Protein marker for diagnosing metabolic syndrome and application and risk assessment model thereof
By screening out protein markers related to metabolic syndrome, UBQLN1, SEZ6L2 and PKHD1, combined with proteomics and machine learning, urine detection methods and risk assessment models are developed, which solves the problem that traditional diagnostic methods cannot achieve early sensitive screening, and improves the problem that traditional diagnostic methods cannot achieve early sensitivity screening. The diagnostic ability and preventive effect of metabolic syndrome are achieved.
Patent Information
- Application Number
- CN202410991167.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-12-03
AI Technical Summary
The diagnosis of traditional metabolic syndrome mainly relies on clinical physical examinations and biochemical examinations, and cannot achieve early sensitive screening and accurate diagnosis, resulting in an increased risk of chronic metabolic diseases.
By studying protein markers related to metabolic syndrome, combining proteomics and machine learning technology, the three protein markers, UBQLN1, SEZ6L2 and PKHD1, were screened out, and non-invasive detection methods and risk assessment models were developed for urine. Realize early diagnosis of metabolic syndrome.
It improves the early diagnosis ability and efficiency of metabolic syndrome, can delay disease progression, reduce the risk of complications of diabetes and cardiovascular disease, and provides convenient and non-invasive methods through urine testing.
Smart Images

Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] The present invention relates to the field of biomedical technology, and in particular to a protein marker for diagnosing metabolic syndrome and its application and risk assessment model. Background Art
[0002] Metabolic syndrome (MetS) is a pathological state in which multiple metabolic components are abnormally aggregated. The Chinese Medical Association Diabetes Branch defines the following four items as the diagnostic criteria for MetS: ① Overweight and / or obesity, BMI ≥ 25.0 kg / m 2 ; ② For patients with hyperglycemia and / or diagnosed diabetes who have been treated, FPG ≥ 6.1 mmol / L and / or 2hPG ≥ 7.8 mmol / L; ③ For patients with hypertension and / or diagnosed hypertension who have been treated, SBP / DBP ≥ 140 / 90 mmHg; ④ For patients with dyslipidemia, fasting blood TG ≥ 1.7 mmol / L and / or fasting blood HDL-c < 0.9 mmol / L for men and < 1.0 mmol / L for women. MetS increases the risk of chronic metabolic diseases such as cardiovascular disease and type 2 diabetes, which are the leading causes of death worldwide. However, the diagnosis of traditional MetS is mainly based on clinical physical examination and biochemical examination, which cannot achieve early sensitive screening and accurate diagnosis.
[0003] Early diagnosis of MetS is an important measure to prevent the occurrence of chronic non-communicable diseases. Summary of the invention
[0004] Based on the role of protein in the occurrence and development of metabolic syndrome, the present invention studies protein markers related to the occurrence and development of metabolic syndrome, thereby providing a new means for the early diagnosis of metabolic syndrome.
[0005] In one aspect, the present invention provides protein markers for diagnosing metabolic syndrome, wherein the protein markers are a combination of UBQLN1, SEZ6L2 and PKHD1.
[0006] Among them, compared with healthy controls, the expression levels of UBQLN1 and SEZ6L2 were downregulated in patients with metabolic syndrome, and the expression level of PKHD1 was upregulated in patients with metabolic syndrome.
[0007] Among them, proteomics, as an emerging diagnostic tool, provides new possibilities for the early diagnosis of MetS. Proteomics refers to the study of all proteins expressed in the genome and their characteristics, mainly including protein structure, protein abundance, protein interactions, etc. Machine learning can be applied to a variety of proteomics studies such as processing proteomics mass spectrometry data, screening protein biomarkers, and predicting protein interactions. Machine learning algorithms can accurately identify biomarkers associated with specific diseases or their subtypes through in-depth analysis of proteomics data, and help scientists determine which proteins or protein combinations are highly relevant in the diagnosis and prediction of specific diseases by building models. In the existing technology, protein markers can be quickly obtained using machine learning algorithms. Proteomics combined with machine learning helps researchers find more accurate and efficient protein markers, and establish diagnostic models to provide new ideas and methods for the timely detection and management of MetS.
[0008] The source of protein is urine, specifically fasting urine (which can be measured according to fasting blood sugar), and more specifically, fasting midstream urine. Urine is a non-invasive diagnostic tool that can provide useful information and is easy to perform in outpatient clinics and hospitals.
