A combination of biomarkers and its screening method
By using nanoparticle-enhanced laser desorption/ionization mass spectrometry and machine learning algorithms, biomarkers such as lactic acid, glutamine, and 3-methylpentenoic acid were screened, solving the problem of matrix interference in traditional methods and enabling efficient early diagnosis and risk assessment of cerebral aneurysms.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2023-02-16
- Publication Date
- 2026-05-26
AI Technical Summary
Existing technologies are insufficient to effectively screen metabolic biomarkers for the early diagnosis of cerebral aneurysms, and traditional electrospray ionization mass spectrometry is greatly affected by matrix interference, making it difficult to achieve efficient cerebral aneurysm risk assessment.
Using nanoparticle-enhanced laser desorption/ionization mass spectrometry combined with machine learning algorithms, biomarkers such as lactic acid, glutamine, and 3-methylpentenoic acid were screened for metabolic analysis and risk assessment of cerebral aneurysms.
It enables rapid and accurate diagnosis and risk assessment of cerebral aneurysms, with fast detection speed, high reproducibility, low sample consumption, and the ability to effectively distinguish between low-risk and high-risk groups.
Smart Images
Figure CN116106401B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of biotechnology, and more particularly to a combination of biomarkers and a method for screening them. Background Technology
[0002] Cerebral aneurysms are life-threatening saccular dilatations of blood vessels in the brain, with a prevalence of 3.2–7.0% in the population. Rupture of a cerebral aneurysm leads to subarachnoid hemorrhage, a serious type of stroke with a poor prognosis. Early diagnosis is crucial for the medical management of cerebral aneurysms. Currently, most cerebral aneurysms are incidentally diagnosed in neurovascular imaging modalities, such as computed tomography angiography and magnetic resonance angiography. However, these methods have drawbacks such as radiation exposure or time consumption, contrast agent toxicity, and poor interpretability. Blood-based liquid biopsy is a promising alternative for early diagnosis, offering high sensitivity, minimal invasiveness, high throughput, and ease of use.
[0003] Advanced omics technologies enable personalized medicine through liquid biopsy. Compared to genomics and proteomics, metabolomics provides real-time insights into the precise pathophysiology of diseases, more closely approximating disease phenotypes. A growing body of research confirms the importance of metabolomics in early diagnosis. However, current metabolomics studies targeting cerebral aneurysms fall far short of the requirements for early diagnosis. The high performance and low cost of analytical techniques such as mass spectrometry have led to their widespread application in clinical research. However, to overcome ionization inhibition caused by matrix interference in complex biological samples, traditional electrospray ionization mass spectrometry is inseparable from liquid / gas chromatography. Notably, the recently developed nanoparticle-enhanced laser desorption / ionization mass spectrometry is a high-performance solid-phase soft ionization technique for direct metabolic analysis of biological samples, offering advantages such as high throughput, good salt / protein tolerance, high sensitivity, and in-situ metabolite enrichment. Therefore, nanoparticle-enhanced laser desorption / ionization mass spectrometry provides an effective metabolic analysis tool for the early diagnosis and biomarker development of cerebral aneurysms.
[0004] Therefore, those skilled in the art are dedicated to developing an effective method and biomarkers for screening biomarkers for metabolic analysis, and to using the biomarkers for assessing the risk of rupture of cerebral aneurysms. Summary of the Invention
[0005] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is how to develop an effective method and biomarkers for screening metabolic analysis, and to use the biomarkers for assessing the risk of rupture of cerebral aneurysms.
[0006] To achieve the above objectives, the present invention provides a combination of biomarkers including lactic acid, glutamine, high arginine and 3-methylpentenoic acid.
[0007] This invention also provides an application of a combination of biomarkers in assessing the risk of cerebral aneurysm rupture.
[0008] This invention also provides a method for screening combinations of biomarkers, comprising the following steps:
[0009] Step 1: Collect metabolic fingerprints of serum from patients with cerebral aneurysms and healthy controls using nanoparticle-enhanced laser desorption / ionization mass spectrometry;
[0010] Step 2: Perform machine learning on the metabolic fingerprints of serum from patients with cerebral aneurysms and healthy controls obtained in Step 1 to screen for combinations of biomarkers.
