Metabolic markers for diagnosing diabetic secondary osteoporosis and use thereof
By screening cytosine nucleotide, ketoglutarate, and triethylamine metabolic markers, and combining them with exosome detection and prediction models, the diagnostic bias problem of diabetic secondary osteoporosis has been solved, and highly accurate risk assessment has been achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-26
- Publication Date
- 2026-04-10
AI Technical Summary
The diagnosis of secondary osteoporosis caused by diabetes in the current technology has risk assessment bias. Traditional bone mineral density testing methods cannot accurately identify high-risk individuals for osteoporosis, especially in patients with type 2 diabetes, which manifests as the bone mineral density paradox.
Three metabolic markers—cytosine nucleotides, ketoglutarate, and triethylamine—were identified. By detecting the levels of these metabolites in exosomes and combining them with algorithms such as orthogonal partial least squares discriminant analysis, a predictive model was constructed for the diagnosis of diabetic secondary osteoporosis.
It improved the diagnostic accuracy and specificity of diabetic secondary osteoporosis, with the AUC of metabolic markers reaching 0.99, significantly improving the accuracy of risk assessment.
Smart Images
Figure CN121410030B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of biotechnology, in particular to a metabolic marker for diagnosing diabetic secondary osteoporosis and application thereof. BACKGROUND
[0002] Long-term hyperglycemia can impair the metabolic function of the body, leading to decreased bone mass and concurrent osteoporosis, i.e., diabetic secondary osteoporosis (DOP), which increases the risk of bone fracture. The prevalence of osteoporosis in patients with type 1 diabetes is about 30-50%, and the risk of hip fracture is 4.4-6 times that of non-diabetic population; the prevalence of osteoporosis in patients with type 2 diabetes is about 20-30%, and the relative risk of fracture at different sites is 1.17 to 2.03 compared with non-diabetic population.
[0003] The diagnosis of DOP currently mainly relies on bone density detection (DXA), but it has obvious limitations: the damage of diabetes to the skeleton is centered on "bone mass deterioration" (disorder of bone material properties and increased cortical porosity), and DXA can only assess bone density, which will inevitably underestimate the risk of fracture due to abnormal bone mass. This defect is particularly pronounced in patients with type 2 diabetes, manifesting as a typical "bone density paradox" phenomenon - although the bone density value may be normal or high, the actual bone mass decreases and the risk of fracture increases significantly. In addition, the commonly used fracture risk assessment tools do not take into account the special damage of diabetes to bone mass, nor do they consider diabetes as an independent risk factor, further leading to bias in risk assessment. Therefore, screening new biomarkers for DOP, establishing early diagnosis models and identifying high-risk individuals are crucial for controlling bone metabolic disorders in diabetic patients, slowing down the deterioration of bone mass in the later stage and reducing the risk of fracture. SUMMARY
[0004] To this end, the technical problem to be solved by the present application is to overcome the risk assessment bias in the diagnosis of DOP in the prior art.
[0005] To solve the above technical problems, the present application provides a metabolic marker for diagnosing diabetic secondary osteoporosis and application thereof. The present application first screens out the significant difference metabolites between the diabetic patients with osteoporosis (DOP) and the diabetic patients without osteoporosis (DM); secondly, the significant difference metabolites between the patients with primary osteoporosis and the healthy control group are screened out, and the intersection of the two groups of difference metabolites is excluded, so that the obtained metabolites exclude the influence of diabetic patients without osteoporosis and primary osteoporosis. Subsequently, the metabolic markers of the present application, cytosine nucleotide (CMP), oxoglutaric acid and triethylamine, are further screened out, and the AUC of the three metabolic markers is greater than 0.8 when they are used alone, and the AUC is 0.99 when they are used in combination, which provides a new diagnostic scheme for the diagnosis of DOP in diabetic population, early screening and prevention of high-risk population of secondary fracture, and effectively makes up for the limitation of traditional bone density detection in identifying osteoporosis in diabetic patient population.
[0006] The first object of the present application is to provide the application of the reagent for detecting the content of the metabolic marker in the preparation of the diagnostic kit for diabetic secondary osteoporosis, and the metabolic marker is selected from one or more of cytosine nucleotide, oxoglutaric acid and triethylamine.
[0007] Further, the detection sample of the diagnostic kit is exosome.