[0009] Urine samples are processed as follows: Centrifuge at 2000g for 10 min at 4°C; aspirate 2 mL of the supernatant and dispense it into 2 mL sterile centrifuge tubes, and quickly freeze the urine samples in a -20°C refrigerator.
[0010] Lysis buffer was added to a sterile centrifuge tube to a final concentration of 1% SDC / 100mM Tris-HCl. After thorough mixing, the protein concentration was determined using the BCA method. An equal amount of protein was taken and all samples were filled to the same volume using 1% SDC / 100mM Tris-HCl solution.
[0011] After adding TCEP and CAA, the mixture was incubated at 60°C for 30 min to complete the reduction and alkylation.
[0012] An equal volume of ddH2O was added to dilute the SDC concentration to below 0.5%, and trypsin was added at a mass ratio of 1:50 between enzyme and protein. The cells were incubated at 37°C with shaking overnight for enzyme digestion.
[0013] The next day, TFA was added to terminate the enzyme cleavage reaction, the sample was centrifuged at 12000g, the supernatant was taken and desalted using a homemade SDB desalting column, vacuum-dried and frozen at -20°C.
[0014] Among them, in the database retrieval process of protein marker screening: the MS1 and MS2 mass tolerance were set to 15ppm; the FDR was set to 1%; in the data cleaning, filtering, conversion and filling processes of protein marker screening: proteins with a missing ratio of more than 80% in samples or a missing ratio of more than 50% in both groups were filtered and logarithmically transformed; a new distribution with a downward shift of 1.8 standard deviations and a width of 0.25 standard deviations was created; in the differential expression analysis process of protein marker screening: the screening criteria were |log2FC|≥0.58, p<0.05.
[0015] On the other hand, the present invention discloses the use of the aforementioned protein markers in a diagnostic kit for metabolic syndrome.
[0016] The kit includes a detection reagent capable of detecting the expression levels of UBQLN1, SEZ6L2 and PKHD1. The detection reagent includes a reagent for detecting protein levels by immunoblotting, enzyme-linked immunosorbent assay, mass spectrometry, radioimmunoassay, radioimmunodiffusion, immunoelectrophoresis, tissue immunostaining, immunoprecipitation analysis, complement fixation analysis, fluorescence activated cell sorting, mass analysis or protein microarray methods. Specifically, the detection reagent is a binding agent that specifically binds to proteins encoded by UBQLN1, SEZ6L2 and PKHD1. More specifically, the protein binding agent includes a peptide, a peptide mimetic, an aptamer, a spiegelmer, a darpin, an ankyrin repeat protein, a Kunitz-type domain, an antibody, a single domain antibody and / or a monovalent antibody fragment, etc.
[0017] On the other hand, the present invention discloses a metabolic syndrome risk assessment model, which uses the expression levels of protein markers UBQLN1, SEZ6L2 and PKHD1 as variables, and uses a nomogram to visualize the model results. Each protein marker corresponds to a corresponding score, and the individual scores corresponding to each protein marker under different values are added together to obtain a total score, which is then corresponded to the probability of a metabolic syndrome event.
[0018] Specifically, in the nomogram, the score range of UBQLN1 is 12.5-7.5, corresponding to 0-31 in the score scale; the score range of SEZ6L2 is 12.4-9.8, corresponding to 0-100 in the score scale; the score range of PKHD1 is 8-16, corresponding to 0-45 in the score scale; the probability of occurrence of metabolic syndrome in the population is 0.01, 0.1, 0.3, 0.5, 0.7 and 0.9, corresponding to 48, 66, 76, 82, 88 and 98 in the total score scale respectively.
