Kidney clear cell carcinoma diagnosis and prognosis prediction system constructed based on PDK1

Abstract

The invention discloses a PDK1-based renal clear cell carcinoma diagnosis and prognosis prediction system, and belongs to the field of medicines.The PDK1-based renal clear cell carcinoma diagnosis and prognosis prediction system comprises a data processing module, a diagnosis and prognosis prediction module and a verification module, the data processing module downloads mRNA expression data of KIRC and preprocesses the mRNA expression data; the diagnosis and prognosis prediction module constructs a dichotomous logistic regression model based on PDK1, constructs a prognosis prediction model through a multi-factor regression analysis method, and calculates the prediction probability of the outcome event of the individual; and the verification module performs internal verification by using a TCGA-KIRC data set, detects PDK1 protein expression conditions in certain cases of renal clear cell carcinoma KIRC tissues and paired para-carcinoma tissue samples through immunohistochemical experiments, verifies differential expression of PDK1, and verifies the relationship between PDK1 and prognosis total lifetime in certain cases of renal clear cell carcinoma KIRC samples.

Description

Technical Field

[0001] The invention belongs to the field of medicine and relates to a renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1. Background Art

[0002] Prediction models play a vital role in the medical field. They can accurately predict the risk of disease occurrence and provide strong support for early intervention; by evaluating treatment effects and prognosis, they provide a scientific basis for clinical decision-making. At the same time, they can optimize the allocation of medical resources, improve medical quality, promote the development of precision medicine, and thus bring better treatment effects and quality of life to patients.

[0003] Renal clear cell carcinoma (KIRC) is the most invasive and common subtype of renal cancer. It has an insidious onset and lacks specific clinical manifestations in the early stage. Data show that about 13%-30% of patients have metastasis at the time of initial diagnosis, and the prognosis of metastatic KIRC is extremely poor, with a 5-year survival rate of only 8%. At present, the clinical diagnosis of KIRC mainly relies on imaging examinations and pathological biopsy, while its prognosis prediction mostly relies on clinical pathological staging. However, these methods have many limitations. For example, ultrasound examination is greatly affected by the operator's level and patient factors, and small lesions are difficult to detect; CT examination has radiation risks and contrast agents may cause adverse reactions, and small lesions are difficult to quantify; MRI examinations are time-consuming, expensive, and insensitive to calcification. At present, there is a lack of a unified and accurate indicator system for the prediction of clinical prognosis of KIRC. The existing indicators, such as pathological staging, have limited accuracy, and due to large individual differences, it is difficult to accurately predict the development of specific patients, and the prediction of treatment response is not accurate enough, which in turn affects the optimization of treatment options. In summary, in clinical work, a more accurate and convenient method is urgently needed to assist in the diagnosis and prognosis prediction of KIRC. In view of this, it is of great clinical significance to explore more molecular markers for early diagnosis and prognosis prediction of KIRC.

[0004] In the early stage, we performed proteomic data analysis on 30 KIRC clinical samples, and combined with bioinformatics methods, we found that pyruvate dehydrogenase kinase 1 (PDK1) showed significant differential expression in KIRC. Previous studies have shown that abnormal glucose metabolism is closely related to the malignant progression of KIRC (AMEMIYAT, SHIBATAK, YAMAGUCHI T, et al. Metabolic Oscillations and Glycolytic Phenotypes of Cancer Cells [J]. IJMS, 2023, 24, 11914). As a key rate-limiting enzyme in the glycolysis process, PDK1 is likely to play an important role in the pathological process of KIRC. In-depth exploration of the diagnostic and prognostic value of PDK1 in KIRC and the construction of a diagnostic and prognostic prediction model based on PDK1 can not only provide a new direction for the early diagnosis of renal cancer, but also improve the quality of life of patients, which is of great clinical significance for vigorously promoting the development of precision medicine for cancer.

