Preoperative risk stratification system based on liver resection after transformation treatment

Abstract

The invention discloses a preoperative risk stratification system based on hepatic resection after transformation treatment in the technical field of biomedicine, and the system comprises a sample module which is used for selecting a proper hepatocellular carcinoma patient as a sample, distributing the patient sample to a training queue and an internal verification queue, and building an external verification queue; the treatment module is used for providing a treatment strategy for the patient based on whether liver resection is performed or not after conversion treatment; the tracking module is used for carrying out regular follow-up visit on the patient subjected to the liver resection or not subjected to the liver resection after the conversion treatment; and the data analysis module is used for determining prediction factors related to recurrence-free survival in the training queue by adopting a minimum absolute contraction and selection operator regression analysis method, further analyzing and screening out independent prognosis factors through multivariable Cox regression, and constructing a column graph based on the factors to predict the postoperative recurrence rate. Through verification of the internal and external verification queues, the method can effectively guide and select a patient who is most likely to benefit from the liver resection operation after transformation treatment.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technologies, and particularly to a preoperative risk stratification system for liver resection based on conversion therapy. Background Art

[0002] Hepatocellular carcinoma is one of the most common malignant tumors globally and poses a serious threat to public health due to its high mortality rate. According to the Barcelona Clinic Liver Cancer (BCLC) staging system and the Chinese Liver Cancer (CNLC) staging system, early-stage and some mid- to late-stage hepatocellular carcinomas (mainly patients with BCLC A / CNLC Ia, Ib, and some BCLC B / CNLC II) are suitable for radical treatments such as surgical resection, local ablation, and liver transplantation, with a median survival time exceeding 5 years. However, some patients are not suitable for surgery due to their inability to tolerate surgical trauma or insufficient future liver remnant (FLR), and thus are considered surgically unresectable. In addition, most patients with hepatocellular carcinoma present at mid- to late-stage (BCLC B and C / CNLC IIb, IIIa, and IIIb). Although some patients may still benefit from surgical resection, most patients are not suitable for direct surgery. Against this background, conversion therapy emerged as an innovative approach, aiming to increase the FLR and induce tumor response to achieve tumor shrinkage or downstaging, thereby converting surgically or oncologically unresectable hepatocellular carcinoma (uHCC) tumors into resectable tumors. This strategy has significantly improved survival outcomes, with a 5-year postoperative survival rate reaching 50% to 60%, comparable to the resection effect of early-stage hepatocellular carcinoma. Therefore, conversion therapy has become an important strategy to improve the treatment effect of uHCC and has received extensive attention in recent years.

[0003] Historically, transarterial chemoembolization (TACE) has been the main method for conversion therapy of uHCC. However, the surgical conversion rate using only TACE is only about 10%. The emergence and wide application of hepatic artery infusion chemotherapy (HAIC), especially represented by the oxaliplatin and 5-fluorouracil regimen, have significantly improved tumor regression, control of tumor thrombus, and prognosis, with a surgical conversion rate reaching 10.2% - 23.9%. Recent advancements in systemic therapy have led to the development of novel multi-combination conversion strategies. Combining interventional therapy with systemic drug therapy can achieve an objective response rate (ORR) of up to 60 - 90% (based on mRECIST), a complete response rate (CR) of 14.7% to 42%, and a conversion surgery rate exceeding 30%. At the same time, these results have also raised an important scientific question: for patients with hepatocellular carcinoma who have successfully undergone conversion, especially those who have achieved CR, is conversion liver resection still necessary?

[0004] Although conversion hepatectomy is a key goal in the treatment of uHCC, long-term survival remains the ultimate aim. The efficacy of conversion therapy is closely related to the degree of tumor remission. Current evidence shows that most patients who achieve a clinical remission may still experience disease progression (PD) over time even if they continue with the medication. The significance of resection after conversion lies in its potential curative effect, which can prolong progression-free survival (PFS) and overall survival (OS). Some studies have reported that resection after conversion therapy can significantly improve long-term survival rates. However, other studies have shown that surgical resection may not be necessary, especially for patients at high risk of recurrence or those who achieve radiological CR. The latest research findings indicate that for patients with radiological CR, a "watch and wait" strategy can still yield good long-term survival outcomes. Therefore, accurately identifying the patient population most likely to benefit from hepatectomy after conversion therapy is crucial.