[0019] The present invention can achieve early diagnosis of metabolic syndrome by detecting the expression levels of UBQLN1, SEZ6L2 and PKHD1, improve detection capability and efficiency, and actively take intervention measures, which can not only delay the progression of metabolic syndrome, but also reduce the risk of complications such as diabetes and cardiovascular diseases. In addition, the use of urine for non-invasive detection is more convenient and easier to promote. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 Build a flow chart for the model; Figure 2 Quantitative protein distribution diagrams for metabolic syndrome group and healthy control group; Figure 3 The quantitative distribution diagram of peptide segments was shown for the metabolic syndrome group and the healthy control group; Figure 4 These are the differentially expressed proteins screened between the metabolic syndrome group and the healthy control group; Figure 5 is the principal component analysis diagram; Figure 6 It is a quantitative heat map of differentially expressed proteins; Figure 7 is the relationship between the fitting error and log(λ); Figure 8 is the LASSO coefficient distribution diagram of differentially expressed proteins; Fig. 9 It is a box plot of protein marker UBQLN1 protein expression; Fig.10 It is a box plot of protein marker SEZ6L2 protein expression; Fig.11 It is a box plot of protein marker PKHD1 protein expression; Fig.12 Receiver operating characteristic curve (ROC) graph of the diagnostic model; Fig.13 This is the confusion matrix diagram of the diagnostic model; Fig.14 It is the calibration curve diagram of the diagnostic model; Fig.15 It is the decision curve (DCA) diagram of the diagnostic model; Fig.16 A nomogram for predicting the risk of metabolic syndrome. DETAILED DESCRIPTION
[0021] In the present invention, the term "metabolic syndrome" refers to a pathological state in which multiple metabolic components are abnormally aggregated, and refers to those who meet three or all of the following four items: ① Overweight and / or obese, BMI ≥ 25.0 kg / m 2 ; ② Patients with high blood sugar and / or diagnosed diabetes who have been treated, FPG ≥ 6.1mmol / L and / or 2hPG ≥ 7.8mmol / L; ③ Hypertension and / or hypertension diagnosed and treated, SBP / DBP ≥ 140 / 90 mmHg; ④ Dyslipidemia, fasting blood TG ≥ 1.7mmol / L, and (or) fasting blood HDL-c < 0.9mmol / L for men and < 1.0mmol / L for women.
[0022] The term "protein marker" broadly refers to any detectable compound present in or derived from a biological sample, such as a protein, peptide, proteoglycan, glycoprotein, lipoprotein, etc.
[0023] In the present invention, protein markers include UBQLN1 (UniprotID: Q9UMX0), SEZ6L2 (UniprotID: Q6UXD5) and PKHD1 (UniprotID: P08F94). UniprotID can be obtained at https: / / www.uniprot.org / .
[0024] In the prior art, receiver operating characteristic (ROC) curves, confusion matrices, calibration curves and decision curves are often used to evaluate diagnostic models. For the receiver operating characteristic curve, the horizontal axis is the false positive rate (1-specificity) and the vertical axis is the true positive rate (sensitivity). Each point on the ROC line graph represents a sensitivity / 1-specificity pair corresponding to a specific decision threshold. A test with perfect discrimination (no overlap between the two result distributions) has an ROC line graph through the upper left corner, where the true positive score is 1.0, or 100% (perfect sensitivity), and the false positive score is 0 (perfect specificity). The theoretical line graph of a test that does not distinguish (the result distributions of the two groups are the same) is a 45° diagonal line from the lower left corner to the upper right corner. Most line graphs fall between these two extremes. Qualitatively, the closer the curve is to the upper left corner, the larger the area under the curve (Area Under Receiver Operating Characteristic Curve, AUC), and the higher the overall accuracy of the model. It is generally believed that when 0.90≤ AUC <1.00, the model has excellent distinguishing ability; 0.75≤ AUC<0.90, the model has good distinguishing ability; 0.60≤ AUC<0.75, the model has certain distinguishing ability, but it is not recommended; AUC<0.60, the model has poor distinguishing ability.
[0025] The confusion matrix, also known as the error matrix, is a standard format for expressing accuracy evaluation. It is commonly expressed in a 2-row, 2-column matrix. Each row of the confusion matrix represents the predicted category, and the total number of each row represents the number of data predicted to be in that category; each column represents the true category of the data, and the total number of data in each column represents the number of data instances in that category. So after obtaining the confusion matrix of the model, you need to see how many observations are in the corresponding positions of the second and fourth quadrants. The more values here, the better; conversely, the fewer observations appearing in the corresponding positions of the first and third quadrants, the better.
