MIBC multi-omics molecular typing method and prediction system

By integrating transcriptome and protein expression data, screening and clustering analysis, a multi-omic molecular classification and subtype prediction system for muscular invasive bladder cancer was constructed, which solved the prognosis judgment problem caused by high heterogeneity of MIBC, and achieved Supports accurate prediction of bladder cancer prognosis and personalized treatment.

CN117577300BActive Publication Date: 2024-11-19SHANGHAI RENDONG MEDICAL LAB CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311494335.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-10
Publication Date
2024-11-19
Estimated Expiration
2043-11-10

AI Technical Summary

Technical Problem

Due to the high heterogeneity of myometrial invasive bladder cancer (MIBC), the prognosis and treatment response vary greatly. It is difficult for the existing technology to accurately judge the prognosis and improve the treatment effect. The single-omic data is not comprehensive enough for disease typing.

Method used

By integrating transcriptome gene abundance expression and protein abundance expression data, MIBC-specific genes were screened out, and a multiomic consensus clustering analysis and multivariate linear regression method were used to construct a predictive model of subtypes of muscle-infiltrating bladder cancer. , provides molecular typing and subtype prediction systems.

Benefits of technology

It achieves accurate prediction of the prognosis of muscular invasive bladder cancer, provides an effective reference for personalized medical care, and improves the accuracy and effectiveness of treatment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117577300B_ABST
    Figure CN117577300B_ABST
Patent Text Reader

Abstract

The present invention discloses a multi-omics molecular typing method for muscle-invasive bladder cancer. The method performs cluster analysis on muscle-invasive bladder cancer based on the transcriptome gene abundance expression characteristics and protein abundance expression characteristics of muscle-invasive bladder cancer patients to obtain molecular subtypes of muscle-invasive bladder cancer associated with prognosis. The present invention also discloses a method and system for predicting multi-omics molecular subtypes of muscle-invasive bladder cancer. The present invention integrates and screens the abundance expression data of transcriptomes and proteomes, and uses a multi-omics consistency clustering method to determine the multi-omics molecular typing of muscle-invasive bladder cancer, and constructs a molecular subtype prediction system, thereby facilitating the accurate prediction of the multi-omics molecular subtypes of muscle-invasive bladder cancer, indicating the prognosis of patients, and providing an effective reference for precision medicine.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of urological oncology medicine, in particular to muscle invasive bladder cancer (MIBC), and more specifically to multi-omics molecular typing of MIBC. Background Art

[0002] Bladder cancer can be divided into invasive bladder cancer (MIBC) and non-invasive bladder cancer, of which MIBC is more invasive and more prone to metastasis. Due to the high heterogeneity of MIBC, the clinical course and prognosis of patients vary. MIBC can be divided into different risk levels according to different risk factors such as tumor number, size, stage, grade, recurrence frequency, and the presence of carcinoma in situ (Tis). Patients will choose appropriate treatment plans based on the risk level grouping. However, even patients with the same clinical and pathological characteristics and the same risk level still have heterogeneity in their diseases, and the disease progression and response to treatment vary greatly. Therefore, how to find effective molecular biomarker detection to accurately judge patient prognosis and improve treatment benefits is a difficult problem that needs to be solved in clinical practice.

[0003] With the development of omics technology, some studies have proposed molecular classification to systematically describe tumors at the molecular level. The University of North Carolina and the Anderson Cancer Center proposed a two-type and three-type system based on gene expression characteristics for MIBC, respectively. In 2017, the TCGA team used transcriptome data to divide MIBC into five subtypes: luminal, luminal-infiltrated, basal-squamous, neural, and luminal-papillary. These five subtypes showed different tissue morphological characteristics and pathological characteristics (Robertson AG, Kim J, Al-Ahmadie H, et al. Comprehensive molecular characterization of Muscle-Invasive bladder cancer. Cell 2018; 174: 1033). In 2019, Aurélie Kamoun et al. integrated the previous multiple MIBC transcriptome typing to obtain a consensus molecular typing (Kamoun A, de Reynies A, Allory Y, et al. A Consensus Molecular Classification of Muscle-invasive Bladder Cancer. Eur Urol 2020; 77(4): 420-33). The consensus molecular typing divides MIBC into six subtypes: LumP, LumNS, LumU, stroma-rich, Ba / Sq and NE-like. There are significant differences in prognosis and molecular characteristics between different subtypes, which can provide a reference for treatment. However, current studies are mostly based on single-omics data to classify the disease, and the understanding of the molecular landscape of the tumor is still incomplete. It is necessary to integrate other levels of omics information to further comprehensively explore the occurrence and development of the disease, help to fully understand the mechanism of tumor occurrence and development and discover potential targets for clinical treatment, and provide new directions for optimizing tumor monitoring and treatment strategies. Summary of the invention