[0005] Therefore, the present invention proposes a risk stratification system before hepatectomy after conversion therapy, which is verified through internal and external validation cohorts to guide the selection of patients most likely to benefit from surgical resection after conversion therapy. Summary of the Invention

[0006] To solve the above problems, the present invention provides a risk stratification system before hepatectomy after conversion therapy, which is verified through internal and external validation cohorts to guide the selection of patients most likely to benefit from surgical resection after conversion therapy.

[0007] To achieve the above object, the technical solution of the present invention is as follows: A risk stratification system before hepatectomy after conversion therapy, comprising: a sample module, which is used to select suitable hepatocellular carcinoma patients as patient samples based on inclusion criteria and exclusion criteria, and allocate the patient samples to a training cohort and an internal validation cohort; and is also used to establish an external validation cohort; collect clinical data of the patients in all cohorts, and the clinical data includes baseline and preoperative characteristics;

[0008] A treatment module, which is used to provide a treatment strategy for the patient on whether to perform hepatectomy after conversion therapy;

[0009] A tracking module, which is used to regularly follow up the patients who have or have not undergone hepatectomy after conversion therapy until the patient's condition recurs or the patient dies; and record the time from the initial treatment to disease progression, recurrence, or the last follow-up of the patient, which is defined as progression-free survival; record the time from the initial treatment to death or the last follow-up, which is defined as overall survival; record the time from surgery to disease recurrence or the last follow-up, which is defined as recurrence-free survival;

[0010] The data analysis module is used to determine the risk factors related to recurrence-free survival in the training cohort based on the least absolute shrinkage and selection operator regression method, and then further analyze the risk factors using a multivariate Cox regression model to screen out independent risk factors. Based on four independent risk factors, namely the number of interventional treatments, the number of baseline tumors, the preoperative AFP level, and the preoperative tumor thrombus activity, a nomogram is established to predict the probability of tumor recurrence after conversion resection. According to the optimal cut-off value in the nomogram, patients are divided into a high-risk group and a low-risk group, and further comparative analysis of recurrence-free survival, progression-free survival, and overall survival of the two groups of patients is carried out.

[0011] Furthermore, the inclusion criteria include at least: conversion therapy includes interventional therapy with or without TKI and / or ICI, the Eastern Cooperative Oncology Group performance status is 0 or 1, successful achievement of R0 resection, and Child-Pugh liver function classification is grade A or B.

[0012] Furthermore, the exclusion criteria include incomplete or lost follow-up of laboratory and imaging evaluation data after treatment, severe dysfunction of important organs, and diagnosis or medical history of other concurrent malignancies.

[0013] Furthermore, interventional therapy includes TACE and / or HAIC.

[0014] Furthermore, the accuracy and performance of the nomogram are evaluated and verified by the concordance index, calibration curve, and internal and external validation methods.

[0015] Furthermore, the data analysis module is also used to evaluate the predictive ability of the nomogram relative to the traditional staging system based on the receiver operating characteristic curve and time-dependent area under the curve comparison.

[0016] Furthermore, in the data analysis module, the multiple imputation method is used to handle the problem of missing data.

[0017] Furthermore, categorical variables in the data analysis module are reported as percentages and analyzed using the chi-square test or Kruskal-Wallis test; continuous variables in the data analysis module are expressed as the median and its interquartile range and evaluated using appropriate parametric or nonparametric statistical tests.

[0018] Furthermore, the data analysis module is also used to perform survival analysis based on the Kaplan-Meier curve and log-rank test.

[0019] The following beneficial effects are obtained by adopting the above scheme:

[0020] Compared with the traditional staging system for conversion therapy after hepatectomy in patients with hepatocellular carcinoma, this protocol uses lasso regression for multivariable selection. The final model only includes four key variables (baseline tumor number, number of interventional treatments, preoperative AFP level, and preoperative tumor thrombus activity), which can stably and accurately predict postoperative recurrence and overall survival after conversion resection. Secondly, a comprehensive analysis of baseline and preoperative indicators was performed to predict postoperative recurrence, which helps to select suitable candidates for conversion resection before surgery. Regardless of whether imaging complete remission is achieved, conversion resection provides significant benefits for patients with low-risk patients.