[0005] At present, there is still a lack of sensitive and specific biomarkers for KIRC diagnosis and prognosis prediction in clinical practice. There is an urgent need for a KIRC diagnosis and prognosis prediction model to achieve efficient and rapid identification of people at high risk of KIRC, develop accurate plans for individualized screening and treatment, and solve the technical problems currently faced by ordinary technicians in this field. Summary of the invention

[0006] The purpose of the present invention is to provide a renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1, explore the diagnostic and prognostic value of PDK1 in KIRC through bioinformatics analysis methods combined with clinical samples, and develop a renal clear cell carcinoma diagnosis and prognosis prediction model based on PDK1, which can achieve efficient and rapid identification of high-risk groups for KIRC, and provide pre-technical support for the formulation of accurate plans for individualized screening and treatment.

[0007] The technical solution adopted by the present invention is as follows:

[0008] A renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1, comprising a data processing module, a diagnosis and prognosis prediction module and a verification module, wherein:

[0009] The data processing module downloads the mRNA expression data and clinical data of renal clear cell carcinoma KIRC in the user's cancer genome atlas TCGA, and performs deduplication and standardization preprocessing on the acquired mRNA expression data and clinical data;

[0010] The diagnosis and prognosis prediction module constructs a binary logistic regression model based on PDK1, and trains the obtained mRNA expression data. During the training process, the parameters are continuously adjusted so that the binary logistic regression model has the best fitting effect on the training data, that is, the difference between the diagnosed probability of illness and the actual illness is minimized; at the same time, the diagnosis and prognosis prediction module constructs a prognosis prediction model through a multivariate regression analysis method, and assigns a score to each value level of each influencing factor according to the degree of influence of each variable in the prognosis prediction model on the outcome event (that is, the size of the regression coefficient), and then adds up the scores to obtain a total score, and calculates the predicted probability of the individual outcome event through the functional conversion relationship between the total score and the probability of occurrence of the outcome event, and finally displays the prediction model in the form of a nomogram;

[0011] The verification module uses the TCGA-KIRC data set for internal verification. The R software in the verification module draws the operating characteristic curve ROC to evaluate the diagnostic value of PDK1 for patients, and the area under the operating characteristic curve ROC AUC measures the ability of the system to distinguish between the case group and the control group under different thresholds; at the same time, the verification module detects the expression of PDK1 protein in renal clear cell carcinoma KIRC tissues and paired adjacent cancer tissue samples of certain cases (such as 120 cases) through immunohistochemistry experiments, verifies the differential expression of PDK1, and verifies the relationship between PDK1 and prognosis of overall survival in certain cases (such as 120 cases) of renal clear cell carcinoma KIRC samples.

[0012] In order to better implement this solution, further, the verification module also uses the GEO database for external verification. The data processing module downloads the data set containing the mRNA expression data of normal tissues and KIRC tissues from the GEO database and obtains external data after preprocessing, and sends it to the verification module. The verification module brings the external data into the binary logistic regression model constructed in the diagnosis and prognosis prediction module, and R software draws the operating characteristic curve ROC to verify the diagnostic value of PDK1 for patients.

[0013] In order to better implement the present scheme, further, the preprocessing of the acquired mRNA expression data and clinical data in the data processing module includes denoising the acquired mRNA expression data and clinical data using wavelet analysis, completing the missing positions of the mRNA expression data and clinical data that should correspond one to one in the data, and on this basis, adding gene types and performing position encoding.

[0014] In order to better implement this solution, further, the binary logistic regression model constructed by the diagnosis and prognosis prediction module based on PDK1 is:

[0015] Set the disease status as: 1 indicates having the disease, 0 indicates normal. During the training process, continuously adjust the parameters to make the fitting effect of the model on the training data the best, that is, the difference between the predicted disease probability and the actual disease situation is the smallest. The diagnostic model formula is:

[0016]

[0017] where p is the probability of the event occurring, β0 is the intercept, and βi is the coefficient of the i-th independent variable x i of.

[0018] To better implement this solution, further, in the binary logistic regression model, the parameters in the model are estimated by the maximum likelihood estimation method.

