Metabolic biomarker for predicting cardiovascular disease risk of senile subclinical hypothyroidism patient and application of metabolic biomarker

Abstract

The invention relates to a metabolic biomarker for predicting the cardiovascular disease risk of a patient with senile subclinical hypothyroidism and application of the metabolic biomarker. The marker comprises inosine, arachidonic acid, arginine, phenylalanine and glycocholic acid. According to the method, metabolite information of each plasma sample in a research queue is obtained by establishing an elderly SCH and concurrent CVD crowd queue and detecting by adopting a high performance liquid chromatography-mass spectrometry technology, and a targeted metabolic content map is established. According to the matrix expression of the metabolic marker, in combination with clinical follow-up visit data, a model is constructed by using machine learning Lasso-cox, and a clinical prediction model of the CVD risk of the elderly SCH patient is constructed to guide precise diagnosis and treatment of the elderly SCH. The invention further provides a kit containing the diagnostic marker suitable for predicting the CVD risk of the elderly SCH, and the kit can be used for predicting the CVD risk of the elderly SCH.

Description

Technical Field

[0001] The present invention belongs to the technical field of clinical laboratory diagnosis, and specifically relates to metabolic biomarkers and applications thereof for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism. Background Art

[0002] According to statistics, more than half of the elderly suffer from thyroid diseases, among which subclinical hypothyroidism (SCH) has become the most common type of thyroid disease in the elderly population due to its high incidence. SCH is characterized by thyroid stimulating hormone (TSH) levels exceeding the upper limit of the normal range, while free thyroxine (FT4) remains in the normal range. This special physiological state makes the management and treatment of elderly patients particularly complicated.

[0003] With the increase of age, the physiological function of the hypothalamus-pituitary-thyroid axis naturally declines, resulting in significant differences in thyroid function between the elderly and young adults, further increasing the difficulty of diagnosis and treatment of SCH. It is particularly noteworthy that there is a close and complex relationship between SCH and cardiovascular disease (CVD). CVD is not only the main complication of elderly SCH patients, but also the main cause of their reduced quality of life and even death. Therefore, for elderly SCH patients, accurately assessing their CVD risk and taking timely intervention measures are of great significance to improving patient prognosis.

[0004] However, in current clinical practice, risk stratification based solely on TSH levels and related traditional risk factors is insufficient, and there is an urgent need to explore new and efficient biological markers to accurately predict the CVD risk of elderly SCH patients. Metabolomics, a research field that has emerged in recent years, provides a powerful tool for analyzing disease mechanisms and mining biomarkers by comprehensively analyzing the changes in small molecule metabolites in organisms. This technology can capture subtle changes in metabolic pathways in organisms under disease states, providing a new perspective for understanding the relationship between SCH and CVD.

[0005] In this context, the introduction of machine learning technology has made it possible to deeply mine metabolomics data. In particular, the Lasso method, as an advanced feature selection technology, can effectively screen out the most predictive features in high-dimensional data, while dealing with the problem of complex collinearity in the data and improving the stability and accuracy of the model. By combining metabolomics with the Lasso machine learning algorithm, this study aims to screen out key metabolic biomarkers from complex metabolite data that can specifically reflect the risk of CVD in elderly SCH patients.

[0006] Specifically, this study will first use metabolomics technology to comprehensively analyze biological samples of elderly SCH patients and screen out potential metabolic biomarkers; then, the Lasso-Cox method will be used to reduce the dimensionality of these markers, build a predictive risk assessment model, and calculate the individualized risk score for each patient. This innovative strategy is not only expected to provide more accurate tools for CVD risk management in elderly SCH patients, but also open up new paths for personalized medicine and precision health management in the future. Summary of the invention

[0007] Technical problem to be solved: In view of the fact that there is currently no clear and effective biological marker to predict the risk of CVD in elderly SCH and a lack of effective liquid diagnostic methods, the present invention provides a metabolic biomarker for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism and its application.