[0021] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flowchart of an embodiment of a preoperative risk stratification system for hepatectomy based on conversion therapy of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0023] The technical solutions of the present invention will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] The following is further detailed through specific embodiments:

[0025] Example:

[0026] As shown in the attached Figure 1As shown in the figure, a preoperative risk stratification system for liver resection based on conversion therapy includes: a sample module, which is used to select suitable hepatocellular carcinoma patients as patient samples based on inclusion criteria and exclusion criteria, and allocate the patient samples to a training cohort and an internal validation cohort; it is also used to establish an external validation cohort; collect clinical data on patients in all cohorts, and the clinical data includes baseline and preoperative characteristics. The inclusion criteria at least include that the conversion therapy adopts interventional therapy (HAIC and / or TACE), which can be combined with or without TKI and / or ICI, the Eastern Cooperative Oncology Group (ECOG) performance status is 0 or 1, R0 resection is successfully achieved, and the Child-Pugh liver function classification is grade A or B. At the same time, the inclusion criteria can also include being diagnosed with hepatocellular carcinoma according to the AASLD criteria; no extrahepatic metastasis; unresectable hepatocellular carcinoma caused by inability to tolerate surgical trauma, insufficient future liver remnant (FLR), insufficient resection margin, high postoperative recurrence risk or advanced disease. The exclusion criteria include incomplete or lost follow-up of laboratory and imaging evaluation data after treatment, severe dysfunction of important organs, and diagnosis or medical history of other concurrent malignancies. Among them, R0 resection is an important evaluation index in liver resection, which is used to judge whether the tumor has been radically resected. Specifically, R0 resection means that during the operation, the tumor lesion is completely resected, and no cancer cells are found when examining the resection margin under the microscope, that is, there is no residual cancer cells visible to the naked eye or under the microscope. This standard indicates that the tumor resection has achieved the optimal treatment effect and is an important basis for judging the success of the operation.

[0027] A treatment module, which is used to provide a treatment strategy for whether to perform liver resection after conversion therapy. Among them, the conversion therapy includes interventional therapy combined with or without TKI and ICI; the interventional therapy includes TACE and / or HAIC.

[0028] A tracking module, which is used to regularly follow up patients who have or have not undergone liver resection after receiving conversion therapy until the patient's condition recurs or the patient dies; record the time from the initial treatment to disease progression, recurrence or the last follow-up of the patient, which is defined as progression-free survival (PFS); record the time from the initial treatment to death or the last follow-up, which is defined as overall survival (OS); record the time from surgery to disease recurrence or the last follow-up, which is defined as recurrence-free survival (RFS).

[0029] The data analysis module is used to determine the risk factors related to RFS in the training cohort based on the Least Absolute Shrinkage and Selection Operator (LASSO) regression method. On this basis, a multivariate Cox regression model is used to further analyze the risk factors and screen out independent risk factors. Finally, based on independent risk factors such as the number of interventional treatments, the number of baseline tumors, the preoperative AFP level, and the preoperative tumor thrombus activity, a nomogram is established to predict the probability of tumor recurrence after conversion resection. According to the optimal cut-off value in the nomogram, the patients are divided into a high-risk group and a low-risk group. Subsequently, further comparative analysis of RFS, PFS, and OS of the two groups of patients is carried out. In addition, the study also included a group of patients who did not undergo conversion resection but met the resectable criteria after conversion therapy. By comparing the differences in PFS and OS between patients with different recurrence risks in the surgical group and the non-surgical group, it aims to provide a scientific basis for clinical treatment decisions.

[0030] In the data analysis module, we included a comprehensive set of clinical indicators, covering the patient's individual characteristics, baseline and preoperative serological markers, oncological characteristics, and individualized treatment plans, totaling 32 variables. By applying the Least Absolute Shrinkage and Selection Operator (LASSO) regression method to screen these variables and conducting iterative analysis in combination with the 1-standard error (1-se) criterion, a model that balances optimal performance and the fewest variables is finally obtained. This model selected four independent risk factors: the frequency of interventional treatment (Frequency of IT), the number of baseline tumors (B. Number), the preoperative AFP level (P. AFP), and the preoperative tumor thrombus activity (P. embolus). Based on these independent risk factors, we established a nomogram to predict the 1-year, 3-year, and 5-year recurrence-free survival (RFS) of hepatocellular carcinoma patients who underwent surgical resection after conversion therapy. This nomogram estimates the probability of postoperative recurrence by adding the scores of each variable and calculating the total score in combination with the scoring scale.

