A preoperative risk assessment prediction method for liver transplantation patients with liver cancer
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG PROVINCIAL PEOPLES HOSPITAL
- Filing Date
- 2026-02-10
- Publication Date
- 2026-06-02
Smart Images
Figure CN122135790A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical technology, specifically to a method for preoperative risk assessment and prediction in liver transplant patients with liver cancer. Background Technology
[0002] Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related death worldwide, with an overall 5-year survival rate of only about 18%. For patients with end-stage liver disease complicated by HCC, total hepatectomy combined with liver transplantation (LT) remains the only potentially curative treatment. The theoretical basis of this treatment strategy is that removing the entire tumor-bearing liver to eliminate all malignant cells, while simultaneously transplanting a liver to restore liver function, thus it is only suitable for patients who have not yet developed vascular invasion or extrahepatic metastasis. Traditional transplant selection criteria, represented by the Milan criteria, primarily screen candidate patients based on morphological indicators such as tumor number and size. While the subsequent UCSF and Hangzhou criteria broadened the indications to some extent, they still primarily relied on imaging and routine clinical parameters. Despite this, under the current screening system, postoperative tumor recurrence still occurs in a significant proportion of patients. Meanwhile, some patients who exceeded the traditional criteria but received transplants achieved long-term disease-free survival, indicating that the existing preoperative recurrence risk assessment system still has significant shortcomings. The root cause of post-transplant recurrence lies in the presence of occult metastases, meaning that before clinical detection, tumor cells have detached from the primary tumor and entered the bloodstream, lurking in distant tissues and eventually forming metastases. Only before tumor cells have spread remotely can the removal of the tumor-burdened liver have true radical significance.
[0003] In summary, a preoperative risk assessment and prediction method for liver transplant patients with liver cancer is needed to address the above issues. Summary of the Invention
[0004] The purpose of this invention is to provide a method for preoperative risk assessment and prediction in liver transplant patients with liver cancer, so as to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: This invention proposes a method for preoperative risk assessment and prediction in liver transplant patients with liver cancer, comprising the following steps: S1. Sample collection: Select plasma samples and corresponding clinicopathological information from liver transplant recipients of hepatocellular carcinoma, and clarify the inclusion and exclusion criteria for samples; S2. Cell-free DNA extraction and whole-genome sequencing from plasma: Cell-free DNA was extracted and quality-controlled from the plasma samples collected in step S1, and sequencing libraries were constructed and low-coverage whole-genome sequencing was performed. S3. Data preprocessing and copy number variation analysis: The raw sequencing data is subjected to quality control, trimming, alignment, and deduplication. Somatic copy number variation analysis is performed based on the processed data. S4. Variable screening: Remove low-frequency copy number variation fragments and screen for copy number variation fragments and clinical variables associated with recurrence after liver transplantation for liver cancer; S5. Predictive Model Construction: Incorporate the screened copy number variation fragments and clinical variables into a multivariate regression model to construct a visual nomogram predictive model; S6. Model Validation: The performance of the prediction model constructed in step S5 is validated using the training set and multiple validation queues; S7. Preoperative risk assessment: Calculate the patient score based on the nomogram prediction model in step S5, assess the risk of recurrence after liver transplantation based on the score results, and determine the selection criteria for transplant candidates. S8. Model Optimization: Based on the validation results of step S6 and the risk assessment data of step S7, the model parameters are calibrated periodically to maintain prediction accuracy.
[0006] Preferably, the implementation process of step S1 is as follows: S1.1. Plasma samples from hepatocellular carcinoma liver transplant recipients from three medical centers in my country were selected, with a total of 260 samples. S1.2. The sample inclusion criteria are: hepatocellular carcinoma patients with complete clinical information who have undergone liver transplantation; S1.3. The sample exclusion criteria are: presence of large blood vessel invasion or distant metastasis, survival time after liver transplantation less than 90 days, tumor recurrence time after liver transplantation less than 60 days, mixed characteristics of hepatocellular-cholangiocarcinoma, and poor quality or insufficient quantity of plasma samples. S1.4. Collect clinicopathological information of the recipients corresponding to the samples, including gender, age, hepatitis B virus infection status, tumor size, number of tumors, tumor pathological differentiation degree, tumor capsule invasion, recurrence time, recurrence site, follow-up time, time of death, cause of death, Child-Pugh score, MELD score, hepatitis B surface antigen, alpha-fetoprotein, carcinoembryonic antigen, carbohydrate antigen 199, abnormal prothrombin, total protein, alanine aminotransferase, aspartate aminotransferase, aspartate aminotransferase / alanine aminotransferase ratio, gamma-glutamyl transferase, alkaline phosphatase, total bilirubin, direct bilirubin, and creatinine.