[0026] Calibration curves are often used to evaluate the consistency between the model's predicted probability and the actual probability of occurrence, also known as goodness of fit. Generally, the categories directly determined by the model (such as whether or not a person is ill) are called relative risk predictions; if it is assumed that everyone has a risk of illness, the difference is that the risk of illness is different, then the specific value of the risk of illness determined by the model, that is, the model's predicted probability, is called an absolute risk prediction. Calibration is an evaluation indicator for evaluating the accuracy of the model's absolute risk prediction. At present, the Hosmer-Lemeshow goodness of fit test method is commonly used to evaluate the calibration of binary classification models. The basic idea of this method is to divide the observed samples into several groups according to the size of the predicted probability, and then calculate the observed frequency (the number of people who are actually ill in the group) and the predicted frequency (the number of people in the group predicted by the model to be ill) of each group, and finally calculate a fitting straight line, and use the fitting straight line to test the difference between the observed frequency and the predicted frequency. The calibration curve is the visualization of the results of the Hosmer-Lemeshow test. The higher the consistency between the model's predicted probability and the actual observation results, the better the calibration of the model.
[0027] The decision curve is a method used to evaluate the clinical application value of a prediction model. It draws a decision curve to compare the clinical benefits of different prediction models at different treatment decision thresholds, helping medical researchers and clinicians make the best choice between different prediction models to achieve better clinical prediction results. In the figure, the probability threshold is the horizontal axis and the net benefit rate is the vertical axis. The red line is the decision curve, which is the change in the net benefit rate brought about by the intervention based on the prediction results of the logistic model with the probability threshold, while the bottom black horizontal line None represents the net benefit rate of 0 when no treatment is performed, and the gray curve All represents the change in the net benefit rate of all treatments with the probability threshold. Within a certain range, the higher the net benefit rate of the model, the greater its clinical utility.
[0028] The screening and verification processes of protein markers are described below with reference to the accompanying drawings.
[0029] 1. Sample collection and storage Urine samples were collected from 15 patients with metabolic syndrome (MetS patient group) and 15 healthy controls of behavioral population (healthy control group). The basic information of the patients and healthy controls and related laboratory test indicators were recorded in detail when collecting samples, including age, gender, height, weight, BMI, triglycerides, high-density lipoprotein, cholesterol, fasting blood sugar, 2h blood sugar, blood pressure and medication history, etc. There was no statistical difference in age, gender and height between the MetS patient group and the healthy control group.
[0030] Inclusion criteria for the MetS patient group: 1. Adult patients aged 18 years and above.
[0031] 2. According to the results of the physical examination, relevant indicators are obtained and meet the diagnostic criteria for MetS - those who meet 3 or all of the following 4 items: ① Overweight and / or obese, BMI ≥ 25.0 kg / m 2 ; ② Patients with hyperglycemia and / or diagnosed diabetes and treated, FPG ≥ 6.1 mmol / L and / or 2hPG ≥ 7.8 mmol / L; ③ Patients with hypertension and / or diagnosed hypertension and treated, SBP / DBP ≥ 140 / 90 mmHg; ④ Dyslipidemia, fasting blood TG ≥ 1.7 mmol / L and / or fasting blood HDL-c < 0.9 mmol / L for men and < 1.0 mmol / L for women.
[0032] 3. The patient or legal representative must sign a complete informed consent form before participating in the study.
[0033] Exclusion criteria for MetS patient group: 1. History of serious illness, such as cancer, autoimmune disease or other serious illness.
[0034] 2. Renal function impairment, positive urine protein or albumin / creatinine ratio <30 mg / g.
[0035] 3. Women who are pregnant or breastfeeding.
[0036] Inclusion criteria for healthy control group: 1. Adult patients aged 18 years and above.
[0037] 2. Those who do not meet the MetS criteria based on relevant indicators obtained based on the results of the physical examination.
[0038] 3. The patient or legal representative must sign a complete informed consent form before participating in the study.
[0039] Exclusion criteria for healthy control group: 1. History of serious illness, such as cancer, autoimmune disease or other serious illness.
[0040] 2. Renal function impairment, positive urine protein or albumin / creatinine ratio <30 mg / g.
[0041] 3. Women who are pregnant or breastfeeding.
[0042] Urine pretreatment methods are: Take 10 mL of fasting midstream urine and centrifuge it at 2000 g for 10 min at 4°C; take 2 mL of the supernatant and divide it into 2 mL sterile centrifuge tubes, and quickly freeze the divided urine samples in a -20°C refrigerator for storage.
[0043] 2. Experimental methods This patent uses the next generation of label-free quantitative proteomics technology to complete the analysis. In the data independent acquisition (DIA) mode, the latest high-resolution mass spectrometry is used to simultaneously acquire peptide ion characteristics in terms of mass number and retention time.
