Biomarkers for benign biliary disease and cholangiocarcinoma discrimination and uses thereof
By screening and modeling lipidomics and peptide biomarkers, the problems of invasiveness in bile collection and low serum sensitivity were solved, achieving highly sensitive and specific cholangiocarcinoma identification and providing intermediate results information for clinical applications.
Patent Information
- Application Number
- CN202310536778.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-10
- Publication Date
- 2025-12-23
- Estimated Expiration
- 2043-05-10
AI Technical Summary
In existing technologies, bile collection is highly invasive, which limits the application of biomarkers for cholangiocarcinoma. Studies of small molecule metabolites in serum have low sensitivity and are difficult to effectively distinguish between benign biliary tract diseases and cholangiocarcinoma.
Lipidome and peptideome biomarkers were used, and fingerprints were obtained using MALDI-MS technology. Specific biomarkers were screened by t-test and OPLS-DA analysis. An artificial neural network model was constructed for discrimination, and a discrimination kit was prepared.
It improves the sensitivity and accuracy of distinguishing between benign biliary tract diseases and bile duct cancer, provides intermediate result information, is suitable for non-invasive data collection, and has better model performance than CA19-9, making it suitable for risk prediction.
Smart Images
Figure CN116559274B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of biotechnology, more particularly, to a biomarker for distinguishing benign biliary disease and cholangiocarcinoma, a screening method and application thereof in preparing a reagent or kit for distinguishing benign biliary disease and cholangiocarcinoma, and application in constructing a risk prediction model for distinguishing benign biliary disease and cholangiocarcinoma. BACKGROUND
[0002] Researchers have made many efforts in determining potential biomarkers of cholangiocarcinoma (CCA), such as using genomics, transcriptomics, proteomics and metabolomics technologies to study potential biomarkers of CCA. In metabolomics research, lipid changes are more stable than other metabolites and can be related to human phenotypes, indicating the development and progression of diseases. So far, lipidomics has shown excellent performance in the early discrimination of various diseases. Polypeptidomics is a specific branch of proteomics, which is the characterization of proteins in biological samples and plays an important regulatory role in various biological processes. Therefore, it can also be used as a candidate biomarker for many diseases in clinical research. Currently, whether it is lipidomics or polypeptidomics, the research objects of CCA discrimination markers are mostly concentrated in bile, which is mainly because bile is closer to the location of CCA and is more likely to obtain CCA-specific biomarkers therefrom. However, bile collection is invasive and not all patients or all stages can perform such analysis, so the application of biomarkers in bile will be subject to many limitations. In contrast, serum is generally considered to be a metabolic pool reflecting metabolic disorders in the whole body of patients, and compared with bile, serum is more easily obtained in clinical practice, more suitable for non-invasive collection, and subject to fewer application limitations. However, current serum metabolomics research on CCA is mostly focused on small molecule metabolites that do not involve lipids, and the discrimination sensitivity is low.
[0003] Therefore, it is of great significance to mine high-sensitivity and high-specificity biomarkers for distinguishing benign biliary disease and cholangiocarcinoma from serum to obtain intermediate result information for guiding the discrimination of CCA. SUMMARY
[0004] In view of this, the purpose of the present application is to provide a biomarker for distinguishing benign biliary disease and cholangiocarcinoma and its application, which can improve the sensitivity of distinguishing benign biliary disease and cholangiocarcinoma and provide intermediate result information or reference for clinical application. The specific scheme is as follows:
[0005] In a first aspect, the present application provides a biomarker for distinguishing benign biliary disease and cholangiocarcinoma, comprising a lipidome and / or a polypeptidome.
[0006] The lipid group comprises: LPA (20:2), LPA (20:3), LPA (20:4), LPE (17:0), LPE (20:4), LPG (18:2), Cer (d40:0), Cer (d40:2), Cer (d41:0), Cer (d41:1), Cer (d41:2), CE (18:1), CE (18:2), CE (20:4), TG (54:3), TG (54:4), DG (38:4), PC (38:3), PC (36:4), and SM (d36:2).