[0019] To better implement this solution, further, the numerical expressions of the area under the receiver operating characteristic curve (AUC) of the working characteristic curve ROC in different ranges are:

[0020] AUC ≤ 0.5 indicates that the model has no predictive ability at all, equivalent to random guessing;

[0021] 0.5 < AUC < 0.6 indicates that the model has a certain predictive ability;

[0022] 0.6 ≤ AUC < 0.8 indicates that the model has good predictive ability;

[0023] AUC ≥ 0.8 indicates that the model has good predictive ability, and this model has high reference value for the diagnosis of diseases.

[0024] To better implement this solution, further, the method for constructing the prognostic prediction model of the training set by the diagnostic and prognostic prediction module through the multi-factor regression analysis method is specifically as follows:

[0025] Randomly divide the dataset formed by the obtained mRNA expression data and clinical data into a training set and a test set according to a certain ratio, where the amount of data in the test set is less than that in the training set; construct a prognostic prediction model based on the training set through the multi-factor regression analysis method, assign scores to each value level of each influencing factor according to the influence degree of each variable in the prognostic prediction model on the outcome event, and then add up the scores to obtain the total score, and calculate the predicted probability of the outcome event of this individual through the function conversion relationship between the total score and the probability of the outcome event occurring.

[0026] To better implement this solution, further, the verification module tests the established prognostic prediction model by establishing an objective function based on the test set, conducts hyperparameter optimization operations, and calculates the prediction accuracy. When the prediction accuracy meets the confidence interval, the prognostic prediction result is output, and the prognostic output result is verified in the form of an ROC curve;

[0027] At the same time, the verification module detected the expression of PDK1 protein in the test set of clear cell renal cell carcinoma KIRC tissues and paired adjacent cancer tissue samples through immunohistochemistry experiments, verified the differential expression of PDK1, and verified the relationship between PDK1 and prognosis of overall survival in clear cell renal cell carcinoma KIRC samples.

[0028] The renal clear cell carcinoma diagnosis and prognosis prediction method formed by the renal clear cell carcinoma diagnosis and prognosis prediction system described in this scheme is:

[0029] Step S1 Data download and preprocessing

[0030] The data download module downloads the mRNA expression data and clinical data of renal clear cell carcinoma KIRC in the Cancer Genome Atlas TCGA from the TCGA database (https: / / TCGAdata.nci.nih.gov / ), and uses the “limma” package in R software (4.2.1) to remove duplicates and standardize the preprocessing to become training data;

[0031] Step S2 Data Analysis

[0032] Data analysis includes differential expression analysis, KIRC diagnostic analysis, and prognostic prediction analysis, including:

[0033] Differential expression analysis: R software in the diagnosis and prognosis prediction module performs differential expression analysis on the training data.

[0034] The KIRC diagnostic analysis is as follows: a binary logistic regression model is constructed based on PDK1, and part of the training data is selected for training. During the training process, the parameters in the model are estimated by methods such as maximum likelihood estimation, and the parameters are continuously adjusted to make the binary logistic regression model fit the training data best, that is, the difference between the diagnosed probability of illness and the actual illness is minimized; the disease status is set as: 1 for illness, 0 for normal, and the final diagnostic model formula is:

[0035]

[0036] Where p is the probability of an event, β0 is the intercept, and βi is the i-th independent variable x. i The coefficient of .

[0037] Data sets containing mRNA expression data of normal tissues and KIRC tissues were downloaded from the GEO database or other case databases. After preprocessing, they were input into the diagnostic model formula obtained in the KIRC diagnostic analysis, and the ROC curve was drawn.

[0038] Prognostic prediction analysis is:

[0039] The preprocessed dataset is divided into a training set and a test set. A prognostic prediction model for the training set is constructed through multivariate regression analysis using the survival and survminer modules in R software, to analyze the prognostic value of PDK1 for KIRC. Furthermore, a prognostic prediction model for KIRC is constructed by combining PDK1 with clinical features. According to the degree of influence (i.e., the magnitude of the regression coefficient) of each variable in the KIRC prognostic prediction model on the outcome event, scores are assigned to each value level of each influencing factor, and then the scores are added together to obtain the total score. The predicted probability of the outcome event for this individual is calculated through the functional conversion relationship between the total score and the probability of the outcome event occurring. Finally, the prediction model is presented in the form of a nomogram.