[0044] The detection and analysis process steps are as follows: 1) Sample preparation Lysis buffer was added to the sample to a final concentration of 1% SDC / 100mM Tris-HCl (pH=8.5), and the protein concentration was determined by the BCA method after thorough mixing. Take an equal amount of protein and use 1% SDC / 100mM Tris-HCl (pH=8.5) solution to make all samples equal in volume. After adding TCEP and CAA, incubate at 60°C for 30 min to complete reduction and alkylation. Add an equal volume of ddH2O to dilute the SDC concentration to less than 0.5%, add trypsin at a mass ratio of 1:50 between enzyme and protein, and incubate at 37°C with shaking overnight for enzyme digestion. On the second day, TFA was added to terminate the enzyme digestion reaction, the sample was centrifuged at 12000g, the supernatant was taken and desalted with a homemade SDB desalting column, vacuum-dried and frozen at -20°C.
[0045] Reagent Description: Tris-HCl: tris(hydroxymethyl)aminomethane hydrochloride; TCEP: tris(2-carbonylethyl)phosphine hydrochloride; CAA: calcium acetylacetonate; ddH2O: double deionized water; SDC: sodium deoxycholate; TFA: trifluoroacetic acid; SDB: Sulfonated styrene-vinylbenzene copolymer.
[0046] 2) Mass spectrometry The mass spectrometry of the samples was detected by UltiMate 3000 RSLCnano nanoliter liquid phase (Thermo) in series with timsTOFPro mass spectrometer (Bruker). The peptide samples were injected by autosampler, bound to C18 trapping column (75µm*2cm, 3µmparticle size, 100 Å pore size, Thermo), and then separated by analytical column (75µm*15cm,1.7µmparticle size, 100 Å pore size, IonOpticks). The mobile phase A (0.1% formic acid) and mobile phase B (0.1% formic acid in ACN) were used to establish the analytical gradient mass spectrometry and the data were collected in diaPASEF mode, with the flow rate set to 300nL / min. The capillary voltage was set to 1500 V. The scanning range of MS1 and MS2 spectra was set to 100-1700 m / z. The ion mobility range was set to 0.6-1.6Vs / cm 2 The accumulation time and ramp time were set to 50 ms. The diaPASEF acquisition window was set using the timsControl software according to the distribution law of mass-to-charge ratio-ion mobility. The collision energy was set from 1 / K0=1.6 Vs / cm 2 The 59eV is linearly reduced to 1 / K0=0.6Vs / cm 2 of 20eV.
[0047] 3) Data analysis 3.1 Database search DIA raw data were analyzed using the library-free mode of DIA-NN (1.8.1) software. The database used for the search was the HUMAN protein sequence database downloaded from Uniprot. The search parameters were mainly set by default, and the key parameters are described as follows: the Precursor ion generation option was enabled for theoretical spectrum library prediction; Trypsin / P was used, and a maximum of 1 missed cleavage site was allowed; Carbamidomethyl (C) modification was set as fixed modification; Oxidaton (M) and Acetylation (protein N-terminal) were set as variable modifications; MS1 and MS2 mass tolerances were set to 15ppm; MBR was enabled; Heuristic protein inference was enabled; FDR was set to 1%. Protein quantitative information was normalized using the MaxLFQ algorithm.
[0048] 3.2 Data cleaning, filtering, transformation and filling Contaminant proteins in the quantitative protein matrix were removed, and proteins with a missing ratio of more than 80% in samples or more than 50% in both groups were filtered and logarithmically transformed. Then, the missing values were filled with values representing a normal distribution near the detection limit of the mass spectrometer. To do this, we determined the mean and standard deviation of the actual intensity distribution and then created a new distribution with a downward shift of 1.8 standard deviations and a width of 0.25 standard deviations. These values were used to impute the missing values in the protein matrix, and the remaining proteins were used for subsequent analysis.
[0049] 3.3 Differential expression analysis The protein quantitative data of the MetS patient group and the healthy control group were compared, and the differentially expressed proteins between the two groups were screened using the fold change (FC) and difference test (two-sided P < 0.05) as the criteria. The screening criteria were |log2FC| ≥ 0.58, p < 0.05. Figure 4 shown.