[0007] Preferably, the plasma sample processing procedure in step S2 is as follows: S2.1. Collect 10 ml of peripheral blood from the patient before liver transplantation in an EDTA anticoagulant tube. S2.2. Within 2 hours after collection, the plasma was separated by centrifugation at 1600×g for 10 minutes at 4℃. The supernatant plasma was then centrifuged again at 16000×g for 10 minutes to remove cell debris. S2.3. Aliquot the clarified plasma and freeze it at -80°C until free DNA is extracted.
[0008] Preferably, the implementation process of cell-free DNA extraction and quality control in step S2 is as follows: S2.4. Use the QIAseq cfDNA Extraction Kit to isolate cell-free DNA from cryopreserved plasma; S2.5. The concentration and purity of the extracted free DNA were determined using Nanodrop and Qubit. S2.6. Assess the integrity and fragmentation of free DNA using 1% agarose gel electrophoresis to ensure that sequencing requirements are met.
[0009] Preferably, the implementation process of step S2, Chinese library construction and whole-genome sequencing, is as follows: S2.7. Take 10 ng of qualified free DNA and use the NEBnext Ultra II DNA Library PrepKit to construct a sequencing library without additional fragmentation. S2.8. After ligating the adapter with an 8bp index sequence into the constructed library, perform PCR amplification and purify the amplified library; S2.9. Perform paired-end sequencing on the Illumina HiSeqXten platform, with the sequencing target being low-coverage whole-genome sequencing at 2-3x.
[0010] Preferably, the implementation process of step S3 is as follows: S3.1. Use FastQC to assess the quality of raw sequencing data, and detect sequencing quality values, base composition, adapter contamination and duplication levels; S3.2. Use trim_galore to trim adapters and low-quality bases from sequencing data; S3.3. The pruned sequence was aligned to the human reference genome GRCh38 using the bwa-mem algorithm; S3.4. Use samtools to compare and convert the format of the result files, sort them, and create an index; S3.5. Use tools in GATK to identify and remove repetitive sequences introduced by PCR amplification; S3.6. Somatic copy number variation analysis was performed on the processed cell-free DNA low-depth whole genome sequencing data using ichorCNA. The analysis process integrated sequence read depth, GC content and fragment size distribution information, and copy number variation was detected based on the hidden Markov model framework. S3.7. Each sample was normalized using cell-free DNA samples from 10 internal healthy controls.
[0011] Preferably, the implementation process of step S4 is as follows: S4.1. Remove low-frequency copy number variation fragments from the sequencing data; S4.2. The LASSO regression method was used to screen for remaining copy number variant fragments and identify copy number variant fragments that were highly associated with recurrence after liver transplantation for hepatocellular carcinoma. S4.3. Univariate Cox regression analysis was used to screen clinical variables associated with recurrence after liver transplantation for hepatocellular carcinoma; S4.4. Perform multivariate analysis to identify copy number variation fragments, abnormal prothrombin, and alpha-fetoprotein as significant predictors of recurrence.
[0012] Preferably, the implementation process of step S5 is as follows: S5.1. Incorporate the significant predictors identified in step S4 into the multivariate Cox regression model; S5.2. Construct a prediction model in the form of a nomogram based on the results of the multivariate Cox regression model. This prediction model is the ZJU standard. S5.3. The model performance was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and optimistic correction C-index. The C-index of the model in the derivation cohort was 0.802, and the areas under the curve for 1 year, 3 years, and 5 years were 0.807, 0.841, and 0.817, respectively.
