Type 2 diabetes mellitus patient metabolic liver fat fraction model and construction method thereof

Abstract

The invention discloses a metabolic liver fat fraction model for patients with type 2 diabetes and a construction method of the metabolic liver fat fraction model, and belongs to the technical field of biological medicines. The metabolic liver fat fraction is equal to MRI-measure level fat + (the value of the level fat-level fat is correspondingly obtained on the optimal linear fitting line of the level fat and the MRI-measure level fat), and the value of the level fat-level fat is correspondingly obtained on the optimal linear fitting line of the level fat and the MRI-measure level fat; the construction method comprises the steps that the metabolite biomarker level used for predicting the metabolic liver fat fraction in baseline plasma of a subject and the liver fat content detected through magnetic resonance are detected, pliver fat is constructed through elastic network linear regression, and then the metabolic liver fat fraction is obtained through the liver fat content detected through magnetic resonance. The cardiovascular metabolism risk stratification of the type 2 diabetes mellitus patient can be realized, and the application of precision medicine in the prognosis of the type 2 diabetes mellitus disease is realized.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology, specifically relating to a metabolic liver fat fraction model for patients with type 2 diabetes and its construction method. Background Technology

[0002] Hepatic steatosis plays a significant role in type 2 diabetes (T2D) because it is closely associated with metabolic disorders and may be used to predict disease prognosis and response to lifestyle interventions. However, studies have shown that hepatic steatosis exhibits significant metabolic heterogeneity, and relying solely on imaging measurements of liver fat content often fails to fully capture this characteristic. Previous studies have estimated the percentage of liver fat using serum lipid profiles composed of three lipids (lipid triads), finding a good correlation between these profiles and liver fat content measured by proton magnetic resonance spectroscopy or liver biopsy (Orešič M, Hyötyläinen T, Kotronen A, et al. Prediction of non-alcoholic fatty-liver disease and liver fat content by serum molecular lipids. Diabetologia. 2013;56(10):2266-2274.). In existing technologies, previous studies have mainly focused on describing the metabolite characteristics of different disease prognostic subtypes in non-alcoholic fatty liver disease (NAFLD) or metabolic dysfunction-related fatty liver disease (Martínez-Arranz I, Bruzzone C, Noureddin M, et al. Metabolic subtypes of patients with NAFLD exhibit distinct cardiovascular risk profiles. Hepatology. Oct 2022;76(4):1121-1134.), while neglecting how to integrate overall metabolic characteristics to construct a metabolic liver fat score for quantitative analysis of the differentiation of different subtypes of liver fat deposition. On the other hand, previous studies have also neglected whether there is heterogeneity in the response of different subtypes of liver fat deposition to lifestyle interventions. Summary of the Invention

[0003] The main objective of this invention is to provide a metabolic liver fat fraction model for type 2 diabetes patients, which can be used for cardiovascular metabolic risk stratification in type 2 diabetes, thereby enabling the application of precision medicine in the management of type 2 diabetes, addressing the problem of predicting cardiovascular metabolic risk and lifestyle intervention response related to liver fat deposition in overweight or obese patients with type 2 diabetes.

[0004] Another objective of this invention is to provide a method for constructing a metabolic liver fat fraction model for patients with type 2 diabetes, which is obtained through metabolomics combined with elastic network analysis modeling.

[0005] Another objective of the present invention is to provide a cardiovascular metabolic risk stratification system for type 2 diabetes, which includes the above-mentioned metabolic liver fat fraction model for type 2 diabetes patients, for reflecting the cardiovascular metabolic risk of overweight or obese type 2 diabetes patients and identifying type 2 diabetes patients who are suitable to benefit from and maintain their condition through lifestyle interventions in the long term.

[0006] Another objective of this invention is to provide a method for stratifying the cardiovascular and metabolic risks of type 2 diabetes, providing effective intervention strategy recommendations for patients with type 2 diabetes with different characteristics.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a metabolic liver fat fraction model for patients with type 2 diabetes, comprising the following calculation formula:

[0009] Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - pliver fat value taken on the best linear fit line between pliver fat and MRI-measured liver fat); where:

[0010] pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network.

[0011] MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging.

[0012] dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

[0013] Preferably, the type 2 diabetes patient is a type 2 diabetes patient who is overweight or obese.

