Bladder cancer prognosis prediction markers, prediction models and construction methods thereof

Abstract

The present invention discloses a group of bladder cancer prognosis prediction markers, including the following genes: BAIAP2, BCL2L14, OAS1, C6orf62, EMP1, TMEM109. The present invention also discloses a method for screening the above markers, a method for constructing a bladder cancer prognosis prediction model based on the above markers, and the resulting model. The present invention uses TCGA-BLCA as a training set, and screens out gene markers that are significantly correlated with the overall survival of bladder cancer patients through the transcriptome expression and corresponding survival data of bladder cancer patients, and establishes a bladder cancer prognosis prediction model based on these gene markers. The model can accurately and effectively predict the overall survival of bladder cancer patients, and stratify patients for prognosis, thereby providing a prognostic reference for patients and doctors.

Description

Technical Field

[0001] The present invention relates to the field of bladder cancer, and in particular to the prognosis prediction of bladder cancer, and more specifically to a prognosis prediction marker for bladder cancer, a prognosis prediction model based on the marker, and a construction method thereof. Background Art

[0002] Bladder cancer is a common malignant tumor, and its prognosis prediction has important clinical significance for the treatment and survival rate of patients. Prognosis prediction is to evaluate the patient's survival or recurrence risk by analyzing the patient's clinical and pathological characteristics, as well as factors such as molecular markers. Traditional bladder cancer prognosis prediction mainly relies on clinical and pathological characteristics. Clinical characteristics include age, gender, symptoms, tumor stage and grade, etc. Pathological characteristics include tumor type, size, depth of invasion, lymph node metastasis and histological grade, etc. These characteristics can be obtained through clinical examination, imaging and histological examination.

[0003] With the development of high-throughput technology, in addition to clinical and pathological characteristics, molecular markers also play an important role in the prognosis prediction of bladder cancer. Molecular markers refer to molecules that change during the occurrence and development of bladder cancer, such as genes, proteins, and non-coding RNA. These molecular markers can be detected through technologies such as gene expression profiling, proteomics, and genetics to evaluate the patient's prognosis. The prognosis prediction of bladder cancer is very important for guiding patients' treatment decisions and personalized treatment. By accurately evaluating the patient's prognosis, the most suitable treatment plan can be selected to improve the patient's survival rate and quality of life. Therefore, bladder cancer prognosis prediction research has important clinical application prospects. Summary of the invention

[0004] One of the technical problems to be solved by the present invention is to provide a group of bladder cancer prognosis prediction markers, which are significantly correlated with the overall survival of bladder cancer patients and can be used to accurately predict the prognosis of bladder cancer patients.

[0005] To solve the above technical problems, the bladder cancer prognosis prediction marker combination of the present invention includes the following genes: BAIAP2, BCL2L14, OAS1, C6orf62, EMP1, and TMEM109.

[0006] The second technical problem to be solved by the present invention is to provide a kit, which contains reagents for detecting the above-mentioned bladder cancer prognosis prediction marker combination.

[0007] The third technical problem to be solved by the present invention is to provide a method for screening the above-mentioned bladder cancer prognosis prediction markers, which method comprises the following steps:

[0008] 1) Obtain the gene expression levels and corresponding overall survival data of bladder cancer patients, remove samples with incomplete overall survival data, and use the remaining qualified samples as training sets;

[0009] 2) Using gene expression as the input matrix and the overall survival of bladder cancer patients as the fitting variable, regression analysis was performed in the training set to screen out genes related to the overall survival of bladder cancer patients;

[0010] 3) Performing multivariate analysis on the genes screened in step 2) to select genes that are independently correlated with the overall survival of bladder cancer patients as prognostic predictors for bladder cancer.

[0011] In the above step 1), the gene expression level can be obtained by the following method: converting the original expression level of the transcriptome into a TPM value, taking the log2(TPM+1) value, performing normalization processing, and obtaining the gene expression level value.

[0012] The fourth technical problem to be solved by the present invention is to provide a method for constructing a bladder cancer prognosis prediction model. The method is based on the above-mentioned method for screening bladder cancer prognosis prediction markers, and further constructs a bladder cancer prognosis prediction model according to the gene expression levels of the screened bladder cancer prognosis prediction markers and the corresponding multivariate analysis coefficients.

[0013] The fifth technical problem to be solved by the present invention is to provide a bladder cancer prognosis prediction model based on the above-mentioned marker combination and constructed by the above-mentioned model construction method, and the model includes the following calculation formula:

[0014] F = 0.17 × BAIAP2 gene expression - 0.095 × BCL2L14 gene expression - 0.13 × OAS1 gene expression - 0.27 × C6orf62 gene expression + 0.1 × EMP1 gene expression + 0.19 × TMEM109 gene expression;

[0015] Among them, F is the bladder cancer risk score, and the gene expression level is the log2(TPM+1) value corresponding to the gene.