[0031] The data analysis module is also used to evaluate the predictive ability of the nomogram relative to the traditional staging system based on the Receiver Operating Characteristic (ROC) curve and the comparison of the time-dependent area under the curve (AUC).

[0032] In the data analysis module, the multiple imputation method is used to handle the problem of missing data.

[0033] In the data analysis module, categorical variables are reported as percentages and analyzed using the chi-square test or the Kruskal-Wallis test; continuous variables are expressed as the median and its interquartile range (IQR) and evaluated using the corresponding parametric or non-parametric statistical tests.

[0034] The data analysis module is also used to perform survival analysis based on the Kaplan-Meier curve and log-rank test.

[0035] The specific implementation process is as follows: 635 patients who underwent conversion resection surgery in our hospital were included in the study. Among them, 445 patients were assigned to the training cohort, and 190 patients were assigned to the internal validation cohort. In addition, 89 patients from two other hospitals were introduced as the external validation cohort. To further use the model to guide clinical treatment decisions, 216 patients who did not undergo conversion resection but met the resectable criteria after conversion therapy were also included in the study.

[0036] According to the treatment needs, transarterial chemoembolization (TACE) was performed on the patients, and the specific treatment frequency was determined by three or more experienced doctors after discussing the patient's condition. Hepatic artery infusion chemotherapy (HAIC) was performed once every 3 - 4 weeks for a maximum of 6 cycles. Subsequently, for patients considered suitable for resection, a multidisciplinary team (MDT) consisting of surgeons, internists, interventional radiologists, radiologists, pathologists, and radiation oncologists discussed the subsequent treatment options. These options usually included surgical resection, systemic treatment, interventional treatment, ablation, stereotactic radiotherapy, or active surveillance. The specific selection was affected by factors such as the patient's tumor burden, overall health status, medical insurance coverage, and the patient's willingness to accept surgery. The surgical safety was comprehensively evaluated by at least two experienced surgeons to ensure that while achieving an R0 resection, 30 - 40% of the residual liver volume was preserved. The preoperative evaluation included liver, kidney, and coagulation function tests, re-evaluation of liver cancer-specific tumor markers, abdominal imaging examinations (CT, MRI, or ultrasound), chest CT scan, cardiopulmonary function assessment, and ECOG performance status score. During the operation, intraoperative ultrasound was used to accurately define the tumor boundary to determine the optimal resection range. The resection of the liver parenchyma was completed using a harmonic scalpel, and strict hemostasis was performed postoperatively. The surgical area and abdominal cavity were thoroughly rinsed with sterile water.

[0037] In the first two years after the patients received conversion liver resection, they were usually followed up every 3 - 4 months, and then every 6 months until recurrence or death occurred. For patients who received non-surgical treatment, they were followed up every 2 - 3 months in the first two years and then every 3 - 6 months. Each follow-up included routine blood tests (including blood routine, liver function tests, and tumor marker assessment) and abdominal imaging examinations (CT, MRI, or ultrasound). The diagnosis of recurrence was based on the following two criteria: (1) new lesions with typical imaging features of liver cancer were found in two imaging examinations; (2) new extrahepatic lesions were found that were not detected before resection.

[0038] Comparison of the tumor characteristics of patients in the training cohort and the internal validation cohort revealed that in the external validation cohort, the median tumor diameter was smaller (8.0 cm vs. 8.0 cm vs. 6.5 cm, p = 0.043), but the incidence of tumor thrombus was significantly higher (25.2% vs. 21.6% vs. 71.9%, p < 0.001), the proportion of AFP positivity was higher (67.9% vs. 67.4% vs. 85.4%, p = 0.003), and the disease stage was later (BCLC C / CNLC III stage, 25.2% vs. 21.6% vs. 71.9%, p < 0.001). Other parameters, including cirrhosis, HBsAg status, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), ALT, AST, TBIL, ALBI score, and PT, showed no significant differences among the three groups.

