figure 2 shows the concentration distribution of 8 urinary metabolic markers between the gestational diabetes group and the non-gestational diabetes group.

[0030] Figure 4 The 8 urinary metabolite markers were grouped by quartiles, and the risk odds ratios of gestational diabetes in different quartiles were calculated (with the first quartile as the reference). Among them, 3 metabolites were positively associated with the risk of gestational diabetes, and 5 metabolites were negatively associated with the risk of gestational diabetes, and the trend test of each metabolite was statistically significant (P<0.05).

[0031] Figure 5Receiver operating characteristic curve analysis was used to evaluate the improvement of urine metabolite markers on the prediction efficiency of traditional risk factors for gestational diabetes. The prediction efficiency of the prediction model based on traditional risk factors for gestational diabetes was 0.648 (95% confidence interval: 0.614, 0.682). After adding 8 urine metabolite markers, the prediction efficiency was significantly improved to 0.692 (95% confidence interval: 0.660, 0.724). DETAILED DESCRIPTION

[0032] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in each embodiment of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0033] The present invention uses nuclear magnetic resonance spectroscopy to identify a total of 51 urine metabolites as candidate metabolic markers for early pregnancy in gestational diabetes, including: amino acids and their derivatives, glycolysis and tricarboxylic acid cycle-related metabolites, purine and pyrimidine metabolism-related metabolites, niacin and nicotinamide metabolism-related metabolites, intestinal flora metabolism-related metabolites, etc. After removing metabolites with a detection limit rate of more than 20% and overlapping characteristic peaks, 39 metabolites were finally included in the analysis. Univariate conditional logistic regression analysis showed that the contents of 20 metabolites were significantly different between the gestational diabetes group and the non-gestational diabetes group, of which 10 metabolites were upregulated in the gestational diabetes group and 10 metabolites were downregulated in the gestational diabetes group; after correcting for multiple comparisons, the contents of 16 metabolites were significantly different between the gestational diabetes group and the non-gestational diabetes group, of which 8 metabolites were upregulated in the gestational diabetes group and 8 metabolites were downregulated in the gestational diabetes group.

[0034] The present invention further uses multivariate conditional logistic regression to discover 8 metabolic markers of gestational diabetes based on the early pregnancy urine metabolome. Among them, valine, 3-hydroxyisobutyric acid and creatine levels are positively associated with the risk of gestational diabetes. Compared with the lowest quartile group, the multivariate-adjusted odds ratios (i.e., relative risks) of gestational diabetes in the highest quartile group are 2.00, 1.83, and 2.37, respectively; pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid levels are negatively associated with the risk of gestational diabetes. Compared with the lowest quartile group, the multivariate-adjusted odds ratios (OR; i.e., relative risks) of gestational diabetes in the highest quartile group are 0.60, 0.60, 0.53, 0.63, and 0.61, respectively.

[0035] The present invention also applies the metabolic markers of gestational diabetes based on the early pregnancy urine metabolome to the optimization and improvement of the traditional prediction model; the traditional early pregnancy prediction model for gestational diabetes is jointly established by age, parity, gestational age, parental diabetes history, gestational diabetes history, pre-pregnancy body mass index and baseline fasting blood glucose, and the area under the curve of the model is 0.648; the newly identified 8 urinary metabolic markers of gestational diabetes are further incorporated into the model as predictive factors, which can significantly improve the area under the curve of the model to 0.692, and obtain significant net reclassification improvement index (NRI, netreclassification improvement) and integrated discrimination improvement index (IDI, integrated discrimination improvement).

[0036] The gestational diabetes mellitus metabolic markers based on the early pregnancy urine metabolome of the present invention can be used as detection targets for preparing a kit for predicting the risk of gestational diabetes mellitus in early pregnancy.

[0037] The gestational diabetes metabolic markers based on the early pregnancy urine metabolome in the present invention can be used as targets in the research and development of drugs related to the prevention and treatment of gestational diabetes; preferably, the drugs involve metabolic pathways such as amino acid metabolism, nucleotide metabolism, and intestinal flora metabolism.