[0007] The polypeptide group comprises: a polypeptide with a mass-to-charge ratio of 1259.6, a polypeptide with a mass-to-charge ratio of 1544.7, a polypeptide with a mass-to-charge ratio of 2084.9, a polypeptide with a mass-to-charge ratio of 3954.2, and a polypeptide with a mass-to-charge ratio of 4282.3.
[0008] Preferably, the polypeptide with a mass-to-charge ratio of 1259.6 is fibrinogen beta chain, and the amino acid sequence thereof is shown as SEQ ID NO: 1.
[0009] Preferably, the polypeptide with a mass-to-charge ratio of 1544.7 is fibrinopeptide A, and the amino acid sequence thereof is shown as SEQ ID NO: 2.
[0010] Preferably, the polypeptide with a mass-to-charge ratio of 3954.2 is intermediate alpha-globulin inhibitor H4, and the amino acid sequence thereof is shown as SEQ ID NO: 3.
[0011] In a second aspect, the present application further provides a use of the biomarker for distinguishing between benign biliary tract disease and cholangiocarcinoma in the preparation of a reagent or kit for distinguishing between benign biliary tract disease and cholangiocarcinoma.
[0012] The composition of the kit can comprise: a biomarker list, a silicon micro-nanoparticle, a lipid chip, a sample incubation and / or extraction solution, a buffer solution, and a matrix solution.
[0013] In a third aspect, the present application further provides a screening method of the biomarker for distinguishing between benign biliary tract disease and cholangiocarcinoma, comprising the following steps:
[0014] Step one, collecting serum of patients with benign biliary tract disease and cholangiocarcinoma as an analysis sample;
[0015] Step two, obtaining the lipidomics and polypeptidomics fingerprint of the analysis sample by MALDI-MS;
[0016] Step three, performing t-test analysis and OPLS-DA analysis on the lipidomics and polypeptidomics fingerprint, respectively, to screen the biomarker for distinguishing between benign biliary tract disease and cholangiocarcinoma in claim 1.
[0017] Preferably, the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma in claim 1 is screened in step three by simultaneously satisfying P < 0.05 in t-test analysis and VIP > 1 in OPLS-DA analysis as the screening standard.
[0018] In a fourth aspect, the application further provides the use of the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma in constructing a risk prediction model for benign biliary tract disease and cholangiocarcinoma.
[0019] Preferably, the risk prediction model for benign biliary tract disease and cholangiocarcinoma is a model constructed based on artificial neural network, least absolute shrinkage and selection operator or support vector machine algorithm.
[0020] Preferably, the risk prediction model for benign biliary tract disease and cholangiocarcinoma is a model constructed based on artificial neural network algorithm.
[0021] Compared with the prior art, the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma provided by the application includes 20 lipids and / or 5 polypeptides, and using a single lipid group or a single polypeptide group as a biomarker to distinguish benign biliary tract disease and cholangiocarcinoma patients has high sensitivity, and using a combination of lipid groups and polypeptide groups as a multi-group biomarker can further improve the discrimination sensitivity and accuracy. Integrated analysis of joint pathways and transcriptomics shows that the disturbance of lipids and polypeptides is an important process of CCA occurrence and progression, and further indicates that the 25 feature molecules provided by the application can be used as potential biomarkers, providing important intermediate result information for clinical application. In addition, the biomarker combination provided by the application exists in serum, which is more easily obtained in clinical practice and more suitable for non-invasive collection, and the identification reagent or kit prepared based on the biomarker combination is more convenient to use.
[0022] Based on the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma described above, the application further constructs a risk prediction model for benign biliary tract disease and cholangiocarcinoma, and three types of models are constructed by selecting artificial neural network (ANN), least absolute shrinkage and selection operator (LASSO) and support vector machine (SVM) three machine learning algorithms, respectively. It is found that the performance of the model constructed by artificial neural network (ANN) algorithm is the best, and the performance of the multi-omics model is better than that of the single lipid group or single polypeptide group model, and the sensitivity, specificity and accuracy of the best multi-omics ANN model are all more than 96%. Compared with the existing CA19-9 biomarker method, the single-omics model and the double-omics model provided by the application both have higher accuracy.