[0008] The second object of the present application is to provide a diagnostic kit for diabetic secondary osteoporosis, which contains a reagent for detecting the content of a metabolic marker, wherein the metabolic marker is the metabolic marker described above.
[0009] Further, the diagnostic kit further comprises a reagent for extracting exosomes from plasma.
[0010] Further, the reagent comprises a reagent for detecting the concentration or content of the metabolic marker in the sample to be tested by nuclear magnetic resonance method, chromatography method, spectroscopy method, mass spectrometry method or combination method thereof.
[0011] The third object of the present application is to provide a reagent for detecting the concentration or content of a metabolic marker in a sample to be tested, wherein the metabolic marker is the metabolic marker described above, and the sample to be tested is exosome.
[0012] The fourth object of the present application is to provide the application of a metabolic marker in the preparation of a diagnostic product for diabetic secondary osteoporosis, and the metabolic marker is the metabolic marker described above.
[0013] Further, the diagnostic product comprises a device for analyzing the content of the metabolic marker.
[0014] A fifth object of the present application is to provide an application of a metabolic marker in constructing a prediction model of diabetes secondary osteoporosis, wherein the metabolic marker is the metabolic marker described above.
[0015] Further, the prediction model comprises a step of analyzing the concentration level data of the metabolic marker by using one or more algorithms selected from the group consisting of orthogonal partial least squares discriminant analysis, partial least squares discriminant analysis, classification and logistic regression, k-nearest neighbor algorithm, naive Bayes, support vector machine, decision tree, random forest, regression tree, gradient boosting decision tree, extreme gradient boosting algorithm, light gradient boosting machine, gradient boosting machine, lasso algorithm and convolutional neural network.
[0016] Further, the prediction model comprises the following steps:
[0017] S1, a data receiving unit, wherein the data comprises concentration levels of metabolic markers in samples, and the metabolic markers are selected from one or more of cytosine nucleotide, ketoglutaric acid and triethylamine;
[0018] S2, a data analysis unit, wherein the data analysis unit comprises analyzing the data in S1 by using one or more algorithms selected from the group consisting of orthogonal partial least squares discriminant analysis, partial least squares discriminant analysis, classification and logistic regression, k-nearest neighbor algorithm, naive Bayes, support vector machine, decision tree, random forest, regression tree, gradient boosting decision tree, extreme gradient boosting algorithm, light gradient boosting machine, gradient boosting machine, lasso algorithm and convolutional neural network.
[0019] The above technical solutions of the present application have the following beneficial effects compared with the prior art:
[0020] The present application screens a group of metabolic markers for diagnosing diabetes secondary osteoporosis, and these markers have good diagnostic effect in the diagnosis of diabetes secondary osteoporosis. Specifically, the cytosine nucleotide, ketoglutaric acid and triethylamine of the present application have an AUC greater than 0.8 when independently diagnosing diabetes secondary osteoporosis, and the AUC of the three metabolic markers is as high as 0.99 when combined diagnosis, which has good prediction effect. The diagnostic markers of the present application can exclude the influence of metabolites in primary osteoporosis on the diagnostic effect, and significantly improve the accuracy and specificity of the diagnosis of diabetes secondary osteoporosis. BRIEF DESCRIPTION OF DRAWINGS
[0021] In order to make the content of the present application more easily understood, the present application will be further described in detail below according to specific embodiments of the present application and in conjunction with the drawings.
