Biomarker combination for glioma risk screening or auxiliary diagnosis and application thereof

Abstract

This invention discloses a biomarker combination for glioma risk screening or auxiliary diagnosis and its application. The biomarker combination comprises creatine, lactate, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate. Based on the biomarker combination provided by this invention, a clinical diagnostic model for assessing glioma risk can be constructed using algorithms such as multivariate logistic regression. The constructed clinical diagnostic model has high sensitivity and specificity, and can quickly and accurately assess the glioma risk of subjects, making it suitable for early screening and auxiliary diagnosis of adult gliomas.

Description

Technical Field

[0001] This invention belongs to the field of biomedical technology and relates to a combination of biomarkers for glioma risk screening or auxiliary diagnosis and their application. Background Technology

[0002] Gliomas are common primary malignant intracranial tumors. Accurate diagnosis is fundamental to the management of glioma patients. Clinical diagnosis of malignant tumors primarily relies on imaging examinations (such as computed tomography (CT) and magnetic resonance imaging (MRI), pathological biopsy, and circulating biomarkers. However, unlike tumors in other organs, intracranial tumor biopsies carry significant risks, especially lesions located in the brainstem or functional areas. Non-invasive neuroimaging remains the primary method for preoperative diagnosis and lesion description of gliomas. However, due to limitations in the precision and resolution of imaging techniques, its diagnostic accuracy in gliomas is only about 70%, a limitation that is more pronounced in early-stage gliomas with smaller tumor volumes. Currently, there are no approved circulating biomarkers for clinical use in gliomas. The blood-brain barrier severely limits the detectable levels of gene mutations originating from intracranial tumors in the bloodstream. Liquid biopsy methods, such as the analysis of circulating tumor cells or cell-free DNA, have low detection rates (<30%), are costly, and require PCR or sequencing platforms, making them difficult to implement in primary care hospitals.

[0003] Nuclear magnetic resonance metabolomics (NMR) testing platforms have achieved high-throughput testing in large-scale population cohorts both domestically and internationally, offering advantages such as low cost and no need for cumbersome pre-treatment. However, there is currently a lack of plasma metabolic biomarkers that can be used for glioma risk screening or auxiliary diagnosis. Summary of the Invention

[0004] The main objective of this invention is to provide a combination of biomarkers for glioma risk screening or auxiliary diagnosis and their application, in order to solve at least one of the above-mentioned technical problems.

[0005] According to a first aspect of the invention, a combination of biomarkers for glioma risk screening or auxiliary diagnosis is provided, comprising creatine, lactic acid, succinic acid, N,N-dimethylglycine, 2-oxoglutaric acid, acetic acid, and glutamic acid.

[0006] The biomarker combination of seven plasma metabolites provided by this invention can be used to rapidly and accurately assess the risk of glioma, and is especially suitable for early screening and auxiliary diagnosis of adult glioma.

[0007] According to a second aspect of the invention, the use of the biomarker combination of the invention in the preparation of products for glioma risk screening or auxiliary diagnosis is provided.

[0008] Based on the biomarker combination provided by this invention, a clinical diagnostic model with strong diagnostic efficacy that can be used to assess the risk of glioma can be constructed through various algorithms such as linear support vector machine, multivariate logistic regression, XGBoost, Gaussian Naive Bayes, and K-nearest neighbors.

[0009] In some implementations, clinical diagnostic models can be stored in an app, system, or computer storage device, and a risk assessment system that can directly output the risk of glioma development can be constructed.

[0010] According to a third aspect of the invention, a product for detecting the metabolic levels of creatine, lactic acid, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate in a sample is provided for use in the preparation of products for glioma risk screening or auxiliary diagnosis.

[0011] In some implementations, products for detecting the metabolic levels of creatine, lactate, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate in a sample can be reagents, kits, chips, and / or instruments suitable for detecting the metabolic levels of various biomarkers in a sample using nuclear magnetic resonance (NMR) spectroscopy. NMR spectroscopy, used for quantifying the metabolic levels of plasma metabolites, not only offers excellent analytical stability but also provides highly comparable results to standard clinical biochemical assays, enabling precise quantification of plasma metabolites without the need for expensive isotope internal standards.