[0040] Step S3 Model Validation:

[0041] Model validation is performed for differential expression analysis, KIRC diagnosis analysis, and prognostic prediction analysis respectively. The validation of differential expression analysis is as follows:

[0042] The expression of PDK1 protein in KIRC tissue and paired adjacent cancer tissue samples of a certain number of cases is detected through immunohistochemistry (IHC) experiments to verify the differential expression of PDK1.

[0043] The validation of KIRC diagnosis analysis includes internal validation and external validation. The internal validation is as follows:

[0044] In the validation module, the data in the training data that are not selected for training the binary logistic regression model are used as the internal validation dataset. R software is used to draw the receiver operating characteristic curve ROC to evaluate the diagnostic value of PDK1 for patients. The area under the receiver operating characteristic curve AUC is used to measure the ability of this system to distinguish between the case group and the control group at different thresholds. The expression of the value of AUC in different ranges is as follows:

[0045] AUC ≤ 0.5 indicates that the model has no predictive ability at all and is equivalent to random guessing;

[0046] 0.5 < AUC < 0.6 indicates that the model has a certain predictive ability;

[0047] 0.6 ≤ AUC < 0.8 indicates that the model has good predictive ability;

[0048] AUC ≥ 0.8 indicates that the model has excellent predictive ability and this model has high reference value for the diagnosis of the disease.

[0049] The external validation is as follows:

[0050] The preprocessing of the acquired mRNA expression data and clinical data in the data processing module includes denoising the acquired mRNA expression data and clinical data using wavelet analysis, completing data at missing positions of the mRNA expression data and clinical data that should have a one-to-one correspondence in the data, and adding gene types on this basis and performing position encoding.

[0051] The validation of the prognostic prediction analysis was:

[0052] Based on the test set, an objective function was established to test the prognostic prediction model of the training set, hyperparameter optimization was performed, and the prediction accuracy was calculated to verify the relationship between PDK1 and prognosis overall survival in certain cases of renal clear cell carcinoma KIRC samples.

[0053] In addition, clinical diagnosis of KIRC mainly relies on imaging diagnosis, and prognosis prediction mainly relies on clinical pathological staging, but these methods have limitations, and clinical prognosis prediction lacks a unified and accurate indicator system, individual differences are large, and predictions are inaccurate, so there is a lack of more accurate and convenient methods to assist in the diagnosis and prognosis prediction of KIRC. The diagnostic model in the present invention performs well, and the AUC of the diagnostic ROC reaches 0.913, and the AUC of the ROC curve when the model is used for external validation is 0.984 and 0.993, respectively, proving that the model has strong discrimination ability, good generalization ability, and high reliability and practicality; the AUC of the nomogram prediction of the prognosis model in predicting the prognosis (such as survival time) of KIRC patients has high accuracy, and the AUCs for predicting 1, 3, and 5 years are 0.859, 0.791, and 0.749.

[0054] This protocol is mainly to build a diagnostic and prognostic prediction model, and differential expression analysis and prognostic analysis are only supplemented by data analysis. This protocol can be divided into a data analysis part and a model construction and verification part. This protocol is mainly to build and verify the diagnostic and prognostic models; and the data analysis part includes the analysis results of the TCGA database such as differential expression and prognostic KM survival curve.

[0055] The diagnostic model of the present invention also performs well in distinguishing target disease states from non-disease states. In terms of sensitivity, a higher AUC value indicates that the model can accurately identify patients with the disease, that is, it has a higher true positive rate and can avoid missed diagnoses to a large extent. In terms of specificity, the higher AUC value also indicates that the model also performs well in identifying non-disease patients, that is, it has a higher true negative rate and can effectively reduce the occurrence of misdiagnosis.

[0056] The prognosis prediction model of the present invention has an AUC of 0.859, 0.791, and 0.749 in the ROC curve for predicting 1-, 3-, and 5-year survival rates, respectively, proving that the model has very high specificity and sensitivity in prognosis prediction at different time points. In terms of specificity, the model can accurately distinguish patients with good prognosis, that is, it has a high true negative rate, and effectively avoids misjudging patients with good prognosis as having poor prognosis. In terms of sensitivity, the model can efficiently identify patients who may have a poor prognosis, that is, it has a high true positive rate, which greatly reduces the risk of missing patients with poor prognosis. By performing differential analysis of PDK1 protein expression and survival analysis on 120 KIRC samples, further evidence is provided for the accuracy and reliability of the prediction model involved in the present invention.