[0016] The critical value of F is preferably -0.79527. If F≥-0.79527, the patient is judged to be at high risk of bladder cancer; if F﹤-0.79527, the patient is judged to be at low risk of bladder cancer.

[0017] The present invention uses TCGA-BLCA as a training set, and screens out gene markers that are significantly correlated with the overall survival of bladder cancer patients through the transcriptome expression levels and corresponding survival data of bladder cancer patients, and establishes a bladder cancer prognosis prediction model based on these gene markers. The model can accurately and effectively predict the overall survival of bladder cancer patients (especially the survival probability within 5 years), stratify patients for prognosis, and thus provide a prognostic reference for patients and doctors. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a time-dependent ROC graph of the 5-year overall survival in the training set of the bladder cancer prognosis prediction model of Example 1;

[0019] Figure 2 This is the Kaplan-Meier overall survival curve of bladder cancer patients in the training set;

[0020] Figure 3 This is the Kaplan-Meier overall survival curve of bladder cancer patients in the validation set. DETAILED DESCRIPTION

[0021] 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 further described in detail in conjunction with the accompanying drawings and specific implementation methods.

[0022] Example 1 Construction of a bladder cancer prognosis prediction model

[0023] 1. Build a dataset

[0024] The training set of this embodiment is from 436 bladder cancer patient samples with prognosis information in the TCGA (The Cancer Genome Atlas)-BLCA cancer genome atlas database, and the original transcriptome expression data of these patient samples and their corresponding total survival data are obtained, and the samples with incomplete total survival data are removed. The original transcriptome expression values ​​of the remaining 424 qualified samples are converted to kilobases per million (TPM) values, and the log2 (TPM + 1) values ​​are taken for normalization. The gene expression amounts described in the present invention all refer to the log2 (TPM + 1) values ​​corresponding to the gene.

[0025] The validation set used 166 bladder cancer patient samples with overall survival data from the sequencing database of the applicant Shanghai Rendong Medical Laboratory Co., Ltd. to obtain the original transcriptome expression data of these patient samples and their corresponding overall survival data. The original transcriptome expression data of the validation set were also normalized according to the above method.

[0026] 2. Screening LASSO features

[0027] Using gene expression as the input matrix and patient OS (overall survival) as the fitting variable, the R package "glmnet" was used to perform LASSO-COX regression analysis in the training set, and 11 characteristic variables with coefficients not equal to 0 (i.e., correlated with OS) were screened out, namely, genes GSDMB, TMEM109, OAS1, C6orf62, NFKBID, BCL2L14, NBAS, EMP1, BAIAP2, IL9R, and ORMDL1. The expression levels of these 11 genes are correlated with the patient's overall survival.

[0028] 3. COX multifactor analysis

[0029] The following COX regression model was used in the training set to perform COX multivariate analysis on the 11 genes screened above:

[0030] F=β1x1+β2x2+…+β n x n

[0031] Where β is the coefficient of the covariate, obtained by maximum likelihood estimation (MLE); x is the value of each variable.

[0032] The characteristic variables that independently affect the overall survival were screened out by p-value (p-value < 0.25), and the following 6 genes that are independently correlated with the overall survival prognosis of patients were obtained: BAIAP2, BCL2L14, OAS1, C6orf62, EMP1, and TMEM109. The expression levels and corresponding coefficients of these genes were used to construct a bladder cancer prognosis prediction model, and the calculation formula of the obtained model was:

[0033] F = 0.17 × BAIAP2 gene expression - 0.095 × BCL2L14 gene expression - 0.13 × OAS1 gene expression - 0.27 × C6orf62 gene expression + 0.1 × EMP1 gene expression + 0.19 × TMEM109 gene expression;

[0034] Among them, F is the bladder cancer risk score, and the gene expression level is the log2(TPM+1) value corresponding to the gene.

[0035] The bladder cancer risk score F of each sample was calculated according to the above formula, and the median of the bladder cancer risk score F of the training set was -0.79527. Using this median as the critical value, all samples in the training set were divided into a high-risk group (bladder cancer risk score F ≥ -0.79527) and a low-risk group (bladder cancer risk score F < -0.79527).

[0036] 4. Use time-dependent ROC and KM survival curves to evaluate the model's ability to predict overall survival prognosis

[0037] The ROC curve of the above bladder cancer prognosis prediction model is as follows Figure 1 As shown, the area under the ROC curve AUC is 0.7581.