[0012] In some embodiments, the sample can be plasma. Therefore, using the biomarker combination provided by this invention to assess the risk of glioma has advantages such as minimal invasiveness to patients, convenient sampling, and low testing cost, making it suitable for large-scale screening and easy to promote and apply in primary hospitals.

[0013] According to a fourth aspect of the present invention, a risk assessment system for glioma risk screening or auxiliary diagnosis is provided, comprising: The input module is used to input the plasma metabolic levels of creatine, lactate, succinate, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate and output them to the calculation module. The calculation module calculates the risk of glioma based on the plasma metabolic levels of each biomarker and the clinical diagnostic model, and outputs the result to the output module. The clinical diagnostic model is a clinical diagnostic model for assessing the risk of glioma constructed based on the plasma metabolic levels of each biomarker in the biomarker combination of the present invention. The output module is used to output the risk of developing glioma.

[0014] In some implementations, the metabolic levels of each biomarker are detected and quantified using nuclear magnetic resonance spectroscopy. This approach offers advantages such as eliminating the need for isotope internal standards, PCR amplification, and aerosol contamination, while also providing convenient sampling, low operational requirements, and low detection costs.

[0015] In some implementations, the clinical diagnostic model used to assess the risk of glioma development can be a clinical diagnostic model constructed using a multivariate logistic regression algorithm. The clinical diagnostic model constructed using the multivariate logistic regression algorithm can also be further transformed into a nomogram that is easily applicable in clinical practice.

[0016] According to a fifth aspect of the present invention, another risk assessment system for glioma risk screening or auxiliary diagnosis is provided, comprising: The data acquisition module is used to read the raw FID file of the nuclear magnetic resonance spectrum and output it to the quantitative module; The quantification module is used to calculate and output the plasma metabolic levels of creatine, lactate, succinate, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate to the assessment module based on the raw FID file of the nuclear magnetic resonance spectrum. The assessment module calculates and outputs the risk of glioma based on the plasma metabolic levels of each biomarker and the clinical diagnostic model; wherein, the clinical diagnostic model is a clinical diagnostic model for assessing the risk of glioma constructed based on the plasma metabolic levels of each biomarker in the biomarker combination of the present invention. Attached Figure Description

[0017] Figures 1-2 ROC curves for 12 plasma metabolites used in the training set to differentiate between glioma patients and healthy controls; Figure 3 The graph shows the coefficient change path of the LASSO regression model, displaying the trajectory of coefficient changes under different regularization intensities. Figure 4 The cross-validation plot for the tuning parameter λ in the LASSO regression model illustrates the process of selecting the optimal λ value using tenfold cross-validation in the LASSO regression model. Figures 5-7ROC curves of the performance of different metabolite combination models constructed by three machine learning algorithms, namely SVM, RF, and PLS-DA, are shown in order. Figure 8 The ROC curves for the diagnostic efficacy of the combination of 7 plasma metabolic markers validated by cross-validation (based on the SVM algorithm) are shown; the black line shows the results of the glioma training set using 70% cross-validation, and the red line shows the results of the Hold-out test set. Figure 9 This is a scatter plot of the probability of each sample class in the SVM-based diagnostic model; black dots represent the training set, red dots represent the hold-out test set; hollow dots represent glioma patients, and solid dots represent healthy control subjects. Figure 10 The prediction accuracy is based on a glioma training set with 100 MCCV runs, and the accuracy is validated on a hold-out test set. Figures 11-14 The following are the ROC curves of the clinical diagnostic models built based on multivariate logistic regression, XGBoost, Gaussian Naive Bayes, and K-nearest neighbors in the training set, respectively. Figure 15 This is a nomogram derived from a multivariate logistic regression model; Figure 16 for Figure 15 The 5-fold cross-validation test of the line graph shown; Figure 17 For the training set, observed and based on Figure 15 The calibrated curves for the predicted glioma prevalence shown in the nomogram; Figure 18 The DCA represents the clinical utility of the clinical diagnostic model in screening for gliomas; the X-axis represents the risk threshold probability, while the Y-axis represents the net benefit; the red line represents the decision curve of plasma metabolic characteristics, while the dashed line represents two reference strategies: "treat-all" (green) and "treat-none" (blue). Figure 19 The ROC curve for the clinical diagnostic model on the test set. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to the embodiments. The embodiments are for illustrative purposes only and do not limit the invention in any way. Unless otherwise specified, the raw materials and reagents used in the embodiments are conventional products that can be obtained commercially; experimental methods that do not specify specific conditions in the embodiments are generally performed under conventional conditions in the art or according to the conditions recommended by the manufacturer.