[0014] Preferably, the metabolite biomarkers used to predict metabolic liver fat fraction in patients with type 2 diabetes are as follows: N-acetylaspartate (NAA), creatinine, 4-hydroxyglutamate, beta-citrylglutamate, N-acetylglutamine, pyroglutamine, S-1-pyrroline-5-carboxylate, 2-hydroxybutyrate / 2-hydroxyisobutyrate, cysteine-glutathione disulfide, 2-methylserine, betaine, dimethylglycine, glycine, N-acetylglycine, N-acetylserine, sarcosine, serine, 4-guanidinobutanoate, N-methylpipecolate, 2-hydroxyhippurate (salicylurate), 4-methylcatechol sulfate, EDTA, perfluorooctanesulfonate (PFOS), sulfate*, and imidazole. lactate, 1-methyl-5-imidazolelactate, 2,6-dihydroxybenzoicacid, 2,3-dihydroxyisovalerate, 3-hydroxystachydrine*, 4-allylphenol sulfate, cinnamoylglycine, dihydroferulate, ergothioneine, N-acetylalliin, quinate, S-allylcysteine, stachhydrine, cotinine, N-lactoyl isoleucine, N-lactoylleucine、isobutyrylglycine、methylsuccinoylcarnitine、N-acetylisoleucine、N2-acetyl,N6,N6-dimethyllysine、N2-acetyl,N6-methyllysine、pipecolate、alpha-ketobutyrate、cystathionine、cystine、N-acetylmethionine、N-acetyltaurine、S-carboxyethylcysteine、S-methylcysteine、S-methylcysteine sulfoxide、S-methylmethionine、N-lactoyl phenylalanine、(N(1) + N(8))-acetylspermidine、indolepropionate、kynurenate、N-formylkynurenine、3-methoxytyrosine、4-methoxyphenol sulfate、homovanillate (HVA)、3-amino-2-piperidone、N-acetylarginine、N-acetylproline、N-methylproline、ornithine、N-acetylglucosaminylasparagine、fructose、maltose / cellobiose、lactate、ascorbicacid 2-sulfate、ascorbic acid 3-sulfate*、heme、1-methylnicotinamide、nicotinamide riboside、gamma-CEHC、beta-cryptoxanthin、carotene diol (1)、retinol(Vitamin A)、phosphate、aconitate[cis or trans]、alpha-ketoglutarate、citraconate / glutaconate、succinylcarnitine (C4-DC)、16a-hydroxy DHEA 3-sulfate、5alpha-androstan-3alpha,17beta-diol disulfate、5alpha-androstan-3beta,17alpha-dioldisulfate、andro steroid monosulfate C19H28O6S (1)*、androstenediol(3beta,17beta) disulfate (2)、deoxycarnitine、palmitoyl-sphingosine-phosphoethanolamine (d18:1 / 16:0)、N-stearoyl-sphingosine (d18:1 / 18:0)*、cortisone、cortolone glucuronide(1)、tetrahydrocortisol glucuronide、tetrahydrocortisone glucuronide (5)、linoleoyl-docosahexaenoyl-glycerol (18:2 / 22:6) [1]*、N-stearoyl-sphinganine (d18:0 / 18:0)*、behenoyl dihydrosphingomyelin(d18:0 / 22:0)*、adrenoylcarnitine(C22:4)*、docosapentaenoylcarnitine(C22:5n3)*、arachidonoylcholine、behenoylcarnitine(C22)*、linolenoylcarnitine(C18:3)*、pimeloylcarnitine / 3-methyladipoylcarnitine(C7-DC)、2-aminooctanoate、2-hydroxyphytanate*、3-carboxy-4-methyl-5-pentyl-2-furanpropionate (3-CMPFP)**、3-carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF)、decadienedioic acid(C10:2-DC)、heptenedioate (C7:1-DC)*、octadecadienedioate (C18:2-DC)*、octadecenedioate (C18:1-DC)*、sebacate (C10-DC)、tetradecanedioate (C14-DC)、undecanedioate(C11-DC)、2-hydroxybehenate、2-hydroxylaurate、glycerol、glycerol3-phosphate、glycosyl-N-behenoyl-sphingadienine (d18:2 / 22:0)*、lactosyl-N-nervonoyl-sphingosine (d18:1 / 24:1)*、myristoleate (14:1n5)、palmitoleate (16:1n7)、1-palmitoyl-GPI (16:0)、1-(1-enyl-palmitoyl)-GPC (P-16:0)*、(2 or 3)-decenoate (10:1n7 or n8)、heptanoate (7:0)、undecanoate (11:0)、1,2-dilinoleoyl-GPC (18:2 / 18:2)、1-linoleoyl-2-linolenoyl-GPC (18:2 / 18:3)*、1-oleoyl-2-docosahexaenoyl-GPC (18:1 / 22:6)*、1-palmitoyl-2-dihomo-linolenoyl-GPC (16:0 / 20:3n3 or 6)*、1-palmitoyl-2-palmitoleoyl-GPC (16:0 / 16:1)*、1-stearoyl-2-docosahexaenoyl-GPE (18:0 / 22:6)*、1-stearoyl-2-oleoyl-GPE (18:0 / 18:1)、1-palmitoyl-2-oleoyl-GPI (16:0 / 18:1)*、1-stearoyl-2-arachidonoyl-GPI (18:0 / 20:4)、glycerophosphoethanolamine、glycerophosphorylcholine (GPC)、phosphoethanolamine、trimethylamine N-oxide、1-(1-enyl-palmitoyl)-2-linoleoyl-GPC (P-16:0 / 18:2)*、1-(1-enyl-palmitoyl)-2-oleoyl-GPC (P-16:0 / 18:1)*、1-(1-enyl-palmitoyl)-2-palmitoleoyl-GPC(P-16:0 / 16:1)*、1-(1-enyl-palmitoyl)-2-palmitoyl-GPC (P-16:0 / 16:0)*、docosapentaenoate (n6 DPA; 22:5n6)、pregnenediolsulfate (C21H34O5S)*、pregnenolone sulfate、chenodeoxycholate、cholate、5alpha-pregnan-3beta,20alpha-diol monosulfate (2)、3b-hydroxy-5-cholenoic acid、deoxycholate、glycocholenate sulfate*、glycodeoxycholate 3-sulfate、glycoursodeoxycholic acid sulfate (1)、lithocholate sulfate (1)、taurocholenatesulfate*、taurolithocholate 3-sulfate、hydroxypalmitoyl sphingomyelin (d18:1 / 16:0(OH))**、sphingomyelin (d18:1 / 18:1, d18:2 / 18:0)、sphingomyelin (d18:1 / 20:0,d16:1 / 22:0)*、sphingomyelin (d18:2 / 14:0, d18:1 / 14:1)*、sphingomyelin (d18:2 / 24:2)*、hexadecasphingosine (d16:1)*、3beta,7alpha-dihydroxy-5-cholestenoate、urate、xanthine、adenine、adenosine 5'-monophosphate (AMP)、N6-carbamoylthreonyladenosine、2'-O-methylcytidine、N4-acetylcytidine、dihydroorotate、3-aminoisobutyrate、2'-deoxyuridine、2'-O-methyluridine、5,6-dihydrouracil、metabolonic lactone sulfate**、branched-chain, straight-chain,or cyclopropyl 12:1 fattyacid**, prolylglycine, gamma-glutamylcitrulline*, gamma-glutamylglutamate, gamma-glutamylglutamine, gamma-glutamylthreonine, gamma-glutamyltryptophan, gamma-glutamylvaline.