[0038] The gestational diabetes prediction model based on the early pregnancy urine metabolome in the present invention is used to identify pregnant women at high risk of gestational diabetes in early pregnancy, and to predict the risk of gestational diabetes in pregnant women in mid-pregnancy by integrating multiple predictive factors (demographic characteristics, lifestyle, and commonly used clinical indicators, etc.); preferably, the model can be presented in the form of software.

[0039] The screening system for gestational diabetes metabolic markers based on early pregnancy urine metabolome of the present invention comprises:

[0040] Sample pre-processing module: The sample pre-processing module is used for the preparation of urine samples;

[0041] Metabolome detection module: The metabolome detection module is used for the collection and processing of urine metabolite spectra, as well as the identification and quantification of metabolites;

[0042] Metabolite concentration pre-processing module: The metabolite concentration pre-processing module is used to normalize and standardize the concentration of each metabolite;

[0043] Differential metabolite screening and analysis module: The differential metabolite screening and analysis module is used to screen through univariate conditional logistic regression, multivariate conditional logistic regression and false discovery rate (FDR) correction methods to obtain metabolites with different concentrations between the gestational diabetes group and the non-gestational diabetes group and metabolites that are significantly associated with the risk of gestational diabetes;

[0044] Gestational diabetes risk prediction module: The gestational diabetes risk prediction module is used to construct a new gestational diabetes prediction model by adding early pregnancy urine metabolites related to the risk of gestational diabetes on the basis of traditional risk factors for gestational diabetes. The predictive efficacy of the established prediction model is evaluated using parameters such as the area under the curve of the receiver operating characteristic curve, the net reclassification improvement index and the comprehensive discrimination improvement index.

[0045] The present invention includes early pregnancy metabolic markers for gestational diabetes based on urine metabolome. The initial screening process of urine metabolic markers is carried out in 51 urine metabolites, mainly covering amino acids and their derivatives, glycolysis and tricarboxylic acid cycle related metabolites, purine and pyrimidine metabolism related metabolites, niacin and nicotinamide metabolism related metabolites, intestinal flora metabolism related metabolites and other categories.

[0046] By sequentially performing univariate conditional logistic regression, multivariate conditional logistic regression, and false discovery rate correction, 16 urinary metabolites with significant concentration differences between the gestational diabetes group and the non-gestational diabetes group, as well as 8 urinary metabolites significantly associated with the risk of gestational diabetes were screened out. Among the 8 urinary metabolites significantly associated with the risk of gestational diabetes, valine, 3-hydroxyisobutyric acid, and creatine were positively associated with the risk of gestational diabetes, while pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate, and p-hydroxyhippuric acid were negatively associated with the risk of gestational diabetes. After the concentrations of the above eight urinary metabolites were grouped according to quartiles, the multivariate-adjusted risk odds ratios (relative risks) and their 95% confidence intervals for gestational diabetes in the highest quartile compared with the lowest quartile were: valine 2.00 (1.32, 3.03), 3-hydroxyisobutyric acid 1.83 (1.22, 2.75), creatine 2.37 (1.57, 3.59), pseudouridine 0.60 (0.41, 0.89), dimethylamine 0.60 (0.40, 0.90), 4-hydroxyphenylacetic acid 0.53 (0.36, 0.78), indoxyl sulfate 0.63 (0.43, 0.92) and p-hydroxyhippuric acid 0.61 (0.41, 0.89).

[0047] The present invention is based on a large birth cohort in China and conducted a prospective nested case-control study. Urine samples were collected from pregnant women when they were included in the cohort in early pregnancy, and a standard 75-gram oral glucose tolerance test (OGTT) was performed in the second trimester. Pregnant women diagnosed with gestational diabetes according to the recommended criteria of the International Diabetes and Pregnancy Study Group (IADPSG) were divided into the case group; and non-gestational diabetic pregnant women were selected according to age (±3 years), sampling gestational age (±4 weeks), and sampling time (±4 weeks), and matched with cases at a ratio of 1:2 to obtain a control group. After the urine samples were prepared, nuclear magnetic resonance spectroscopy was used to detect the urine metabolome and quantify the concentration of urine metabolites. The association between urine metabolite concentration and the risk of gestational diabetes was analyzed using statistical methods such as multivariate conditional logistic regression. The research protocol involved in the present invention was approved by the Ethics Committee of the Medical College, and all pregnant women who participated in the cohort and provided samples signed informed consent.