[0019] Example 1: Validation of the diagnostic efficacy of a combination of biomarkers for glioma risk screening or auxiliary diagnosis. 1. Participant Clinical Information The training set included 56 adult glioma patients and 56 healthy adult controls recruited between January 2022 and March 2023. All 56 glioma patients subsequently underwent surgical treatment, and pathological diagnosis (gold standard) confirmed glioma. Among the 56 glioma patients, 24 had glioblastoma, 19 had astrocytoma, 12 had oligodendroglioma, and 1 had mixed-type glioma.

[0020] The test set included 26 adult patients with gliomas and 26 healthy adult controls enrolled between March 2023 and August 2025. All 26 glioma patients subsequently underwent surgical treatment, and pathological diagnosis (gold standard) confirmed glioma. Among the 26 glioma patients, 10 had glioblastomas, 7 had astrocytomas, 3 had oligodendrogliomas, 4 had diffuse midline gliomas, and 1 had an ependymoma.

[0021] The demographic and clinical characteristics of glioma patients and healthy controls in the training and test sets are shown in Tables 1 and 2 below, respectively.

[0022] Table 1. Demographic and clinical characteristics of glioma patients in the training set.

[0023] Table 2. Demographic and clinical characteristics of glioma patients in the test set.

[0024] 2. Plasma-targeted metabolomics detection based on NMR platform Qualitative and quantitative detection methods for plasma metabolites include the following steps: (1) Sample collection: Subjects fasted for more than 8 hours, and 3 mL of venous blood was collected using a vacuum anticoagulation blood collection tube containing EDTA-K2. The blood was then stored at 4℃ and 1600× g Centrifuge for 15 min to separate the plasma, and store the plasma at -80℃.

[0025] (2) Sample detection: Take plasma sample, and after it is completely thawed, take 340 μL of plasma and 340 μL of NMR buffer (Plasma buffer, Bruker, H170046) and mix them 1:1. After mixing thoroughly, take 600 μL and place it in a 5 mm NMR tube. Place it in the NMR tube rack in the order of loading the instrument and load it into the autosampler to wait for detection.

[0026] The plasma testing procedure was performed using a Bruker Biospin AG nuclear magnetic resonance spectrometer (600 MHz). Samples were fed one by one into the magnet via an autosampler. A standard temperature control procedure was used, with the autosampler set to 280K to ensure sample quality. Samples were preheated by airflow before injection to ensure they reached the detection temperature (310K) before testing. 1 H NMR spectroscopy scan.

[0027] (3) Qualitative and quantitative analysis of metabolites: The NMR spectra were normalized using the QuantRef management system built into the Bruker Topspin software to normalize the intensity of the spectra to proton concentrations in mmol / L. Chemical shifts were corrected using the bimodal signals of TSP (3-Trimethylsilylpropanoic acid) and alanine at 1.48 ppm. Qualitative analysis of the NMR spectra was performed based on the Bruker NMR library. Quantification of metabolites was achieved using the integral of the signal cluster at specific chemical shift positions.

[0028] Targeted metabolomics analysis was performed on plasma samples from 56 adult glioma patients in the training set and plasma samples from 56 healthy adult volunteers of similar age and sex to glioma patients using a 600 MHz NMR quantitative metabolomics platform.

[0029] Ultimately, 153 plasma metabolites, including carbon metabolites such as creatine, lactic acid, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamic acid, as well as blood lipids, were identified and quantified.