[0057] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0058] 1. The present invention discloses a renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1. A KIRC diagnosis and prognosis prediction model based on PDK1 is constructed and presented in the form of a nomogram and a mathematical formula. The model of this scheme has strong discrimination ability, good generalization ability, and high reliability and practicality. The AUC predicted by the nomogram in predicting the prognosis (such as survival time) of KIRC patients has high accuracy, and the AUCs predicted for 1, 3, and 5 years are 0.859, 0.791, and 0.749 respectively.

[0059] 2. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 described in the present invention uses internal and external data sets to verify the models involved, and the results are more reliable. Its good predictive efficiency can assist clinicians in judging the disease progression and outcome of patients, and achieve early warning for high-risk groups and precise stratified treatment for KIRC patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative work, among which:

[0061] Figure 1 It is the expression of PDK1 in TCGA-KIRC and the prognosis survival curve in one set of embodiments of this scheme, wherein A is the differential expression of PDK1 in the TCGA-KIRC cohort; B is the KM survival curve of the overall survival (OS) of PDK1 in the TCGA-KIRC cohort;

[0062] Figure 2 This is the diagnostic ROC curve of PDK1 in TCGA-KIRC in one set of embodiments of this scheme;

[0063] Figure 3 is the ROC curve of the diagnostic prediction model based on the GSE36895 dataset PDK1 in one set of embodiments of this scheme;

[0064] Figure 4 is the ROC curve of the diagnostic prediction model based on the GSE53757 dataset PDK1 in one set of embodiments of this scheme;

[0065] Figure 5 This is a representative image of the expression of PDK1 in 120 cases of renal clear cell carcinoma tissues and adjacent tissues detected by immunohistochemical experiments in a group of embodiments of this protocol;

[0066] Figure 6 In one embodiment of this protocol, immunohistochemical experiments were performed to detect the difference in PDK1 protein expression between cancer and adjacent tissue samples from 120 KIRC cases;

[0067] Figure 7 This is a schematic diagram of a nomogram of a renal clear cell carcinoma prognosis prediction model based on PDK1 combined with clinical pathological characteristics in one embodiment of the present scheme;

[0068] Figure 8 1 is a schematic diagram of the ROC curve of the accuracy of the nomogram in predicting the 1-, 3-, and 5-year survival probabilities in a group of embodiments of the present scheme;

[0069] Fig. 9 This is a KM survival curve drawn by dividing 120 KIRC cases into a high expression group and a low expression group according to the PDK1 expression level in one embodiment of this scheme. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the embodiments described are only part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0071] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention.

[0072] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0073] The features and performance of the present invention are further described in detail below in conjunction with the embodiments.

[0074] Embodiment 1

[0075] The preferred embodiment of the present invention provides a renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1. In this embodiment, 120 KIRC tissue samples were collected from the Affiliated Tumor Hospital of Xinjiang Medical University. This research plan has been approved by the Ethics Committee of the Affiliated Tumor Hospital of Xinjiang Medical University. The patients and their families have signed informed consent before surgery. PDK1 antibody was purchased from Proteintech, Chicago, USA.

[0076] The diagnostic and prognostic value analysis of KIRC based on the gene and protein expression of PDK1 in KIRC in the aforementioned samples and the diagnostic and prognostic prediction model based on PDK1 expression were performed as follows:

[0077] Step 1: Download PDK1 mRNA expression profiles and corresponding clinical data of 613 renal clear cell carcinoma tissues and 72 normal tissues from the TCGA database (https: / / TCGAdata.nci.nih.gov / ), of which 537 patients had complete clinical data. The “limma” package of R software (4.2.1) was used to uniformly convert the RNA sequencing data in FPKM format in the dataset into TPM format. Cases with incomplete information were eliminated, and finally the TPM data of 537 patients were included for subsequent analysis.