[0048] The specific steps of this study are:

[0049] (1) Urine sample collection and storage: After guidance from the staff, pregnant women use sterile urine cups to collect midstream urine, mix well, and then divide the urine into smaller packages and store them in a -40°C refrigerator for testing.

[0050] (2) Preparation of urine samples: The urine sample was mixed with heavy water and potassium fluoride solution in sequence, and after centrifugation, the supernatant was transferred to an NMR tube, and then deuterated ethylenediaminetetraacetic acid solution was added and mixed, and then phosphate buffer was added and mixed before testing.

[0051] (3) NMR spectrum acquisition: Place the NMR tube in the NMR spectrometer, perform tuning, field locking, and field shimming, select the standard pulse sequence, set the sampling parameters, and acquire the NMR spectrum.

[0052] (4) NMR spectrum processing: Use professional software to perform phase correction, baseline correction, chemical shift calibration, and chemical shift offset correction on the collected NMR spectrum, and then integrate the metabolite characteristic signals.

[0053] (5) Screening of urine metabolic markers, establishment and evaluation of prediction models: Using common statistical methods in epidemiology, we screened urine metabolic markers in early pregnancy that were significantly associated with the occurrence of gestational diabetes, and established and evaluated a prediction model for gestational diabetes.

[0054] The analysis process of this study is as follows Figure 1 shown.

[0055] The specific steps of this study will be described in detail below in conjunction with the embodiments.

[0056] Example 1: Urine sample preparation

[0057] The stored urine sample was taken out, thawed and mixed at room temperature, 300 μL of urine was taken, 200 μL of heavy water and 5M potassium fluoride solution (final concentration 74.3 mM) were added in sequence, and the mixture was allowed to stand for 10 minutes after mixing; after centrifugation (14 500×g, 4°C, 10 minutes), 450 μL of supernatant was taken into a 5 mm NMR tube; then 0.12 M deuterated ethylenediaminetetraacetic acid solution (final concentration 1.2 mM) was added and mixed thoroughly; finally, 45 μL of phosphate buffer (1.5 M, pH 7.40, dipotassium hydrogen phosphate / sodium dihydrogen phosphate = 4 / 1, containing 0.05% sodium 3-trimethylsilyldeuterated propionate and 0.01% sodium azide, prepared in heavy water) was added to the NMR tube, and the mixture was mixed thoroughly before testing.

[0058] Example 2: Nuclear Magnetic Resonance Spectrum Collection

[0059] The data were acquired using a Bruker AVANCE III HD 600 MHz NMR spectrometer ( 1 H resonance frequency was 600.25 MHz, BBO probe, Bruker Biospin, Germany), and all samples were subjected to one-dimensional 1 H NMR spectrum acquisition. The sequence used was NoesyGppr1d (recycle delay-G1-90°-T1-90°-T m -G2-90°-acquisition), where the cycle delay (2s) and mixing time (t m , 100ms) for pre-saturation water compression. The main parameters of the sequence are: 96 cumulative times, 32768 sampling points, 12019Hz spectrum width, 90° pulse width of about 14μs, t1 4μs, and constant receiver gain. Select representative samples and collect a series of two-dimensional nuclear magnetic resonance spectra for metabolite identification. A total of 51 urine metabolites were identified in this study, such as Figure 2 shown.

[0060] Example 3: Nuclear magnetic resonance spectrum processing

[0061] The acquired one-dimensional 1 The H NMR spectra were Fourier transformed with a 0.3 Hz broadening factor and 131,072 data points filled with zeros, and then phase and baseline corrections were performed. Chemical shift calibration (δ0.00) was performed with the internal standard sodium 3-trimethylsilyldeuterated propionate, followed by chemical shift offset correction. The characteristic peaks of metabolites were integrated using AMIX 3.9.15 software (Bruker Biospin, Germany).