[0008] Technical solution: Metabolic biomarkers for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism, the markers including inosine, arachidonic acid, arginine, phenylalanine and glycocholic acid.

[0009] Preferably, the above markers consist of inosine, arachidonic acid, arginine, phenylalanine and glycocholic acid.

[0010] A method for constructing a diagnostic model for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism based on the markers, comprising the following steps: ① collecting and analyzing samples, covering elderly patients with subclinical hypothyroidism of different age groups, genders and disease courses; ② using high performance liquid chromatography-mass spectrometry technology to perform targeted and non-targeted analysis of comprehensive metabolic markers for each plasma sample to construct a metabolic profile map; ③ using data preprocessing technology to optimize metabolite data, and then combining clinical follow-up data to construct a prediction model using a machine learning algorithm; the machine learning method includes Lasso-Cox regression, random forest or support vector machine; ④ evaluating and optimizing model performance through cross-validation, external validation and other methods.

[0011] The above construction method obtains a diagnostic model for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism.

[0012] A diagnostic system for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism, comprising: (1) an acquisition unit, used to obtain marker information of a sample to be tested; (2) a detection unit, used to use the marker information of the sample to be tested as an input feature of a machine learning model, and then use the prediction model trained by the method of claim 3 to obtain the result of the sample to be tested; the result is a risk score of the sample to be tested.

[0013] An electronic device comprises: a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program to apply the system.

[0014] An executable storage medium stores computer program instructions. When the computer program instructions are executed on a processor, the processor executes the system.

[0015] A kit based on the marker combination, comprising reagents for detecting inosine, arachidonic acid, arginine, phenylalanine, glycocholic acid and auxiliary reagents.

[0016] A remote medical service system based on the diagnostic model includes a server and a client. The server stores the diagnostic model and has data processing capabilities. The client is used to receive sample data uploaded by the user. The server runs the diagnostic model according to the received data, generates a risk assessment report, and sends the report back to the client via the network, thereby realizing remote and real-time cardiovascular disease risk assessment services.

[0017] A drug development method based on the marker combination, which uses these markers as drug targets or efficacy evaluation indicators to screen and verify effective drug candidates for the prevention or treatment of cardiovascular diseases in elderly patients with subclinical hypothyroidism through in vitro experiments, animal model experiments and clinical trials.

[0018] Beneficial effects: 1. The present invention provides a method of establishing a cohort of elderly people with SCH and concurrent CVD, and using high performance liquid chromatography-mass spectrometry to detect the metabolite information of each plasma sample in the research cohort, and establishing a targeted metabolic content map. For plasma samples of 331 elderly people with subclinical hypothyroidism, 19 metabolic biomarkers were detected using high performance liquid chromatography-mass spectrometry to obtain a targeted metabolic content map. Each line of the data is metabolite information, which is used for further machine learning model establishment. 2. According to the matrix expression of the metabolite marker, the present invention uses the machine learning Lasso-cox model to construct a clinical prediction model for the CVD risk of elderly SCH patients in combination with clinical follow-up data, and guides the precise diagnosis and treatment of elderly SCH. 3. The present invention also provides a diagnostic marker kit containing the above-mentioned diagnostic markers suitable for predicting the risk of CVD in elderly SCH, which can be used to predict the risk of CVD in elderly SCH. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 .Use Lasso to screen 5 non-zero regression coefficient feature variables.

[0020] Figure 2.To establish a nomogram model for predicting the risk of CVD in elderly patients with SCH.

[0021] Figure 3 . AUROC of the training and test sets of the model for predicting the 3-year and 5-year risk of CVD in elderly SCH.

[0022] Figure 4 .AUPRC of the training and test sets of the 3-year and 5-year morbidity models for predicting the risk of CVD in elderly SCH.

[0023] Figure 5 DCA of the training and test sets of the 3-year and 5-year morbidity models for predicting the risk of CVD in elderly SCH.

