A molecular marker combination for predicting the prognosis of patients with cervical cancer and its application

Abstract

The present invention discloses a molecular marker combination for predicting the prognosis of patients with cervical cancer and its application. The present invention uses the three NRGs screened including CHMP4C, SLC25A5 and VDAC1 as a molecular marker combination, establishes a risk prediction model with 3 NRGs, and divides cervical cancer patients into high-risk and low-risk groups. The risk prediction model in the present invention has an AUC value of 0.694 when predicting a 1-year survival period, an AUC value of 0.703 when predicting a 3-year survival period, and an AUC value of 0.664 when predicting a 5-year survival period, which has better accuracy and specificity. And the predictive performance of the risk prediction model in the present invention has been verified in the validation set. The model in the present invention can effectively predict the prognosis of patients with cervical cancer, which is helpful for the selection of clinical treatment plans, and thus realizes early intervention and early treatment of cervical cancer.

Description

Technical Field

[0001] The invention belongs to the field of medical detection, and in particular relates to a molecular marker combination for predicting the prognosis of cervical cancer patients and an application thereof. Background Art

[0002] Cervical cancer is a common gynecological disease and one of the leading causes of cancer death in women. Currently, the main treatment for cervical cancer is radical hysterectomy combined with radiotherapy and cisplatin chemotherapy, but some patients still relapse after surgery or radiotherapy, and the prognosis of relapsed patients is poor. In order to further improve the treatment effect of cervical cancer and develop precise treatment strategies, oncologists need to predict the prognosis of cervical cancer patients.

[0003] At present, the FIGO (Federation International of Gynecology and Obstetrics) staging system is commonly used in clinical practice to predict the prognosis of patients with cervical cancer. For surgical patients, the FIGO staging system is mainly based on imaging or pathological examinations; for some non-surgical patients, the FIGO staging is based on the doctor's subjective judgment of the patient's imaging and physical examination results. However, if the patient has pelvic inflammatory disease, endometriosis or obesity, it will lead to misdiagnosis. In addition, due to the significant difference between the clinically determined staging and the surgical pathology results, there are significant differences in the prognosis of FIGO staged patients.

[0004] In recent years, many studies have confirmed the predictive role of genes in tumor prognosis. Research on cancer at the molecular level is more conducive to individualized treatment and prognosis assessment. At present, the pathogenesis of cervical cancer is not fully understood, and there are no effective molecular markers for the diagnosis and treatment of cervical cancer in clinical practice. Therefore, the discovery of new prognostic markers and the construction of new prediction models are of great significance for the mechanism research and clinical application of cervical cancer. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the above-mentioned prior art. To this end, the present invention proposes a molecular marker combination for predicting the prognosis of cervical cancer patients and its application. In the present invention, a risk prediction model with a combination of CHMP4C, SLC25A5, and VDAC1 human necroptosis-related genes (NRGs) is established, and cervical cancer patients are divided into high-risk and low-risk groups according to the risk score. In the training set, the risk score of cervical cancer patients is significantly correlated with the overall survival rate of patients (P<0.05). The receiver operating characteristic curve (ROC) curve analysis of the risk prediction model in the present invention shows that the AUCs are 0.694, 0.703, and 0.664, respectively, in 1-year, 3-year, and 5-year follow-up. They are all higher than the AUC values ​​of the risk prediction model established using CHMP4C, SLC25A5, and VDAC1 alone. Compared with FIGO staging, the risk prediction model in the present invention has better accuracy and specificity, thereby achieving early intervention and early treatment of cervical cancer, and providing a valuable NRGs targeted treatment approach for cervical cancer.

[0006] The first aspect of the present invention provides a molecular marker combination, which is a combination of CHMP4C, SLC25A5 and VDAC1.

[0007] The second aspect of the present invention provides an application of the molecular marker combination described in the first aspect of the present invention in the preparation of a product for predicting the prognosis risk and / or diagnosing cervical cancer.

[0008] According to the second aspect of the present invention, in some embodiments of the present invention, the product includes reagents, kits, and the like.