[0078] Step 2: Perform box plot analysis of PDK1 mRNA expression differences in TCGA-KIRC (N=72; T=530). The logarithm of PDK1 mRNA expression in normal tissues is about 2, while the logarithm of PDK1 mRNA expression in KIRC tissues is about 4. Figure 1As shown in A, it can be seen that in the vast majority of KIRC patients, the expression level of PDK1 mRNA in cancer tissues is significantly higher than that in adjacent normal tissues.

[0079] Step 3: Combined with the above analysis, PDK1 is significantly differentially expressed in cancer and adjacent tissues, and has potential diagnostic value. Therefore, a binary logistic regression model was constructed based on PDK1, where disease status: 1 indicates disease and 0 indicates normal; parameter estimation: the parameters in the model were estimated by methods such as maximum likelihood estimation.

[0080] Step 4: The diagnostic model formula obtained from the above steps is as follows:

[0081]

[0082] Where: p represents the predicted probability, and x represents the expression level of PDK1.

[0083] Step 5: Using the TCGA-KIRC dataset for internal validation, the ROC curve in this example selects the point with the highest sum of specificity and sensitivity, 1.125, as the critical value of PDK1 expression (H-score), and divides the PDK1 expression in the above samples into two levels: high expression and low expression. When the specificity -1 is equal to 0.2, the sensitivity level of PDK1 has exceeded 0.91 ( Figure 2 ).

[0084] Step 6: Download the GSE36895 and GSE53757 datasets as external validation sets for the diagnostic model. The AUC of the external validation ROC curve of this model is 0.984 and 0.993 respectively ( Figure 3-4 ), which proves that the model has strong discrimination ability, good generalization ability, high reliability and practicality.

[0085] Step 7: To further verify the protein expression of PDK1, 120 renal clear cell carcinoma tissues and corresponding adjacent tissues were collected from patients who underwent surgery in the Urology Department of the Affiliated Tumor Hospital of Xinjiang Medical University from January 2011 to December 2015; none of them received anti-tumor treatment or had other tumors before surgery, and the postoperative pathological reports were all clear cell renal carcinoma. Sample inclusion criteria:

[0086] (1) Patients diagnosed with clear cell renal cell carcinoma;

[0087] (2) All patients did not receive chemotherapy, radiotherapy, or targeted therapy before surgery;

[0088] (3) Postoperative pathological diagnosis of renal clear cell carcinoma;

[0089] (4) no history of other or combined malignant tumors;

[0090] (5) Complete clinical information

[0091] The characteristics of the 120 KIRC tissue samples are shown in Table 1:

[0092] Table 1 Basic data of 120 patients with renal clear cell carcinoma

[0093]

[0094] Paraffin sections of 120 KIRC tissues were prepared, and the protein expression of PDK1 was detected by IHC experiment, and the results were interpreted by ImageJ software. The results showed that PDKI was significantly differentially expressed in cancer and adjacent tissues, which was consistent with the results of database analysis. Figure 5-Figure 6 .

[0095] Figure 1 B is the Kaplan-Meier (KM) survival analysis of PDK1 and OS in TCGA-KIRC; the prognostic results showed that the overall survival (OS, HR=0.53; P<0.001) of the PDK1 high expression group (H-score≥1.125) in KIRC patients was significantly longer than that of the PDK1 low expression group (H-score<1.125).

[0096] Combined with the above prognostic analysis results, high expression of PDK1 usually indicates a good prognosis for KIRC patients, suggesting that it has a certain prognostic value. Based on this, PDK1 combined with clinical characteristics to construct a KIRC prognosis prediction model. The steps are as follows:

[0097] Step S1: Data preprocessing: The TCGA-KIRC cohort data were denoised using wavelet analysis, missing data positions in the cohort were filled, and gene types were added and positional coding was performed on this basis;

[0098] Step S2: The PDK1 mRNA dataset is randomly divided into a training set and a test set in a ratio of 4:1, wherein 80% of the samples in the PDK1 mRNA dataset are used as the training set, and the remaining 20% ​​are used as the validation set;

[0099] Step S3: First, a multi-factor regression model is constructed based on the training set. According to the degree of influence of each variable in the model on the outcome event (i.e., the size of the regression coefficient), each value level of each influencing factor is scored. Then, each score is added up to get the total score, and then the predicted probability of the individual outcome event is calculated through the functional conversion relationship between the total score and the probability of the outcome event. Finally, the prediction model is displayed in the form of a nomogram;

[0100] Step S4: Based on the test set, establish the objective function to test the target model, perform hyperparameter optimization operations, and calculate the prediction accuracy.