[0004] One of the technical problems to be solved by the present invention is to provide a multi-omics molecular typing method for muscle-invasive bladder cancer, which can assist the prognosis of muscle-invasive bladder cancer.

[0005] In order to solve the above technical problems, the multi-omics molecular typing method of muscle-invasive bladder cancer of the present invention performs multi-omics molecular typing of muscle-invasive bladder cancer based on the transcriptome gene abundance expression characteristics and protein abundance expression characteristics of muscle-invasive bladder cancer patients. Specifically, it comprises the following steps:

[0006] 1) Obtain transcriptome gene abundance expression data and protein abundance expression data of MIBC patients and normal control tissues;

[0007] 2) Analyze the differences in transcriptome gene abundance expression and protein abundance expression between tumor tissues and adjacent tissues, and screen out genes with the same significant change trend in transcriptome gene abundance expression and protein abundance expression as MIBC-specific genes;

[0008] 3) Integrate and cluster the abundance expression characteristics of the MIBC-specific genes and their corresponding proteins to obtain the best MIBC typing results.

[0009] In the above step 1), for protein abundance expression data, the KNN method can be used to fill missing values; for transcriptome gene abundance expression data, the TPM standardization method can be selected for processing. The TPM standardization conversion formula is:

[0010]

[0011] Where i is the sample number, j is the gene number, and R ij is the read count value of gene j in sample i, F ij is the FPKM value of gene j in sample i, L j is the length of the coding region of gene j, T i is the number of sequencing reads of sample i.

[0012] In the above step 2), the abundance expression of transcriptome genes in tumor tissue and adjacent tissue is differentially analyzed to screen out differential genes, and then the abundance expression of proteins corresponding to the differential genes in tumor tissue and normal human bladder tissue is differentially analyzed to screen out differential proteins, and genes that are significantly overexpressed or significantly underexpressed at both the transcriptome and protein levels are retained as MIBC-specific genes, and the proteins corresponding to the MIBC-specific genes are retained.

[0013] In the above step 3), before integrating the abundance expression characteristics of MIBC-specific genes and their corresponding proteins, the transcriptome abundance expression data of MIBC-specific genes can be log2 transformed. Cluster analysis preferably adopts a multi-omics consistency cluster analysis method, selects the point with the highest Gap statistic and subtype prediction index as the best cluster, and obtains clustering results using more than three clustering algorithms, and then integrates these clustering results to obtain the best typing result.

[0014] The second technical problem to be solved by the present invention is to provide a method for predicting molecular subtypes of muscle-invasive bladder cancer, which also predicts molecular subtypes of muscle-invasive bladder cancer based on transcriptome gene abundance expression characteristics and protein abundance expression characteristics of patients with muscle-invasive bladder cancer. Specifically, after screening out MIBC-specific genes according to the aforementioned method, the following steps are then performed:

[0015] 1) The sample set with both transcriptome gene abundance expression data and protein abundance expression data was used as a training set to screen MIBC-specific genes and their corresponding proteins to obtain molecular subtype predictive markers for muscle-invasive bladder cancer;

[0016] 2) Input the expression abundance of the muscle-invasive bladder cancer molecular subtype prediction markers and the real molecular subtype information of each sample in the training set, perform model training, and obtain a muscle-invasive bladder cancer molecular subtype prediction model;

[0017] 3) Inputting the expression abundance of the muscle-invasive bladder cancer molecular subtype prediction marker of the sample to be tested into the muscle-invasive bladder cancer molecular subtype prediction model to obtain the muscle-invasive bladder cancer molecular subtype prediction result of the sample to be tested.