[0023] The application provides biomarkers for distinguishing benign biliary diseases and cholangiocarcinoma, a screening method and application of the biomarkers in constructing a risk prediction model for distinguishing benign biliary diseases and cholangiocarcinoma, which are all information methods as intermediate results and provide references for clinical application. BRIEF DESCRIPTION OF DRAWINGS
[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the provided drawings.
[0025] Figure 1 Principal component analysis of discovery set and quality control (QC) samples;
[0026] Figure 2 Pathway analysis by MetaboAnalyst 5.0 showing metabolic pathway changes in CCA and BBD patients;
[0027] Figure 3 ChemRICH enrichment statistics chart, each node reflects a significantly changed lipid cluster, and the node size represents the total number of lipids in each cluster;
[0028] Figure 4 OPLS-DA score plot in the discovery set lipidomics analysis;
[0029] Figure 5 OPLS-DA score plot in the discovery set polypeptideomics analysis;
[0030] Figure 6 OPLS-DA score plot in the discovery set multi-omics analysis;
[0031] Figure 7 Relative intensity change diagram of 25 potential biomarkers between CCA group and BBD group in the discovery set;
[0032] Figure 8 Relative intensity change diagram of 25 potential biomarkers between CCA group and BBD group in the verification set;
[0033] Figure 9 Correlation analysis diagram of potential biomarkers and serum clinical indicators, wherein, *: p<0.05, **: p<0.01, ***: p<0.001;
[0034] Figure 10 Joint pathway analysis diagram of 25 characteristic molecules;
[0035] Figure 11 Schematic diagram for biological process analysis;
[0036] Figure 12 Schematic diagram for molecular function;
[0037] Figure 13 Schematic diagram for cellular component analysis;
[0038] Figure 14 Schematic diagram for KEGG pathway analysis;
[0039] Figure 15 Schematic diagram for protein-protein interaction network of integrated genes;
[0040] Figure 16 Schematic diagram for the top 10 hub genes in connectivity in PPI network;
[0041] Figure 17 Schematic diagram for sensitivity, specificity and accuracy of risk prediction models based on different machine learning algorithms;
[0042] Figure 18 ROC curve of different risk prediction models and CA19-9. DETAILED DESCRIPTION
[0043] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0044] Example 1 Screening and functional analysis and verification of biomarkers
[0045] Step one, collect and analyze samples.
[0046] A total of 215 serum samples (ex vivo samples) were collected from the First Affiliated Hospital of Zhejiang University School of Medicine and the Sir Y.K. Pao Institute of Common Diseases, Zhejiang University School of Medicine, including 111 patients with benign biliary tract disease (BBD) and 104 patients with malignant cholangiocarcinoma (CCA). Serum samples from different hospitals were randomly assigned to the discovery group and the verification group, and the demographic and clinical information of the patients is shown in Table 1, in which: HDL represents high-density lipoprotein; LDL represents low-density lipoprotein; VLDL represents very low-density lipoprotein; GGT represents gamma-glutamyltransferase; AST represents aspartate aminotransferase; ALT represents alanine aminotransferase; AFP represents alpha-fetoprotein; CEA represents carcinoembryonic antigen; P value: between CCA and BBD cohorts.
[0047] Table 1. Patient demographics and serum clinical indicators
[0048]
[0049]
[0050] Step 2: Obtain lipidomics and peptidomics information of the analyzed samples.
[0051] Lipidomics and peptidomics fingerprints of serum samples were obtained using a high-throughput laser desorption / ionization mass spectrometry (MALDI-MS) platform.