[0062] Metabolites with a rate below the detection limit of less than 20% and no overlap with other signals were selected for subsequent analysis. Among the 51 metabolites identified in this study, 4 metabolites had a rate below the detection limit of more than 20%, and the characteristic signals of 8 metabolites overlapped with other signals. Finally, 39 metabolites were included in the subsequent analysis, among which the integrals of metabolites below the detection limit were filled with half of the minimum integral value of the metabolite in all samples. The integrals of the metabolites after the above processing were then subjected to probabilistic quotient normalization (PQN), and the normalized integral value represented the content of the metabolite and was used for subsequent statistical analysis.

[0063] Example 4: Screening of Urine Metabolism Markers

[0064] Based on Stata / SE 15.1 software (StataCorp LLC, College Station, Texas) and R4.2.1 software (The R Foundation for Statistical Computing), common statistical methods in epidemiology were used to screen urine metabolic markers in early pregnancy that were significantly associated with the occurrence of gestational diabetes, and to establish and evaluate a prediction model for gestational diabetes. The analysis process is detailed as follows:

[0065] (1) Normalization and standardization of metabolite content

[0066] Since metabolite concentrations generally show a right-skewed distribution, they need to be natural logarithm transformed and Z-score standardized before data analysis.

[0067] (2) Screening of urine metabolic markers

[0068] Univariate conditional logistic regression and multivariate conditional logistic regression methods were used, combined with Benjamini–Hochberg false positive rate correction, to screen urine metabolite markers. Specifically, univariate conditional logistic regression analysis showed that the levels of 20 metabolites were significantly different between the gestational diabetes group and the non-gestational diabetes group, of which 10 metabolites were upregulated in the gestational diabetes group and 10 metabolites were downregulated in the gestational diabetes group (P < 0.05); after correction for multiple comparisons, the levels of 16 metabolites were significantly different between the gestational diabetes group and the non-gestational diabetes group, of which 8 metabolites were upregulated in the gestational diabetes group and 8 metabolites were downregulated in the gestational diabetes group (FDR-corrected P < 0.05). In addition, multivariate conditional logistic regression analysis showed that the levels of 16 metabolites were significantly associated with the risk of gestational diabetes, of which 6 metabolites were positively associated with the risk of gestational diabetes, and 10 metabolites were negatively associated with the risk of gestational diabetes (P 趋势 <0.05); after correction for multiple comparisons, the contents of 8 metabolites were significantly associated with the risk of gestational diabetes, among which valine, 3-hydroxyisobutyric acid, and creatine were positively associated with the risk of gestational diabetes, while pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate, and p-hydroxyhippuric acid were negatively associated with the risk of gestational diabetes (FDR-corrected P 趋势 <0.05), these 8 metabolites are urine metabolic markers, and their concentration distribution is as follows Figure 3 The concentrations of urine metabolite markers were grouped by quartiles. Compared with the lowest quartile group, the multivariate-adjusted odds ratios (relative risks) of gestational diabetes mellitus in the highest quartile group and their 95% confidence intervals were: valine 2.00 (1.32, 3.03), 3-hydroxyisobutyric acid 1.83 (1.22, 2.75), creatine 2.37 (1.57, 3.59), pseudouridine 0.60 (0.41, 0.89), dimethylamine 0.60 (0.40, 0.90), 4-hydroxyphenylacetic acid 0.53 (0.36, 0.78), indoxyl sulfate 0.63 (0.43, 0.92) and p-hydroxyhippuric acid 0.61 (0.41, 0.89) ( Figure 4 The covariates adjusted for in the multivariate conditional logistic regression model included age, gestational age, parity, history of gestational diabetes, parental history of diabetes, education level, smoking status, drinking status, physical activity level, pre-pregnancy body mass index, systolic blood pressure, fasting blood glucose, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides.

