Prediction model of M2 with early-stage hepatocellular carcinoma based on multiple machine learning

Prediction model of M2 with early-stage hepatocellular carcinoma based on multiple machine learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Prediction model of M2 with early-stage hepatocellular carcinoma based on multiple machine learning Guoyi Xia, Zeyan Yu, Shaolong Lu, Xiaobo Wang, Yuanquan Zhao, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4410132/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Microvascular invasion (MVI) is a crucial factor for early recurrence and poor outcomes in hepatocellular carcinoma (HCC). However, there are few studies on M2 classification. We aimed to build a predictive model for M2 in early-stage HCC, assisting clinical decision-making. Methods: We retrospectively enrolled 451 patients with early-stage HCC and employed multiple machine learning algorithms to identify the risk factors influencing the robustness of M2. Model performance was evaluated using receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA)，and clinical impact curve (CIC). Results: There were 363 M0 and 88 M2 cases. Differences in recurrence-free survival (RFS) and overall survival(OS) between the M0 and M2 groups were statistically significant ( P 5cm, incomplete tumor capsule, and Edmondson-Steiner stage III-IV were independent risk factors for M2.The prediction model showed an area under the receiver operating characteristic curve(AUROC) of 0.765 and 0.807 in the training and validation groups, respectively. Calibration curves showed good agreement between actual and predicted M2 risks, and the DCA and CIC showed a significant clinical efficacy. Conclusion: The nomogram-based model had a good predictive effect for M2 in patients with early-stage HCC ,providing guidance for treatment decisions. Hepatocellular carcinoma (HCC) early-stage machine learning nomogram Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Hepatocellular carcinoma (HCC) is the most common liver malignancy, ranking sixth and third in terms of global morbidity and mortality, respectively. [ 1 , 2 ] It is well known that HCC in stage Barcelona Clinic Liver Cancer-A(BCLC-A) is defined as early-stage. [ 3 ] In the 2022 update of the BCLC prognosis prediction and treatment strategy, stage BCLC 0 (normal liver function, single tumor, 2 cm diameter, no vascular invasion or extrahepatic metastasis) was defined as very early-stage. The first-line treatment strategy for HCC in stage BCLC 0/A is surgical resection, ablation and liver transplantation. [ 4 ] Despite some improvements in the diagnosis and treatment of HCC, the proportion of patients with early-stage HCC is < 30%;together with the high invasion and heterogeneity of HCC, the overall recurrence rate at 3 years remains as high as 30–50%, leading to a poor prognosis. [ 5 , 6 ] Microvascular invasion (MVI) mainly refers to the cancer cell nests that can only be observed under a microscope in the lumen of the tiny vessels lined with endothelial cells, mostly in the tumor envelope and adjacent liver tissue. [ 7 ] According to the quantity and distribution of cancer cell nests, MVI can be categorized into three levels: M0 level, where no MVI is detected; M1 level, characterized by ≤ 5 MVIs occurring in the peri-cancerous liver tissue region (≤ 1cm); and M2 level, defined by > 5 MVIs or MVIs occurring in the distant peri-cancerous liver tissue region (> 1cm). [ 8 , 9 ] MVI is associated with various factors, including the invasive ability of tumor cells, the secretion of angiogenic factors, and immune escape. Tumor cells degrade the vascular basement membrane via secreted proteases, which then invade the vascular lumen to form cancer cell nests. Simultaneously, tumor cells secrete angiogenic factors, promote the formation of neovascularization, and provide conditions for tumor growth and metastasis. Numerous studies have confirmed that MVI is an important factor affecting the recurrence and survival of patients with liver cancer, and personalized treatment based on MVI is of great significance for improve the surgical outcomes of patients with HCC.