[0009] The third aspect of the present invention provides an application of the molecular marker combination described in the first aspect of the present invention in constructing a prognostic risk prediction model for cervical cancer patients.

[0010] According to the content of the third part of the present invention, in some embodiments of the present invention, the method for constructing the prognostic risk prediction model for cervical cancer patients comprises the following steps:

[0011] (1) Gene expression data of cervical cancer samples and normal samples, as well as clinical information data of corresponding patients, were collected and included in the training set. The gene expression data in the training set were then standardized to obtain a gene expression matrix;

[0012] (2) using the R package limma to perform differential expression analysis on the human necroptosis-related genes in the gene expression matrix in step (1), and screen the gene expression profiles of differentially expressed human necroptosis-related genes, with the screening criteria set to P < 0.05 and the absolute value of the difference fold > 2;

[0013] (3) The differentially expressed human necroptosis-related gene expression profile obtained in step (2) was combined with the survival data in the clinical information data of the corresponding patient, and a univariate Cox regression analysis was performed to screen out human necroptosis-related genes related to the prognosis of cervical cancer, and the screening standard was P<0.05; then a LASSO Cox regression model was constructed to obtain the prognosis risk prediction model for cervical cancer patients of the present invention, and the calculation formula of the risk prediction model was as follows:

[0014]

[0015] Where n is the number of prognostic genes, exp i is the expression value of gene i, β i is the regression coefficient of gene i.

[0016] In some more preferred embodiments of the present invention, the gene expression data of the cervical cancer samples and normal samples are downloaded from the TCGA database.

[0017] In some more preferred embodiments of the present invention, human necroptosis-related gene sets are downloaded from the KEGG database and related literature.

[0018] In some preferred embodiments of the present invention, the method further comprises: evaluating the predictive performance of the risk prediction model based on the training set data.

[0019] In some more preferred embodiments of the present invention, based on the training set data, evaluating the predictive performance of the risk prediction model comprises the following steps:

[0020] Calculate the risk score of each cervical cancer patient in the training set in step (1) based on the risk score calculation formula of the risk prediction model;

[0021] Based on the obtained risk scores, the cervical cancer samples in the training set were divided into high-risk and low-risk groups according to the median of the risk scores;

[0022] The Kaplan-Meier curve of the training set data was used to assess whether there was a significant difference in survival between the high-risk group and the low-risk group.

[0023] The ROC curve analysis of the training set data was used to evaluate the goodness of fit of the risk prediction model.

[0024] In some preferred embodiments of the present invention, the ROC curve is drawn using the R software "survival" "timeROC" package software.

[0025] In some preferred embodiments of the present invention, the prognostic risk prediction model for cervical cancer patients described in the present invention can be used for the 1-year, 3-year and 5-year prognosis of the patients.

[0026] In some more preferred embodiments of the present invention, the prognostic risk prediction model for cervical cancer patients described in the present invention can be used for the 3-year prognosis of the patients.

[0027] In some preferred embodiments of the present invention, the method further comprises: obtaining validation set data, and evaluating the effectiveness of the risk prediction model based on the validation set data.

[0028] In some more preferred embodiments of the present invention, the validation set is downloaded from the GEO database, and standardized gene expression profile data and clinical information are extracted therefrom, and ultimately 300 samples are included in the validation set.

[0029] In some more preferred embodiments of the present invention, the evaluating the effectiveness of the risk prediction model based on the validation set data comprises the following steps:

[0030] Calculate the risk score for each cervical cancer patient in the validation set based on the risk score calculation formula;

[0031] Based on the obtained risk scores, the samples in the validation set were divided into high-risk and low-risk groups according to the median risk score of the training set;

[0032] The Kaplan-Meier curve of the validation set data was used to assess whether there was a significant difference in survival between the high-risk group and the low-risk group.

[0033] The ROC curve analysis of the validation dataset was used to evaluate the goodness of fit of the risk prediction model.