[0030] 3. Validation of the diagnostic efficacy of plasma metabolic biomarkers and their combinations in glioma patients. (1) Assessment of the diagnostic efficacy of plasma metabolites for individuals The individual diagnostic efficacy of 12 plasma metabolites differentially expressed in the plasma of glioma patients and healthy adult volunteers was evaluated. The 12 plasma metabolites were creatine, acetic acid, 2-oxoglutarate, D-glucoside, H4FC, IDAB, lactate, N,N-dimethylglycine, succinate, glutamate, IDPN, and glutamine.

[0031] In the training set, all 12 plasma metabolites showed individual diagnostic potential, with their areas under the ROC curve (AUC) exceeding 0.75. Figures 1-2 ).

[0032] The coefficient path plot of the Lasso regression model shows the trajectory of the coefficients of the top 12 metabolites as a function of log(λ). Figure 3It can be seen that as the penalty parameter log(λ) increases, the coefficient of metabolites gradually shrinks.

[0033] Cross-validation curves show that the model bias is minimal when λmin = 0.033; and the model is most concise while maintaining a good fit when λ1se = 0.01. Figure 4 ).

[0034] Based on the ease of clinical application, this invention further optimizes the model and reduces the number of included metabolites. Seven plasma metabolites were selected for constructing the clinical diagnostic model based on strict screening criteria: creatine, lactate, succinate, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate. The screening criteria were as follows: (1) common differences in three adult glioma subtypes; (2) mean precision decrease > 0.010 in the random forest model; (3) meeting the λmin criterion in Lasso regression. Through dual screening using RF and LASSO regression algorithms, the potential interference from inter-individual heterogeneity can be effectively mitigated.

[0035] (2) Determination of plasma metabolite composition Based on (1) above, a diagnostic model was constructed by combining seven metabolites (creatine, lactate, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate) using three algorithms: linear support vector machine (SVM), random forest (RF), and partial least squares discriminant analysis (PLS-DA). The results are as follows: Figures 5-7 As shown, in the training set, the AUCs of the models of the three machine learning algorithms are similar and all have good diagnostic efficacy.

[0036] Algorithm comparisons show that the SVM algorithm achieves the highest diagnostic efficacy with 10 features, while the RF and PLS-DA models require 25 features. Therefore, this invention ultimately determines a diagnostic model based on SVM that integrates the aforementioned seven plasma metabolites, which exhibits strong diagnostic performance. Specifically, this diagnostic model containing combinations of seven plasma metabolites achieved an AUC of 0.978 (95% CI = 0.932-1) in 70% cross-validation (CV) and an AUC of 0.972 in the test set. Figures 8-9 The results of 100 Monte Carlo Cross-Validations (MCCVs) showed that the average accuracy of the metabolite combination diagnostic model was 0.917, and the accuracy on the test set was 0.942. Figure 10 ).

[0037] (3) Construction and validation of clinical diagnostic models To facilitate clinical translation, this invention constructs a clinical diagnostic model based on the plasma metabolite combination consisting of creatine, lactate, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamate, using four machine learning algorithms: multivariable logistic regression, XGBoost, Gaussian Naive Bayes, and K-Nearest Neighbors.

[0038] like Figures 11-14 As shown, all four algorithms can construct clinical diagnostic models with strong diagnostic efficacy. Among them, the multivariate logistic regression model has the best diagnostic efficacy, with an AUC of 0.996 (95% CI: 0.987-1.000).

[0039] To facilitate clinical application, the multivariate logistic regression model was further transformed into an easy-to-use nomogram. Figure 15 In a nomogram, each predictor variable is assigned a weighted score based on its regression coefficient. The total score corresponds to the individual's risk ratio for developing glioma, thus providing a visual assessment of an individual's glioma risk. When using a nomogram to assess glioma risk, the quantitative absolute values ​​of plasma metabolites measured by NMR need to be log10 transformed and Z-score normalized. To determine the score associated with a particular metabolite value, the measured metabolic level of that metabolite is plotted on the corresponding scale line, then a vertical line is drawn to the "Score" axis, and the sum of the individual scores for all metabolites is calculated. Next, the total score is plotted on the "Total Score" axis, and a vertical line is drawn downwards to the "Risk" axis to obtain the odds ratio of an individual's glioma risk.