[0015] A second aspect of the present invention provides a method for constructing the metabolic liver fat fraction model of type 2 diabetic patients, comprising the following steps:

[0016] (1) Detect the levels of metabolite biomarkers in the baseline plasma of the subjects used to predict the metabolic liver fat fraction in patients with type 2 diabetes and the liver fat content obtained by magnetic resonance imaging.

[0017] (2) The levels of the metabolite biomarkers, combined with the age and sex of the subjects, are used to generate a predictive liver fat content obtained by MRI-measured liver detection through linear regression of an elastic network. The predicted result for each subject is a predictive liver fat score, from which a metabolic liver fat score model is obtained, which is calculated using the following formula:

[0018] Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - value of liver fat on the best-fit line between liver fat and MRI-measured liver fat); where:

[0019] pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network.

[0020] MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging.

[0021] dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

[0022] Preferably, in step (1), the method for detecting the level of the metabolite biomarker includes: using non-targeted metabolomics analysis, and using ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS / MS) to detect the level of the metabolite biomarker;

[0023] The liver fat content of the subjects was detected by abdominal MRI-PDFF using a 3.0T magnetic resonance scanner (Ingenia, Philips Healthcare).

[0024] As a preferred embodiment, in step (2), before modeling, the levels of the metabolite biomarkers, together with the age and gender of the subjects, are first standardized by Z-score based on the baseline mean and standard deviation; then, an Elastic-Net regression model containing age, gender, and metabolite biomarker features is constructed to predict liver fat content; at the same time, 10-fold cross-validation is adopted, that is, in each iteration, the model is trained on 90% of the samples and tested on the remaining 10% of the samples, and the regularization parameter (lambda) is optimized by the cv.glmnet R package to minimize the mean square error and reduce the risk of overfitting.

[0025] A third aspect of the present invention provides a cardiovascular metabolic risk stratification and decision-making system for type 2 diabetes, comprising a metabolic liver fat fraction model of the type 2 diabetes patient, for outputting their cardiovascular metabolic risk stratification, and simultaneously outputting whether the subject of the type 2 diabetes patient is suitable to benefit from and maintain long-term from lifestyle interventions.

[0026] A fourth aspect of the present invention provides a method for stratifying cardiovascular metabolic risk in type 2 diabetes, comprising: integrating age, sex, and metabolite information using elastic network linear regression analysis based on the age, sex, liver fat content obtained from MRI scans, and 190 metabolites involved in predicting liver fat content in type 2 diabetes patients; simultaneously, obtaining a metabolic liver fat score based on the calculation formula of the metabolic liver fat score model for type 2 diabetes patients, and calculating the difference between the metabolic liver fat score and MRI-measured liver fat (dliver fat) for quantifying risk scoring; specifically including:

[0027] The degree of glucose and lipid metabolism disorders is quantified by the difference between metabolic liver fat fraction and MRI-measured liver fat (dliver fat). The higher the dliver fat value, the more severe the glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain the effect for up to 48 weeks.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] 1. In response to the significant metabolic heterogeneity of liver fat in overweight or obese type 2 diabetes patients, this invention, through metabolomics analysis combined with elastic network regression modeling, has for the first time discovered and verified 190 metabolite combinations as biomarkers to predict residual cardiovascular metabolic risk in type 2 diabetes patients beyond simple liver fat deposition. This effectively stratifies cardiovascular metabolic risk in type 2 diabetes patients and realizes the application of precision medicine in the prognosis of type 2 diabetes.