[ 10 , 11 ] Studies have shown that patients with M2 have a poorer prognosis after radical resection than those with M0 and M1 [ 8 , 12 ] , and M2 is a high-risk factor for postoperative residual cancer recurrence and intrahepatic metastasis [ 13 , 14 ] .Therefore, we should focus not only on the presence or absence of MVI, but also on the M2. If preoperative patients requiring liver resection are judged to have a higher risk of M2, it is recommended to expand the margin or advance the intervention for transformation treatment to prevent early postoperative recurrence and metastasis thus improving patient prognosis. [ 15 , 16 ] Machine learning has been widely used in research on HCC. Machine learning is more accurate, more stable, and more fully high-throughput and multi-dimensional clinical data mining, so it can improve the performance of prediction models. [ 17 ] In recent years, with the development of radiomics, an increasing number of radiomic features have been considered to predict MVI and judge the prognosis of HCC. [ 18 , 19 ] Although many prediction models have been used to predict MVI, there is a large heterogeneity in the criteria for HCC inclusion, and the risk prediction studies on M2 remain scarce. [ 20–22 ] In this study, we screened clinical indicators of HCC through machine learning, established a prediction model for early-stage HCC patients with M2, and drew a nomogram for stratified intervention to reduce postoperative recurrence, which could be beneficial for clinicians. Methods Patients and data collection The clinical data of 888 patients diagnosed with HCC from December 2012 to December 2018 at the Affiliated Cancer Hospital of Guangxi Medical University were collected. The inclusion criteria were as follows: 1) BCLC 0/A stage, 2) hepatitis B surface antigen (HBsAg) (+), and 3) MVI grade M0 and M2.The exclusion criteria were as follows: 1) imaging and blood index examination 1 week before surgery, 2) non-R0 resection, 3) preoperative surgery, ablation, radiofrequency, or neoadjuvant therapy, 4) MVI status not assessed, and 5) incomplete clinical data. Finally, the clinical data of 451 patients were included, and the number of seeds was set using R software. The patients were divided into training and validation groups at a 7:3 ratio. The training group was used to build the prediction model, and the validation group was used for internal validation ( Fig. 1 ) . This was a retrospective and controlled study. This study adhered to the Declaration of Helsinki and was approved by the Ethics Committee of Guangxi Medical University Cancer Hospital(Batch No:KY2024454). Data collection and processing Basic patient information was collected, including data on gender, age, C-reactive protein, D-dimer, complement C3, T helper cells, inhibitory T cells, natural killer cells, B-lymphocytes, total-bilirubin, total-protein, glutamic-pyruvic transaminase, glutamic-oxalacetic transaminase, alpha fetoprotein (AFP), platelet count, absolute neutrophil values, absolute monocyte values, absolute lymphocyte values, prothrombin time, cirrhosis, ascites, tumor size, number of tumors, tumor capsule, BCLC-stages, Edmondson-Steiner classification, recurrence, recurrence-free survival(RFS), survival and overall survival(OS). Postoperative tissue specimens were further examined pathologically to confirm the presence of MVI and subsequently graded. Liver cirrhosis was defined as B-ultrasound liver transient elastography with an elasticity value (or hardness value) > 7.1 kPa. RFS as defined as the time interval from the start of surgical treatment to the first recurrence of the tumor. OS was defined as the time from the diagnosis of HCC to death from any cause. After tumor resection, all patients were followed-up regularly according to clinical guidelines. Follow-ups included laboratory tests (serum AFP level and liver function tests) and imaging studies (ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI)) once every 3 months after surgery. The diagnosis of HCC recurrence was based on two or more examination methods, including ultrasonography, CT, MRI, and hepatic arteriography, with elevated serum AFP levels. The patients with confirmed recurrent HCC underwent further evaluation by a multidisciplinary team. The last follow-up was conducted on 31 December 2020. Screening for indicators related to M2 The Least Absolute Shrinkage and Selection Operator (LASSO) is a regularization algorithm used for linear regression and related problems. It achieves a sparseness of the model coefficients by adding L1 regular terms to the loss function, thus prompting the model to select relatively few features. The Boruta algorithm is a machine learning algorithm used for feature selection, which is especially suitable for the random forest model. The algorithm determines the threshold by ranking the feature importance. This threshold will be used to distinguish which features are "important" and which are "not". XGBoost (eXtreme Gradient Boosting) is a machine learning algorithm for gradient-boosting trees that has achieved significant improvements in prediction performance and