[0013] Preferably, the implementation process of step S6 is as follows: S6.1. Select a training set and three independent validation queues to validate the prediction model. The three independent validation queues include one internal validation queue and two external validation queues. S6.2. The areas under the curve for the internal validation cohort at 1 year, 3 years, and 5 years are 0.821, 0.758, and 0.766, respectively. S6.3. The areas under the curve for the first external validation cohort over 1 year, 3 years, and 5 years are 0.802, 0.759, and 0.745, respectively. S6.4. The areas under the curve for the second external validation cohort at 1 year and 3 years were 0.864 and 0.815, respectively. This cohort was not validated for 5 years due to insufficient follow-up time.
[0014] Preferably, the implementation process of step S7 is as follows: S7.1. Calculate the patient's score based on the nomogram prediction model, and use a score ≤100 as the cutoff value for screening hepatocellular carcinoma patients suitable for liver transplantation; S7.2. Patients with a score ≤100 have a 5-year overall survival rate ≥73% and a recurrence rate ≤22%; patients with a score >100 have a 5-year overall survival rate ≤38% and a recurrence rate ≥59%. S7.3. After completing risk assessments for every 10 batches of patients, the parameters of the prediction model are recalibrated based on the patients' follow-up results and recurrence status. The weights of each predictor in the nomogram are adjusted to achieve stable cyclic optimization of the model.
[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention utilizes plasma-extracted cfDNA for whole-genome sequencing and combines it with clinical testing information to construct a preoperative risk assessment and prediction model for recurrence after liver transplantation in hepatocellular carcinoma recipients based on non-invasive testing. This model can be used to predict the probability of recurrence-free survival after liver transplantation. In the model derivation cohort, clinical records and circulating tumor DNA data are integrated for preoperative recurrence risk prediction. For circulating tumor DNA data, insignificant copy number variation fragments are removed during the preprocessing stage. Subsequently, LASSO regression is used to screen the remaining candidate fragments, ultimately identifying two fragments highly correlated with recurrence after liver transplantation. To identify important clinical variables for predicting recurrence after liver transplantation in hepatocellular carcinoma, univariate Cox regression analysis is performed. After screening out important clinical variables, multivariate variable analysis is conducted. In the multivariate analysis, only Fragment 1, Fragment 2, PIVKA-II, and AFP remain significant predictors of recurrence. Based on this, the ZJU criteria based on nomograms proposed in this invention present the prediction model in a visual form, facilitating its clinical implementation in personalized risk assessment. Attached Figure Description
[0016] Figure 1 The nomogram of the present invention is shown; Figure 2 The prediction results of the training set of the present invention are shown in the figure; Figure 3 The training set calibration curve of the present invention is shown; Figure 4 The ROC curve of the verification set of the internal verification queue of the present invention is shown. Figure 5 The ROC curve of the verification set of the external verification queue 1 of the present invention is shown; Figure 6 The ROC curve of the verification set of the external verification queue 2 of the present invention is shown; Figure 7 The graphs showing the overall survival time and recurrence rate of patients with different scores in the model derivation cohort of this invention are illustrated. Figure 8 The graphs show the overall survival and recurrence rate of patients with different scores in the internal validation cohort of this invention; Figure 9 The graphs showing overall survival and recurrence rate of patients with different scores in the external validation cohort 1 of this invention are shown. Figure 10 The graphs showing the overall survival and recurrence rate of patients with different scores in the external validation cohort 2 of this invention are illustrated. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Example 1, please refer to Figures 1 to 10 This invention proposes a method for preoperative risk assessment and prediction in liver transplant patients with liver cancer, comprising the following steps: S1. Sample collection: Select plasma samples and corresponding clinicopathological information from liver transplant recipients of hepatocellular carcinoma, and clarify the inclusion and exclusion criteria for samples; S2. Cell-free DNA extraction and whole-genome sequencing from plasma: Cell-free DNA was extracted and quality-controlled from the plasma samples collected in step S1, and sequencing libraries were constructed and low-coverage whole-genome sequencing was performed. S3. Data preprocessing and copy number variation analysis: The raw sequencing data is subjected to quality control, trimming, alignment, and deduplication. Somatic copy number variation analysis is performed based on the processed data. S4. Variable screening: Remove low-frequency copy number variation fragments and screen for copy number variation fragments and clinical variables associated with recurrence after liver transplantation for liver cancer; S5. Predictive Model Construction: Incorporate the screened copy number variation fragments and clinical variables into a multivariate regression model to construct a visual nomogram predictive model; S6. Model Validation: The performance of the prediction model constructed in step S5 is validated using the training set and multiple validation queues; S7. Preoperative risk assessment: Calculate the patient score based on the nomogram prediction model in step S5, assess the risk of recurrence after liver transplantation based on the score results, and determine the selection criteria for transplant candidates. S8. Model Optimization: Based on the validation results of step S6 and the risk assessment data of step S7, the model parameters are calibrated periodically to maintain prediction accuracy.