[0030] 2. Based on the aforementioned 190 metabolite combinations as biomarkers, this invention constructs a metabolic liver fat fraction model for type 2 diabetes patients by combining patient age, gender, and liver fat content measured by magnetic resonance imaging. This model provides predictive suggestions for the future intervention effects for type 2 diabetes patients who may require lifestyle interventions, helping clinicians identify type 2 diabetes patients who are more likely to benefit and maintain long-term health, avoiding ineffective interventions for low-responders, improving treatment efficiency, and reducing unnecessary waste of medical resources. Attached Figure Description

[0031] Figure 1 The blue scatter plot shows the correlation between liver fat measured by MRI at baseline and liver fat measured by metabolomics in the example. The blue and green histograms represent the distribution of liver fat and liver fat, respectively.

[0032] Figure 2 To verify the linear correlation between baseline liver fat (measured by MRI), liver fat (measured by metabolomics), and dliver fat and cardiovascular metabolic characteristics in the examples; colors represent regression coefficients, *P<0.05, **P<0.01, ***P<0.001 (after Benjamini-Hochberg correction); the abbreviations are as follows: ADA, Adaptation Index; BMI, Body Mass Index; DBP, Diastolic Blood Pressure; DI, Disposal Index; FIB-4, Fibrosis-4 Index; FPG, Fasting Plasma Glucose; HbA1c, Glycated Hemoglobin; HDL, High-Density Lipoprotein; HOMA-IR, Homeostasis Model of Insulin Resistance Assessment; ISI, Insulin Sensitivity Index; LDL, Low-Density Lipoprotein; OGIS, Oral Glucose Insulin Sensitivity; PPG, Post-Load Plasma Glucose; SBP, Systolic Blood Pressure; TG, Triglycerides.

[0033] Figure 3In this example, based on baseline liver fat stratification, a linear model was used to estimate changes in cardiometabolic characteristics in the dietary intervention group relative to the control group from baseline to 12 and 48 weeks; where the abbreviations are as follows: BMI, body mass index; DI, metabolic index; HbA1c, glycated hemoglobin; ISI, insulin sensitivity index.

[0034] Figure 4 The image shows the correlation between liver fat content measured by MRI and liver fat content measured by metabolomics at 12 weeks in this example (blue scatter plot). The blue and green histograms represent the distribution of liver fat and liver fat, respectively.

[0035] Figure 5 To verify the linear correlation between liver fat content (measured by MRI), liver fat content (mliver fat), and liver fat (dliver fat) at 12 weeks and cardiovascular metabolic characteristics in the examples; colors represent regression coefficients, *P<0.05, **P<0.01, ***P<0.001 (after Benjamini-Hochberg correction); the abbreviations are as follows: ADA, Adaptation Index; BMI, Body Mass Index; DBP, Diastolic Blood Pressure; DI, Disposal Index; FIB-4, Fibrosis-4 Index; FPG, Fasting Plasma Glucose; HbA1c, Glycated Hemoglobin; HDL, High-Density Lipoprotein; HOMA-IR, Homeostasis Model of Insulin Resistance Assessment; ISI, Insulin Sensitivity Index; LDL, Low-Density Lipoprotein; OGIS, Oral Glucose Insulin Sensitivity; PPG, Post-Load Plasma Glucose; SBP, Systolic Blood Pressure; TG, Triglycerides. Detailed Implementation

[0036] To more fully understand and demonstrate the technical solutions, objectives, and advantages of the present invention, the technical effects produced by the present invention will be further described in detail and completely below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be noted that other embodiments obtained by those skilled in the art without departing from the concept of the present invention are all within the protection scope of the present invention.

[0037] The following examples present a set of metabolite biomarkers for constructing a metabolic liver fat fraction model in patients with type 2 diabetes, consisting of 190 metabolites listed in Table 1.

[0038] The following examples also propose a metabolic liver fat fraction model for patients with type 2 diabetes, which includes the following calculation formula:

[0039] Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - pliver fat value taken on the best linear fit line between pliver fat and MRI-measured liver fat); where:

[0040] pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network.

[0041] MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging.

[0042] dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

[0043] The above-mentioned metabolic liver fat fraction model for patients with type 2 diabetes was obtained through a construction method including the following steps:

[0044] (1) Detect the levels of metabolite biomarkers in the baseline plasma of the subjects used to predict the metabolic liver fat fraction in patients with type 2 diabetes and the liver fat content obtained by magnetic resonance imaging.

[0045] (2) The levels of the metabolite biomarkers, combined with the age and sex of the subjects, are used to generate a predictive liver fat content obtained by MRI-measured liver detection through linear regression of an elastic network. The predicted result for each subject is a predictive liver fat score, from which a metabolic liver fat score model is obtained, which is calculated using the following formula:

[0046] Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - the value of pliver fat on the best fit line between pliver fat and MRI-measured liver fat).

[0047] In some embodiments, the detection method for the above-mentioned metabolite biomarker levels includes: using non-targeted metabolomics analysis, and employing ultra-high performance liquid chromatography-tandem mass spectrometry (UPLC-MS / MS) to detect the biomarker levels, which is performed by the metabolomics platform of the Calibra laboratory of Dian Diagnostics (Hangzhou, Zhejiang, China).

[0048] In some embodiments, the constructed metabolic liver fat fraction and dliverfat of type 2 diabetic patients are also used to capture cardiometabolic disorders other than simple liver deposition, and the association between metabolic liver fat fraction and dliver fat and multiple cardiometabolic indicators is analyzed by linear regression.