computing efficiency. This provides a means for assessing the importance of these features. The relative importance of each feature can be obtained by measuring the number of splittings constructing the tree or the contribution of the feature to the target variable during splitting. The Best subset algorithm is a feature-selection method designed to select the best subset from a given set of features to build a linear regression model. The algorithm evaluates the performance of each subset by exhaustively imposing all possible feature combinations and selecting the subset with the most influence on the target. SHAP (SHapley Additive exPlanations) is a model used to interpret machine learning model prediction results. The SHAP values are based on the Shapley-value concept in game theory, which provides a fair, consistent, and efficient way to assign contributing values to each feature, thereby explaining how the model's output for each sample is formed. For each feature, the SHAP value considers all possible subset combinations that the feature may have formed with other features and calculates the average contribution of the feature to the output in these combinations. The clinical characteristic factors derived from the four machine algorithms were intersected, and a Venn diagram was created. Ultimately, the analysis identified the definitive predictive factors associated with M2. These factors were used to construct the nomogram. The nomogram measures each regression coefficient in the logistic regression on a scale of 0–100 points. The points for each independent variable were summed, and the predicted probabilities were obtained from the total points. The predictive performance and accuracy of the nomograms were evaluated using the area under the receiver operating characteristic curve(AUROC) and calibration curves, respectively. Decision curve analysis (DCA) and clinical impact curve (CIC) were used to assess the clinical utility of the nomogram by calculating the net benefit of different threshold probability points. Statistical analysis Normally distributed continuous variable data were presented as the mean ± standard deviation (SD), and comparisons between the two groups were performed using the Student’s t-test. For non-normally distributed data, the median and interquartile range (M (Q1, Q3)) was used to present the data, and the Mann–Whitney U test was used for comparisons between the two groups. Counts were described as numbers and percentages (%), and comparisons between the two groups were performed using the χ2 test. Logistic regression was used to conduct the univariate and multivariate analyses. RFS and OS were calculated using Kaplan–Meier curves, which were compared using the log-rank test. All statistical tests were two-tailed, and P < 0.05 was considered statistically significant. The data were processed using version SPSS 26.0 (IBM, New York, USA) and version R software (version 4.3.2, R Project for Statistical Computing). Results Baseline characteristics In total, 451 patients with HCC were retrospectively included in this study. There were 363 (80.49%) M0 patients with a mean age of 51.1 years, and 88 (19.51%) M2 patients with a mean age of 52.0 years. Detailed clinical case characteristics are shown in (Table 1 ) . A baseline comparison of the clinical case characteristics between the training group (n = 317) and validation group (n = 134) showed that the differences in all metrics were not statistically significant (P > 0.05) ( Table 2 ) . Table 1 Comparison of baseline data among M0-M2 stage HCC patients Variables Levels M0 (N = 363) M2 (N = 88) Age(year) Mean ± SD 51.1 ± 11.6 52.0 ± 10.4 CRP(mg/L) Mean ± SD 7.2 ± 14.2 8.8 ± 12.7 D.dimer(µg/mL) Mean ± SD 1.1 ± 1.7 1.1 ± 1.3 C3(g/L) Mean ± SD 1.0 ± 0.3 0.9 ± 0.2 Th Cells(%) Mean ± SD 38.9 ± 8.0 38.2 ± 7.9 Treg Cells(%) Mean ± SD 21.0 ± 6.9 20.4 ± 5.6 NK Cells(%) Mean ± SD 15.1 ± 7.9 14.5 ± 8.6 B.Lymphocytes(%) Mean ± SD 12.8 ± 5.4 12.6 ± 5.1 TBIL(µmol/L) Mean ± SD 14.8 ± 8.3 15.4 ± 9.8 ALB(g/L) Mean ± SD 38.7 ± 4.1 37.7 ± 3.8 ALT(U/L) Mean ± SD 46.0 ± 50.6 37.3 ± 19.5 AST(U/L) Mean ± SD 46.5 ± 38.0 45.8 ± 22.6 Platelet(*10 9 /L) Mean ± SD 196.8 ± 75.4 203.0 ± 70.8 Neutrophil(*10 9 /L) Mean ± SD 3.6 ± 1.5 3.8 ± 1.5 Monocyte(*10 9 /L) Mean ± SD 0.5 ± 0.4 0.5 ± 0.2 Lymphocytes (*10 9 /L) Mean ± SD 1.8 ± 0.6 1.7 ± 0.6 AFP(ng/ml) ≤ 400 264 (72.7%) 49 (55.7%) >400 99 (27.3%) 39 (44.3%) HBV-DNA(IU/ml) ≤ 10 3 179 (49.3%) 43 (48.9%) >10 3 184 (50.7%) 45 (51.1%) Gender Female 61 (16.8%) 7 (8%) Male 302 (83.2%) 81 (92%) PT(s) ≤ 13 230 (63.4%) 62 (70.5%) >13 133 (36.6%) 26 (29.5%) Ascites(ml) ≤ 20 322 (88.7%) 70 (79.5%) >20 41 (11.3%) 18 (20.5%) Tumor-Size(cm) ≤ 5 220 (60.6%) 30 (34.1%) >5 143 (39.4%) 58 (65.9%) Cirrhosis No 86 (23.7%) 11 (12.5%) Yes 277 (76.3%) 77 (87.5%) Tumor-Number Single 340 (93.7%) 87 (98.9%) Multiple 23 (6.3%) 1 (1.1%) Tumor-Capsule Complete 326 (89.8%) 65 (73.9%) Incomplete 37 (10.2%) 23 (26.1%) BCLC-stage 0-Stage 29 (8%) 1 (1.1%) 1-Stage 334 (92%) 87 (98.9%) Edmondson-stage I-II 204 (56.2%) 23 (26.1%) III-IV 159 (43.8%) 65 (73.9%) Recurrence No 227 (62.5%) 40 (45.5%) Yes 136 (37.5%) 48 (54.5%) RFS(mon) Mean ± SD 16.0 ± 14.2 10.4 ± 11.2 Survival Live 304 (83.7%) 50 (56.8%) Dead 59 (16.3%) 38 (43.2%) OS(mon) Mean ± SD 38.4 ± 15.7 25.8 ± 12.7 Abbreviations: CRP, C-reactive protein;C3, Complement C3; Th Cells, CD3 + CD4 + T Cells; Treg Cells, CD3 + CD8 + T Cells; NK Cells, Natural Killer Cells; AFP, alpha fetoprotein; Edmondson-stage, Edmondson-Steiner stage;RFS, recurrence-free survival;OS, overall survival. Table 2 Comparison of baseline data between training group and validation group for HCC patients Variables Number(%)/Mean(SD) p -value Training-Group (N = 317) Validation-Group (N = 134) Age(year) 51.2 (11.4) 51.4 (11.4) 0.842 CRP(mg/L) 7.38 (13.9) 7.80 (14.3) 0.779 D.dimer(µg/mL) 1.13 (1.83) 0.89 (1.06) 0.081 C3(g/L) 0.94 (0.25) 0.95 (0.27) 0.821 Th Cells(%) 38.9 (7.93) 38.5 (8.09) 0.642 Treg Cells(%) 21.2 (6.79) 20.3 (6.31) 0.224 NK Cells(%) 14.9 (7.68) 15.4 (8.87) 0.576 B.Lymphocytes(%) 12.9 (5.50) 12.6 (5.08) 0.640 TBIL(µmol/L) 14.6 (7.88) 15.9 (10.0) 0.170 ALB(g/L) 38.5 (4.00) 38.4 (4.26) 0.732 ALT(U/L) 45.1 (52.2) 42.3 (28.1) 0.457 AST(U/L) 47.7 (40.2) 43.4 (20.5) 0.137 Platelet(*10 9 /ml) 201 (78.7) 191 (63.2) 0.169 Neutrophil(*10 9 /ml) 3.68 (1.51) 3.49 (1.37) 0.176 Monocyte(*10 9 /ml) 0.47 (0.17) 0.51 (0.57) 0.385 Lymphocytes(*10 9 /ml) 1.82 (0.64) 1.79 (0.62) 0.562 AFP(ng/ml) 0.738 ≤ 400 222 (70.0%) 91 (67.9%) >400 95 (30.0%) 43 (32.1%) HBV-DNA(IU/ml) 0.612 ≤ 10 3 159 (50.2%) 63 (47.0%) >10 3 158 (49.8%) 71 (53.0%) Gender 1.000 Female 48 (15.1%) 20 (14.9%) Male 269 (84.9%) 114 (85.1%) PT(s) 0.786 ≤ 13 207 (65.3%) 85 (63.4%) >13 110 (34.7%) 49 (36.6%) Ascites(ml) 0.535 ≤ 20 273 (86.1%) 119 (88.8%) >20 44 (13.9%) 15 (11.2%) Tumor-Size(cm) 0.382 ≤ 5 171 (53.9%) 79 (59.0%) >5 146 (46.1%) 55 (41.0%) Cirrhosis 0.502 No 65 (20.5%) 32 (23.9%) Yes 252 (79.5%) 102 (76.1%) Tumor-Number 1.000 Single 300 (94.6%) 127 (94.8%) Multiple 17 (5.36%) 7 (5.22%) Tumor-Capsule 0.838 Complete 276 (87.1%) 115 (85.8%) Incomplete 41 (12.9%) 19 (14.2%) BCLC-stage 0.138 0 17 (5.36%) 13 (9.70%) 1 300 (94.6%) 121 (90.3%) Edmondson-stage 0.416 I-II 164 (51.7%) 63 (47.0%) III-IV 153 (48.3%) 71 (53.0%) Comparison of RFS and OS M0-M2 stage HCC patients. The median follow-up time was 42.0 months (range: 2–63 months). The median RFS in the M0 group was 30.8 months, and the median OS was not achieved. The median RFS in the M2 group was 10.4 months and the median OS was 43.1 months. The hazard ratio (HR) of M0 RFS compared to M2 was 2.149 (95% confidence interval (CI): 1.542–2.996, P < 0.0001), and the HR of M0 OS compared to M2 was 4.348 (95% CI: 2.852–6.63, P < 0.0001) ( Fig. 2 ) . Screening feature variables based on machine learning The LASSO algorithm screening variables ( Fig. 3 A, 3 B ) , Boruta algorithm screening variables ( Fig. 3 C ) , XGboost algorithm screening variables ( Fig. 3 D ) , and Best_subset algorithm screening variables ( Fig. 3 E ) were used to finally determine the C3 level, tumor size, tumor membrane integrity, and Edmondson–Steiner classification as the four key variables ( Fig. 3 F ) . Construction and validation of model The four key variables screened using the training set data were included in the multivariate logistic regression analysis and prediction model based on the odds ratio (OR) value and 95% CI ( Fig. 4 A ) , and a prediction model based on the multivariate logistic regression analysis ( Fig. 4 B ) . The AUC in the training set was 0.765 (95% CI: 0.696–0.843) ( Fig. 5 A ) and that in the validation set was 0.807 (95% CI: 0.712–0.903) ( Fig. 5 B ) . The prediction results of the calibration curve were close to the actual results in both the training and validation sets, with good consistency (R 2 = 0.220 and 0.292, respectively; P = 0.864 and 0.590, respectively) ( Fig. 5 C, 5 D ) . The best cut-off probability obtained using the ROC curve was 0.185, with MVI corresponding to the optimal critical probability value and a total score of 120 points. Evaluation and application of the model Finally, the clinical utility of the nomogram model was assessed using DCA and CIC. The results showed that use of a nomogram prediction model within the 0.10–0.60 