[0018] Wherein, the muscle-invasive bladder cancer molecular subtype prediction markers preferably include the IDs ENSG00000133392.18, ENSG00000108773.11, ENSG00000105974.13, ENSG00000150593.18, ENSG00000073050.12, ENSG0000013504 6.14, ENSG00000164305.19, ENSG00000082701.17, ENSG00000204389.10, ENSG00000168610.16, ENSG00000064012.22 and 6 protein markers, MYH11, GCN5L2, PDCD4, GSK3_pS9, HSP70, and SRC_pY527.

[0019] The third technical problem to be solved by the present invention is to provide a muscle-invasive bladder cancer molecular subtype prediction system, which includes:

[0020] An input module, used for inputting the abundance expression data of the muscle-invasive bladder cancer molecular subtype prediction gene markers and protein markers of the muscle-invasive bladder cancer patients;

[0021] A prediction module, wherein the prediction module has a built-in muscle-invasive bladder cancer molecular subtype prediction model, and is used to calculate the patient's muscle-invasive bladder cancer molecular subtype prediction result based on the patient's muscle-invasive bladder cancer molecular subtype prediction gene marker and protein marker abundance expression data;

[0022] The output module is used to output the prediction results of the muscle-invasive bladder cancer molecular subtypes calculated by the prediction module.

[0023] The fourth technical problem to be solved by the present invention is to provide a kit, which includes reagents for detecting the abundance expression of the above-mentioned muscle-invasive bladder cancer molecular subtype prediction markers.

[0024] The present invention integrates and screens the abundance expression characteristics of transcriptome genes and proteins in patients with muscle-invasive bladder cancer, and uses a multi-omics consistency clustering analysis method to determine the molecular typing associated with the prognosis of muscle-invasive bladder cancer. The multivariate linear regression method is further used to screen typing prediction markers, and a nearest neighbor algorithm is used to construct a system that can accurately predict the multi-omics molecular subtypes of muscle-invasive bladder cancer, which helps to accurately predict the prognosis of patients with muscle-invasive bladder cancer and provide an effective reference for personalized precision medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is the optimal cluster number selection diagram of Example 1 of the present invention.

[0026] Figure 2 It is a heat map of the consistent clustering results obtained by integrating 10 methods in Example 1 of the present invention.

[0027] Figure 3 Example 1 of the present invention is based on the fact that multi-omics subtypes are significantly correlated with disease-specific survival. DETAILED DESCRIPTION

[0028] In order to have a more specific understanding of the technical content, characteristics and effects of the present invention, the technical solution of the present invention is now described in detail in conjunction with the accompanying drawings and specific embodiments.

[0029] Example 1 Multi-omics molecular typing of muscle-invasive bladder cancer

[0030] 1. Collect gene abundance expression data and protein abundance expression data of MIBC and normal control tissues

[0031] Transcriptome gene abundance expression data and protein abundance expression data corresponding to the BLCA dataset were downloaded from the TCGA database. In addition, protein abundance expression data of normal human bladder tissue were collected from the literature (Jung M, Lee C, Han D, et al. Proteomic-Based Machine Learning Analysis Reveals PYGB as a Novel Immunohistochemical Biomarker to Distinguish Inverted Urothelial Papilloma From Low-Grade Papillary Urothelial Carcinoma With Inverted Growth. Front Oncol. 2022 Mar 24; 12: 841398). A total of 412 patient transcriptome data, 19 paracancerous transcriptome data, 344 patient proteome data, and 10 normal human bladder tissue proteome data were included, of which 339 patients had both transcriptome and proteome data. TPM normalization was performed on the transcriptome data, and the KNN method was used to supplement the missing values ​​of the proteome data.

