Non-invasive diagnosis and prediction model for non-muscular invasive urinary tract epithelium cancer and construction method of non-invasive diagnosis and prediction model

Abstract

The invention provides a non-invasive diagnosis and prediction model for non-muscular invasive urinary tract epithelial carcinoma (NMIBC) based on urine tumor DNA (utDNA) and a construction method of the non-invasive diagnosis and prediction model. According to the model, sequencing analysis is performed on utDNA of an NMIBC patient by combining a low-depth whole genome sequencing technology and a high-depth targeted sequencing technology, and an NMIBC diagnosis score is calculated by utilizing a specific algorithm so as to predict the recurrence / progress risk of the patient. The invention aims to solve the problems of invasiveness, high cost, high missed diagnosis rate and the like of the traditional cystoscopy, and provides a non-invasive, accurate and economical NMIBC recurrence / progress monitoring method. The non-invasive diagnosis and prediction model has a wide application prospect, the postoperative life quality of the NMIBC patient can be improved, and powerful support is provided for formulating a personalized treatment scheme.

Description

Technical Field

[0001] The present invention relates to the field of medical diagnosis models, and in particular to a non-invasive diagnosis prediction model for non-muscle invasive urothelial carcinoma and a construction method thereof. Background Art

[0002] Non-muscle invasive urothelial carcinoma (NMIBC) is the most common pathological type of bladder cancer, accounting for over 80% of new bladder cancer cases. NMIBC patients face a high risk of recurrence and progression after treatment, particularly after transurethral resection of bladder tumors (TURBT). The recurrence / progression rate for high-risk NMIBC patients exceeds 50% within two years, placing significant strain on their physical and mental health and financial burden.

[0003] Currently, cystoscopy is the gold standard for monitoring recurrence / progression after surgery for NMIBC. However, as an invasive procedure, cystoscopy not only causes pain and significant psychological stress for patients, but can also damage the bladder and urethral mucosa, increasing the risk of complications such as hematuria and urethral stricture. Furthermore, cystoscopy is expensive and has a high rate of missed diagnoses, especially for small lesions, which can delay subsequent treatment.

[0004] With advances in medical technology, particularly next-generation sequencing, liquid biopsy-based minimal residual disease (MRD) detection has gradually emerged. In recent years, MRD detection based on circulating tumor cell (ctDNA) in peripheral blood has demonstrated significant clinical value in recurrence monitoring, patient stratification, and treatment guidance for various malignant solid tumors. However, the application of MRD testing in NMIBC remains exploratory.

[0005] Studies have shown that free tumor DNA in NMIBC patients is more likely to be enriched in urine than in peripheral blood. Therefore, urine tumor DNA (utDNA)-based MRD testing offers a new approach for noninvasively monitoring postoperative recurrence / progression in NMIBC patients. Currently, some studies have attempted to use utDNA for MRD detection in bladder cancer, with some success. However, these studies are mostly in the technical exploration stage, and a mature, reliable noninvasive diagnostic prediction model has yet to be established.

[0006] In order to solve the above problems, the applicant proposed a non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma and a construction method thereof. Summary of the Invention

[0007] The purpose of the present invention is to provide a non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma and a construction method thereof, so as to solve the problems in the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solutions: a non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma based on urine tumor DNA, which is used to predict the recurrence / progression risk of patients with non-muscle invasive urothelial carcinoma, comprising:

[0009] Sequencing tumor DNA in urine samples to obtain sequencing data;

[0010] Bioinformatics analysis was performed on the sequencing data to screen for mutated genes and copy number variations associated with recurrence / progression of non-muscle invasive urothelial carcinoma;

[0011] Based on the screened mutant genes and copy number variations, a specific algorithm is used to calculate the NMIBC diagnostic score that reflects the patient's risk of recurrence / progression.

[0012] Optionally, the specific algorithm includes but is not limited to linear regression, random forest, support vector machine or deep learning model.

[0013] A method for constructing a non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma based on urine tumor DNA, comprising the following steps:

[0014] Urine samples were collected from patients with non-muscle invasive urothelial carcinoma;

[0015] Extracting tumor DNA from urine samples;

[0016] Perform low-depth whole-genome sequencing and high-depth targeted sequencing on the extracted tumor DNA to obtain sequencing data;

[0017] Bioinformatics analysis was performed on the sequencing data to screen for mutated genes and copy number variations associated with recurrence / progression of non-muscle invasive urothelial carcinoma;

[0018] Based on the screened mutant genes and copy number variations, a non-invasive diagnostic prediction model is constructed through a specific algorithm. This model can calculate the NMIBC diagnostic score that reflects the patient's recurrence / progression risk.

[0019] Optionally, the specific algorithm includes but is not limited to linear regression, random forest, support vector machine or deep learning model.