[0024] Figure 6 Scores of the training and test sets of the 3-year and 5-year morbidity models for predicting the CVD risk of elderly SCH.

[0025] Figure 7 . Grouping of high-risk and low-risk patients in the training and test sets of the 3-year and 5-year morbidity models for predicting the risk of CVD in elderly SCH. DETAILED DESCRIPTION

[0026] The present invention provides a prediction model based on a group of biological metabolites, which is used to predict the risk of CVD in SCH. The model of the present invention can be used for diagnosis only by drawing blood, which is convenient, fast and non-invasive. It has high sensitivity and good specificity for predicting the risk of CVD in elderly SCH, and provides support for clinicians to quickly and accurately predict the risk of cardiovascular disease in patients and formulate precise prevention and treatment plans.

[0027] Example 1

[0028] A screening method suitable for predicting the risk of CVD in elderly SCH includes the following steps:

[0029] Plasma samples from elderly SCH patients were collected as analytical samples;

[0030] Inclusion criteria: (a) elderly people aged 65 years or above; (b) signed informed consent; (c) complete medical history data; (d) TSH ≥ upper limit of TSH reference range.

[0031] Exclusion criteria: (a) patients with clear hyperthyroidism or hypothyroidism; (b) patients taking contraceptives or estrogen; (c) patients taking glucocorticoids, dopamine, dobutamine, etc.; (d) patients taking anti-epileptic drugs (phenytoin sodium, carbamazepine, etc.); (e) patients with adrenal insufficiency, renal insufficiency (failure) or other serious systemic diseases; (f) patients who have undergone iodinated contrast examination and those who have undergone iodinated contrast examination. Coronary heart disease includes acute ST-segment elevation myocardial infarction, acute non-ST-segment elevation myocardial infarction and unstable angina, chronic stable exertional angina and stable stage after ACS.

[0032] In this study, there were 331 elderly patients with SCH, and 66 of them developed CVD during the 5-year follow-up period.

[0033] High performance liquid chromatography-mass spectrometry was used to perform targeted metabolomics analysis on each population sample to obtain the original targeted metabolic fingerprint of each plasma sample.

[0034] Among them, through the bioinformatics resource database, literature reports and mass spectrometry multiple reaction monitoring (MRM) technology prediction software, the parent ion-daughter ion pairs are selected according to the mass number and abundance of the fragment ions, and multiple MRM channels are determined for each screened metabolite.

[0035] Amino acids rich in heavy isotopes such as C13 were used as internal standards.

[0036] By adjusting the sample preparation process and chromatography-mass spectrometry conditions, including selecting parent ion-daughter ion pairs, improving chromatographic peak shape, and increasing the mass spectrometry signal-to-noise ratio, 19 metabolites were simultaneously quantitatively detected.

[0037] When each sample was analyzed by high performance liquid chromatography-mass spectrometry for plasma targeted metabolism, the chromatographic column used was Phenomenex C18 2.1x 100mm, 2.6μm chromatographic mobile phase: mobile phase A (aqueous phase) is 0.1% formic acid in water, mobile phase B (organic phase) is 0.1% formic acid in acetonitrile, and the program elution gradient is set as follows: 0-5min, 10%-70% B; 5-7min, 70%-95% B; 7-9min, 95% B; 9-9.1min, 95%-10% B; 9.1-10min, 10% B. The flow rate was set to 0.3mL / min, and the autosampler temperature was set to 4°C.

[0038] In the above method, mass spectrometry multiple reaction monitoring technology is used for each analysis sample, and an electrospray ionization cation and anion are electrospray ionized in positive ion mode of an electrospray ion source.