[0022] Figure 1is an exosome characteristic identification chart, wherein A is an exosome electron microscope chart, and B is an exosome particle size chart;
[0023] Figure 2 is a core metabolic marker result chart identified by the SVM-RFE and RF models, wherein A is a relationship between the number of SVM-RFE features and model errors, and 20 features are determined as the optimal number of features, and B is an important feature sorting screened based on the RF model, and the features are arranged in descending order according to the average reduction amount of the Gini coefficient, and the higher the value, the greater the importance of the corresponding feature in the model classification decision;
[0024] Figure 3 is an intersection chart of key metabolites screened by the SVM-RFE and RF models;
[0025] Figure 4 is an ROC curve of N-acetyl-D-galactosamine;
[0026] Figure 5 is an ROC curve of N-acetyl-L-lysine;
[0027] Figure 6 is an ROC curve of cytosine nucleotides;
[0028] Figure 7 is an ROC curve of docosahexaenoic ethanolamide;
[0029] Figure 8 is an ROC curve of heptanedioic acid;
[0030] Figure 9 is an ROC curve of methylmalonic acid;
[0031] Figure 10 is an ROC curve of pyridoxal phosphate;
[0032] Figure 11 is an ROC curve of aldosterone;
[0033] Figure 12 is an ROC curve of triethylamine;
[0034] Figure 13 is an ROC curve of ketoglutaric acid;
[0035] Figure 14 is an ROC curve of oroxin;
[0036] Figure 15 is an ROC curve of cinchonidine;
[0037] Figure 16 is an ROC curve of oleic acid;
[0038] Figure 17Figures showing the differential expression of three specific diagnostic marker metabolites (cytosine nucleotides, ketoglutarate, triethylamine) between the case-control groups, wherein A-C are the differential expression figures in the discovery group of Example 1, and D-F are the differential expression figures in the control group and the case group of Example 2;
[0039] Figure 18 Figures showing the receiver operating characteristic (ROC) curves of the three specific diagnostic marker metabolites (cytosine nucleotides, ketoglutarate, triethylamine) in identifying DOP patients in the diabetic population, wherein A is the ROC curve of the three metabolite markers used alone in the discovery group of Example 1, B is the ROC curve of the three metabolite markers used in combination in the discovery group of Example 1, C is the ROC curve of the three metabolite markers used alone in Example 2, and D is the ROC curve of the three metabolite markers used in combination in Example 2. DETAILED DESCRIPTION
[0040] The application will be further described below in conjunction with the drawings and specific examples so that those skilled in the art can better understand the application and implement it, but the examples are not intended to limit the application.
[0041] Example 1: Screening of differential metabolites in plasma exosomes between diabetic patients with secondary osteoporosis and diabetic patients without osteoporosis
[0042] 1. Research subjects and research design
[0043] All subjects obtained written informed consent before receiving the study. The subjects' venous blood was collected in the morning on an empty stomach, the plasma was separated, and the plasma was stored in a -80°C refrigerator.
[0044] The research subjects of this example include two groups: (a) a discovery group (diabetic group): 20 diabetic patients without osteoporosis (DM) constitute a control group, and 20 diabetic patients with secondary osteoporosis (DOP) constitute a case group; (b) a discrimination group (non-diabetic group): containing 50 healthy subjects and 50 subjects with primary osteoporosis.
[0045] The inclusion and exclusion criteria for the subjects are as follows:
[0046] (1) Female subjects aged ≥65 years were included;
[0047] (2) The inclusion criteria for diabetes were typical symptoms of diabetes (polydipsia, polyphagia, polyuria, unexplained weight loss) + fasting plasma glucose (FPG) ≥7 mmol / L;
[0048] (3) The inclusion criteria for osteoporosis were based on bone mineral density measured by dual-energy X-ray absorptiometry (bone mineral density T-value of the femoral neck was less than -2.5);
[0049] (4) Patients with a history of fracture, liver and kidney dysfunction, abnormal thyroid function, chronic inflammatory infection, malignant tumor and other diseases were excluded.
[0050] 2. Plasma exosome isolation and identification
[0051] Exosomes in plasma samples were isolated using a plasma exosome affinity extraction kit, particle size was calculated using nanoparticle tracking analysis, and exosome morphology was detected by transmission electron microscopy. The morphology of exosomes showed a typical "tea tray" structure, with a size of 60-80 nm (see Figure 1 ).
[0052] 3. Exosome metabolite extraction
[0053] (1) Move the sample to a 2 mL centrifuge tube and add 100 mg of glass beads;
[0054] (2) Add 1.0 mL of acetonitrile:methanol:H2O mixed solution (2:2:1) and vortex for 30 s;
[0055] (3) Place the centrifuge tube in the 2 mL adapter provided with the instrument, immerse it in liquid nitrogen for 5 min, then thaw at room temperature, place the centrifuge tube in the 2 mL adapter again, and install it in the tissue grinder, grind at 60 Hz for 2 min, repeat twice;
[0056] (4) Take out the centrifuge tube, process the sample at 12000 rpm, 4°C for 10 min, collect the supernatant and concentrate and dry;
[0057] (5) Add 300 µL of 2-chloro-L-phenylalanine solution (2.5 μg / mL) prepared with acetonitrile:0.1% formic acid (1:9) to reconstitute the sample, filter it with a 0.22 μm membrane, and then perform liquid chromatography-mass spectrometry (LC-MS) detection.