[0049] In some embodiments, derived dliver fat is also used to predict the intervention effect in subjects receiving dietary interventions, thereby identifying individuals who are more likely to benefit from lifestyle interventions, and the association between baseline dliver fat levels and the effect of dietary interventions is analyzed by linear regression.

[0050] Table 1: Metabolites involved in metabolic liver fat fraction

[0051]

[0052]

[0053]

[0054]

[0055]

[0056]

[0057]

[0058]

[0059] Example 1

[0060] 1. Research subjects and intervention program

[0061] Target population: Overweight or obese patients with type 2 diabetes. This example includes 324 participants from the randomized controlled clinical trial (IDEATE study) who received dietary intervention, exercise or health education (control).

[0062] Intervention plan:

[0063] (1) Dietary intervention: restrict energy intake to 790 kcal / day for 2 days a week (usually two consecutive days), and maintain a regular diet for the other 5 days.

[0064] (2) Exercise intervention: supervised training twice a week (usually on non-consecutive days) at a medical center, specifically including:

[0065] High-intensity interval training (HIIT): This is done using a stationary bike and includes a 5-minute warm-up and a 5-minute cool-down. The core training consists of 4 minutes of high-intensity interval training (reaching 85-90% of the age-predicted maximum heart rate).

[0066] Resistance training: Use a multi-strength training machine to complete 4 machine training exercises, 8-12 repetitions per set, for a total of 2 sets (intensity is 80% of the maximum repetition weight).

[0067] During training, heart rate is monitored in real time via a GEONAUTE Bluetooth heart rate chest strap, and the intensity of resistance training for each body part is recorded. The compliance standard is set at ≥85% of maximum heart rate for HIIT and 80% of 1-RM intensity for resistance training. During certain periods, the exercise program is adjusted to be performed at home. HIIT can be implemented using a stationary bike, treadmill, or running in place, while resistance training is replaced with bodyweight strength training, remotely guided by a physician via real-time audio and video conferencing.

[0068] (3) Health education: Randomized physicians conducted health education in the same manner for all intervention and control groups, providing instructions on healthy eating and exercise according to the "Guidelines for the Prevention and Treatment of Type 2 Diabetes in China". Physicians provided lifestyle advice to participants weekly via telephone or WeChat, as well as monthly face-to-face education.

[0069] Intervention duration: 12 weeks, with follow-up continuing until week 48 after the intervention.

[0070] Follow-up time points: baseline, early intervention (week 4), end of intervention (week 12), and end of follow-up (week 48).

[0071] 2. Cardiovascular metabolic risk indicators

[0072] Plasma glucose was measured using either the glucose oxidase method or the hexokinase method; HbA1c was detected using high-performance liquid chromatography (BIORAD, Hercules, CA, USA) and completed within two hours of blood collection in a local laboratory, strictly adhering to quality control procedures. Insulin concentration was uniformly measured by the central laboratory using an automated analyzer (Siemens Atellica Solution, Siemens Corporation). Insulin sensitivity was assessed using the Oral Glucose Insulin Sensitivity Index (OGIS), the Insulin Sensitivity Index (ISI), and the Homeostasis Model Assessment of Insulin Resistance (HOMA-IR). Beta cell function was assessed using the Disposal Index (DI) and the Adaptation Index. Lipid profiles were uniformly measured by enzymatic methods in the central laboratory using an automated analyzer (Roche Cobas 800 c 701, Mannheim, Germany). Liver enzymes (alanine aminotransferase and aspartate aminotransferase) were detected in the local laboratory using an automated biochemical analyzer. Weight, BMI, and waist circumference were measured by trained research nurses according to standardized procedures. Bioelectrical impedance analysis (SEEHiGHER H-Key350, Beijing, China) was used to estimate the subject's body fat mass and lean body mass. Blood pressure was measured in a seated position using an automated electronic blood pressure monitor (Omron HEM-752 FUZZY, Tokyo, Japan), and the average of three readings was recorded.

[0073] 3. Measurement of liver fat content

[0074] Liver fat content was assessed using an abdominal MRI-PDFF scanner (Ingenia, Philips Healthcare) with a 3.0T MRI scanner. Multi-echo images were acquired using the mDIXON-Quant sequence to separate water and lipids, generating water maps, fat maps, T2* maps, and fat fraction maps.

[0075] In the quantitative analysis, nine circular regions of interest (ROIs; 3 cm² each) corresponding to the Couinaud liver segments were selected on the PDFF image. ROIs were placed in the central region of each liver segment, avoiding major blood vessels, liver margins, and imaging artifacts. PDFF values ​​were extracted for each ROI, and the mean PDFF value of the nine liver segments was used as the overall liver PDFF. All MRI-PDFF measurements were performed by technicians whose participant grouping was unknown.

[0076] 4. Evaluation indicators of intervention effectiveness

[0077] Changes in weight, BMI, waist circumference, body fat, lean body mass, liver fat, glycated hemoglobin, insulin sensitivity index, and treatment index from baseline at 12 and 48 weeks.

[0078] 5. Methods for constructing a metabolic liver fat fraction model

[0079] 5.1 Baseline clinical variable collection for the predictive model:

[0080] Demographic characteristics: sex, age;

[0081] 5.2 Determination of Model Metabolites for Calculating Metabolic Liver Fat Fraction

[0082] Non-targeted metabolomics analysis was performed on the baseline fasting plasma of subjects participating in the dietary intervention. Metabolites with a baseline missing rate of more than 20% were removed, and 771 metabolites were finally measured.