[0052] Serum lipidomics detection: Lipidomics fingerprinting and analysis were performed using an UltrafleXtreme MALDI-TOF-MS instrument (Bruker Daltonics Corp.) equipped with a 355nm Nd:YAG laser beam. Serum lipid samples extracted with organic solvents were added to the detection wells of a mass spectrometry lipid chip. After sample drying, the lipid chip was fed into the mass spectrometer for detection. Mass spectrometry data were obtained in both positive and negative ion reflectance modes, with a mass range of m / z = 400-1000. Other instrument parameters were set as follows: 100µm laser spot, 100ns pulsed ion extraction, 19kV (ion source 1), and 16.75kV (ion source 2) accelerating voltage. 500 laser irradiations were performed at a single point in the detection well, superimposed four times to generate the mass spectrum.
[0053] Serum peptidomics detection: Peptidomics fingerprinting and analysis was performed using porous silicon micro / nanoparticles to selectively capture serum peptides and analyze them using MALDI-TOF-MS (…). Data acquisition and analysis were performed using an instrument (I, Huijian Technology, China). The detection mass range was 0.6–20 kDa. Peptide mass spectrometry data were acquired in linear positive ion mode, using delayed extraction with an accelerating voltage of 25 kV and a pulsed ion extraction time of 220 ns.
[0054] Principal component analysis (PCA) was performed on the lipidomics fingerprint, and the results are as follows: Figure 1 As shown, CCA patients and BBD patients overlapped and were distributed over a wide area, while the quality control (QC) samples, a mixture of 50 healthy samples, clustered in areas distinct from both BBD and CCA patients. This indicates that the method for obtaining lipidomics fingerprints based on the MALDI-MS platform in this invention has good stability and reproducibility. Furthermore, the wide distribution of BBD and CCA samples suggests heterogeneity among individual patients.
[0055] Although PCA failed to separate BBD and CCA groups, 93 P<0.05 feature lipids were found by analysis. Functional analysis of these feature lipids was performed using MetaboAnalyst 5.0, and the results are shown in FIG. 1, where changes in glycerophospholipid metabolism, sphingolipid metabolism, glycerolipid metabolism, and linoleic acid metabolism had the most significant impact on the lipidome of CCA. Figure 2 Further alternative enrichment analysis based on chemical similarity (ChemRICH) showed that the metabolism of several classes of lipids was disturbed, as shown in FIG. 2. Figure 3
[0056] Further biomarkers were explored based on orthogonal partial least squares discriminant analysis (OPLS-DA), as shown in FIG. 3 and FIG. 4, based on single lipidome or single peptidome, BBD and CCA groups could be partially clustered and separated, while as shown in FIG. 5, based on comprehensive analysis of lipidome and peptidome, compared with single lipidome or single peptidome, BBD and CCA groups could be more accurately clustered and separated, which further proved the excellent performance of multi-omics in the discrimination of benign biliary tract disease and cholangiocarcinoma. Figure 4 5 Figure 6
[0057] Step three, screening specific biomarkers.
[0058] t-test analysis and OPLS-DA analysis were performed on the lipidomic and peptidomic fingerprints, respectively, to meet the screening criteria of P<0.05 in t-test analysis and VIP>1 in OPLS-DA analysis at the same time, and 25 potential biomarkers were screened out, including 20 lipids and 5 polypeptides.
[0059] The statistical analysis information of the biomarkers is shown in Table 2, and the biomarkers are arranged in descending order of VIP, where: LPE represents lysophosphatidylethanolamine; LPA represents lysophosphatidic acid; LPG represents lysophosphatidylglycerol; TG represents triglyceride; CE represents cholesteryl ester; DG represents diglyceride; Cer represents ceramide; PC represents phosphatidylcholine; and SM represents sphingomyelin.
[0060] Table 2. Biomarker statistical analysis information
[0061]
[0062]
[0063] The specific information of the 5 polypeptides (Peptide 1-5) is shown in Table 3, and the mass-to-charge ratios of Peptide 1-5 are 1259.6, 1544.7, 2084.9, 3954.2, and 4282.3, respectively. After identification, Peptide 1 is Fibrinogen beta chain, the amino acid sequence of which is shown in SEQ ID NO: 1; Peptide 2 is Fibrinopeptide A, the amino acid sequence of which is shown in SEQ ID NO: 2; and Peptide 4 is Inter-alpha (Globulin) inhibitor H4, the amino acid sequence of which is shown in SEQ ID NO: 4.