[0058] 4. Metabolite LC-MS analysis
[0059] LC chromatography conditions: ACQUITY UPLC® HSS T3 (2.1 x 100 mm, 1.8 pm) column was used, the flow rate was 0.3 mL / min, the column temperature was 40 °C, and the injection volume was 2.0 pL. In positive mode, the mobile phase was 0.1% formic acid in acetonitrile (B1) and 0.1% formic acid in water (A1), and the gradient elution program was as follows: 0-1 min, 8% B1; 1-8 min, 8%-98% B1; 8-10 min, 98% B1; 10-10.1 min, 98%-8% B1; 10.1-12 min, 8% B1. In negative mode, the mobile phase was acetonitrile (B2) and 5 mM ammonium formate water (A2), and the gradient elution program was as follows: 0-1 min, 8% B2; 1-8 min, 8%-98% B2; 8-10 min, 98% B2; 10-10.1 min, 98%-8% B2; 10.1-12 min, 8% B2.
[0060] MS mass spectrometry conditions: The positive ion spray voltage was 3.50 kV, the negative ion spray voltage was -2.50 kV, the sheath gas was 40 arb, and the auxiliary gas was 10 arb. The capillary temperature was 325 °C, the full scan was performed at a resolution of 60,000, the ion scan range was m / z 100-1000, the secondary fragmentation was performed by high-energy collisional dissociation (HCD) at a collision energy of 30%, the secondary resolution was 15,000, the signals of the top 4 ions were collected for fragmentation, and unnecessary tandem mass spectrometry (MS / MS) information was removed by dynamic exclusion.
[0061] 5. Data preprocessing
[0062] (1) The LC-MS raw data was converted to mzXML format by the MSConvert tool in the Proteo Wizard software, and the peak identification, peak extraction, peak alignment and integration were processed by the XCMS toolkit based on R language. The Compound Discover 3.1 software was used to remove mass spectrum peaks with a relative standard deviation (RSD) <30%.
[0063] (2) The metabolites in the analyzed samples were identified and annotated by comparing the secondary mass spectrum with multiple databases such as KEGG database, LIPIDMaps database, HMDB database, MassBank database, MzClou database and self-built standard library.
[0064] (3) Multivariate statistical analysis: First, unsupervised principal component analysis (PCA) was used to observe the natural clustering trend and outliers among samples, and to evaluate the overall data quality. Then, supervised partial least squares discriminant analysis (PLS-DA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were used to maximize the differences between groups and to find the metabolite markers that best distinguished each group. To effectively avoid overfitting problems that may occur during modeling, we used the permutation test method, which was specifically performed 100 times (n = 100) to enhance the stability and reliability of the model.
[0065] 6. Metabolic marker screening
[0066] (1) t-test was performed on the data, and the variable importance projection (VIP) value and the adjusted significance probability value (P FDR ) of the independent variables between groups were calculated. Based on P FDR <0.05, VIP > 1, significant metabolites between the diabetic patients with osteoporosis (DOP) and the diabetic patients without osteoporosis (DM) were screened; secondly, significant metabolites between the primary osteoporosis patients and the healthy control group were screened. After excluding the intersection of the two groups of differential metabolites, 30 differential metabolites were finally obtained for the discovery group. These potential DOP-specific metabolic markers were included in the subsequent analysis and further screening.
[0067] (2) To reduce the data dimension, prevent model overfitting, and identify the most biologically meaningful characteristic biomarkers, this study used two feature selection methods: support vector machine-recursive feature elimination (SVM-RFE) and random forest (RF). Cross-validation was used to ensure the robustness of the feature set selected by the two methods. Feature selection first used SVM-RFE, which removed the least contributing features to the model through a recursive loop, thereby obtaining the optimal feature subset. We used linear kernel SVM as the base learner, and used ten-fold cross-validation recursive feature elimination (RFECV) to automatically determine the optimal number of features. When 20 features were included, the CV error value was the smallest, which was 0.00158. Secondly, we used the RF algorithm to evaluate the feature importance, and the number of decision trees (ntree) was set to 500, and the rest of the parameters used the default value. The optimal feature subset determined by SVM-RFE cross-validation and the 18 features selected by random forest that had a significant contribution to the model accuracy were retained. Finally, the 13 feature metabolites common to the above two sets were identified as the key biomarkers of DOP (see Table 2). Figures 2-3 ).