[0039] The obtained patient two-dimensional mass spectrometry sample data matrix was randomly divided into training set and test set at a ratio of 70% and 30%, respectively. The training set and test set included 231 and 100 elderly SCH patients, of which 46 and 20 had clinical outcomes with CVD. The Lasso regression variable screening was performed using the glmnet package in R, and the variables were constrained by L1 norm regularization. The convergence model complexity was adjusted by optimizing the λ parameter. The number of iterations was 200, and λ.1se=0.1 was determined. A 10-fold cross validation was used to screen the characteristic variables according to the minimum error of the model. Finally, five non-zero regression coefficient characteristic variables including inosine, arachidonic acid, arginine, phenylalanine, and glycocholic acid were determined ( Figure 1 ), the coxph function in the survival package was used to perform multifactor fitting regression analysis on the selected characteristic variables, and a nomogram model for predicting the occurrence of CVD in elderly SCH was established ( Figure 2 ).

[0040] The timeROC and ggDCA packages were used for ROC, PR analysis, and DCA to evaluate the predictive efficacy and clinical benefit of the model, respectively. The 5-year AUROC of the training set and test set was 0.938 and 0.966, and the 3-year AUROC was 0.914 and 0.935 ( Figure 3 ), the 5-year AUPRC of the training set and test set are 0.742 and 0.848, and the 3-year AUPRC are 0.532 and 0.453 ( Figure 4 ). DCA analysis showed that the training set and test set showed high clinical yield in terms of 5-year and 3-year CVD risk prediction ability ( Figure 5 ).

[0041] According to the Lasso-cox model established above, the nomogram for predicting the 3-year and 5-year CVD risk of elderly SCH was established using the rms and regplot packages. The Total score of individual patients was calculated based on the obtained Total score linear equation (Total score = 0.37 × inosine - 0.24 × arachidonic acid - 0.14 × arginine - 0.07 × phenylalanine - 5 × glycocholic acid + 277.9). The optimal threshold point was determined to be the Newton value = 231.3 ( Figure 6), individual patient scores above 231.3 were defined as high-risk elderly SCH patients with CVD, and scores below 231.3 were defined as low-risk elderly SCH patients with CVD. According to the patient risk level, patients in the SCH+CVD+ group had a higher Totalscore. Statistics showed that the proportion of patients in the SCH+CVD+ group who were defined as high risk was significantly higher than that in the SCH+CVD- group ( Figure 7 ).

[0042] In the above screening method, the analytical sample and the quality control sample are pre-treated as follows before injection:

[0043] (a) Use a pipette to draw 150 μL of plasma and place it in a 2.0 mL EP tube.

[0044] (b) Add 450 μL of -20°C precooled methanol, vortex for 5 min to mix, and place at -20°C for 1 hour.

[0045] (c) Centrifuge at 14,000 rpm for 15 min at 4°C. Transfer the supernatant to a new EP tube and store at -80°C.

[0046] (d) Pipette 100 μL of each sample into an EP tube, add 800 μL of -80°C precooled methanol, vortex thoroughly for 5 min, and incubate in a -80°C refrigerator for 8 h to fully precipitate proteins and extract small molecule metabolites.

[0047] (e) Centrifuge at 16,000 g, 4°C for 15 min. Carefully remove the sample and transfer the supernatant to a new centrifuge tube.

[0048] (f) 200 μL of 80% methanol water precooled at -80°C was added to the precipitate, and the mixture was sonicated at 4°C for 5 min and vortexed vigorously for 5 min to fully resuspend the precipitate. The mixture was centrifuged at 16,000 g at 4°C for 15 min and the supernatant was collected.

[0049] (g) The supernatants of the two samples were combined and concentrated to dryness by vacuum centrifugation at 4°C. The sample was reconstituted with 100 μL of 80% methanol aqueous solution and centrifuged at 18,000 g for 15 min at 4°C. The supernatant was collected and stored in a mass spectrometer bottle at 4°C until loading.

[0050] Table 2. Mass spectrometry information of 19 metabolites.