[0019] In this example, it should also be noted that the implementation process of step S1 is as follows: S1.1. Plasma samples from hepatocellular carcinoma liver transplant recipients from three medical centers in my country were selected, with a total of 260 samples. S1.2. The sample inclusion criteria are: hepatocellular carcinoma patients with complete clinical information who have undergone liver transplantation; S1.3. The sample exclusion criteria are: presence of large blood vessel invasion or distant metastasis, survival time after liver transplantation less than 90 days, tumor recurrence time after liver transplantation less than 60 days, mixed characteristics of hepatocellular-cholangiocarcinoma, and poor quality or insufficient quantity of plasma samples. S1.4. Collect clinicopathological information of the recipients corresponding to the samples, including gender, age, hepatitis B virus infection status, tumor size, number of tumors, tumor pathological differentiation degree, tumor capsule invasion, recurrence time, recurrence site, follow-up time, time of death, cause of death, Child-Pugh score, MELD score, hepatitis B surface antigen, alpha-fetoprotein, carcinoembryonic antigen, carbohydrate antigen 199, abnormal prothrombin, total protein, alanine aminotransferase, aspartate aminotransferase, aspartate aminotransferase / alanine aminotransferase ratio, gamma-glutamyl transferase, alkaline phosphatase, total bilirubin, direct bilirubin, and creatinine.
[0020] In this example, it should also be noted that the plasma sample processing procedure in step S2 is as follows: S2.1. Collect 10 ml of peripheral blood from the patient before liver transplantation in an EDTA anticoagulant tube. S2.2. Within 2 hours after collection, the plasma was separated by centrifugation at 1600×g for 10 minutes at 4℃. The supernatant plasma was then centrifuged again at 16000×g for 10 minutes to remove cell debris. S2.3. Aliquot the clarified plasma and freeze it at -80°C until free DNA is extracted.
[0021] In this example, it should also be noted that the implementation process for cell-free DNA extraction and quality control in step S2 is as follows: S2.4. Use the QIAseq cfDNA Extraction Kit to isolate cell-free DNA from cryopreserved plasma; S2.5. The concentration and purity of the extracted free DNA were determined using Nanodrop and Qubit. S2.6. Assess the integrity and fragmentation of free DNA using 1% agarose gel electrophoresis to ensure that sequencing requirements are met.
[0022] In this example, it should also be noted that the implementation process of step S2, Chinese library construction and whole-genome sequencing, is as follows: S2.7. Take 10 ng of qualified free DNA and use the NEBnext Ultra II DNA Library PrepKit to construct a sequencing library without additional fragmentation. S2.8. After ligating the adapter with an 8bp index sequence into the constructed library, perform PCR amplification and purify the amplified library; S2.9. Perform paired-end sequencing on the Illumina HiSeqXten platform, with the sequencing target being low-coverage whole-genome sequencing at 2-3x.
[0023] In this example, it should also be noted that the implementation process of step S3 is as follows: S3.1. Use FastQC to assess the quality of raw sequencing data, and detect sequencing quality values, base composition, adapter contamination and duplication levels; S3.2. Use trim_galore to trim adapters and low-quality bases from sequencing data; S3.3. The pruned sequence was aligned to the human reference genome GRCh38 using the bwa-mem algorithm; S3.4. Use samtools to compare and convert the format of the result files, sort them, and create an index; S3.5. Use tools in GATK to identify and remove repetitive sequences introduced by PCR amplification; S3.6. Somatic copy number variation analysis was performed on the processed cell-free DNA low-depth whole genome sequencing data using ichorCNA. The analysis process integrated sequence read depth, GC content and fragment size distribution information, and copy number variation was detected based on the hidden Markov model framework. S3.7. Each sample was normalized using cell-free DNA samples from 10 internal healthy controls.