[0039] Among the 635 patients included, 474 (74.6%) received a treatment regimen based on hepatic artery infusion chemotherapy (HAIC), either alone or in combination with transcatheter arterial chemoembolization (TACE). Most patients (60.9%) received 2 or fewer interventional treatments, and 39.1% of the patients received more than 2 treatments. In contrast, 71.9% of the patients in the external validation cohort received only 1 - 2 TACE treatments. In addition, 33.1% of the patients in this cohort received combined systemic therapy, while this proportion was significantly higher in the external validation cohort (67.4%), which was consistent with its later tumor stage.

[0040] After receiving conversion therapy, tumor characteristics improved in each cohort compared with baseline. Specifically, tumor size: the median tumor size decreased from 8.0 cm to 5.7 cm (training cohort), from 8.0 cm to 5.95 cm (internal validation cohort), and from 6.5 cm to 4.5 cm (external validation cohort); number of tumors: the proportion of patients with ≤ 3 lesions increased from 76% to 78.2% (training cohort), from 74.7% to 76.8% (internal validation cohort), and from 84.3% to 86.5% (external validation cohort); proportion of cancer thrombus activity: decreased from 25.2% to 20.2% (training cohort), from 21.6% to 20.5% (internal validation cohort), and from 71.9% to 46.1% (external validation cohort); proportion of AFP positivity: decreased from 67.9% to 46.7% (training cohort), from 67.4% to 45.8% (internal validation cohort), and from 85.4% to 80.9% (external validation cohort). Overall, the decrease in the proportion of AFP positivity was particularly significant in the internal cohort (from 67.7% to 46.5%), while the reduction in the proportion of cancer thrombus activity was more prominent in the external validation cohort (from 71.9% to 46.1%). The above results indicate that conversion therapy plays an important role in reducing tumor burden and improving tumor biological characteristics.

[0041] Harrell's C-index is an important indicator for measuring the discrimination ability of a prediction model, and its value ranges from 0.5 (no discrimination ability) to 1 (perfect discrimination ability). In this study, the C-index was used to evaluate the prediction performance of the prediction model for recurrence-free survival (RFS). The results showed that it was 0.78 (95% confidence interval: 0.72 - 0.81) in the training cohort, 0.73 (95% confidence interval: 0.63 - 0.78) in the internal validation cohort, and 0.88 (95% confidence interval: 0.82 - 0.93) in the external validation cohort. In contrast, the C-indexes for RFS prediction of six commonly used staging systems were: 0.59 for the Barcelona Clinic Liver Cancer (BCLC) staging, 0.55 for the American Joint Committee on Cancer (AJCC) staging, 0.57 for the albumin-bilirubin (ALBI) grade, 0.51 for the Eastern Cooperative Oncology Group (ECOG) performance status score, 0.50 for the Child-Pugh score, and 0.60 for the Chinese Liver Cancer Staging (CNLC). These results indicate that the prognostic model proposed in this study is significantly superior to the six commonly used staging systems in terms of RFS prediction performance.

[0042] The study further compared the clinical outcomes of patients who did not undergo conversion resection but met the resectability criteria after conversion therapy (non-surgical group, n = 216) with those of patients in the internal and external validation cohorts who underwent surgical resection during the same period (surgical group, n = 279). Among low-risk patients, the disease progression rate was significantly lower in patients who underwent surgical resection (p < 0.0001), and the overall survival (OS) was significantly higher (p = 0.033). In contrast, among high-risk patients, there were no significant differences in progression-free survival (PFS) or overall survival (OS) between the surgical and non-surgical groups (p = 0.47 and p = 0.13). Additionally, the pathologic complete response rate (pCR) was 30.7% in the low-risk group, which was significantly higher than 10.6% in the high-risk group (p < 0.0001).

[0043] Previous studies have shown that patients who achieve imaging complete response (rCR) after conversion therapy may still achieve long-term survival even without surgical resection. To make our analysis more precise, we conducted a detailed subgroup analysis, classifying patients into the imaging complete response group and the non-complete response group. In the rCR subgroup, for low-risk patients, surgical resection provided significant benefits in terms of median progression-free survival (PFS) and median overall survival (OS). Specifically, the mPFS of the surgical and non-surgical low-risk group patients was not reached vs. 30.7 months (p = 0.0067), the median OS was not reached vs. 60.5 months (p = 0.082), and the 5-year OS rates were 76.3% vs. 58.9%, respectively. In the non-CR subgroup, surgical resection also showed a significant survival advantage, with mPFS not reached (surgical resection group) compared to 5.4 months (non-surgical group, p < 0.0001), mOS not reached (surgical resection group) compared to not reached (non-surgical group, p = 0.0076), and 5-year OS rates of 70.9% (surgical resection group) and 54.6% (non-surgical group), respectively. However, among high-risk patients, surgical resection did not significantly improve PFS or OS in either the CR subgroup or the non-CR subgroup. These findings further validate the clinical utility of this system in patient stratification, which can effectively identify the best patients suitable for conversion hepatectomy.