[0034] In some preferred embodiments of the present invention, the differentially expressed NRGs in step (3) specifically include 1DH1, TNFRSF10B, CHMP4C, PGAM5, EZH2, STAT5B, TRADD, IL33, KLF9, AXL, HSP90AA1, SLC25A5, CDKN2A, IDH2, VDAC1, CAPN1, and PARP1.

[0035] In some more preferred embodiments of the present invention, the NRGs associated with the prognosis of cervical cancer screened in step (4) are CHMP4C, SLC25A5, and VDAC1.

[0036] In some preferred embodiments of the present invention, the risk score formula of the risk prediction model in step (3) is: risk score = 0.0659 × CHMP4C gene expression level - 0.0030 × SLC25A5 gene expression level + 0.0102 × VDAC1 gene expression level.

[0037] The beneficial effects of the present invention are:

[0038] (1) The present invention establishes a risk prediction model with a combination of CHMP4C, SLC25A5, and VDAC1, and divides cervical cancer patients into high-risk and low-risk groups based on the risk score. The risk score of cervical cancer patients in the training set is significantly correlated with the overall survival rate of the patients (P<0.05). ROC curve analysis shows that the AUC of the risk prediction model in the present invention was 0.694, 0.703, and 0.664 in the 1-year, 3-year, and 5-year follow-up, respectively. They are all higher than the AUC values ​​of the risk prediction model established using CHMP4C, SLC25A5, and VDAC1 alone. Compared with FIGO staging, the risk prediction model in the present invention has better accuracy and specificity;

[0039] (2) The predictive performance of the risk prediction model in the present invention has been verified in the validation set. The risk prediction model can effectively predict the prognosis of patients with cervical cancer, which is helpful for the selection of clinical treatment plans and thus achieves early intervention and early treatment of cervical cancer;

[0040] (3) The risk prediction model in the present invention systematically studied for the first time the relationship between the expression pattern of NRGs in cervical cancer and the prognosis of patients, providing new insights into the role of NRGs in the occurrence and development of cervical cancer and a valuable NRGs targeted treatment approach for cervical cancer patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is the heat map of 17 differentially expressed NRGs;

[0042] Figure 2 Kaplan-Meier survival analysis curves of cervical cancer patients in the high-risk group and low-risk group in the training set;

[0043] Figure 3 The time-dependent ROC curves of the risk prediction model for 1, 3, and 5 years in the training set;

[0044] Figure 4 Kaplan-Meier survival analysis curves of cervical cancer patients in the high-risk and low-risk groups in the validation set GEO44001;

[0045] Figure 5The time-dependent ROC curves of the risk prediction model for 1, 3, and 5 years in the validation set;

[0046] Figure 6 Cox regression analysis was used to evaluate the independent prognostic value of the risk prediction model in patients with cervical cancer (univariate and multivariate cox regression analysis was performed on age (Age), histological grade (Grade), clinical stage (Stage), T stage and risk score (Risk score), respectively). Figure 6 A is univariate cox regression analysis; Figure 6 B is multivariate cox regression analysis;

[0047] Figure 7 The result diagram of ROC comparative analysis of the prognostic risk prediction model and clinically relevant pathological information of patients with cervical cancer, where Age is the patient's age, Grade is the histological grade, Stage is the clinical stage, T represents the T stage, and Riskscore represents the risk score;

[0048] Figure 8 The nomogram and nomogram calibration curve were constructed by combining clinical characteristics and risk scores, among which, Figure 8 A nomogram that combines risk scores and clinical information features, Figure 8 B Calibration graph shows that the 3-year survival probability predicted by the nomogram corresponds to the actual observed 3-year survival probability. Figure 8 C calibration graph shows that the nomogram predicted 5-year survival probability corresponds to the actually observed 5-year survival probability;

[0049] Fig. 9 1-, 3-, and 5-year ROC curves of the risk prediction model constructed for gene CHMP4C;

[0050] Fig.10 1-, 3-, and 5-year ROC curves of the risk prediction model constructed for gene SLC25A5;

[0051] Fig.11 1-, 3-, and 5-year ROC curves of the risk prediction model constructed for gene VDAC1. DETAILED DESCRIPTION

[0052] The present invention will be further described below in conjunction with specific embodiments, and the advantages and features of the present invention will become clearer as the description proceeds. However, these embodiments are exemplary only and do not constitute any limitation to the scope of the present invention. It should be understood by those skilled in the art that the details and forms of the technical solution of the present invention may be modified or replaced without departing from the spirit and scope of the present invention, but these modifications and replacements all fall within the scope of protection of the present invention.