[0024] In this example, it should also be noted that the implementation process of step S4 is as follows: S4.1. Remove low-frequency copy number variation fragments from the sequencing data; S4.2. The LASSO regression method was used to screen for remaining copy number variant fragments and identify copy number variant fragments that were highly associated with recurrence after liver transplantation for hepatocellular carcinoma. S4.3. Univariate Cox regression analysis was used to screen clinical variables associated with recurrence after liver transplantation for hepatocellular carcinoma; S4.4. Perform multivariate analysis to identify copy number variation fragments, abnormal prothrombin, and alpha-fetoprotein as significant predictors of recurrence.
[0025] In this example, it should also be noted that the implementation process of step S5 is as follows: S5.1. Incorporate the significant predictors identified in step S4 into the multivariate Cox regression model; S5.2. Construct a prediction model in the form of a nomogram based on the results of the multivariate Cox regression model. This prediction model is the ZJU standard. S5.3. The model performance was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and optimistic correction C-index. The C-index of the model in the derivation cohort was 0.802, and the areas under the curve for 1 year, 3 years, and 5 years were 0.807, 0.841, and 0.817, respectively.
[0026] In this example, it should also be noted that the implementation process of step S6 is as follows: S6.1. Select a training set and three independent validation queues to validate the prediction model. The three independent validation queues include one internal validation queue and two external validation queues. S6.2. The areas under the curve for the internal validation cohort at 1 year, 3 years, and 5 years are 0.821, 0.758, and 0.766, respectively. S6.3. The areas under the curve for the first external validation cohort over 1 year, 3 years, and 5 years are 0.802, 0.759, and 0.745, respectively. S6.4. The areas under the curve for the second external validation cohort at 1 year and 3 years were 0.864 and 0.815, respectively. This cohort was not validated for 5 years due to insufficient follow-up time.
[0027] In this example, it should also be noted that the implementation process of step S7 is as follows: S7.1. Calculate the patient's score based on the nomogram prediction model, and use a score ≤100 as the cutoff value for screening hepatocellular carcinoma patients suitable for liver transplantation; S7.2. Patients with a score ≤100 have a 5-year overall survival rate ≥73% and a recurrence rate ≤22%; patients with a score >100 have a 5-year overall survival rate ≤38% and a recurrence rate ≥59%. S7.3. After completing risk assessments for every 10 batches of patients, the parameters of the prediction model are recalibrated based on the patients' follow-up results and recurrence status. The weights of each predictor in the nomogram are adjusted to achieve stable cyclic optimization of the model.
[0028] Example 2, please refer to Figures 1 to 10 In practical applications, the preoperative risk assessment and prediction method for liver cancer liver transplant patients of this invention specifically includes the following steps: Sample collection: This invention incorporates plasma samples from 260 liver transplant recipients with hepatocellular carcinoma from three centers in my country; Sample inclusion and exclusion criteria: Inclusion criteria include plasma samples from hepatocellular carcinoma liver transplant recipients with complete clinical information and who have undergone liver transplantation; Exclusion criteria include: presence of large blood vessel invasion or distant metastasis, survival time of less than 90 days after liver transplantation, tumor recurrence time of less than 60 days after liver transplantation, and mixed features of hepatocellular-cholangiocarcinoma. In addition, plasma samples with poor quality or insufficient quantity were also excluded; Collect clinicopathological information of the recipients corresponding to the above specimens, including: gender, age, hepatitis B (HBV) infection status, tumor characteristics (size, number, pathological differentiation degree, capsule invasion), recurrence time, recurrence site, follow-up time, time of death, cause of death, Child-Pugh score, MELD score, hepatitis B surface antigen, alpha-fetoprotein, carcinoembryonic antigen, carbohydrate antigen 199, abnormal prothrombin, total protein, alanine aminotransferase, aspartate aminotransferase, AST / ALT ratio, gamma-glutamyl transferase, alkaline phosphatase, total bilirubin, direct bilirubin, and creatinine; Plasma cfDNA WGS sequencing: (1) Plasma sample collection: Before the transplant, 10 ml of peripheral blood was collected from each patient using EDTA anticoagulant tubes; Within 2 hours after sample collection, plasma was separated by centrifugation at 1600×g for 10 minutes at 4°C. Collect the supernatant plasma and centrifuge it again at 16000×g for 10 minutes to completely remove residual cell debris; The clarified plasma was aliquoted and frozen at -80°C until free DNA extraction was performed. (2) DNA extraction and quantification: Cell-free DNA extraction: Cell-free DNA was isolated from plasma using the QIAseqcfDNAExtraction