[0011] Furthermore, step 1 also includes:
[0012] Step 1.1: Dilute the serum sample with deionized water to obtain a pretreated serum sample;
[0013] Step 1.2: Preparation of iron oxide inorganic nanoparticles;
[0014] Step 1.3: Sample preparation. Spot the pretreated serum sample obtained in step 1.1 onto the mass spectrometry target plate and allow it to dry naturally to obtain the mass spectrometry target plate after sample loading.
[0015] Step 1.4: Prepare the matrix. Prepare a matrix solution of 1 mg / mL by dissolving the iron oxide inorganic nanoparticles obtained in Step 1.2. Spot the matrix solution onto the mass spectrometry target plate obtained in Step 1.3 after sample loading. Allow the solution to dry naturally to obtain the mass spectrometry target plate with added sample and matrix.
[0016] Step 1.5: Detect the mass spectrometry target plate with added sample and matrix obtained in step 1.4 in a matrix-assisted laser desorption / ionization mass spectrometer to collect the metabolic fingerprint of serum samples.
[0017] Furthermore, in step 1.3, the pretreated serum sample spotting volume is 1 μL; in step 1.4, the matrix solution spotting volume is 1 μL.
[0018] Further, the method for preparing iron oxide inorganic nanoparticles in step 1.2 is as follows: 0.60g of trisodium citrate, 2.40g of ferric chloride and 3.84g of sodium acetate are added sequentially to 80mL of ethylene glycol solution and stirred and dispersed (600rpm). The solution is transferred to a Teflon high-pressure reactor and reacted in an oven at 200 degrees Celsius for 10 hours. The product is washed three times with ethanol and deionized water, and finally dried in an oven at 60 degrees Celsius.
[0019] Furthermore, the detection conditions of the matrix-assisted laser desorption / ionization mass spectrometer in step 1.5 are as follows: the mass spectrometry detection of metabolic fingerprints adopts reflectance mode, positive ion detection, the detection range is set to 100-1000 Da, the laser wavelength is 355 nm, the laser frequency is 2 kHz; the delay time is 150 ns; the accelerating voltage is 20 kV, the repetition rate of the delay extraction is 1 kHz; and 2000 laser irradiations are superimposed for each analysis.
[0020] Furthermore, the machine learning in step 2 includes:
[0021] Step 2.1: Divide the serum samples into training set and test set, perform signal preprocessing on the collected metabolic fingerprints, and obtain the preprocessed metabolic fingerprints of the training set and the preprocessed metabolic fingerprints of the test set, respectively.
[0022] Step 2.2: Use the gradient boosting decision tree algorithm on Orange 3.33.3 to perform feature selection and machine learning on the metabolic fingerprint of the preprocessed training set obtained in Step 2.1 to build a model;
[0023] Step 2.3: Use the gradient boosting decision tree algorithm on Orange 3.33.3 to process the metabolic fingerprint of the preprocessed test set obtained in Step 2.1, and verify the performance of the model constructed in Step 2.3.
[0024] Step 2.4: Establish biomarker screening criteria based on the results of the feature selection method, and screen out combinations of biomarkers.
[0025] Furthermore, the signal preprocessing in step 2.1 specifically includes spectral smoothing, baseline correction, and spectral peak alignment.
[0026] Furthermore, the biomarker screening criteria in step 2.5 are as follows: the biomarker must simultaneously meet the following conditions: the corrected P-value < 0.05, the area under the curve of the single indicator > 0.7, and the corresponding coefficient of the gradient boosting decision tree algorithm must be ranked in the top 25.
[0027] In a preferred embodiment of the present invention, the process of collecting serum metabolic fingerprints based on nanoparticle-enhanced laser desorption / ionization mass spectrometry is described in detail.
[0028] In another preferred embodiment 2 of the present invention, the process of performing machine learning on the serum metabolic fingerprints of patients with cerebral aneurysms and healthy controls, screening biomarkers and applying them is described in detail.