[0044] It should be noted that the above examples are only used to clearly illustrate the research results and do not constitute a limitation on the implementation methods. For ordinary technicians in the field, based on the above description, various forms of adjustments or changes can be made to the implementation methods. It is not necessary and impossible to enumerate all possible changes, and the obvious adjustments or changes resulting therefrom still fall within the protection scope of the present invention.

Claims

1. A risk stratification system for hepatectomy after conversion therapy, characterized by: A sample module, used to select appropriate hepatocellular carcinoma patients as patient samples based on inclusion criteria and exclusion criteria, and to allocate the patient samples to a training cohort and an internal validation cohort; It is also used to establish external validation cohorts; Clinical data were collected for patients in all cohorts, including baseline and preoperative characteristics; A treatment module to provide treatment strategies for patients based on whether liver resection is performed after conversion therapy; The tracking module is used to regularly follow up patients who have undergone or have not undergone liver resection after conversion therapy until the patient's disease relapses or dies; and to record the time from initial treatment to disease progression, relapse or last follow-up, which is defined as progression-free survival; the time from initial treatment to death or last follow-up, which is defined as overall survival; and the time from surgery to disease relapse or last follow-up, which is defined as relapse-free survival; The data analysis module is used to determine the risk factors associated with recurrence-free survival in the training cohort based on the minimum absolute shrinkage and selection operator regression methods, and then use the multivariate Cox regression model to further analyze the risk factors, screen out independent risk factors, and establish a nomogram based on at least four independent risk factors including the number of interventional treatments, baseline tumor number, preoperative AFP level and preoperative tumor thrombus activity to predict the probability of tumor recurrence after conversion resection; according to the optimal cutoff value in the nomogram, the patients are divided into high-risk group and low-risk group, and the recurrence-free survival, progression-free survival and overall survival of the two groups of patients are further compared and analyzed.

2. The risk stratification system for preoperative liver resection based on conversion therapy according to claim 1, characterized in that: Inclusion criteria included at least: conversion therapy using interventional therapy with or without tyrosine kinase inhibitors and / or immune checkpoint inhibitors, Eastern Cooperative Oncology Group performance status score of 0 or 1, successful R0 resection, and Child-Pugh liver function grade A or B.

3. The risk stratification system for preoperative liver resection based on conversion therapy according to claim 2, characterized in that: Exclusion criteria included incomplete data of laboratory and imaging evaluations after treatment or loss to follow-up, severe dysfunction of vital organs, and diagnosis or history of other concurrent malignancies.

4. The risk stratification system for preoperative liver resection based on conversion therapy according to claim 3, characterized in that: Interventional treatment includes TACE and / or HAIC.

5. The risk stratification system for preoperative liver resection after conversion therapy according to claim 4, characterized in that: The accuracy and performance of the nomogram were evaluated and validated by using the consistency index, calibration curves, and internal and external validation methods.

6. The risk stratification system for preoperative liver resection based on conversion therapy according to claim 5, characterized in that: The data analysis module was also used to evaluate the predictive ability of the nomogram relative to the traditional staging system based on the receiver operating characteristic curve and time-dependent area under the curve comparison.

7. The risk stratification system for preoperative liver resection after conversion therapy according to claim 6, characterized in that: The data analysis module handles the problem of missing data based on the multiple imputation method.

8. The risk stratification system for preoperative liver resection based on conversion therapy according to claim 7, characterized in that: Categorical variables in the data analysis module were reported as percentages and analyzed using the chi-square test or Kruskal-Wallis test; continuous variables in the data analysis module were expressed as medians and their interquartile ranges and evaluated using corresponding parametric or nonparametric statistical tests.

9. The risk stratification system for preoperative liver resection after conversion therapy according to claim 8, characterized in that: The data analysis module was also used to perform survival analysis based on Kaplan-Meier curves and log-rank tests.

Images