probability threshold in the training set was beneficial for patients. Furthermore, use of the nomogram prediction model within the 0.05–0.85 probability threshold in the validation set was beneficial for patients ( Fig. 6 A, 6 B ) . The CIC found that the predictive power of the nomogram model was significant ( Fig. 6 C, 6 D ) . Discussion Our results indicated that M2 stage in HCC may be associated with lower complement C3 levels, tumor size > 5 cm, incomplete tumor capsules, and late Edmondson–Steiner grade. Using these influencing factors, we developed a nomogram model that could effectively distinguish the presence of M2 in patients with HCC. The long-term prognosis of patients with HCC remains poor in the early to middle stages after treatment, which is mainly due to the high recurrence rate after primary resection caused by de novo tumors resulting from intrahepatic metastasis or underlying liver pathologies. [ 2 , 6 ] MVI, considered an important marker of aggressive behavior in HCC, can significantly affect the intrahepatic metastasis of tumor cells through the portal circulation, leading to tumor recurrence after curative surgery. [ 23 ] For MVI to occur, the cancer cells must acquire the ability to disrupt ligation/adhesion, decompose the extracellular matrix, and migrate to and utilize alternative energy sources. These changes are commonly observed during epithelial-to-mesenchymal transition (EMT), in which malignant cells de-differentiate from a polar, adherent phenotype to a mobile mesenchymal state with a more aggressive and resilient biology. [ 24 ] Increasing evidence suggests that the presence of MVI is associated with detection of the EMT phenotype in primary tumors. In MVI-positive cases, M2 grade is an obvious indicator of poor prognosis in HCC. [ 25 ] Myeloid cells and macrophages in the tumor microenvironment of HCC promote MVI and tumor progression through midkine proteins, providing favorable conditions for rapid tumor growth and invasion. [ 26 ] Unfortunately, M2 classification can only be diagnosed by histopathological examination after surgical resection; therefore, we need to look for important risk factors for MVI and its M2 grade and develop predictive models to make optimal clinical decisions, which may have important implications for postoperative recurrence detection and anti-recurrence strategies in patients with HCC. In this study, approximately 19.51% of patients with HCC (88/451) were classified as M2 grade. Our results suggesting significant differences in OS and RFS between M0 and M2 patients, with a shorter RFS and OS observed in M2 patients. These results are comparable with Yao et al. [ 9 ] who reported similar findings, which may be related to the similar baseline settings in both studies. We used LASSO, Boruta, XGboost, and Best_subset to screen the four machine learning methods by five-fold cross-validation and used the intersection to derive the most important factors related to M2. Complement C3 is a plasma protein synthesized by the liver and macrophages and is a key component of the complement system. C3 has the highest complement content and plays a central role. The complement system is an important component of the immune system and can be activated through multiple pathways. C3b binds to the surface of the pathogen and facilitates its clearance through multiple pathways. In addition, complement C3 participates in processes such as cytotoxicity, the inflammatory response, and the removal of pathogenic microorganisms to enhance the body's ability to resist infection. Moreover, complement C3 increases vascular permeability, virus neutralization, and cytolysis. C3 is an important component of the immune system and plays an important role in maintaining immune balance and resisting infection. Studies have shown that serum fatty acids, adipose tissue-derived cytokines, and gut-derived endotoxins participate in complement activation. Following complement activation, C3 interacts with different types of hepatic innate immune cells, ultimately participating in the pathogenesis of non-alcoholic fatty liver disease (NAFLD). [ 27 ] Other studies have shown that the liver synthesizes most blood proteins, except γ -globin, liver damage reduces the synthesis of C3 and C4, and that hepatitis B virus (HBV) infection may induce the formation of various antigen–antibody complexes, which activate the complement system, leading to excessive consumption of complement components including C3 and C4. It is well known that HBV infection, as well as NAFLD, are important factors in the development of HCC. [ 28 ] Other studies have shown that PIWIL1 induces