[0069] Example 5: Establishment and evaluation of prediction model

[0070] In the present invention, the traditional early pregnancy prediction model for gestational diabetes is composed of age, parity, gestational age, parental diabetes history, gestational diabetes history, pre-pregnancy body mass index and baseline fasting blood sugar. The present invention uses the receiver operating characteristic curve (ROC) to evaluate the predictive efficiency of the prediction model. The abscissa of the receiver operating characteristic curve represents "1-specificity" (i.e., false positive rate), and the ordinate represents "sensitivity" (i.e., true positive rate); the area under the curve (AUC) represents the accuracy of the prediction, ranging from 0 to 1, and the larger the area under the curve, the higher the accuracy of the prediction. When evaluating the predictive ability of the binary logistic regression model, the value of C-statistic is equal to the area under the curve. The statistical differences in the area under the receiver operating characteristic curve between different prediction models were compared by the DeLong test. In addition, the net reclassification improvement index (NRI) and the integrated discrimination improvement index (IDI) were used to further evaluate the differences in the predictive ability of different prediction models, wherein NRI includes two indicators, continuous NRI and classification NRI.

[0071] In the present invention, the C-statistic (95% confidence interval) of the traditional early pregnancy prediction model for gestational diabetes is 0.648 (0.614, 0.682). After adding the above 8 urine metabolic markers as predictive factors to the traditional prediction model, its C-statistic (95% confidence interval) can be increased to 0.692 (0.660, 0.724) ( Figure 5 ); DeLong test showed that the improvement of the C-statistic was significant (P < 0.001). In addition, the improvement of the C-statistic corresponded to a continuous net reclassification improvement index of 36.9%, a classification net reclassification improvement index of 5.2%, and an absolute comprehensive discrimination improvement index of 3.3% (all P < 0.001). This shows that the urine metabolic markers in the present invention can significantly improve the predictive efficiency of the traditional prediction model.

[0072] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Use of early pregnancy urine metabolic markers for gestational diabetes in the preparation of a kit; the kit is used to predict the risk of gestational diabetes in pregnant women in the second trimester during early pregnancy; The metabolic marker is at least one of valine, 3-hydroxyisobutyric acid, creatine, pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid; The concentrations of valine, 3-hydroxyisobutyric acid and creatine in the urine of the gestational diabetes patients in early pregnancy were higher than those of the non-patient group; The concentrations of pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid in the early pregnancy urine of the gestational diabetes mellitus group were lower than those of the non-patient group.

2. The use according to claim 1, characterized in that The valine, 3-hydroxyisobutyrate and creatine are each independently positively associated with the risk of gestational diabetes.

3. The use according to claim 1, characterized in that The pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid are each independently and negatively associated with the risk of gestational diabetes.

4. Application of early pregnancy urine metabolic markers of gestational diabetes mellitus in the preparation of drugs related to the prevention and / or treatment of gestational diabetes mellitus; The metabolic marker is at least one of valine, 3-hydroxyisobutyric acid, creatine, pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid; The concentrations of valine, 3-hydroxyisobutyric acid and creatine in the urine of the gestational diabetes patients in early pregnancy were higher than those of the non-patient group; The concentrations of pseudouridine, dimethylamine, 4-hydroxyphenylacetic acid, indoxyl sulfate and p-hydroxyhippuric acid in the early pregnancy urine of the gestational diabetes mellitus group were lower than those of the non-patient group.

5. The use according to claim 4, characterized in that The drug is an amino acid metabolism drug, a nucleotide metabolism drug or an intestinal flora metabolism drug.

6. A screening system for early pregnancy urine metabolic markers for gestational diabetes, characterized in that: include: Metabolome detection module: The metabolome detection module is used for the collection and processing of urine metabolite spectra, as well as the identification and quantification of metabolites; Metabolite concentration pre-processing module: The metabolite concentration pre-processing module is used to normalize and standardize the concentration of each metabolite; Differential metabolite screening and analysis module: The differential metabolite screening and analysis module is used to screen through univariate conditional logistic regression, multivariate conditional logistic regression and false discovery rate correction methods to obtain metabolites with different concentrations between the gestational diabetes group and the non-gestational diabetes group and metabolites that are significantly associated with the risk of gestational diabetes.

7. The screening system according to claim 6, characterized in that Also includes: Gestational diabetes risk prediction module: The gestational diabetes risk prediction module is used to construct a new gestational diabetes prediction model by adding early pregnancy urine metabolites related to the risk of gestational diabetes on the basis of traditional risk factors for gestational diabetes. The predictive efficacy of the established prediction model is evaluated by the area under the curve of the receiver operating characteristic curve, the net reclassification improvement index and the comprehensive discrimination improvement index.

Images