[0032] MIBC-specific gene screening

[0033] The R package limma was used to perform differential analysis on the abundance expression of transcriptome genes in tumor tissues and adjacent tissues, and genes with p values ​​less than 0.05 and logFC greater than 1 or less than -1 were screened as differential genes. Then, the Wilcoxon test method was used to perform differential analysis on the abundance expression of proteins corresponding to differential genes in tumor tissues and normal human bladder tissues, and proteins with p values ​​less than 0.05 were screened as differential proteins. Genes and their corresponding proteins that were significantly over-expressed or significantly under-expressed at both the transcriptome and protein levels were retained, and a total of 19 genes and 24 proteins were obtained, as shown in Table 1:

[0034] Table 1 MIBC-specific genes and corresponding proteins

[0035]

[0036] 3. Multi-omics consensus cluster analysis determines multi-omics molecular subtypes of muscle-invasive bladder cancer

[0037] The abundance expression of the 19 genes in Table 1 was log2 transformed, and the abundance expression data of the 24 proteins in Table 1 were integrated and input into the getClustNum function in the MOVICS package of the R language to calculate the Gap statistic and subtype prediction index (such as Figure 1As shown in the figure, the highest point (k = 3) is selected as the best clustering; secondly, the getMOIC function is used, and a total of 10 clustering algorithms, including SNF, PINSPlus, NEMO, COCA, LRAcluster, ConsensusClustering, IntNMF, MoCluster, iClusterBayes and CIMLR, are selected, and N.clust is set to 3 to obtain the typing results corresponding to each algorithm. Finally, the getConsensusMOIC function is used to integrate the consistency of these 10 clustering results to obtain the final typing result, as shown in the figure. Figure 2 As shown, muscle-invasive bladder cancer can be divided into three molecular subtypes.

[0038] 4. Validation of the correlation between multi-omics molecular subtypes and prognosis in muscle-invasive bladder cancer

[0039] Survival analysis was performed on the multi-omics molecular subtypes of muscle-invasive bladder cancer obtained above. Among them, 15 patients did not have complete survival information (missing survival time or missing survival status), so they were eliminated from the survival analysis. The remaining 324 patients were subjected to Kaplan-Meier survival analysis, and the p value was 0.0014, indicating that the multi-omics molecular subtypes of muscle-invasive bladder cancer were significantly correlated with prognosis. According to the length of survival, the three subtypes were named long survival type, intermediate survival type and short survival type, as shown in Figure 2. Figure 3 shown.

[0040] Example 2 Construction of a multi-omics molecular subtype prediction system for muscle-invasive bladder cancer

[0041] 1. Screening of gene markers and protein markers

[0042] The 339 samples with both transcriptome gene abundance expression data and protein abundance expression data were randomly divided into 70% training set and 30% test set. In the training set, the MIBC-specific genes and their corresponding proteins were screened using the MultiTaskLasso function in the python package scikit-learn, with the parameter alpha = 0.006, and genes or proteins with coefficients greater than 0 were retained as markers. Finally, 11 gene markers and 6 protein markers were obtained, as shown in Table 2:

[0043] Table 2 Gene and protein markers

[0044] Markers coefficient type MYH11 0.0596 protein GCN5L2 0.0021 protein PDCD4 0.0093 protein GSK3_pS9 0.0051 protein HSP70 0.0130 protein SRC_pY527 0.0486 protein ENSG00000133392.18 0.1052 Gene ENSG00000108773.11 0.0063 Gene ENSG00000105974.13 0.0053 Gene ENSG00000150593.18 0.0270 Gene ENSG00000073050.12 0.0057 Gene ENSG00000135046.14 0.0317 Gene ENSG00000164305.19 0.0023 Gene ENSG00000082701.17 0.0414 Gene ENSG00000204389.10 0.0004 Gene ENSG00000168610.16 0.0754 Gene ENSG00000064012.22 0.0078 Gene

[0045] 2. Multi-omics prediction of muscle-invasive bladder cancer subtypes

[0046] The nearest neighbor model was constructed using the NearestCentroid function in the python package scikit-learn. The marker expression abundance and multi-omics subtype information of the training sample were input, and the parameter metric = 'correlation' was set to train the model. Then the marker expression abundance of the test sample was input into the trained model to obtain the multi-omics subtype prediction results of the test sample, which were compared with the actual multi-omics subtypes of the sample, and the classification accuracy was 91.07%, as shown in Table 3.