[0051]

[0052] According to the above-mentioned scoring formula Total score = 0.37 × inosine - 0.24 × arachidonic acid - 0.14 × arginine - 0.07 × phenylalanine - 5 × glycocholic acid + 277.9, the present invention calculated the Total score of plasma samples obtained from 10 patients with subclinical hypothyroidism in the elderly, and the results are shown in Table 3. Among them, the Total score of 3 patients was greater than the cutoff value of 231.3, and they were determined to be high-risk patients for CVD, and the Total score of 7 patients was less than the cutoff value of 231.3, and they were determined to be low-risk patients for CVD. Follow-up of these 10 patients was performed, and the clinical follow-up results and the predicted results were compared, and it was found that the two were consistent. The results are shown in Table 3.

[0053] Table 3 Total point, risk, follow-up time and outcome of patient plasma samples

[0054]

[0055] In summary, the advantages of the present invention are that the clinical prediction model for the occurrence of CVD in elderly SCH patients is constructed by using plasma metabolomics technology and artificial intelligence data analysis technology, and the precise diagnosis and treatment of elderly SCH is guided. The diagnostic marker screening method of the present invention is highly operable, the model construction method is simple, the obtained diagnostic model has good effect, and the diagnosis can be performed only by blood sampling, which is convenient, fast and non-invasive. It has high sensitivity and good specificity for predicting the risk of CVD in elderly SCH, and has good clinical application value.

Claims

1. Metabolic biomarkers for predicting cardiovascular disease risk in elderly patients with subclinical hypothyroidism, characterized in that: The markers include inosine, arachidonic acid, arginine, phenylalanine and glycocholic acid.

2. The metabolic biomarker for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism according to claim 1, characterized in that: The markers consist of inosine, arachidonic acid, arginine, phenylalanine and glycocholic acid.

3. A method for constructing a diagnostic model for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism based on the markers described in claim 1 or 2, characterized in that: The following steps are involved: ① Collect and analyze samples from elderly patients with subclinical hypothyroidism of different age groups, genders and disease durations; ②Use high performance liquid chromatography-mass spectrometry to conduct targeted and non-targeted analysis of comprehensive metabolic markers for each plasma sample and construct a metabolic profile map; ③Use data preprocessing technology to optimize metabolite data, and then use machine learning algorithms to build a prediction model based on clinical follow-up data; the machine learning methods include Lasso-Cox regression, random forest or support vector machine; ④Evaluate and optimize model performance through cross-validation, external validation and other methods.

4. A diagnostic model for predicting the risk of cardiovascular disease in elderly patients with subclinical hypothyroidism obtained by the construction method of claim 3.

5. A diagnostic system for predicting cardiovascular disease risk in elderly patients with subclinical hypothyroidism, characterized in that: include: (1) an acquisition unit, used to acquire marker information of the sample to be tested; (2) A detection unit, used to use the marker information of the sample to be tested as an input feature of a machine learning model, and then use the prediction model trained by the method of claim 3 to obtain the result of the sample to be tested; the result is the risk score of the sample to be tested.

6. An electronic device, characterized in that: include: Memory and processor; The memory is used to store computer programs; The processor is configured to execute a computer program to apply the system according to claim 5.

7. An executable storage medium, characterized in that: The storage medium stores computer program instructions, which, when executed on a processor, cause the processor to execute the system as claimed in claim 5 .

8. A kit based on the marker combination according to claim 1, characterized in that: The kit contains reagents for detecting inosine, arachidonic acid, arginine, phenylalanine, glycocholic acid and auxiliary reagents.

9. A remote medical service system based on the diagnostic model of claim 4, characterized in that: The system includes a server and a client. The server stores the diagnostic model and has data processing capabilities. The client is used to receive the sample data to be tested uploaded by the user. The server runs the diagnostic model based on the received data, generates a risk assessment report, and sends the report back to the client through the network, thereby realizing remote and real-time cardiovascular disease risk assessment services.

10. A drug development method based on the marker combination of claim 1, characterized in that: This method uses these markers as drug targets or efficacy evaluation indicators to screen and verify effective drug candidates for the prevention or treatment of cardiovascular disease in elderly patients with subclinical hypothyroidism through in vitro experiments, animal model experiments and clinical trials.

Images