[0068] 7. ROC analysis
[0069] To further verify the specificity of the above 13 metabolites in the diagnosis of DOP, univariate receiver operating characteristic (ROC) curve analysis was performed on the markers, and the results are shown in Figures 4-16 : The area under the curve (AUC) of 12 of the 13 metabolites was greater than 0.80 (the AUC value of N-acetyl-L-lysine was less than 0.8, excluded), with diagnostic value. Among them, cytosine nucleotide (AUC = 0.98), ketoglutaric acid (AUC = 0.91), and triethylamine (AUC = 0.81) can effectively distinguish the case group from the control group, and show high accuracy and discrimination ability.
[0070] Example 2: Verification of plasma exosome differential metabolites between diabetic secondary osteoporosis patients and diabetic non-osteoporosis patients
[0071] 1. DOP-related exosome metabolite verification
[0072] This example contains 17 diabetic non-osteoporosis patients (control group) and 9 diabetic secondary osteoporosis patients (case group). Using the foregoing detection method and statistical analysis method, a stable feature spectrum containing three metabolites (cytosine nucleotide, ketoglutaric acid, and triethylamine) was identified and verified in plasma exosomes. The three metabolites reproduced significant case-control group differences and were consistent with the change trend of the case-control group in Example 1 (p<0.05) Figure 17 ), while the change trend of the remaining 9 metabolites was inconsistent with the change trend of the case-control group in Example 1 or the trend was consistent but the group difference was not significant, so the remaining 9 metabolites were excluded.
[0073] 2. ROC analysis
[0074] The ROC analysis results show that cytosine nucleotide, ketoglutaric acid, and triethylamine have good specificity and sensitivity in predicting DOP. As shown in Figure 18 , the area under the curve of each metabolite is greater than 0.8, and the area under the curve of the diagnostic model when used in combination is as high as 0.99. This shows that the above three metabolites can be used as specific metabolic markers for DOP and for the diagnosis of DOP.
[0075] Obviously, the above examples are merely examples for clarity and do not limit the embodiments. Based on the above description, other different forms of changes or variations can be made by those of ordinary skill in the art. Here, it is not necessary or possible to exhaust all embodiments. The obvious changes or variations derived therefrom are still within the protection scope of the present application.
Claims
1. Use of a reagent for detecting the content of a metabolic marker in the manufacture of a diagnostic kit for secondary osteoporosis in diabetes, characterized in that, The metabolic marker is selected from any one of the following components: (1) cytosine nucleotides; (2) triethylamine; (3) cytosine nucleotides, ketoglutarate and triethylamine.
2. Use according to claim 1, characterized in that, The detection sample of the diagnostic kit is an exosome.
3. A diagnostic kit for diabetic secondary osteoporosis, characterized by, The diagnostic kit contains reagents for detecting the content of metabolic markers, wherein the metabolic markers are the metabolic markers described in claim 1.
4. The diagnostic kit according to claim 3, characterized in that, The diagnostic kit further comprises reagents for extracting exosomes from plasma.
5. The diagnostic kit according to claim 3, characterized in that, The reagents include reagents for detecting the concentration or content of the metabolic markers in the test sample by nuclear magnetic resonance method, chromatography method, spectroscopy method, mass spectrometry method or combination thereof.
6. A reagent for detecting the concentration or content of a metabolic marker in a test sample, characterized by, The metabolic marker is the metabolic marker described in claim 1, and the test sample is an exosome.
7. Use of a metabolic marker for the manufacture of a diagnostic product for secondary osteoporosis in diabetes, characterized in that, The metabolic marker is the metabolic marker described in claim 1.
8. Use according to claim 7, characterized in that, The diagnostic product includes equipment for analyzing the content of the metabolic markers.
9. Use of metabolic markers in constructing a prediction model for diabetes secondary osteoporosis, characterized in that, The metabolic marker is the metabolic marker described in claim 1.
10. Use according to claim 9, characterized in that, The prediction model includes a step of analyzing the concentration level of the metabolic markers using one or more algorithms selected from the group consisting of orthogonal partial least squares discriminant analysis, partial least squares discriminant analysis, classification and logistic regression, k-nearest neighbor algorithm, naive Bayes, support vector machine, decision tree, random forest, regression tree, gradient boosting decision tree, extreme gradient boosting algorithm, light gradient boosting machine, gradient boosting machine, lasso algorithm and convolutional neural network.