[0050] 3.4 Key protein screening The total samples were randomly divided into training set and test set in a ratio of 7:3. In the training set, the Lasso regression model was used for five-fold cross validation to screen out N differential protein features; then, the N proteins were randomly combined to perform Logistic regression analysis to construct a diagnostic model, and the optimal diagnostic model was selected by the discrimination index AUC and calibration index Brill score calculated by three-fold cross validation to select the optimal protein marker combination. The differentially expressed proteins in the optimal combination are the disease-related protein markers.
[0051] 3.5 Model construction and evaluation Logistic regression and cross-validation were used to randomly select protein combinations from protein markers in the training set to construct multiple prediction models, and the optimal model was selected based on the area under the receiver operating characteristic curve (AUC) and the Brill score. The optimal prediction model was then evaluated. The discrimination, calibration, and clinical utility performance of the model in the training set and the test set were evaluated by the receiver operating characteristic (ROC) curve, calibration curve, and decision curve in the test set and the training set, respectively. It is generally believed that when 0.90≤AUC<1.00, the model has excellent discrimination ability; when 0.75≤AUC<0.90, the model has good discrimination ability; when 0.60≤AUC<0.75, the model has certain discrimination ability, but it is not recommended; when AUC<0.60, the model has poor discrimination ability. All statistical analyses were performed using R (version 4.3.2) and Python (version 3.10.12).
[0052] 3. Results and analysis 1) From the physical examination population, 15 pairs of information-matched MetS patients and healthy controls were selected for proteomic study, and a diagnostic model was established. Figure 1 A flowchart describing the construction of the MetS diagnostic model is shown.
[0053] 2) The demographic and clinical characteristics of the study population were summarized, and the results are shown in Table 1.
[0054] Table 1 variable total (n =30) MetS Patient group (n = 15) Healthy control group (n= 15) Statistics P Gender, n(%) - 0.143 female 15 (50.00) 10 (66.67) 5 (33.33) male 15 (50.00) 5 (33.33) 10 (66.67) <![CDATA[Age, M (Q1, Q3)]]> 49.43 ± 12.92 50.27 ± 12.33 48.60 ± 13.86 t=0.35 0.730 Height (cm), Mean ± SD 164.00 (160.00,172.25) 164.00 (159.00, 173.00) 161.00 (162.50, 170.00) Z=-0.89 0.371 Body weight (kg), Mean ± SD 68.15 ± 12.88 77.16 ± 10.14 59.13 ± 8.16 t=5.36 <.001 BMI (kg / m^2), Mean ±SD 24.98 (21.66,27.20) 27.33 (25.45, 28.70) 21.50 (19.70, 22.98) Z=-4.65 <.001 <![CDATA[Triglyceride (mmol / L), M(Q1, Q3)]]> 1.54 (0.75, 2.22) 2.19 (1.58, 3.37) 0.72 (0.60, 1.14) Z=-3.86 <.001 <![CDATA[Systolic blood pressure (mmHg), M (Q1,Q3)]]> 123.50 (116.50,135.50) 137.00 (126.00, 145.00) 117.00 (110.00, 125.00) Z=-3.10 0.002 Diastolic blood pressure (mmHg), Mean± SD 79.21 ± 14.43 88.36 ± 13.68 71.46 ± 10.06 t=3.48 0.002 <![CDATA[HDL-c(mmol / L), M(Q1, Q3)]]> 1.14 (0.90, 1.43) 0.92 (0.90, 1.08) 1.45 (1.30, 1.83) Z=-3.63 <.001 <![CDATA[Fasting blood glucose (mmol / L), M(Q1, Q3)]]> 5.06 (4.87, 5.42) 5.33 (5.14, 6.03) 4.88 (4.72, 4.96) Z=-3.06 0.002 Wherein, (1) t: t-test, Z: Mann-Whitney test; (2) SD: standard deviation; M: median; Q1: first quartile; Q3: third quartile.
[0055] 3) Mass spectrometry quality control information (1) Using data independent acquisition (DIA) technology, we obtained urine proteome profiles of 15 MetS patients and 15 healthy controls. Using DIA technology, a total of 2255 proteins and 124,841 peptides were quantified in the entire cohort. Figure 2 and Figure 3 shown.
[0056] (2) The expression of 77 proteins in the MetS patient group was significantly different from that in the healthy control group. Compared with the healthy control group, the expression of 44 proteins in the MetS patient group was upregulated and the expression of 33 proteins was downregulated. Figure 4 shown.