[0047] Table 3 Test results of the multi-omics subtype prediction system for muscle-invasive bladder cancer

[0048]

[0049] The above embodiments are merely feasible or preferred embodiments of the present invention and are used to illustrate the present invention rather than to limit the scope of the patent application of the present invention. Therefore, all equivalent changes and modifications made according to the scope of the patent application of the present invention should fall within the scope covered by the patent of the present invention.

Claims

1. A multi-omics molecular typing method for muscle-invasive bladder cancer, which is not intended for the diagnosis and treatment of the disease. It is characterized in that Multi-omics molecular typing of muscle-invasive bladder cancer is performed based on the transcriptome gene abundance expression characteristics and protein abundance expression characteristics of patients with muscle-invasive bladder cancer, including the following steps: 1) Obtain transcriptome gene abundance expression data and protein abundance expression data of MIBC patients and normal control tissues; 2) Analyze the differences in transcriptome gene abundance expression and protein abundance expression between tumor tissues and adjacent tissues, and screen out genes with the same significant change trend in transcriptome gene abundance expression and protein abundance expression as MIBC-specific genes; 3) Integrate and cluster the abundance expression characteristics of the MIBC-specific genes and their corresponding proteins to obtain the best MIBC typing results; In the above step 2), the abundance expression of transcriptome genes in tumor tissue and adjacent tissue is differentially analyzed to screen out differential genes, and then the abundance expression of proteins corresponding to differential genes in tumor tissue and normal human bladder tissue is differentially analyzed to screen out differential proteins, and genes with significant high expression or significant low expression at both transcriptome and protein levels are retained as MIBC-specific genes, and proteins corresponding to the MIBC-specific genes are retained; the MIBC-specific genes and corresponding proteins are shown in Table 1: Table 1 MIBC-specific genes and corresponding proteins 2. The method according to claim 1, It is characterized in that Muscle-invasive bladder cancer is divided into three molecular subtypes: long-survival type, intermediate-survival type, and short-survival type.

3. The method according to claim 1, It is characterized in that Step 1), use the KNN method to fill missing values ​​in the protein abundance expression data; perform TPM standardization on the transcriptome gene abundance expression data, and the TPM standardization conversion formula is: Where i is the sample number, j is the gene number, and R ij is the read count value of gene j in sample i, F ij is the FPKM value of gene j in sample i, L j is the length of the coding region of gene j, T i is the number of sequencing reads of sample i.

4. The method according to claim 1, It is characterized in that Step 3), before integrating the abundance expression characteristics of MIBC-specific genes and their corresponding proteins, log2 transformation of the transcriptome abundance expression data of MIBC-specific genes was performed.

5. The method according to claim 1, It is characterized in that Step 3) adopts a multi-omics consistency clustering analysis method, obtains clustering results using three or more clustering algorithms, and then integrates these clustering results to obtain the best typing result.

6. The method according to claim 5, It is characterized in that Step 3), the clustering algorithms include SNF, PINSPlus, NEMO, COCA, LRAcluster, ConsensusClustering, IntNMF, MoCluster, iClusterBayes, and CIMLR.

7. The method according to claim 1 or 5, It is characterized in that Step 3), select the point with the highest Gap statistic and subtype prediction index as the best cluster.