HCC cells to secrete complement C3, which mediates HCC through the interaction of cells and myeloid-derived suppressor cells (MDSCs), initiating the expression of the immunosuppressive cytokine IL-10. Neutralizing IL-10 secretion reduces the immunosuppressive activity of MDSCs in the HCC microenvironment with PIWIL1 expression, thereby promoting the initiation, development, and progression of HCC. [ 29 ] However, few studies have investigated the association between complement C3 and MVI. In our study, the OR for complement C3 was < 1, indicating that complement C3 is a protective factor. Studies have shown that tumor aggressiveness increases with an increase in tumor size, and that tumor size is correlated with MVI. [ 30 ] A study from an international multicentre database showed that the incidence of MVI increased with the size of the resected HCC tumors. [ 31 , 32 ] Histological examination revealed that venous invasion-positive tumors had a strong tendency for invasion, resulting in an irregular tumor margin irregular and incomplete capsule. A radiomics-based study demonstrated that tumor size and incomplete capsules were highly reliable predictors of MVI. [ 33 ] In a prediction model, Chen et al. [ 34 ] showed that an incomplete tumor capsule is an important factor for predicting M2. Similarly, in our study, an incomplete tumor capsule was shown to be an independent risk factor for M2. The Edmondson-Steiner grade is used to evaluate the malignancy of HCC based on the morphological and histological characteristics of tumor cells, and grades I and II represent high levels of differentiation, while grades III and IV represent poorly differentiated and undifferentiated tumors, respectively. The Edmondson-Steiner grade has been shown to be associated with the prognosis of HCC. [ 35 ] MVI and Edmondson-Steiner grades are important components of the pathological grade of HCC. Several predictive models have linked the Edmondson–Steiner classification to MVI. [ 36 ] Our study found that clinical Edmondson–Steiner type was significantly associated with M2 and was an independent predictor of M2 grade, implying that a Edmondson–Steiner classification indicating poor cellular differentiation corresponds to a worse MVI grade. In a recent randomized controlled study of BCLC-A stage, a tumor diameter of 5 cm was associated with high-risk MVI. The disease-free survival rates at 1, 2, 3, and 5 years were 86.7%, 76.7%, 60.0%, and 56.3%, respectively. In contrast, in the intention-to-treat population, the survival rates in the surgery-alone group were 90.0%, 66.7%, 52.8%, and 45.7%, respectively (P = 0.448). No statistically significant difference in the survival outcome was observed between neoadjuvant radiotherapy and upfront surgery in patients with early HCC and those with a high MVI risk. [ 37 ] Studies have also shown that postoperative adjuvant transcatheter arterial chemoembolization(TACE) is beneficial for patients with middle- and advanced-stage HCC and MVI. [ 38 ] Ueshima K et al. [ 39 ] showed that hepatic arterial infusion chemotherapy(HAIC) is a potential first-line treatment option for patients with advanced HCC and MVI but without distant metastasis. In a recent study, Wang K et al. [ 40 ] showed significantly prolonged RFS with MVI (median RFS: 27.7 months vs 15.5 months; HR: 0.534, 95% CI: 0.360–0.792; P = 0.002). Many studies have shown that for MVI and high-risk, resectable HCC patients, active preoperative neoadjuvant therapy other than radiotherapy may improve RFS and OS; therefore, more aggressive treatment options should be adopted for patients with M2 grade HCC. [ 41 ] The nomogram is considered a user-friendly and practical prediction tool with high accuracy and good discriminative power and is widely used for the evaluation of prognostic or prognostic events. [ 42 ] Therefore, we developed a nomogram combining C3 level, tumor size, tumor envelope integrity, and Edmondson–Steiner grade to predict M2. According to the optimal calibration curve, the predicted probabilities were in good agreement with the actual observed values. In training set, the AUC was 0.765 (95% CI: 0.696–0.843), and in validation set, it was 0.807 (95% CI: 0.712–0.903). The optimal cut-off probability obtained using the ROC curve was 0.185, and the M2 total score was calculated according to the nomogram. The DCA results showed that the use of a nomogram to predict M2 may add more benefit than treating all patients, with or without any patients in the training and validation sets. The CIC results indicated that the predictive power of the nomogram model was significant. It is important to note that