[0020] Optionally, the method further includes a step of validating the constructed non-invasive diagnostic prediction model, including processing the urine samples in the validation set according to the method, calculating the NMIBC diagnostic score of each sample, and then comparing the scores with the actual recurrence / progression to evaluate the accuracy of the model.

[0021] Beneficial effects: Non-invasive: The present invention provides a non-invasive NMIBC recurrence / progression monitoring method, avoiding the pain and complication risks brought to patients by traditional cystoscopy.

[0022] Accuracy: By combining low-depth whole-genome sequencing and high-depth targeted sequencing technologies, as well as the calculation of specific algorithms, the present invention can accurately predict the recurrence / progression risk of NMIBC patients and improve the accuracy of diagnosis.

[0023] Economical: Compared with traditional cystoscopy, the non-invasive diagnostic prediction model of the present invention has lower costs, reducing the financial burden on patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is an MRD assessment diagram based on urine tumor cell (utDNA) detection in an embodiment of the present invention;

[0025] Figure 2 This is an MRD assessment graph based on peripheral blood circulating tumor cell (ctDNA) detection in an embodiment of the present invention;

[0026] Figure 3 This is a diagram showing the construction of the NMIBC diagnostic score based on the MRD algorithm model of urine tumor cells (utDNA) according to an embodiment of the present invention; DETAILED DESCRIPTION

[0027] The following describes preferred embodiments of the present invention with reference to the accompanying drawings to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0028] The purpose of the present invention is to provide a non-invasive diagnostic prediction model for NMIBC based on utDNA and a method for constructing the same. The model can accurately predict the recurrence / progression risk of NMIBC patients and provide strong support for clinical decision-making and personalized treatment.

[0029] Sample collection and processing

[0030] (1) Sample collection: Collect urine samples from NMIBC patients. The collection of urine samples should follow standardized operating procedures to ensure the purity and integrity of the sample. It is recommended to collect midstream urine after the patient's first urination in the morning to avoid contamination and interference.

[0031] (2) Sample processing: The collected urine sample is centrifuged to separate the cellular components in the urine. The centrifugation speed and time should be optimized according to the specific experimental conditions to ensure sufficient separation of the cellular components. After centrifugation, the supernatant is collected for subsequent processing or stored under appropriate conditions for future use.

[0032] DNA extraction and sequencing

[0033] (1) DNA extraction: Use appropriate DNA extraction reagents and methods to extract total DNA from the cellular components of the urine sample. During the extraction process, care should be taken to avoid DNA degradation and contamination to ensure the quality and purity of the extracted DNA.

[0034] (2) Sequencing: The extracted DNA is sequenced and analyzed. The present invention adopts a strategy that combines low-depth whole genome sequencing (1WGS) and high-depth targeted sequencing (HTS). 1WGS is used to comprehensively detect mutations and variations across the genome, while HTS performs deep sequencing on specific genes or regions to improve the sensitivity and accuracy of detection.

[0035] Bioinformatics analysis

[0036] (1) Data preprocessing: Perform quality control and filtering on the raw data obtained by sequencing to remove low-quality sequences and contamination sequences to ensure the accuracy and reliability of the data.

[0037] (2) Mutation detection: Using bioinformatics software and algorithms, mutation detection is performed on pre-processed data. By comparing the sequence differences between patient samples and normal control samples, the mutant genes and mutation sites in the patient samples are identified.

[0038] (3) Copy Number Variation Analysis: In addition to mutation detection, the present invention also performs copy number variation (CNV) analysis. CNV refers to the phenomenon of DNA copy number variation in a specific region of the genome, which is closely related to the occurrence and development of tumors. CNV analysis can further understand the genomic structural variation in patient samples.

[0039] Model building

[0040] (1) Feature selection: Based on the results of bioinformatics analysis, mutant genes and copy number variations associated with NMIBC recurrence / progression are screened as model features. Feature selection should follow the principles of scientificity, rationality, and operability to ensure that the selected features can accurately reflect the patient's risk of recurrence / progression.

[0041] (2) Algorithm selection: Based on the results of feature selection, an appropriate algorithm is selected to construct a non-invasive diagnostic prediction model. The present invention can use algorithms such as linear regression, random forest, support vector machine, or deep learning to construct the model. Different algorithms have different advantages and applicable scenarios, and the selection should be based on specific experimental data and requirements.

[0042] (3) Model training and validation: The selected algorithm is trained using the training set data to obtain a preliminary non-invasive diagnostic prediction model. The model is then validated and evaluated using the validation set data to ensure its accuracy and reliability. Strategies such as cross-validation and leave-one-out validation should be used during the validation process to avoid overfitting and underfitting.

[0043] Model application and evaluation

[0044] (1) Model application: The constructed non-invasive diagnostic prediction model is applied in clinical practice to predict the recurrence / progression risk of NMIBC patients. Based on the output of the model, patients are divided into low-risk and high-risk groups, providing strong support for clinical decision-making and personalized treatment.