[0083] 5.3 Construction of a metabolic liver fat fraction model

[0084] Before modeling, 771 metabolites, along with age and sex, were standardized using Z-scores based on baseline mean and standard deviation. An Elastic-Net regression model incorporating age, sex, and metabolite features was then constructed. The final model selected 190 metabolites (as shown in Table 1) to predict liver fat content. To reduce the risk of overfitting, 10-fold cross-validation was used: in each iteration, the model was trained on 90% of the samples and tested on the remaining 10%. The regularization parameter (lambda) was optimized using the cv.glmnet R package to minimize the mean squared error.

[0085] The formula for calculating metabolic liver fat fraction is as follows:

[0086] Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - the value of pliver fat corresponding to the best linear fit line between pliver fat and MRI-measured liver fat), where:

[0087] pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network.

[0088] MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging.

[0089] dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

[0090] 6. Association between metabolic liver fat fraction and cardiovascular risk indicators

[0091] Linear regression analysis was used to analyze the association between liver fat, metabolic liver fat fraction, and dliver fat with cardiometabolic parameters. P-values ​​were corrected for multiple comparisons using the Benjamini-Hochberg (BH) method. For ease of comparison, all cardiometabolic parameters were standardized using Z-scores.

[0092] 7. Association between metabolic liver fat fraction and lifestyle interventions

[0093] Within each tertile of liver fat, metabolic liver fat fraction, or dliver fat, linear regression analysis was used to assess differences in changes in cardiometabolic parameters at 12 and 48 weeks between the diet or exercise intervention groups and the control group. The models were adjusted for sex, age, center of study, baseline cardiometabolic parameter values, and baseline liver fat, and interaction terms between the tertiles and intervention groups (for diet intervention and exercise intervention, respectively) were added to the models to test for interaction effects.

[0094] 8. Final Result

[0095] The results are as follows Figure 1-3 As shown, metabolic liver fat fraction was highly correlated with liver fat measured by MRI, while dliver fat was not correlated with MRI-measured liver fat but was closely associated with various cardiometabolic characteristics, including body composition, blood glucose levels, insulin sensitivity, and triglycerides. Participants with a metabolically healthy phenotype (metabolic liver fat fraction < MRI-measured liver fat) responded better to dietary interventions and showed sustained improvements in weight, insulin sensitivity, and beta cell function during long-term follow-up after intervention.

[0096] 9. Model Validation

[0097] To verify the robustness of the constructed metabolic liver fat fraction model, this model was further applied to the levels of 190 metabolites measured at 12 weeks and the liver fat content obtained by magnetic resonance imaging (MRI) in this population. This was to verify the close correlation between the difference between the metabolic liver fat fraction and the MRI-measured liver fat content (dliverfat) and the cardiovascular risk of the subjects. The results are as follows: Figure 4 and 5 As shown in the figure, liver fat in the validation population was not associated with liver fat measured by MRI, but was closely related to various cardiometabolic characteristics, including body fat, blood glucose levels, insulin sensitivity, and triglycerides, indicating that the constructed metabolic liver fat fraction model has good robustness.

[0098] In summary, this invention utilizes a metabolic liver fat fraction model constructed from clinical indicators and metabolites of subjects in randomized controlled clinical trials. This model is used to capture systemic metabolic disorders beyond liver fat in patients with type 2 diabetes, providing a new technical approach for cardiometabolic risk stratification. It can also identify type 2 diabetes patients who are more likely to benefit from lifestyle interventions, helping to optimize the selection of individualized lifestyle strategies and has significant clinical application value.

[0099] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A metabolic liver fat fraction model for patients with type 2 diabetes, characterized in that, It includes the following calculation formula: Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - pliver fat value taken on the best linear fit line between pliver fat and MRI-measured liver fat); where: pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network. MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging. dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

2. The metabolic liver fat fraction model for type 2 diabetic patients according to claim 1, characterized in that, The type 2 diabetes patients are those who are overweight or obese.