this study had several limitations. First, all patients with HCC in our study tested positive for HBsAg. Although the established nomogram showed satisfactory discriminatory performance, our prediction model may only be applicable to HCC caused by HBV and does not consider HCC caused by hepatitis C virus or non-alcoholic/alcoholic hepatitis. [ 43 ] Second, this study had a small sample size, was retrospective, and inevitably had case–selection bias. Therefore, a prospective study with a balanced population and a large sample size is required to confirm the reliability of our nomogram. Third, our study was conducted in one study unit without any validation; it is necessary to validate our results using data from multiple centres. Finally, because the model was based on clinicopathological data, the mechanisms leading to early relapse and metastasis in M2 patients remain unclear, and the accuracy may be further improved if specific markers of M2 are added. In conclusion, complement C3, tumor size > 5 cm, incomplete tumor capsule, and Edmondson–Steiner stages III–IV were identified as important predictors for the development of M2 in patients with HCC. We then combined the above variables to create a practical nomogram to make the individualized prediction of the M2 rank more objective and accurate. According to the nomogram scoring system, if patients are considered to be at a high risk of M2, more aggressive and precise treatment is tailored to reduce the potential risk of recurrence. Finally, our nomogram could improve individualized treatment designs and facilitate the selection of surveillance plans to develop more effective treatment options for patients with HCC. Abbreviations AFP: Alpha fetoprotein ALP: Alkaline phosphatase ALT ： Alanine aminotransferase AST ： Aspartate aminotransferase AUROC ： Area Under the Receiver Operating Characteristic Curve BCLC ： Barcelona Clinic Liver Cancer CI ： Confidence interval CIC ： Clinical Impact Curve CT ： Computed tomography DCA ： Decisive cure analysis HAIC ： Hepatic Artery Infusion Chemotherapy HCC ： Hepetacullar carcinoma HR ： Hazard Ratio MRI ： Magnetic resonance imaging MVI ： Microvascular invasion OR ： Odds ratio RBC ： Red blood cells ROC ： Receiver operating characteristic TACE: Transcatheter arterial chemoembolization WBC: White blood cells Declarations Data availability declaration The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Acknowledgments We would like to acknowledge the statistical assistance offered by Qingyuan Zhang，Jie Lin，Peng Zhu，Hualin Wu. Funding Declaration This work was supported by grants from the National Natural Science Foundation of China (82260345), the Key Research and Development Program of Guangxi (AB22080066), the "139" Plan for Training High-level Backbone Medical Talents in Guangxi (G202003008), the Guangxi Medical University Outstanding Young Talents Training Program , Scientific Research Project of Hunan Provincial Health Commission(No: 202204013436),and the ScientificResearch Project of shaoyang city Science and Technology Bureau(NO:2022GZ4161) Author information Guoyi Xia and Zeyan Yu have contributed equally to this work. Authors and Affiliations Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital, Nanning, 530021 Guangxi, China. Guoyi Xia,Zeyan Yu,Shaolong Lu,Xiaobo Wang&Jie Chen Department of Hepatobiliary Surgery, The Central Hospital of Shaoyang, Shaoyang,422000 Hunan,China. Guoyi Xia Department of Hepatobiliary Surgery, Guangxi Zhuang Autonomous Region People's Hospital, Nanning, 530021 Guangxi, China. Yuanquan Zhao Consent to Participate declaration Guoyi Xia，Zeyan Yu and Jie Chen contributed to the conception and design of the study and drafted the manuscript; Shaolong lu participated in data collection and literature research; and Xiaobo Wang ，Yuanquan Zhao contributed to data analysis and interpretation. All authors read and approved the final manuscript. Corresponding author Correspondence to Jie Chen. Email: [email protected] Ethics declarations Human Ethics and Consent to Participate declarations This was a retrospective and controlled study. This study adhered to the Declaration of Helsinki and was approved by the Ethics Committee of Guangxi Medical University Cancer Hospital (Clinical Trial Number:KY2024454). Conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. References Yang JD, Hainaut P, Gores GJ, Amadou A, Plymoth A, Roberts LR: A global view of hepatocellular carcinoma: trends, risk, prevention and management . Nat Rev Gastroenterol Hepatol 2019, 16 (10):589-604. Llovet JM, Kelley RK, Villanueva A, Singal AG, Pikarsky E, Roayaie S, Lencioni R, Koike K, Zucman-Rossi