[0053] Example 1

[0054] Construction of a prognostic risk prediction model for cervical cancer

[0055] 1. Data download and organization

[0056] The gene expression data of cervical cancer samples and normal samples, as well as the clinical information data of the corresponding cervical cancer patients, were downloaded from the TCGA database, including a total of 306 cancer tissue samples and 3 normal tissue samples as training sets.

[0057] NRGs (hsa04217) were downloaded from KEGG (Kyoto Encyclopedia of Genes and Genomes), and combined with literature reports, 211 NRGs were finally included for subsequent analysis. The perl script was used to extract the expression data of NRGs from the expression profile data of the training set, and the gene expression data was standardized to obtain the NRGs expression matrix; the R package limma was used to analyze the differential expression of NRGs in the training set, with P < 0.05 and the absolute value of the difference multiple > 2 as the cutoff value for screening differential genes. Finally, a total of 17 differentially expressed NRGs were obtained, and the differentially expressed NRGs were visualized using the R language heatmap package. The results are shown in Figure 2. Figure 1 As shown, from Figure 1 It can be seen that TRADD, TNFRSF10B, PGAM5, PARP1, IDH1, HSP90AA1, VDAC1, CAPN1, EZH2, CHMP4C, SLC25A5, IDH2 and CDKN2A were significantly upregulated in cancer tissues, while KLF9, IL33, STAT5B and AXL were significantly downregulated in cancer tissues.

[0058] 2. Univariate Cox and Lasso Cox regression analysis

[0059] The 17 differentially expressed NRGs screened were subjected to univariate Cox analysis to screen out genes related to the survival of cervical cancer patients. Among them, genes with P < 0.05 were considered to have an impact on the survival of cervical cancer patients. Finally, a total of 3 genes related to the prognosis of cervical cancer were screened, as shown in Table 1, namely CHMP4C, SLC25A5 and VDAC1. Then Lasso Cox regression analysis was performed to construct a Lasso Cox regression model. The linear combination of the regression coefficient of the Lasso Cox regression model and the expression level of the gene expression matrix was used to construct a prognostic gene signature to form a risk score formula. The risk score of each gene was calculated, and all cervical cancer samples were divided into high-risk group and low-risk group according to the median of the risk score.

[0060] The calculation formula of risk score is:

[0061] Where n is the number of prognostic genes, exp i is the expression value of gene i, β i is the regression coefficient of gene i.

[0062] That is: risk score = 0.0659 × CHMP4C gene expression level - 0.0030 × SLC25A5 gene expression level + 0.0102 × VDAC1 gene expression level.

[0063] Table 1 Three NRGs screened by the risk prediction model

[0064]

[0065] HR in univariate Cox analysis is used to characterize relative risk, where an HR value greater than 1 indicates that the expression value of the corresponding gene is positively correlated with the risk score, so that the corresponding Lasso Cox regression coefficient is greater than 0, and an HR value less than 1 indicates that the expression value of the corresponding gene is negatively correlated with the risk score, so that the corresponding Lasso Cox regression coefficient is less than 0. In addition, HR.95L and HR.95H represent 95% confidence intervals.

[0066] Validation of a prognostic risk prediction model for cervical cancer

[0067] The Series Matrix File of cervical cancer was obtained from the GEO database, from which standardized gene expression profile data and clinical information were extracted. Finally, 300 samples were included in the validation set.