[0029] Nanoparticle-enhanced laser desorption / ionization mass spectrometry (NDEMS) is a high-performance solid-phase soft ionization technique for direct metabolic analysis of biological samples, offering advantages such as high throughput, good salt / protein tolerance, high sensitivity, and in-situ metabolite enrichment. NDEMS provides an effective metabolic analysis tool for the early diagnosis of cerebral aneurysms and the development of biomarkers. This invention utilizes NDEMS to screen lactate, glutamine, high arginine, and 3-methylpentenoic acid as metabolic biomarkers, and uses these biomarkers to assess the rupture risk of cerebral aneurysms. The beneficial technical effects of this invention are as follows:
[0030] 1. The biomarker screening method of the present invention has the advantages of fast detection and analysis speed (about 30 seconds per sample), high reproducibility (characteristic intensity variation coefficient of sample detection signal <20%) and low sample consumption (about 100 nL of raw serum).
[0031] 2. By using machine learning to analyze serum metabolic fingerprints, we achieved high diagnostic performance for cerebral aneurysms, with an area under the curve of 0.842 on the training set and 0.817 on the test set.
[0032] 3. Lactic acid, glutamine, arginine, and 3-methylpentenoic acid were selected as metabolic biomarkers, with an area under the curve (AUC) of 0.812 for cerebral aneurysm diagnosis. These biomarkers were then used to assess the risk of cerebral aneurysm rupture, showing a significant difference between the low-risk and high-risk groups (p<0.05).
[0033] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description
[0034] Figure 1 This is a representative metabolic mass spectrometry fingerprint of serum from a patient with a cerebral aneurysm and a healthy control group, which is a preferred embodiment of the present invention.
[0035] Figure 2 This is a heatmap of 122 m / z signals corresponding to serum fingerprint preprocessing of 291 samples, which is a preferred embodiment 2 of the present invention.
[0036] Figure 3 This is a performance diagram of a cerebral aneurysm with 122 m / z signals, which is a preferred embodiment 2 of the present invention;
[0037] Figure 4 This is a biomarker difference diagram of a patient with a cerebral aneurysm and a healthy control group in a preferred embodiment 2 of the present invention. Detailed Implementation
[0038] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.
[0039] Example 1: Serum metabolic fingerprinting based on nanoparticle-enhanced laser desorption / ionization mass spectrometry:
[0040] Step 1: Serum sample pretreatment, specifically by diluting the serum sample 10 times with deionized water;
[0041] Step 2, the preparation method of iron oxide inorganic nanoparticle matrix, is as follows: 0.60g of trisodium citrate, 2.40g of ferric chloride and 3.84g of sodium acetate are added sequentially to 80mL of ethylene glycol solution and stirred and dispersed (600rpm). The solution is transferred to a Teflon high-pressure reactor and reacted in an oven at 200 degrees Celsius for 10 hours. The product is washed three times with ethanol and deionized water, and finally dried in an oven at 60 degrees Celsius to obtain iron oxide inorganic nanoparticles as matrix material.
[0042] Step 3: Sample preparation on the mass spectrometry target plate. Spot 1 μL of pretreated serum sample and allow it to air dry.
[0043] Step 4: Preparation of matrix on mass spectrometry target plate. Prepare a matrix solution of 1 mg / mL iron oxide inorganic nanoparticles, spot 1 μL of the solution, and allow it to air dry.
[0044] Step 5: Serum metabolic fingerprints were collected using matrix-assisted laser desorption / ionization mass spectrometry (MALS). The results are shown in the attached figure. Figure 1 The serum metabolic mass spectrum acquired by matrix-assisted laser desorption / ionization mass spectrometry is shown in the figure. Figure 1 The upper part represents patients with cerebral aneurysms, and the lower part represents healthy controls. Metabolic fingerprinting was performed using mass spectrometry in reflectance mode with positive ion detection. The detection range was set to 100-1000 Da. The instrument model was Autoflex MALDI-TOF( / TOF)-MS (Bruker Autoflex Speed), with the following parameters: laser wavelength 355 nm, laser frequency 2 kHz; delay time 150 ns; accelerating voltage 20 kV; delay extraction repetition rate 1 kHz; and 2000 laser irradiations were superimposed per analysis.