[0101] Step S5: When the prediction accuracy meets the confidence interval, the prognosis prediction result is output, and the prognosis output result is verified in the form of ROC curve. Figure 7 As shown in the figure, three independent variables affecting prognosis, including patient age, clinical stage, and PDK1 expression, were screened out by COX regression analysis, and a nomogram was constructed to predict the overall survival of KIRC patients alone. The R package (survivalROC) was used to evaluate the accuracy of the nomogram prognostic prediction in the validation set. Figure 8 As shown, the AUC areas for 1, 3, and 5 years were 0.859, 0.791, and 0.749, respectively, indicating that the model had good prognostic prediction accuracy.

[0102] Step S6: The protein expression data of PDK1 in 120 KIRC samples were obtained through step 3, and then the survival information of these samples was collected, and the samples with missing follow-up information were eliminated to verify the relationship between PDK1 and the prognosis of KIRC patients. Fig. 9 As shown in the figure, high expression of PDK1 was correlated with good prognosis, with a P value of 0.00047, which was consistent with the analysis results of the TCGA database.

[0103] The above studies show that we have successfully constructed a diagnosis and prognosis prediction model based on PDK1, which shows good predictive efficacy.

[0104] In summary, the present invention screened out PDK1 as a research target through comprehensive analysis of proteomics, transcriptome data and clinical data of KIRC clinical cases, and analyzed the value of PDK1 in KIRC diagnosis and prognosis prediction. The prediction model constructed based on PDK1 showed good predictive efficiency. The KIRC diagnosis and prognosis prediction model developed by the present invention through the expression level of PDK1 can realize the efficient and rapid identification of people at high risk of KIRC, and formulate accurate plans for individualized screening and treatment.

[0105] It is worth mentioning that the samples in this embodiment are from inpatients of the Affiliated Tumor Hospital of Xinjiang Medical University who were diagnosed as KIRC by pathology. The postoperative tissue specimens are uniformly preserved by the Xinjiang High-Incidence Tumor Biological Sample Bank. The sample bank was established in 2010 and was approved by the Ministry of Science and Technology for the administrative license for human genetic resources management-human genetic resources preservation approval (National Science and Heritage Office Examination No.

[2016] 620) in 2016. In 2022, it was again approved for the administrative license for human genetic resources management-human genetic resources preservation approval (National Science and Heritage Office Examination No.

[2022] BC0077). In 2023, it passed the sample bank quality inspection and became the only sample bank in Xinjiang that passed the national quality inspection and obtained the administrative license for human genetic resources management-human genetic resources preservation approval (National Science and Heritage Office Examination No.