[0064] Table 3: Information of characteristic polypeptides
[0065]
[0066] From Figure 7 and 8 it can be seen that the relative intensity changes of the 25 different biomarker molecules in the discovery group ( Figure 7 ) are consistent with those in the validation group ( Figure 8 ), indicating that the screened biomarker molecules have high reliability.
[0067] Further correlation analysis found that the 25 potential biomarkers screened exist correlations with serum clinical indicators (as shown in Figure 9 ), although the existing serum clinical indicators (bilirubin, triglyceride, low-density lipoprotein (LDL), very low-density lipoprotein (VLDL), gamma-glutamyl transferase (GGT), aspartate aminotransferase (AST), and alanine aminotransferase (ALT)) cannot accurately diagnose CCA and BBD, but they also have statistical differences between CCA and BBD (all p<0.05), further indicating that the 25 potential biomarkers are closely related to the progression of CCA.
[0068] Lipid and polypeptide joint pathway analysis showed that the biomarker molecules play an important role in various biological pathways such as fat digestion and absorption, platelet activation, cholesterol metabolism, complement and coagulation cascade (as shown in Figure 10 ).
[0069] To further understand the regulation process of biomarkers in the genome-scale metabolic network, the lipids and polypeptides in the biomarkers were mapped to their corresponding genes, thereby identifying 334 genes. Among them, 75 genes were consistent with the genes that changed significantly in TCGA. Functional enrichment analysis was performed on these genes to reveal their potential functions. The results showed that these genes were significantly enriched in biological processes such as glycerolipid metabolism, triglyceride metabolism, neutral lipid metabolism, plasma lipoprotein particle level regulation, and protein-lipid complex remodeling (as shown in Figure 11 ); the molecular function (MF) was rich in lipase activity, phospholipid binding, and phospholipid transport protein activity (as shown in Figure 12 ); and the genes in the cell component (CC) group were mainly enriched in high-density lipoprotein particles, plasma lipoprotein particles, protein-lipid complexes, and chylomicrons (as shown in Figure 13 ).
[0070] KEGG analysis showed that, in addition to the pathways involved in the previous pathway analysis (as shown in Figure 2 ), the lipids and polypeptides in the biomarkers were also related to cholesterol metabolism, fat digestion and absorption, and PPAR signaling pathways (as shown in Figure 14 ). PPAR belongs to the nuclear hormone receptor family, and according to recent studies, it plays an important role in the occurrence and development of CCA and is closely related to lipid metabolism. The above research results show that the genes that change during CCA significantly affect a variety of lipid and protein-related pathways, which is also consistent with our results of lipidomics and polypeptidomics. In addition, PPI network analysis was performed on the aforementioned 75 genes (as shown in Figure 15 ), and CytoHubba was used to identify central genes, which illustrated the connectivity of the top 10 genes in the PPI network (as shown in Figure 16 ). These genes were also mainly related to the coagulation cascade and apolipoprotein.
[0071] The above analysis fully proves from different angles that the lipids and / or polypeptides screened in this embodiment have an important indicating effect on the occurrence and development of CCA, and can be applied to the preparation of a benign biliary disease and cholangiocarcinoma discrimination reagent or kit as a biomarker, which has very high practical application value.
[0072] Example 2: Construction of a benign biliary disease and cholangiocarcinoma risk prediction model
[0073] Three machine learning algorithms—Artificial Neural Network (ANN), Least Absolute Shrinkage Selection Operator (LASSO), and Support Vector Machine (SVM)—were employed to construct different risk prediction models for 20 biomarkers in single-lipomics, 5 biomarkers in single-peptidomics, and 25 biomarkers in multi-omics. In the ANN pattern recognition model, a multilayer perceptual structure with 10 hidden neurons was constructed. Under the LASSO model, based on a generalized linear model, the model parameters were optimized to find the optimal variable values. Under the SVM model, based on the principle of structural risk minimization, the regularization parameters and kernel function parameters were adjusted to construct the optimal risk prediction model.