[0045] Example 2: Machine learning was used to screen biomarkers in the serum metabolic fingerprints of patients with cerebral aneurysms and healthy controls, and these biomarkers were then selected and applied.
[0046] Step 1: Pretreatment of 291 serum samples, specifically by diluting the serum samples 10 times with deionized water;
[0047] Step 2, the preparation method of iron oxide inorganic nanoparticle matrix, is as follows: 0.60g of trisodium citrate, 2.40g of ferric chloride and 3.84g of sodium acetate are added sequentially to 80mL of ethylene glycol solution and stirred and dispersed (600rpm). The solution is transferred to a Teflon high-pressure reactor and reacted in an oven at 200 degrees Celsius for 10 hours. The product is washed three times with ethanol and deionized water, and finally dried in an oven at 60 degrees Celsius to obtain iron oxide inorganic nanoparticles as matrix material.
[0048] Step 3: Sample preparation on the mass spectrometry target plate. Spot 1 μL of pretreated serum sample and allow it to air dry.
[0049] Step 4: Preparation of matrix on mass spectrometry target plate. Prepare a matrix solution of 1 mg / mL iron oxide inorganic nanoparticles, spot 1 μL of the solution, and allow it to air dry.
[0050] Step 5: Serum metabolic fingerprints were acquired using matrix-assisted laser desorption / ionization mass spectrometry (MAMS). The metabolic fingerprint was detected in reflectance mode with positive ion detection, within a detection range of 100-1000 Da. The instrument used was an Autoflex MALDI-TOF( / TOF)-MS (Bruker Autoflex Speed), with the following parameters: laser wavelength 355 nm, laser frequency 2 kHz; accelerating voltage 20 kV; delay extraction repetition rate 1 kHz; delay time 150 ns; and 2000 laser irradiations were superimposed per analysis.
[0051] Step 6: Perform signal preprocessing on the collected metabolic fingerprints, specifically including spectral smoothing, baseline correction, and peak alignment. The heatmap corresponding to the 122 m / z signals after preprocessing of the 291 serum metabolic fingerprints is attached. Figure 2 As shown, the metabolic fingerprint heatmaps correspond to 122 m / z signals after preprocessing of serum metabolic fingerprints from 291 samples.
[0052] Step 7: Divide the 291 samples into a training set (194 cases) and a test set (97 cases);
[0053] Step 8: Use the gradient boosting decision tree algorithm on Orange 3.33.3 to perform feature selection and machine learning on the training set, build the model, and the area under the curve of the training set is 0.842. The performance graph is attached. Figure 3 As shown, the diagnostic performance of cerebral aneurysms is plotted on the training set and the test set, with the area under the curve (AUC) of 0.842 for the training set and 0.817 for the test set.
[0054] Step 9: Validate the model's performance on the test set using the gradient boosting decision tree algorithm on Orange 3.33.3. The area under the curve on the test set is 0.817. The performance graph is attached. Figure 3 As shown;
[0055] Step 10: Compile three feature selection methods as biomarker screening criteria, which must simultaneously meet the following conditions: 1) corrected p-value < 0.05, 2) area under the curve of a single indicator > 0.7, and 3) the corresponding coefficients of the gradient boosting decision tree algorithm are ranked in the top 25. A graph showing the differences in biomarkers between patients with cerebral aneurysms and healthy controls is attached. Figure 4 As shown, combinations of biomarkers were screened, including lactate, glutamine, arginine, and 3-methylpentenoic acid. The area under the curve for biomarker-based cerebral aneurysms was 0.812. The biomarkers were used to assess the risk of cerebral aneurysm rupture, and there was a significant difference between the low-risk and high-risk groups (P < 0.05).
[0056] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.
Claims
1. A combination of biomarkers, characterized in that, The biomarker combination includes lactic acid, glutamine, high arginine, and 3-methylpentenoic acid.