[0057] (3) The principal component analysis (PCA) results showed the intensities of 77 differentially expressed proteins, indicating significant differences between the MetS patient group and the healthy control group. Figure 5 shown.
[0058] (4) A heat map was created using the expression levels of 77 differentially expressed proteins, showing that there were significant differences in the expression levels of differentially expressed proteins between the MetS patient group and the healthy control group. Figure 6 shown.
[0059] 4) Protein screening The total samples were randomly divided into training set and test set in a ratio of 7:3. In the training set, 13 important proteins were screened out from 77 differentially expressed proteins using Lasso regression model for five-fold cross validation. These 13 important proteins were randomly combined for Logistic regression analysis to construct a diagnostic model, and the optimal diagnostic model was selected by the discrimination index AUC and calibration index Brier score calculated by three-fold cross validation. The screening results are shown in Figure 2. Figure 7 and Figure 8Finally, UBQLN1, SEZ6L2 and PKHD1 were selected as the best protein marker combination, and the clinical risk model was established. The results of the logistic regression model are shown in Table 2.
[0060] Table 2
[0061] The expression levels of the three protein markers UBQLN1, SEZ6L2 and PKHD1 in the two groups of people were plotted as box plots to visualize the differences in the expression of these three protein markers in the urine of the MetS patient group and the healthy control group. The box plots showed that compared with the healthy control group, the expression levels of UBQLN1 and SEZ6L2 in the MetS patient group were downregulated, and the expression level of PKHD1 in the MetS patient group was upregulated. Figure 9-11 shown.
[0062] 4. Model Evaluation When evaluating the diagnostic model composed of protein markers UBQLN1, SEZ6L2, and PKHD1, 30 samples were divided into two sets according to the ratio, of which 70% of the data was used as a training set (n=20) and 30% of the data was used as a test set (n=10). The training set and the test set were used for model evaluation.
[0063] 1) Receiver operating characteristic curve (ROC) plot The results are as follows Fig.12 As shown in the figure, the ROC curve is used to evaluate the discrimination of the diagnostic model. The area under the ROC curve (AUC) is generally used to evaluate the model prediction performance. Evaluation criteria: AUC ≥ 0.90, the model has excellent discrimination ability; 0.90> AUC ≥ 0.75, the model has good discrimination ability; 0.75> AUC ≥ 0.60, the model has a certain discrimination ability, but it is not recommended; AUC < 0.60, the model has poor discrimination ability. The ROC curve shows that the AUC (95% CI) of the training set is 0.940 (0.841-1), and the AUC (95% CI) of the test set is 0.920 (0.736-1). The ROC curve shows that the MetS diagnostic model has good discrimination ability.
[0064] 2) Confusion Matrix The results are as follows Fig.13 As shown in the figure, the confusion matrix diagram shows the classification of the test set by the diagnostic model. The horizontal axis represents the actual category, and the vertical axis represents the predicted category. The people in the upper left quadrant and the lower right quadrant are classified into the correct category. The confusion matrix diagram results show that the model accuracy is 83.33% (25 / 30), the sensitivity is 80% (12 / 15), and the specificity is 86.67% (13 / 15), indicating that the diagnostic model has good predictive ability.
[0065] 3) Calibration curve The results are as follows Fig.14 As shown in the figure, the calibration curve is used to evaluate the calibration of the diagnostic model, that is, the consistency between the predicted probability of the outcome and the actual observed frequency of occurrence. The dotted line is the ideal curve, the red line is the actual curve of the model, and the blue line is the curve after resampling calibration. The closer the calibration curve is to the ideal curve, the better the model effect. The results of the calibration curve show that the predicted probability and observed probability of the MetS diagnostic model are consistent.
[0066] 4) Decision curve The results are as follows Fig.15 As shown in the figure, the decision curve is used to evaluate the clinical utility of the model. The net rate of return reflects the clinical value of the model. The larger the net rate of return, the better the model effect. In the decision curve, the black horizontal line indicates that all people are not treated, and the net benefit is 0; the gray oblique line indicates that all people are treated, and the net benefit is a reverse oblique line with a negative slope; the red curve is the net benefit curve of the current diagnostic model. The higher the net rate of return of the model, the greater its clinical utility. The decision curve shows that within a certain range, the net rate of return of the diagnostic model is higher, and is higher than the two options of all treatment and no treatment, proving that the clinical utility of the diagnostic model of MetS is greater.