[2022] BC0077). Proteomics testing is implemented by Hangzhou Jingjie Biotechnology Co., Ltd., a well-known company in the industry. Therefore, the test results are scientific and credible.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions and improvements made by any technician familiar with the field within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A PDK1-based renal clear cell carcinoma diagnosis and prognosis prediction system, characterized in that It includes a data processing module, a diagnosis and prognosis prediction module, and a verification module, where: The data processing module downloads the mRNA expression data and clinical data of kidney clear cell carcinoma (KIRC) in The Cancer Genome Atlas (TCGA) of the user, and preprocesses the obtained mRNA expression data and clinical data; The diagnosis and prognosis prediction module constructs a binary logistic regression model based on PDK1 and trains it for the preprocessed data. During the training process, the parameters are continuously adjusted to make the fitting effect of the binary logistic regression model on the training data the best; at the same time, the diagnosis and prognosis prediction module constructs a prognosis prediction model through multivariate regression analysis. According to the influence degree of each variable in the prognosis prediction model on the outcome event, each value level of each influencing factor is scored, and then the scores are added up to obtain the total score. The predicted probability of the individual outcome event is calculated through the functional conversion relationship between the total score and the occurrence probability of the outcome event; The verification module uses the TCGA-KIRC dataset for internal verification. The R software in the verification module plots the receiver operating characteristic curve (ROC) to evaluate the diagnostic value of PDK1 for patients, and the area under the receiver operating characteristic curve (AUC) is used to measure the ability of this system to distinguish the case group and the control group at different thresholds; at the same time, the verification module detects the PDK1 protein expression in the kidney clear cell carcinoma (KIRC) tissue and paired adjacent cancer tissue samples of a certain number of cases through immunohistochemistry experiments, verifies the differential expression of PDK1, and verifies the relationship between PDK1 and the overall survival of prognosis in a certain number of kidney clear cell carcinoma (KIRC) samples.

2. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 1, characterized in that: The verification module also uses the GEO database for external verification. The data processing module downloads the mRNA expression data containing normal tissues and KIRC tissues from the GEO database, preprocesses it to obtain external data, and sends it to the verification module. The verification module brings the external data into the binary logistic regression model constructed in the diagnosis and prognosis prediction module, and the R software plots the receiver operating characteristic curve (ROC) to verify the diagnostic value of PDK1 for patients.

3. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 1, characterized in that: The preprocessing of the obtained mRNA expression data and clinical data in the data processing module includes denoising the obtained mRNA expression data and clinical data by wavelet analysis method, filling in the missing positions of the mRNA expression data and clinical data that should correspond one by one in the data, and adding gene types and performing position encoding on this basis.

4. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 1, characterized in that: The binary logistic regression model constructed by the diagnosis and prognosis prediction module based on PDK1 is: Set the disease state as: 1 represents diseased, 0 represents normal. During the training process, the parameters are continuously adjusted to make the fitting effect of the model on the training data the best. The diagnostic model formula is: Where p is the probability of an event, β0 is the intercept, and βi is the i-th independent variable x. i The coefficient of .

5. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 4, characterized in that: In the binary logistic regression model, the parameters in the model are estimated by the maximum likelihood estimation method.

6. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 1, characterized in that: The expression of the value of the area under the receiver operating characteristic curve (AUC) in different ranges is: AUC ≤ 0.5 indicates that the model has no predictive ability at all, equivalent to random guessing; 0.5 < AUC < 0.6 indicates that the model has a certain predictive ability; 0.6≤AUC<0.8 indicates that the model has good predictive ability; AUC ≥ 0.8 indicates that the model has good predictive ability and has a high reference value for disease diagnosis.

7. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 1, characterized in that: The method for constructing a prognostic prediction model of a training set by the diagnosis and prognosis prediction module through multi-factor regression analysis is specifically as follows: The data set formed by the acquired mRNA expression data and clinical data was randomly divided into a training set and a test set according to a certain ratio, wherein the amount of data in the test set was less than that in the training set. A prognosis prediction model was constructed based on the training set by multivariate regression analysis. According to the degree of influence of each variable in the prognosis prediction model on the outcome event, each value level of each influencing factor was scored. Then, each score was added together to obtain a total score. The predicted probability of the outcome event for the individual was calculated through the functional conversion relationship between the total score and the probability of occurrence of the outcome event.

8. The renal clear cell carcinoma diagnosis and prognosis prediction system based on PDK1 according to claim 7, characterized in that: The verification module tests the established prognosis prediction model by establishing an objective function based on the test set, performs a hyperparameter optimization operation, and calculates the prediction accuracy. When the prediction accuracy satisfies the confidence interval, the prognosis prediction result is output, and the prognosis output result is verified in the form of an ROC curve; At the same time, the verification module detected the expression of PDK1 protein in the renal clear cell carcinoma KIRC samples in the test set through immunohistochemistry experiments, verified the differential expression of PDK1, and verified the relationship between PDK1 and prognosis of overall survival in renal clear cell carcinoma KIRC samples.

Images