3. The metabolic liver fat fraction model for type 2 diabetic patients according to claim 1, characterized in that, The metabolite biomarkers for predicting the metabolic liver fat fraction in type 2 diabetes patients are as follows: N-acetylaspartate (NAA), creatinine, 4-hydroxyglutamate, beta-citrylglutamate, N-acetylglutamine, pyroglutamine, S-1-pyrroline-5-carboxylate, 2-hydroxybutyrate / 2-hydroxyisobutyrate, cysteine-glutathione disulfide, 2-methylserine, betaine, dimethylglycine, glycine, N-acetylglycine, N-acetylserine, sarcosine, serine, 4-guanidinobutanoate, N-methylpipecolate, 2-hydroxyhippurate (salicylurate), 4-methylcatechol sulfate, EDTA, perfluorooctanesulfonate (PFOS), sulfate*, imidazole lactate, 1-methyl-5-imidazolelactate, 2,6-dihydroxybenzoic acid, 2,3-dihydroxyisovalerate, 3-hydroxystachydrine*, 4-allylphenol sulfate, cinnamoylglycine, dihydroferulate, ergothioneine, N-acetylalliin, quinate, S-allylcysteine, stachydrine, cotinine, N-lactoyl isoleucine, N-lactoylleucine、isobutyrylglycine、methylsuccinoylcarnitine、N-acetylisoleucine、N2-acetyl,N6,N6-dimethyllysine、N2-acetyl,N6-methyllysine、pipecolate、alpha-ketobutyrate、cystathionine、cystine、N-acetylmethionine、N-acetyltaurine、S-carboxyethylcysteine、S-methylcysteine、S-methylcysteine sulfoxide、S-methylmethionine、N-lactoyl phenylalanine、(N(1) + N(8))-acetylspermidine、indolepropionate、kynurenate、N-formylkynurenine、3-methoxytyrosine、4-methoxyphenol sulfate、homovanillate (HVA)、3-amino-2-piperidone、N-acetylarginine、N-acetylproline、N-methylproline、ornithine、N-acetylglucosaminylasparagine、fructose、maltose / cellobiose、lactate、ascorbicacid 2-sulfate、ascorbic acid 3-sulfate*、heme、1-methylnicotinamide、nicotinamide riboside、gamma-CEHC、beta-cryptoxanthin、carotene diol (1)、retinol(Vitamin A)、phosphate、aconitate[cis or trans]、alpha-ketoglutarate、citraconate / glutaconate、succinylcarnitine (C4-DC)、16a-hydroxy DHEA 3-sulfate、5alpha-androstan-3alpha,17beta-diol disulfate、5alpha-androstan-3beta,17alpha-dioldisulfate、andro steroid monosulfate C19H28O6S (1)*、androstenediol(3beta,17beta) disulfate (2)、deoxycarnitine、palmitoyl-sphingosine-phosphoethanolamine (d18:1 / 16:0)、N-stearoyl-sphingosine (d18:1 / 18:0)*、cortisone、cortolone glucuronide(1)、tetrahydrocortisol glucuronide、tetrahydrocortisone glucuronide (5)、linoleoyl-docosahexaenoyl-glycerol (18:2 / 22:6) [1]*、N-stearoyl-sphinganine (d18:0 / 18:0)*、behenoyl dihydrosphingomyelin(d18:0 / 22:0)*、adrenoylcarnitine(C22:4)*、docosapentaenoylcarnitine(C22:5n3)*、arachidonoylcholine、behenoylcarnitine(C22)*、linolenoylcarnitine(C18:3)*、pimeloylcarnitine / 3-methyladipoylcarnitine(C7-DC)、2-aminooctanoate、2-hydroxyphytanate*、3-carboxy-4-methyl-5-pentyl-2-furanpropionate (3-CMPFP)**、3-carboxy-4-methyl-5-propyl-2-furanpropanoate (CMPF)、decadienedioic acid(C10:2-DC)、heptenedioate (C7:1-DC)*、octadecadienedioate (C18:2-DC)*、octadecenedioate (C18:1-DC)*、sebacate (C10-DC)、tetradecanedioate (C14-DC)、undecanedioate(C11-DC)、2-hydroxybehenate、2-hydroxylaurate、glycerol、glycerol3-phosphate、glycosyl-N-behenoyl-sphingadienine (d18:2 / 22:0)*、lactosyl-N-nervonoyl-sphingosine (d18:1 / 24:1)*、myristoleate (14:1n5)、palmitoleate (16:1n7)、1-palmitoyl-GPI (16:0)、1-(1-enyl-palmitoyl)-GPC (P-16:0)*、(2 or 3)-decenoate (10:1n7 or n8)、heptanoate (7:0)、undecanoate (11:0)、1,2-dilinoleoyl-GPC (18:2 / 18:2)、1-linoleoyl-2-linolenoyl-GPC (18:2 / 18:3)*、1-oleoyl-2-docosahexaenoyl-GPC (18:1 / 22:6)*、1-palmitoyl-2-dihomo-linolenoyl-GPC (16:0 / 20:3n3 or 6)*、1-palmitoyl-2-palmitoleoyl-GPC (16:0 / 16:1)*、1-stearoyl-2-docosahexaenoyl-GPE (18:0 / 22:6)*、1-stearoyl-2-oleoyl-GPE (18:0 / 18:1)、1-palmitoyl-2-oleoyl-GPI (16:0 / 18:1)*、1-stearoyl-2-arachidonoyl-GPI (18:0 / 20:4)、glycerophosphoethanolamine、glycerophosphorylcholine (GPC)、phosphoethanolamine、trimethylamine N-oxide、1-(1-enyl-palmitoyl)-2-linoleoyl-GPC (P-16:0 / 18:2)*、1-(1-enyl-palmitoyl)-2-oleoyl-GPC (P-16:0 / 18:1)*、1-(1-enyl-palmitoyl)-2-palmitoleoyl-GPC(P-16:0 / 16:1)*、1-(1-enyl-palmitoyl)-2-palmitoyl-GPC (P-16:0 / 16:0)*、docosapentaenoate (n6 DPA; 22:5n6)、pregnenediolsulfate (C21H34O5S)*、pregnenolone sulfate、chenodeoxycholate、cholate、5alpha-pregnan-3beta,20alpha-diol monosulfate (2)、3b-hydroxy-5-cholenoic acid、deoxycholate、glycocholenate sulfate*、glycodeoxycholate 3-sulfate、glycoursodeoxycholic acid sulfate (1)、lithocholate sulfate (1)、taurocholenatesulfate*、taurolithocholate 3-sulfate、hydroxypalmitoyl sphingomyelin (d18:1 / 16:0(OH))**、sphingomyelin (d18:1 / 18:1, d18:2 / 18:0)、sphingomyelin (d18:1 / 20:0,d16:1 / 22:0)*、sphingomyelin (d18:2 / 14:0, d18:1 / 14:1)*、sphingomyelin (d18:2 / 24:2)*、hexadecasphingosine (d16:1)*、3beta,7alpha-dihydroxy-5-cholestenoate、urate、xanthine、adenine、adenosine 5'-monophosphate (AMP)、N6-carbamoylthreonyladenosine、2'-O-methylcytidine、N4-acetylcytidine、dihydroorotate、3-aminoisobutyrate、2'-deoxyuridine、2'-O-methyluridine、5,6-dihydrouracil、metabolonic lactone sulfate**、branched-chain, straight-chain,or cyclopropyl 12:1 fattyacid**、prolylglycine、gamma-glutamylcitrulline*、gamma-glutamylglutamate、gamma-glutamylglutamine、gamma-glutamylthreonine、gamma-glutamyltryptophan、gamma-glutamylvaline。 4. The method for constructing the metabolic liver fat fraction model for patients with type 2 diabetes as described in any one of claims 1-3, characterized in that, Includes the following steps: (1) Detect the levels of metabolite biomarkers in the baseline plasma of the subjects used to predict the metabolic liver fat fraction in patients with type 2 diabetes and the liver fat content obtained by magnetic resonance imaging. (2) The levels of the metabolite biomarkers, combined with the age and sex of the subjects, are used to generate a predictive liver fat content obtained from magnetic resonance imaging through linear regression of an elastic network. The predicted result for each subject is a predictive liver fat score, from which a metabolic liver fat score model is obtained, which is calculated using the following formula: Metabolic liver fat fraction = MRI-measured liver fat + (pliver fat - value of pliver fat on the best-fit line between pliver fat and MRI-measured liver fat); where: pliver fat represents the predicted liver fat content obtained from magnetic resonance imaging (MRI) by combining the levels of metabolite biomarkers used to predict the metabolic liver fat fraction in patients with type 2 diabetes with the patient's age and sex through linear regression using an elastic network. MRI-measured liver fat refers to the amount of fat in the liver obtained by magnetic resonance imaging. dliver fat represents the difference between metabolic liver fat fraction and MRI-measured liver fat, used to quantify the degree of blood glucose and lipid metabolism disorders. That is, the higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity with lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