8. A method for predicting molecular subtypes of muscle-invasive bladder cancer, which method is not intended for diagnosis or treatment of the disease. It is characterized in that Predicting the molecular subtype of muscle-invasive bladder cancer based on transcriptome gene abundance expression characteristics and protein abundance expression characteristics of patients with muscle-invasive bladder cancer includes the following steps: 1) Obtain transcriptome gene abundance expression data and protein abundance expression data of MIBC patients and normal control tissues; 2) Analyze the differences in transcriptome gene abundance expression and protein abundance expression between tumor tissues and adjacent tissues, and screen out genes with the same significant change trend in transcriptome gene abundance expression and protein abundance expression as MIBC-specific genes; 3) The samples with both transcriptome gene abundance expression data and protein abundance expression data were used as training sets to screen MIBC-specific genes and their corresponding proteins to obtain molecular subtype predictive markers for muscle-invasive bladder cancer; 4) Input the expression abundance of the muscle-invasive bladder cancer molecular subtype prediction markers and the real molecular subtype information of each sample in the training set, perform model training, and obtain a muscle-invasive bladder cancer molecular subtype prediction model; 5) inputting the expression abundance of the muscle-invasive bladder cancer molecular subtype prediction marker of the sample to be tested into the muscle-invasive bladder cancer molecular subtype prediction model to obtain the muscle-invasive bladder cancer molecular subtype prediction result of the sample to be tested; In the above step 2), the abundance expression of transcriptome genes in tumor tissue and adjacent tissue is differentially analyzed to screen out differential genes, and then the abundance expression of proteins corresponding to differential genes in tumor tissue and normal human bladder tissue is differentially analyzed to screen out differential proteins, and genes with significant high expression or significant low expression at both transcriptome and protein levels are retained as MIBC-specific genes, and proteins corresponding to the MIBC-specific genes are retained; the MIBC-specific genes and corresponding proteins are shown in Table 1: Table 1 MIBC-specific genes and corresponding proteins 9. The prediction method according to claim 8, It is characterized in that Step 1), use the KNN method to fill missing values ​​in the protein abundance expression data; perform TPM standardization on the transcriptome gene abundance expression data, and the TPM standardization conversion formula is: Where i is the sample number, j is the gene number, and R ij is the read count value of gene j in sample i, F ij is the FPKM value of gene j in sample i, L j is the length of the coding region of gene j, T i is the number of sequencing reads of sample i.

10. The prediction method according to claim 8, It is characterized in that The muscle-invasive bladder cancer molecular subtype prediction markers include the following IDs in the Ensembl database: ENSG00000133392.18, ENSG00000108773.11, ENSG00000105974.13, ENSG00000150593.18, ENSG00000073050.12, ENSG00000135046. 14, 11 gene markers including ENSG00000164305.19, ENSG00000082701.17, ENSG00000204389.10, ENSG00000168610.16, and ENSG00000064012.22, and 6 protein markers including MYH11, GCN5L2, PDCD4, GSK3_pS9, HSP70, and SRC_pY527.

11. Muscle-invasive bladder cancer molecular subtype prediction system, Features: include: An input module, used for inputting the abundance expression data of the muscle-invasive bladder cancer molecular subtype prediction gene markers and protein markers of the muscle-invasive bladder cancer patients; A prediction module, wherein the prediction module has a built-in muscle-invasive bladder cancer molecular subtype prediction model, and is used to calculate the patient's muscle-invasive bladder cancer molecular subtype prediction result based on the patient's muscle-invasive bladder cancer molecular subtype prediction gene marker and protein marker abundance expression data; An output module, used for outputting the prediction results of muscle-invasive bladder cancer molecular subtypes calculated by the prediction module; The gene markers include those in the Ensembl database with IDs ENSG00000133392.18, The 11 genes of ENSG00000108773.11, ENSG00000105974.13, ENSG00000150593.18, ENSG00000073050.12, ENSG00000135046.14, ENSG00000164305.19, ENSG00000082701.17, ENSG00000204389.10, ENSG00000168610.16, and ENSG00000064012.22, and the protein markers include MYH11, GCN5L2, PDCD4, GSK3_pS9, HSP70, and SRC_pY527.

12. The prediction system according to claim 11, Features: The subtypes include long-survival type, intermediate-survival type and short-survival type.

13. Test kit, It is characterized in that The invention comprises a reagent for detecting the abundance expression of a muscle-invasive bladder cancer molecular subtype prediction marker, wherein the muscle-invasive bladder cancer molecular subtype prediction marker comprises a gene marker and a protein marker, and the gene marker comprises the IDs ENSG00000133392.18, ENSG00000108773.11, ENSG00000105974.13, ENSG00000150593.18, ENSG0000007305 in the Ensembl database 0.12, ENSG00000135046.14, ENSG00000164305.19, ENSG00000082701.17, ENSG00000204389.10, ENSG00000168610.16, ENSG00000064012.22, and the protein markers included MYH11, GCN5L2, PDCD4, GSK3_pS9, HSP70, and SRC_pY527.