Kit; DNA quality control: DNA concentration and purity were determined using Nanodrop and Qubit. DNA integrity and fragmentation were assessed by 1% agarose gel electrophoresis. (3) Library construction and sequencing: Free DNA sequencing library: Take 10 ng of free DNA and construct a library using NEB next Ultra II DNA Library Prep Kit. No additional fragmentation is required. PCR amplification was performed after ligating a adapter with an 8-bp index sequence; The purified library was subjected to paired-end sequencing on the Illumina HiSeqXten platform, with the goal of low-coverage whole-genome sequencing at 2-3x speeds. Data preprocessing (1) Data quality control and preprocessing: FastQC was used to assess the quality of raw sequencing data, checking key indicators such as sequencing quality value, base composition, adapter contamination and duplication level. Use trim_galore to trim joints and low-quality bases; The quality-controlled sequences were aligned to the human reference genome GRCh38 using the bwa-mem algorithm. Use samtools to convert the format of the alignment result file, sort it, and create an index; Use tools in GATK to identify and remove repetitive sequences introduced by PCR amplification; (2) Copy number variation analysis: Low-depth whole-genome sequencing data of cell-free DNA: somatic copy number variation analysis using ichorCNA; This tool detects copy number mutations by integrating sequence read length depth, GC content, and fragment size distribution information, based on a hidden Markov model framework. Each sample was normalized using cfDNA samples from 10 internal healthy controls. Model building: First, low-frequency copy number variation fragments were removed, and then LASSO regression was used to screen for important fragments. Key clinical variables were screened using Cox regression. The selected clinical variables and copy number variation fragments were incorporated into a multivariate Cox regression model, and a nomogram was constructed based on the results (see [link to model]). Figure 1 ), Figure 1 For nomograms: PIVKA-II: high, >40 mAU / mL; low, ≤40 mAU / mL. AFP: high, >60 ng / mL; low, ≤60 ng / mL; Fragment 1 is increased in the 2000001-3000000 segment of chromosome 7; Fragment 2 is deleted in the 47000001-48000000 segment of chromosome 16. For the training set data, receiver operating characteristics (ROC) curves were used (see [link]). Figure 2 ) and calibration curve (see Calibration curve, see Figure 3 The model performance was evaluated, and the overfitting of the model was assessed using the optimism-corrected C-index, which showed that the model had very low overfitting. We proposed the ZJU standard based on nomograms; This standard presents the predictive model in a visual format, which helps to achieve clinical translation in personalized risk assessment; In the model derivation queue, the model exhibits strong discriminative ability, with a C-index of 0.802; Time-dependent receiver operating characteristic (ROC curve) analysis showed that the area under the curve for the model at 1 year, 3 years, and 5 years were 0.807, 0.841, and 0.817, respectively (see [link to analysis]). Figure 2 ); Validate the model using validation set data; When using data from three different validation queues for validation, good prediction results were obtained (see...). Figure 4 (5, 6) Prognostic analysis: We proposed the ZJU standard based on nomogram; The standard presents the predictive model in a visual format, which facilitates its clinical implementation in personalized risk assessment; Specifically, we recommend using a nomogram score ≤100 as the cutoff value for screening hepatocellular carcinoma patients suitable for liver transplantation; We divided patients into a training set and a validation set, with the validation set data coming from three different medical centers; In the model-derived cohort, patients with a nomogram score >100 had a significantly worse prognosis, with a 5-year overall survival rate of 36% and a hepatocellular carcinoma recurrence rate of 59%. Conversely, patients with a score ≤100 showed significantly improved survival outcomes, with a 5-year overall survival rate of 84% and a relapse rate of only 17%. Figure 7 ); In the internal validation cohort, patients with high scores had a lower 5-year overall survival and a higher relapse rate, while patients with low scores achieved a 76% 5-year overall survival and a 21% relapse rate. Figure 8 ); In the external validation cohort I, prognostic stratification remained robust: the 5-year overall survival rate for high-scoring patients was 38%, and the relapse rate was 63%. The overall survival rate for patients with low scores was 73%, and the recurrence rate was 22%. Figure 9 ); In the external validation cohort II, since all patients in this cohort were enrolled after 2021 and the follow-up time was insufficient, the accuracy of the model in predicting recurrence within 5 years could not be verified. In this cohort, patients with a nomogram score >100 had a significantly worse prognosis, with a 3-year overall survival rate of 60% and a hepatocellular carcinoma recurrence rate of 45%. Patients with a score ≤100 showed significantly improved survival outcomes, with a 3-year overall survival rate of 86% and a relapse rate of only 16%. Figure 10 ).