[0045] (2) Model evaluation: Regularly evaluate and adjust the application effect of the model. Evaluation indicators may include accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (ROC curve) (Area Under the Curve, AUC). The evaluation results can help us understand the performance changes of the model in a timely manner and make necessary optimization and adjustments to the model.

[0046] Example 1: Sample collection and processing

[0047] Sample Collection: 100 NMIBC patients were selected for the study, and midstream urine samples were collected after their first morning urination. Standardized procedures were followed during sample collection to ensure sample purity and integrity.

[0048] Sample processing: The collected urine samples were centrifuged at 3000 rpm for 10 minutes. After centrifugation, the supernatant was collected for subsequent processing.

[0049] Example 2: DNA extraction and sequencing

[0050] DNA extraction: A commercial DNA extraction kit was used to extract total DNA from the cellular components of the urine sample. The extraction process was performed strictly according to the kit instructions to ensure DNA quality and purity.

[0051] Sequencing: Extracted DNA undergoes low-depth whole-genome sequencing and high-depth targeted sequencing. lWGS utilizes the Illumina sequencing platform at a sequencing depth of 0.5×; HTS performs deep sequencing of known bladder cancer-related genes at a sequencing depth of 500×. Strict quality control is maintained throughout the sequencing process to ensure data accuracy and reliability.

[0052] Example 3: Bioinformatics analysis

[0053] Data preprocessing: Quality control and filtering of the raw sequencing data were performed to remove low-quality and contaminating sequences. FastQC software was used for quality control analysis to ensure that the data quality met the requirements for subsequent analysis.

[0054] Mutation Detection: GATK (Genome Analysis Toolkit) software is used for mutation detection. By comparing sequence differences between patient samples and normal control samples, the mutated genes and variant sites in the patient samples are identified. The false positive and false negative rates are strictly controlled during the mutation detection process to ensure the accuracy of the results.

[0055] Copy Number Variation Analysis: CNVkit software was used for copy number variation analysis. By comparing the copy number differences between patient samples and normal controls, regions of copy number variation in patient samples were identified. During CNV analysis, noise and interference were carefully controlled to improve the reliability of the results.

[0056] Example 4: Model construction

[0057] Feature selection: Based on bioinformatics analysis, we screened for mutant genes and copy number variations associated with NMIBC recurrence / progression as model features. The feature selection process adhered to the principles of scientificity, rationality, and operability to ensure that the selected features accurately reflect the patient's risk of recurrence / progression.

[0058] Algorithm selection: Based on the results of feature selection, the random forest algorithm was selected to construct a non-invasive diagnostic prediction model. The random forest algorithm has advantages such as processing high-dimensional data and strong resistance to overfitting, making it suitable for the model construction of the present invention.

[0059] Model Training and Validation: A random forest algorithm was trained using data from 80 patients as the training set to generate a preliminary noninvasive diagnostic prediction model. The model was then validated and evaluated using data from the remaining 20 patients as the validation set. A cross-validation strategy was employed during validation to avoid overfitting and underfitting. Validation results demonstrated that the model achieved an accuracy of 85%, a sensitivity of 80%, a specificity of 90%, and an AUC of 0.88, demonstrating good predictive performance.

[0060] Example 5: Model application and evaluation

[0061] Model Application: The constructed noninvasive diagnostic prediction model will be applied in clinical practice to predict the risk of recurrence / progression in NMIBC patients. Based on the model's output, patients will be divided into low-risk and high-risk groups. Regular follow-up and monitoring are recommended for patients in the low-risk group, while active treatment and intervention are recommended for patients in the high-risk group.

[0062] Model evaluation: Regularly assess and adjust the model's effectiveness. Models are evaluated every six months using metrics such as accuracy, sensitivity, specificity, and AUC. Evaluation results provide timely insights into model performance changes and allow for necessary optimization and adjustments. Evaluation results demonstrate that the model maintains good stability and predictive performance over extended periods of application.

[0063] Non-invasive: The present invention uses urine samples for diagnosis and prediction, avoiding the invasiveness and pain of traditional cystoscopy, and improving patient acceptance and comfort.

[0064] Accuracy: This method combines low-depth whole-genome sequencing and high-depth targeted sequencing technologies, as well as advanced algorithms such as random forests, to accurately predict the recurrence / progression risk of NMIBC patients, providing strong support for clinical decision-making and personalized treatment.

[0065] Economical: Compared to traditional cystoscopy, the non-invasive diagnostic prediction model of this invention is much cheaper, reducing the financial burden on patients. Furthermore, through early prediction and intervention, it can reduce treatment costs and medical resource consumption.