5. The method for constructing a metabolic liver fat fraction model for patients with type 2 diabetes according to claim 4, characterized in that, In step (1), the method for detecting the level of the metabolite biomarker includes: using non-targeted metabolomics analysis, and using ultra-high performance liquid chromatography-tandem mass spectrometry to detect the level of the metabolite biomarker; The liver fat content of the subjects was detected by abdominal MRI-PDFF using a 3.0T magnetic resonance scanner.

6. The method for constructing a metabolic liver fat fraction model for patients with type 2 diabetes according to claim 4, characterized in that, In step (2), before modeling, the levels of the metabolite biomarkers, along with the age and gender of the subjects, are standardized using Z-scores based on the baseline mean and standard deviation. Then, an Elastic-Net regression model containing age, gender, and metabolite biomarker features is constructed to predict liver fat content. At the same time, 10-fold cross-validation is used, that is, in each iteration, the model is trained on 90% of the samples and tested on the remaining 10% of the samples. The regularization parameter (lambda) is optimized using the cv.glmnetR package to minimize the mean squared error and reduce the risk of overfitting.

7. A cardiovascular metabolic risk stratification and decision-making system for type 2 diabetes, characterized in that, It includes the metabolic liver fat fraction model for patients with type 2 diabetes as described in any one of claims 1-3, for outputting their cardiovascular metabolic risk stratification, and for outputting whether the subjects of type 2 diabetes are suitable to benefit from and maintain long-term from lifestyle interventions.

8. A method for stratifying cardiovascular and metabolic risks in type 2 diabetes, characterized in that, include: Based on the age, sex, and liver fat content obtained from MRI scans of type 2 diabetic patients, as well as 190 metabolites involved in predicting liver fat content, elastic network linear regression analysis was used to integrate age, sex, and metabolite information. Simultaneously, a metabolic liver fat score was obtained based on the calculation formula of the metabolic liver fat score model for type 2 diabetic patients as described in any one of claims 1-3, and the difference between the metabolic liver fat score and MRI-measured liver fat (dliver fat) was calculated for quantifying risk scoring; specifically including: dliver fat is used to quantify the degree of blood glucose and lipid metabolism disorders. The higher the dliver fat value, the more severe the blood glucose and lipid metabolism disorders in the subject. Patients with type 2 diabetes and overweight / obesity who have lower dliver fat values ​​are more likely to lose weight, improve insulin sensitivity and beta cell function through diet and maintain these effects in the long term.

Images