[0040] like Figure 16 As shown, the robustness of the multivariate logistic regression model was verified by a 5-fold cross-validation test. The calibration curves showed good agreement between the observed and predicted glioma prevalence. Figure 17 Decision curve analysis (DCA) results show that the constructed multivariate logistic regression model can provide higher net benefits over a wide range of threshold probabilities, validating its high accuracy and clinical applicability. Figure 18 ).

[0041] like Figure 19As shown, further validation results demonstrate that the multivariate logistic regression model maintains high sensitivity and specificity on the test set, with an AUC of 0.964, accuracy of 0.923, sensitivity of 0.885, and specificity of 0.962, proving that this clinical diagnostic model has strong diagnostic efficacy for gliomas. The diagnostic efficacy of the clinical diagnostic model provided by this invention is significantly better than that of cerebrospinal fluid cell-free DNA (detection rate of approximately 80%, https: / / doi.org / 10.1038 / s41586-019-0882-3) and plasma circulating tumor DNA (detection rate of only 14% in gliomas, https: / / doi.org / 10.1016 / j.annonc.2022.07.413).

[0042] The above descriptions are merely some embodiments of the present invention. Those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these all fall within the scope of protection of the present invention.

Claims

1. A biomarker combination for glioma risk screening or aid diagnosis, characterized in that, The biomarker combination consists of creatine, lactic acid, succinic acid, N,N-dimethylglycine, 2-oxoglutarate, acetic acid, and glutamic acid.

2. The use of the biomarker combination according to claim 1 in the preparation of products for glioma risk screening or auxiliary diagnosis.

3. The application of a product for detecting the metabolic levels of each biomarker in the biomarker combination according to claim 1 in a sample in the preparation of products for glioma risk screening or auxiliary diagnosis.

4. The application according to claim 3, characterized in that, The sample was plasma.

5. The application according to claim 3 or 4, characterized in that, The product used to detect the metabolic levels of each biomarker in the biomarker combination according to claim 1 in the sample is a reagent, kit, chip, and / or instrument suitable for detecting the metabolic levels of each biomarker in the biomarker combination according to claim 1 in the sample by nuclear magnetic resonance spectroscopy.

6. A risk assessment system for glioma risk screening or auxiliary diagnosis, characterized in that, Its components include: An input module is used to input the plasma metabolic level of each biomarker in the biomarker combination of claim 1 and output it to the calculation module; The calculation module calculates the risk of glioma based on the plasma metabolic levels of each biomarker in the biomarker combination of claim 1 and the clinical diagnostic model, and outputs the result to the output module; the clinical diagnostic model is a clinical diagnostic model for assessing the risk of glioma constructed based on the plasma metabolic levels of each biomarker in the biomarker combination of claim 1. The output module is used to output the risk of developing glioma.

7. The risk assessment system for glioma risk screening or auxiliary diagnosis according to claim 6, characterized in that, The plasma metabolic levels of each biomarker in the biomarker combination of claim 1 are detected and quantified by nuclear magnetic resonance spectroscopy.

8. A risk assessment system for glioma risk screening or auxiliary diagnosis, characterized in that, include: The data acquisition module is used to read the raw FID file of the nuclear magnetic resonance spectrum and output it to the quantitative module; A quantitative module is used to calculate and output the plasma metabolic level of each biomarker in the biomarker combination of claim 1 to the evaluation module based on the original FID file of the nuclear magnetic resonance spectrum. The assessment module calculates and outputs the risk of glioma based on the plasma metabolic levels of each biomarker in the biomarker combination of claim 1 and a clinical diagnostic model; the clinical diagnostic model is a clinical diagnostic model for assessing the risk of glioma constructed based on the plasma metabolic levels of each biomarker in the biomarker combination of claim 1.

Images