[0038] The total survival data of the training set patients were input into the R package "survival", and the relationship between the grouping by the bladder cancer risk score F value and the patient's survival days was calculated to produce a Kaplan-Meier survival curve, such as Figure 2 As shown, the x-axis is the patient survival time (months), and the y-axis is the cumulative survival rate. At the beginning, the survival time is 0 and the survival rate is 1. As the survival time increases, the cumulative survival rates of the high-risk group and the low-risk group will decrease, but the log-rank test shows that there is a significant difference in the decline rate between the two groups. The cumulative survival rate of the high-risk group is significantly greater than that of the low-risk group, and the OS of the high-risk group is significantly worse than that of the low-risk group, indicating that the prognosis of the low-risk group is significantly better than that of the high-risk group, which also proves the prognostic prediction ability of the model constructed in this embodiment for the overall survival of bladder cancer.

[0039] 5. Validation of the prognostic prediction model for bladder cancer

[0040] The bladder cancer prognosis prediction model constructed above was used to calculate the bladder cancer risk score F for each bladder cancer patient sample in the validation set, and the prognosis of the bladder cancer patients was grouped according to the value of F.

[0041] Taking the patient numbered C02720 as an example, the expression of the BAIAP2 gene of the patient is about 9.9199, the expression of the BCL2L14 gene is about 9.7688, the expression of the OAS1 gene is about 12.1556, the expression of the C6orf62 gene is about 11.8321, the expression of the EMP1 gene is about 9.6297, and the expression of the TMEM109 gene is about 10.8265. According to the calculation formula of the above-mentioned bladder cancer overall survival prognosis prediction model, the bladder cancer risk score F=-1.5952 can be calculated. Since F<-0.79527, the patient belongs to the low-risk group (the actual overall survival of the patient is 34.23 months, which is greater than the median survival time of 15.13 months in the validation set, indicating that the patient has a good prognosis, which is consistent with the grouping calculated according to the model). The bladder cancer risk score F of other samples is calculated and grouped in the same way.

[0042] Then the relationship between the grouping by bladder cancer risk score F value and the patient's survival days was calculated to produce a Kaplan-Meier survival curve, such as Figure 3 As shown. Figure 3It can be seen that in the validation set, there are still significant differences in the Kaplan-Meier survival curve between the high-risk group and the low-risk group calculated according to the model of this embodiment, indicating that the prognostic prediction model of this embodiment has a certain universality in bladder cancer patients and can effectively predict the overall survival of bladder cancer patients, thereby contributing to the auxiliary diagnosis of bladder cancer and promoting personalized medicine.

[0043] 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 combination of markers for predicting prognosis of bladder cancer, characterized in that: It is composed of the following genes: BAIAP2, BCL2L14, OAS1, C6orf62, EMP1, TMEM109.

2. A kit, characterized in that: Contains reagents for detecting the marker combination according to claim 1.

3. A method for constructing a bladder cancer prognosis prediction model, characterized in that the steps include: 1) Obtain the gene expression levels and corresponding overall survival data of bladder cancer patients, remove samples with incomplete overall survival data, and use the remaining qualified samples as training sets; 2) Using gene expression as the input matrix and the overall survival of bladder cancer patients as the fitting variable, regression analysis was performed in the training set to screen out genes related to the overall survival of bladder cancer patients; 3) Performing multi-factor analysis on the genes screened in step 2) to screen out genes that are independently correlated with the overall survival of bladder cancer patients as bladder cancer prognosis prediction markers; the bladder cancer prognosis prediction markers include the following genes: BAIAP2, BCL2L14, OAS1, C6orf62, EMP1, TMEM109; 4) Using the expression levels of the genes screened in step 3) and the corresponding multivariate analysis coefficients, a bladder cancer prognosis prediction model is constructed; the bladder cancer prognosis prediction model includes the following calculation formula: F = 0.17 × BAIAP2 gene expression - 0.095 × BCL2L14 gene expression - 0.13 × OAS1 gene expression - 0.27 × C6orf62 gene expression + 0.1 × EMP1 gene expression + 0.19 × TMEM109 gene expression; Among them, F is the bladder cancer risk score, F≥-0.79527 indicates a high risk of bladder cancer, and F﹤-0.79527 indicates a low risk of bladder cancer; the gene expression level is the log2(TPM+1) value corresponding to the gene.

4. The method according to claim 3, characterized in that Step 1), the gene expression level is obtained by the following method: converting the original expression level value of the transcriptome into a TPM value, taking the log2(TPM+1) value, performing normalization processing, and obtaining the gene expression level value.

5. The method according to claim 3, characterized in that: Step 3), COX multivariate analysis method was used to screen genes independently associated with the overall survival of bladder cancer patients according to p-value < 0.25.

Images