[0068] 1. Survival curve analysis

[0069] The survival analysis and survival curves of the high-risk and low-risk groups of cervical cancer patients in the training set were performed using the R software "survival" and "survminer" packages, and the differences between the groups were compared using the log-rank test. The results showed that in the training group, the survival time of the high-risk group patients was significantly shorter than that of the low-risk group. Figure 2 Similarly, in the validation set GEO44001, the survival time of patients in the high-risk group was significantly shorter than that in the low-risk group, as shown in Figure 4 shown.

[0070] 2. ROC curve analysis

[0071] In order to evaluate the accuracy of the risk prediction model in predicting the prognosis of cervical cancer, the R software "survival" "timeROC" package was used to detect the 1-year, 3-year, and 5-year prognostic efficacy of the cervical cancer prognosis risk prediction model in the embodiment of the present invention using the ROC curve. The results showed that the AUC of the cervical cancer prognosis risk prediction model in the embodiment of the present invention was 0.694, 0.703, and 0.664 in the 1-year, 3-year, and 5-year follow-up, respectively. Figure 3 As shown in Figure 2. The AUC value shows that the risk prediction model composed of three NRGs has a good discriminative performance for the prognosis of cervical cancer patients. Similarly, in the validation set, the risk prediction model composed of three NRGs also has a good discriminative performance for the prognosis of cervical cancer patients, as shown in Figure 2. Figure 5 shown.

[0072] 3. Prognostic value of risk score in multivariate regression analysis

[0073] Clinical characteristics of cervical cancer patients were collected from the TCGA database, including survival time, survival status, age (AGE), histological grade (Grade), clinical stage (Stage), and T stage. Univariate Cox and multivariate Cox regression analyses were performed using clinical data and risk scores to evaluate whether the risk score is an independent prognostic factor for cervical cancer patients. Values ​​of p < 0.05 were considered statistically significant. The results showed that the risk score was an independent prognostic factor for cervical cancer patients (P < 0.001), as Figure 6 A (univariate cox regression analysis results) and 6B (multivariate cox regression analysis results). Figure 7 It is a comparison chart of the ROC curves of the risk prediction model and other clinical pathological characteristics. It can be seen from the figure that the AUC value corresponding to the cervical cancer prognosis risk prediction model in the embodiment of the present invention is 0.713, which is higher than FIGO staging (i.e. Stage, AUC value = 0.654) and other clinical pathological characteristics. Therefore, in general, the Risk score in the embodiment of the present invention has certain advantages in predicting survival. Compared with FIGO staging, the risk prediction model based on 3 NRGs in the embodiment of the present invention has better accuracy and specificity.

[0074] 4. Constructing a nomogram

[0075] The R package “rms” was used to construct a nomogram based on age, histological grade, clinical stage, T stage, and risk score, as shown in Figure 8 As shown in A, age, histological grade, clinical stage, and T stage are the most important factors among various clinical parameters. Then draw a calibration curve, such as Figure 8 B and Figure 8C, in order to evaluate the consistency between the actual survival results and the expected survival results, it can be used to predict the survival probability of cervical cancer patients. The study found that the calibration curve shows that the actual survival rate and the expected survival rate have a high consistency, indicating that the risk prediction model in the embodiment of the present invention has a high credibility.

[0076] Comparative Example 1

[0077] In Comparative Example 1, the same model construction method as in Example 1 was used to construct a prognostic risk prediction model for cervical cancer patients based on the CHMP4C gene. The AUC value corresponding to the prediction of 1-year survival period by the constructed prognostic risk prediction model was 0.632, the AUC value corresponding to the prediction of 3-year survival period was 0.619, and the AUC value corresponding to the prediction of 5-year survival period was 0.587. Fig. 9 shown.

[0078] Comparative Example 2

[0079] In Comparative Example 2, the same model construction method as in Example 1 was used to construct a prognostic risk prediction model for cervical cancer patients based on the SLC25A5 gene. The AUC value corresponding to the prediction of 1-year survival was 0.341, the AUC value corresponding to the prediction of 3-year survival was 0.417, and the AUC value corresponding to the prediction of 5-year survival was 0.391. Fig.10 shown.