[0066] Operability: The non-invasive diagnostic prediction model construction method of the present invention is simple to understand, easy to operate, and suitable for wide application in clinical laboratories and medical institutions.

[0067] Scalability: The non-invasive diagnostic prediction model of the present invention can be further expanded and applied to the diagnosis and prediction of other types of tumors, and has broad application prospects and clinical value.

[0068] This study successfully developed an efficient and accurate noninvasive diagnostic prediction model for NMIBC by integrating a series of technical processes, including urine sample collection, DNA extraction and sequencing, bioinformatics analysis, and model construction and application. This model not only overcomes the limitations of traditional cystoscopy, such as invasiveness, high cost, and low patient acceptance, but also significantly improves the accuracy of predicting the risk of NMIBC recurrence and progression.

[0069] Specifically, the technical effects of the present invention are embodied in the following aspects:

[0070] Improved diagnostic efficiency: Traditional cystoscopy requires patients to undergo a painful examination, and the results are limited by the doctor's experience and equipment. However, the non-invasive diagnostic prediction model of this invention only requires the patient to provide a urine sample to quickly and accurately assess the risk of recurrence / progression, greatly improving diagnostic efficiency.

[0071] Personalized treatment guidance: Using this model, doctors can more accurately understand a patient's risk of recurrence / progression, enabling them to develop personalized treatment plans. For low-risk patients, more conservative treatment strategies can be adopted to minimize unnecessary medical interventions; for high-risk patients, proactive treatment measures can be implemented promptly to prevent progression.

[0072] Reduced medical costs: The non-invasive diagnostic prediction model of this invention is low-cost and can effectively reduce patient treatment costs and medical resource consumption through early prediction and intervention. This is of great significance for alleviating the financial burden on patients and improving the efficiency of medical resource utilization.

[0073] Promoting the development of liquid biopsy technology: This utDNA-based noninvasive diagnostic and prediction model for NMIBC represents a significant advancement in the clinical application of liquid biopsy technology. Its successful implementation provides valuable insights and lessons for the diagnosis and prediction of other tumor types using liquid biopsy technology.

[0074] Promoting medical research and advancement: During the implementation of this invention, a large amount of urine samples and gene sequencing data from NMIBC patients have been accumulated, providing a rich resource for subsequent medical research and advancement. This data can be used to further explore the pathogenesis of NMIBC, identify new therapeutic targets, and develop new drugs.

[0075] In summary, the utDNA-based noninvasive diagnostic prediction model for NMIBC and its construction method proposed in this paper have significant technical advantages and clinical application value. It not only provides new ideas and methods for the diagnosis and treatment of NMIBC patients, but also opens new avenues for the application and development of liquid biopsy technology in the medical field. We believe that with the continuous advancement and improvement of technology, the noninvasive diagnostic prediction model of this invention will play an even more important role in clinical practice, bringing benefits to more patients.

[0076] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from all points of view, the embodiments should be regarded as illustrative and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and range of equivalents of the claims are included in the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0077] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma based on urine tumor DNA, characterized by: The model is used to predict the risk of recurrence / progression in patients with non-muscle invasive urothelial carcinoma, including: Sequencing tumor DNA in urine samples to obtain sequencing data; Bioinformatics analysis was performed on the sequencing data to screen for mutated genes and copy number variations associated with recurrence / progression of non-muscle invasive urothelial carcinoma; Based on the screened mutant genes and copy number variations, a specific algorithm is used to calculate the NMIBC diagnostic score that reflects the patient's risk of recurrence / progression.

2. The non-invasive diagnostic prediction model according to claim 1, characterized in that: The specific algorithms include but are not limited to linear regression, random forest, support vector machine or deep learning model.

3. A method for constructing a non-invasive diagnostic prediction model for non-muscle invasive urothelial carcinoma based on urine tumor DNA, characterized in that: The following steps are involved: Urine samples were collected from patients with non-muscle invasive urothelial carcinoma; Extracting tumor DNA from urine samples; Perform low-depth whole-genome sequencing and high-depth targeted sequencing on the extracted tumor DNA to obtain sequencing data; Bioinformatics analysis was performed on the sequencing data to screen for mutated genes and copy number variations associated with recurrence / progression of non-muscle invasive urothelial carcinoma; Based on the screened mutant genes and copy number variations, a non-invasive diagnostic prediction model was constructed through an algorithm, which can calculate the NMIBC diagnostic score that reflects the patient's recurrence / progression risk.

4. The method according to claim 3, characterized in that The specific algorithms include but are not limited to linear regression, random forest, support vector machine or deep learning model.

5. The method according to claim 3 or 4, characterized in that The method also includes a step of validating the constructed non-invasive diagnostic prediction model, including processing the urine samples in the validation set according to the method, calculating the NMIBC diagnostic score for each sample, and then comparing the score with the actual recurrence / progression situation to evaluate the accuracy of the model.

Images