[0029] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for preoperative risk assessment and prediction in liver transplant patients with liver cancer, characterized in that, Includes the following steps: S1. Sample collection: Select plasma samples and corresponding clinicopathological information from liver transplant recipients of hepatocellular carcinoma, and clarify the inclusion and exclusion criteria for samples; S2. Cell-free DNA extraction and whole-genome sequencing from plasma: Cell-free DNA was extracted and quality-controlled from the plasma samples collected in step S1, and sequencing libraries were constructed and low-coverage whole-genome sequencing was performed. S3. Data preprocessing and copy number variation analysis: The raw sequencing data is subjected to quality control, trimming, alignment, and deduplication. Somatic copy number variation analysis is performed based on the processed data. S4. Variable screening: Remove low-frequency copy number variation fragments and screen for copy number variation fragments and clinical variables associated with recurrence after liver transplantation for liver cancer; S5. Predictive Model Construction: Incorporate the screened copy number variation fragments and clinical variables into a multivariate regression model to construct a visual nomogram predictive model; S6. Model Validation: The performance of the prediction model constructed in step S5 is validated using the training set and multiple validation queues; S7. Preoperative risk assessment: Calculate the patient score based on the nomogram prediction model in step S5, assess the risk of recurrence after liver transplantation based on the score results, and determine the selection criteria for transplant candidates. S8. Model Optimization: Based on the validation results of step S6 and the risk assessment data of step S7, the model parameters are calibrated periodically to maintain prediction accuracy.
2. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 1, characterized in that, The implementation process of step S1 is as follows: S1.
1. Plasma samples from hepatocellular carcinoma liver transplant recipients from three medical centers in my country were selected, with a total of 260 samples. S1.
2. The sample inclusion criteria are: hepatocellular carcinoma patients with complete clinical information who have undergone liver transplantation; S1.
3. The sample exclusion criteria are: presence of large blood vessel invasion or distant metastasis, survival time after liver transplantation less than 90 days, tumor recurrence time after liver transplantation less than 60 days, mixed characteristics of hepatocellular-cholangiocarcinoma, and poor quality or insufficient quantity of plasma samples. S1.
4. Collect clinicopathological information of the recipients corresponding to the samples, including gender, age, hepatitis B virus infection status, tumor size, number of tumors, tumor pathological differentiation degree, tumor capsule invasion, recurrence time, recurrence site, follow-up time, time of death, cause of death, Child-Pugh score, MELD score, hepatitis B surface antigen, alpha-fetoprotein, carcinoembryonic antigen, carbohydrate antigen 199, abnormal prothrombin, total protein, alanine aminotransferase, aspartate aminotransferase, aspartate aminotransferase / alanine aminotransferase ratio, gamma-glutamyl transferase, alkaline phosphatase, total bilirubin, direct bilirubin, and creatinine.
3. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 2, characterized in that, The plasma sample processing procedure in step S2 is as follows: S2.
1. Collect 10 ml of peripheral blood from the patient before liver transplantation in an EDTA anticoagulant tube. S2.
2. Within 2 hours after collection, the plasma was separated by centrifugation at 1600×g for 10 minutes at 4℃. The supernatant plasma was then centrifuged again at 16000×g for 10 minutes to remove cell debris. S2.
3. Aliquot the clarified plasma and freeze it at -80°C until free DNA is extracted.
4. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 3, characterized in that, The implementation process for free DNA extraction and quality control in step S2 is as follows: S2.
4. Use the QIAseq cfDNA Extraction Kit to isolate cell-free DNA from cryopreserved plasma; S2.
5. The concentration and purity of the extracted free DNA were determined using Nanodrop and Qubit. S2.
6. Assess the integrity and fragmentation of free DNA using 1% agarose gel electrophoresis to ensure that sequencing requirements are met.
5. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 4, characterized in that, The implementation process for step S2, Chinese library construction and whole-genome sequencing, is as follows: S2.
7. Take 10 ng of qualified free DNA and use the NEBnext Ultra II DNA Library Prep Kit to construct a sequencing library without additional fragmentation. S2.
8. After ligating the adapter with an 8bp index sequence into the constructed library, perform PCR amplification and purify the amplified library; S2.
9. Perform paired-end sequencing on the Illumina HiSeqXten platform, with the sequencing target being low-coverage whole-genome sequencing at 2-3x.
6. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 5, characterized in that, The implementation process of step S3 is as follows: S3.
1. Use FastQC to assess the quality of raw sequencing data, and detect sequencing quality values, base composition, adapter contamination and duplication levels; S3.
2. Use trim_galore to trim adapters and low-quality bases from sequencing data; S3.
3. The pruned sequence was aligned to the human reference genome GRCh38 using the bwa-mem algorithm; S3.
4. Use samtools to compare and convert the format of the result files, sort them, and create an index; S3.
5. Use tools in GATK to identify and remove repetitive sequences introduced by PCR amplification; S3.
6. Somatic copy number variation analysis was performed on the processed cell-free DNA low-depth whole genome sequencing data using ichorCNA. The analysis process integrated sequence read depth, GC content and fragment size distribution information, and copy number variation was detected based on the hidden Markov model framework. S3.
7. Each sample was normalized using cell-free DNA samples from 10 internal healthy controls.
7. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 6, characterized in that, The implementation process of step S4 is as follows: S4.
1. Remove low-frequency copy number variation fragments from the sequencing data; S4.
2. The LASSO regression method was used to screen for remaining copy number variant fragments and identify copy number variant fragments that were highly associated with recurrence after liver transplantation for hepatocellular carcinoma. S4.
3. Univariate Cox regression analysis was used to screen clinical variables associated with recurrence after liver transplantation for hepatocellular carcinoma; S4.
4. Perform multivariate analysis to identify copy number variation fragments, abnormal prothrombin, and alpha-fetoprotein as significant predictors of recurrence.
8. The method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 7, characterized in that, The implementation process of step S5 is as follows: S5.
1. Incorporate the significant predictors identified in step S4 into the multivariate Cox regression model; S5.
2. Construct a prediction model in the form of a nomogram based on the results of the multivariate Cox regression model. This prediction model is the ZJU standard. S5.
3. The model performance was evaluated using receiver operating characteristic (ROC) curves, calibration curves, and optimistic correction C-index. The C-index of the model in the derivation cohort was 0.802, and the areas under the curve for 1 year, 3 years, and 5 years were 0.807, 0.841, and 0.817, respectively.
9. A method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 8, characterized in that, The implementation process of step S6 is as follows: S6.
1. Select a training set and three independent validation queues to validate the prediction model. The three independent validation queues include one internal validation queue and two external validation queues. S6.
2. The areas under the curve for the internal validation cohort at 1 year, 3 years, and 5 years are 0.821, 0.758, and 0.766, respectively. S6.
3. The areas under the curve for the first external validation cohort over 1 year, 3 years, and 5 years are 0.802, 0.759, and 0.745, respectively. S6.
4. The areas under the curve for the second external validation cohort at 1 year and 3 years were 0.864 and 0.815, respectively. This cohort was not validated for 5 years due to insufficient follow-up time.
10. A method for preoperative risk assessment and prediction in liver transplant patients with liver cancer according to claim 9, characterized in that, The implementation process of step S7 is as follows: S7.
1. Calculate the patient's score based on the nomogram prediction model, and use a score ≤100 as the cutoff value for screening hepatocellular carcinoma patients suitable for liver transplantation; S7.
2. Patients with a score ≤100 have a 5-year overall survival rate ≥73% and a recurrence rate ≤22%; patients with a score >100 have a 5-year overall survival rate ≤38% and a recurrence rate ≥59%. S7.
3. After completing risk assessments for every 10 batches of patients, the parameters of the prediction model are recalibrated based on the patients' follow-up results and recurrence status. The weights of each predictor in the nomogram are adjusted to achieve stable cyclic optimization of the model.