[0067] 5. Model Visualization The results are as follows Fig.16 As shown in the figure, the model results are visualized using a nomogram to facilitate the prediction of disease risk. The individual scores (Points) corresponding to each protein marker at different values are added together to obtain the total score (Total Points), which is then converted to the Risk score, which is the probability of healthy people developing metabolic syndrome events. Specifically, in the nomogram, the score range of UBQLN1 is 12.5-7.5, corresponding to 0-31 in the score scale; the score range of SEZ6L2 is 12.4-9.8, corresponding to 0-100 in the score scale; the score range of PKHD1 is 8-16, corresponding to 0-45 in the score scale; the probability of occurrence of metabolic syndrome in the population is 0.01, 0.1, 0.3, 0.5, 0.7 and 0.9, corresponding to 48, 66, 76, 82, 88 and 98 in the total score scale, respectively.
[0068] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims
1. A protein marker for diagnosing metabolic syndrome, characterized in that: The protein markers are a combination of UBQLN1, SEZ6L2 and PKHD1.
2. The protein marker according to claim 1, characterized in that Compared with healthy controls, the expression levels of UBQLN1 and SEZ6L2 were downregulated in patients with metabolic syndrome, and the expression level of PKHD1 was upregulated in patients with metabolic syndrome.
3. The protein marker according to claim 1, characterized in that The source of protein is urine.
4. The protein marker according to claim 1, characterized in that The source of protein is fasting urine.
5. The protein marker according to claim 1, characterized in that Urine samples are processed as follows: Centrifuge at 2000g for 10 min at 4°C; aspirate 2 mL of the supernatant and dispense it into a 2 mL sterile centrifuge tube, and quickly freeze the urine sample in a -20°C refrigerator; Add lysis buffer to a sterile centrifuge tube to a final concentration of 1% SDC / 100 mM Tris-HCl. Mix thoroughly and determine the protein concentration using the BCA method. Take an equal amount of protein and use 1% SDC / 100 mM Tris-HCl solution to make all samples the same volume. After adding TCEP and CAA, the mixture was incubated at 60 °C for 30 min to complete the reduction and alkylation; Add an equal volume of ddH2O to dilute the SDC concentration to less than 0.5%, add trypsin at a mass ratio of 1:50 between enzyme and protein, and incubate at 37°C with shaking overnight for enzyme digestion; The next day, TFA was added to terminate the enzyme cleavage reaction, the sample was centrifuged at 12000g, the supernatant was taken and desalted using a homemade SDB desalting column, vacuum-dried and frozen at -20°C.
6. The protein marker according to claim 1, characterized in that In the database search process for protein marker screening: MS1 and MS2 mass tolerances were set to 15 ppm; FDR was set to 1%; In the data cleaning, filtering, transformation and filling steps of protein marker screening: proteins with a missing ratio of more than 80% or more than 50% in both groups were filtered and logarithmically transformed; a new distribution with a downward shift of 1.8 standard deviations and a width of 0.25 standard deviations was created; In the differential expression analysis process of protein marker screening: the screening criteria are |log2FC|≥0.58, p<0.
05.
7. Use of the protein marker as claimed in claim 1 in a diagnostic kit for metabolic syndrome.
8. The use according to claim 7, characterized in that: The diagnostic kit includes detection reagents capable of detecting the expression levels of UBQLN1, SEZ6L2 and PKHD1.
9. A metabolic syndrome risk assessment model, characterized in that: The risk assessment model uses the expression level of the protein marker described in claim 1 as a variable, and uses a nomogram to visualize the model results. The individual scores corresponding to each protein marker at different values are added together to obtain a total score, which is then corresponding to the probability of healthy people developing metabolic syndrome events.
10. The evaluation model according to claim 9, characterized in that In the nomogram, the score range of UBQLN1 is 12.5-7.5, corresponding to the score scale of 0-31; the score range of SEZ6L2 is 12.4-9.8, corresponding to the score scale of 0-100; the score range of PKHD1 is 8-16, corresponding to the score scale of 0-45; the probability of occurrence of metabolic syndrome in the population is 0.01, 0.1, 0.3, 0.5, 0.7 and 0.9, corresponding to the total score scale of 48, 66, 76, 82, 88 and 98 respectively.