[0080] Comparative Example 3

[0081] In Comparative Example 3, the same model construction method as in Example 1 was used to construct a prognostic risk prediction model for cervical cancer patients based on the VDAC1 gene. The AUC value corresponding to the prediction of 1-year survival period by the constructed prognostic risk prediction model was 0.503, the AUC value corresponding to the prediction of 3-year survival period was 0.67, and the AUC value corresponding to the prediction of 5-year survival period was 0.574. Fig.11 shown.

[0082] Therefore, in this embodiment, considering the important significance of NRGs in the biology of cervical cancer, a risk prediction model related to NRGs was determined to predict the prognosis of cervical cancer patients, providing a theoretical basis for the prognosis and treatment of cervical cancer patients, and the prognosis results are more accurate.

[0083] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A molecular marker combination, characterized in that: The molecular marker combination is a combination of CHMP4C, SLC25A5 and VDAC1.

2. Use of the molecular marker combination described in claim 1 in the preparation of cervical cancer prognosis risk prediction and / or diagnosis products.

3. Use of the molecular marker combination described in claim 1 in constructing a prognostic risk prediction model for patients with cervical cancer.

4. The use according to claim 3, characterized in that: The method for constructing the prognostic risk prediction model for cervical cancer patients comprises the following steps: (1) Gene expression data of cervical cancer samples and normal samples and clinical information data of corresponding patients were collected and included in the training set. The gene expression data in the training set were then standardized to obtain a gene expression matrix; (2) Use the R package limma to perform differential expression analysis on the human necroptosis-related genes in the gene expression matrix in step (1), and screen the gene expression profiles of differentially expressed human necroptosis-related genes. The screening criteria are set as P <0.05 and the absolute value of the fold difference >2; (3) The differentially expressed human necroptosis-related gene expression profiles obtained in step (2) were combined with the survival data in the clinical information data of the corresponding patients, and a univariate Cox regression analysis was performed to screen out human necroptosis-related genes associated with the prognosis of cervical cancer. The screening criteria were: P <0.05; then the LASSO Cox regression model was constructed to obtain the prognostic risk prediction model for cervical cancer patients; The risk score of the risk prediction model in step (3) is: Risk score = 0.0659 × CHMP4C gene expression level - 0.0030 × SLC25A5 gene expression level + 0.0102 × VDAC1 gene expression level.

5. The use according to claim 4, characterized in that: The method further comprises: Based on the training set data, the predictive performance of the risk prediction model is evaluated.

6. The use according to claim 5, characterized in that: The step of evaluating the prediction performance of the model based on the training set data comprises the following steps: Calculate the risk score of each cervical cancer patient in the training set in step (1) based on the risk score calculation formula; Based on the obtained risk scores, the cervical cancer samples in the training set were divided into high-risk and low-risk groups according to the median of the risk scores; Using the Kaplan-Meier curve of the training set to evaluate whether there is a significant difference in survival between the high-risk group and the low-risk group; The receiver operating characteristic curve analysis of the training set data was used to evaluate the goodness of fit of the risk prediction model.

7. The use according to claim 4, characterized in that: The method further comprises: Acquire validation set data, and evaluate the effectiveness of the risk prediction model based on the validation set data.

8. The use according to claim 7, characterized in that: The step of evaluating the effectiveness of the risk prediction model based on the validation set data comprises the following steps: Calculate the risk score for each cervical cancer patient in the validation set based on the risk score calculation formula; Based on the obtained risk scores, the samples in the validation set were divided into high-risk and low-risk groups according to the median risk score of the training set; The Kaplan-Meier curve of the validation set data was used to evaluate whether there was a significant difference in survival between the high-risk group and the low-risk group; The receiver operating characteristic curve analysis of the validation set data was used to evaluate the goodness of fit of the risk prediction model.

9. The use according to claim 4, characterized in that: The differentially expressed human necroptosis-related genes in step (3) specifically include 1DH1, TNFRSF10B, CHMP4C, PGAM5, EZH2, STAT5B, TRADD, IL33, KLF9, AXL, HSP90AA1, SLC25A5, CDKN2A, IDH2, VDAC1, CAPN1 and PARP1.

Images