2. A method for screening biomarker combinations as described in claim 1, characterized in that, The method includes the following steps: Step 1: Collect metabolic fingerprints of serum from patients with cerebral aneurysms and healthy controls using nanoparticle-enhanced laser desorption / ionization mass spectrometry; Step 2: Perform machine learning on the metabolic fingerprints of serum from patients with cerebral aneurysms and healthy controls obtained in Step 1 to screen the biomarker combinations.
3. The screening method as described in claim 2, characterized in that, Step 1 further includes: Step 1.1: Dilute the serum sample with deionized water to obtain a pretreated serum sample; Step 1.2: Preparation of iron oxide inorganic nanoparticles; Step 1.3: Sample preparation. Spot the pretreated serum sample obtained in step 1.1 onto the mass spectrometry target plate and allow it to dry naturally to obtain the mass spectrometry target plate after sample loading. Step 1.4: Prepare the matrix. Prepare a matrix solution by dissolving the iron oxide inorganic nanoparticles obtained in Step 1.
2. Spot the solution onto the mass spectrometry target plate obtained in Step 1.3 after sample loading. Allow the solution to dry naturally to obtain the mass spectrometry target plate with added sample and matrix. Step 1.5: Detect the mass spectrometry target plate with added sample and matrix obtained in step 1.4 in a matrix-assisted laser desorption / ionization mass spectrometer to collect the metabolic fingerprint of the serum sample.
4. The screening method as described in claim 3, characterized in that, In step 1.3, the pretreated serum sample spotting volume is 1 µL; in step 1.4, the matrix solution concentration is 1 mg / mL, and the spotting volume is 1 µL.
5. The screening method as described in claim 3, characterized in that, The method for preparing iron oxide inorganic nanoparticles in step 1.2 is as follows: 0.60 g of trisodium citrate, 2.40 g of ferric chloride and 3.84 g of sodium acetate are added sequentially to 80 mL of ethylene glycol solution and stirred to disperse. The solution is transferred to a Teflon high-pressure reactor and reacted in an oven at 200 degrees Celsius for 10 hours. The product is washed three times with ethanol and deionized water, and finally dried in an oven at 60 degrees Celsius.
6. The screening method as described in claim 3, characterized in that, The detection conditions of the matrix-assisted laser desorption / ionization mass spectrometer in step 1.5 are as follows: the mass spectrometry detection of metabolic fingerprints adopts reflection mode, positive ion detection, the detection range is set to 100-1000 Da, the laser wavelength is 355 nm, the laser frequency is 2 kHz; the delay time is 150 ns; the accelerating voltage is 20 kV, the repetition rate of the delayed extraction is 1 kHz; and 2000 laser irradiations are superimposed for each analysis.
7. The screening method as described in claim 3, characterized in that, The machine learning mentioned in step 2 includes: Step 2.1: Divide the serum samples into a training set and a test set, and preprocess the collected metabolic fingerprints to obtain the preprocessed metabolic fingerprints of the training set and the preprocessed metabolic fingerprints of the test set, respectively. Step 2.2: Use the gradient boosting decision tree algorithm on Orange 3.33.3 to perform feature selection and machine learning on the metabolic fingerprint of the preprocessed training set obtained in Step 2.1 to build a model; Step 2.3: Use the gradient boosting decision tree algorithm described above on Orange 3.33.3 to process the metabolic fingerprint of the preprocessed test set obtained in Step 2.1, and verify the performance of the model constructed in Step 2.
3. Step 2.4: Establish biomarker screening criteria based on the results of the feature selection method, and screen out combinations of biomarkers.
8. The screening method as described in claim 7, characterized in that, The signal preprocessing described in step 2.1 specifically includes spectral smoothing, baseline correction, and spectral peak alignment.
9. The screening method as described in claim 7, characterized in that, The biomarker screening criteria in step 2.4 are as follows: they must simultaneously meet the following conditions: corrected P-value < 0.05, area under the curve of a single indicator > 0.7, and the corresponding coefficient of the gradient boosting decision tree algorithm must be ranked in the top 25.