[0074] Based on the validation set samples, the sensitivity, specificity, and accuracy of the above model predictions were tested using a 10-fold cross-validation method. The results are as follows: Figure 17 As shown, the three peptide proteomics models all have a sensitivity, specificity, and accuracy exceeding 80%, the three lipidomics models all have a sensitivity, specificity, and accuracy exceeding 88%, and the three dual-omics models all have a sensitivity, specificity, and accuracy exceeding 92%. This indicates that both single-omics and dual-omics models have excellent performance, with the dual-omics model showing better predictive performance.
[0075] By comparing the performance of models built using different algorithms, it can be seen that the model built using artificial neural networks (ANN) has the best performance.
[0076] The dual-omics model constructed using artificial neural networks (ANN) achieved a prediction sensitivity of 96.53%, a specificity of 96.35%, and an accuracy of 96.44%.
[0077] CA19-9 is an important serum biomarker for CCA discrimination, and it was further compared with single-omics and dual-omics models. Figure 18 As shown, ROC analysis revealed that the AUC value of the dual-omics model was 0.99, the AUC value of the lipidomics model was 0.936, and the AUC value of the peptideomics model was 0.884, all of which demonstrated better predictive performance compared to CA19-9.
[0078] The foregoing description of the disclosed embodiments enables a person skilled in the art to make or use the application. Modifications of these embodiments will occur to persons of skill in the art, and, while certain modifications are discussed, it is desired to be protected in accordance with the spirit and scope of the application. Therefore, the application is not limited to the specific embodiments shown and described, but only by the scope of the appended claims, unless otherwise specified.
Claims
1. A biomarker for discriminating between benign biliary disease and cholangiocarcinoma, characterized by, The biomarker is a lipidome; The lipidome is: LPA (20:2), LPA (20:3), LPA (20:4), LPE (17:0), LPE (20:4), LPG (18:2), Cer (d40:0), Cer (d40:2), Cer (d41:0), Cer (d41:1), Cer (d41:2), CE (18:1), CE (18:2), CE (20:4), TG (54:3), TG (54:4), DG (38:4), PC (38:3), PC (36:4), and SM (d36:2).
2. Use of the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma in claim 1 in the preparation of a reagent or kit for distinguishing benign biliary tract disease and cholangiocarcinoma.
3. The use according to claim 2, comprising subjecting the serum sample to be tested to MALDI-MS detection, and analyzing the lipidome fingerprint.
4. The use of claim 3, further comprising analyzing a polypeptide set fingerprint, the polypeptide set comprising: polypeptide with a mass-to-charge ratio of 1259.6, a polypeptide with a mass-to-charge ratio of 1544.7, a polypeptide with a mass-to-charge ratio of 2084.9, a polypeptide with a mass-to-charge ratio of 3954.2, and a polypeptide with a mass-to-charge ratio of 4282.
3.
5. The use according to claim 4, wherein the polypeptide with a mass-to-charge ratio of 1259.6 is fibrinogen beta chain, the amino acid sequence of which is shown as SEQ ID NO: 1; the polypeptide with a mass-to-charge ratio of 1544.7 is fibrinopeptide A, the amino acid sequence of which is shown as SEQ ID NO: 2; and the polypeptide with a mass-to-charge ratio of 3954.2 is intermediate alpha-globulin inhibitor H4, the amino acid sequence of which is shown as SEQ ID NO:
3.
6. Use of the biomarker for distinguishing benign biliary tract disease and cholangiocarcinoma in claim 1 in the construction of a benign biliary tract disease and cholangiocarcinoma risk prediction model, wherein the benign biliary tract disease and cholangiocarcinoma risk prediction model is a model constructed based on an artificial neural network, a least absolute shrinkage and selection operator, or a support vector machine algorithm.