J, Finn RS: Hepatocellular carcinoma . Nat Rev Dis Primers 2021, 7 (1):6. Llovet JM BC, Bruix J.: Prognosis of hepatocellular carcinoma: the BCLC staging classification . Semin Liver Dis 1999, 19 (3):329-338. Reig M, Forner A, Rimola J, Ferrer-Fàbrega J, Burrel M, Garcia-Criado Á, Kelley RK, Galle PR, Mazzaferro V, Salem R et al : BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update . J Hepatol , 76 (3):681-693. Brown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM: Management of Hepatocellular Carcinoma: A Review . JAMA Surg 2023, 158 (4):410-420. Llovet JM, Pinyol R, Kelley RK, El-Khoueiry A, Reeves HL, Wang XW, Gores GJ, Villanueva A: Molecular pathogenesis and systemic therapies for hepatocellular carcinoma . Nat Cancer 2022, 3 (4):386-401. Erstad DJ, Tanabe KK: Prognostic and Therapeutic Implications of Microvascular Invasion in Hepatocellular Carcinoma . Ann Surg Oncol 2019, 26 (5):1474-1493. Cong WM, Bu H, Chen J, Dong H, Zhu YY, Feng LH, Chen J, Guideline C: Practice guidelines for the pathological diagnosis of primary liver cancer: 2015 update . World J Gastroenterol 2016, 22 (42):9279-9287. Yao LQ, Li C, Diao YK, Liang L, Jia HD, Tang SC, Zeng YY, Wu H, Wang MD, Gu LH et al : Grading severity of microscopic vascular invasion was independently associated with recurrence and survival following hepatectomy for solitary hepatocellular carcinoma . Hepatobiliary Surg Nutr 2024, 13 (1):16-28. Huo T-I, Ho S-Y, Liao J-I: Predicting post-resection early recurrence of hepatocellular carcinoma: Defining the role of microvascular invasion . Liver Int 2023, 43 (12):2826-2827. Yao L-Q, Chen Z-L, Feng Z-H, Diao Y-K, Li C, Sun H-Y, Zhong J-H, Chen T-H, Gu W-M, Zhou Y-H et al : Clinical Features of Recurrence After Hepatic Resection for Early-Stage Hepatocellular Carcinoma and Long-Term Survival Outcomes of Patients with Recurrence: A Multi-institutional Analysis . Ann Surg Oncol 2022. Wang H, Liu R, Mo H, Li R, Lian J, Liu Q, Han S: A novel nomogram predicting the early recurrence of hepatocellular carcinoma patients after R0 resection . Front Oncol 2023, 13 :1133807. Wang K, Xiang Y, Yan J, Zhu Y, Chen H, Yu H, Cheng Y, Li X, Dong W, Ji Y et al : A deep learning model with incorporation of microvascular invasion area as a factor in predicting prognosis of hepatocellular carcinoma after R0 hepatectomy . Hepatol Int 2022, 16 (5):1188-1198. Sheng X, Ji Y, Ren G-P, Lu C-L, Yun J-P, Chen L-H, Meng B, Qu L-J, Duan G-J, Sun Q et al : A standardized pathological proposal for evaluating microvascular invasion of hepatocellular carcinoma: a multicenter study by LCPGC . Hepatology International 2020, 14 (6):1034-1047. Yang P, Si A, Yang J, Cheng Z, Wang K, Li J, Xia Y, Zhang B, Pawlik TM, Lau WY et al : A wide-margin liver resection improves long-term outcomes for patients with HBV-related hepatocellular carcinoma with microvascular invasion . Surgery 2019, 165 (4):721-730. Wang H, Du P-C, Wu M-C, Cong W-M: Postoperative adjuvant transarterial chemoembolization for multinodular hepatocellular carcinoma within the Barcelona Clinic Liver Cancer early stage and microvascular invasion . Hepatobiliary Surgery and Nutrition 2018, 7 (6):418-428. Lee HW, Kim H, Park T, Park SY, Chon YE, Seo YS, Lee JS, Park JY, Kim DY, Ahn SH et al : A machine learning model for predicting hepatocellular carcinoma risk in patients with chronic hepatitis B . Liver Int 2023, 43 (8):1813-1821. Hwang YJ, Bae JS, Lee Y, Hur BY, Lee DH, Kim H: Classification of microvascular invasion of hepatocellular carcinoma: correlation with prognosis and magnetic resonance imaging . Clin Mol Hepatol 2023, 29 (3):733-746. Xu Y, Chen Y, Wu J, Pan J, Liao C, Su H: The utility of Vision Transformer in preoperatively predicting microvascular invasion status of hepatocellular carcinoma . Hpb 2023, 25 (5):533-542. Zhang J, Zeng F, Jiang S, Tang H, Zhang J: Preoperative prediction model of microvascular invasion in patients with hepatocellular carcinoma . HPB (Oxford) 2023, 25 (1):45-53. Wang H, Lu Y, Liu R, Wang L, Liu Q, Han S: A Non-Invasive Nomogram for Preoperative Prediction of Microvascular Invasion Risk in Hepatocellular Carcinoma . Front Oncol 2021, 11 :745085. Li J, Su X, Xu X, Zhao C, Liu A, Yang L, Song B, Song H, Li Z, Hao X: Preoperative prediction and risk assessment of microvascular invasion in hepatocellular carcinoma . Crit Rev Oncol Hematol 2023, 190 :104107. Rodriguez-Peralvarez M, Luong TV, Andreana L, Meyer T, Dhillon AP, Burroughs AK: A systematic review of microvascular invasion in hepatocellular carcinoma: diagnostic and prognostic variability . Ann Surg Oncol 2013, 20 (1):325-339. Mima K, Hayashi H, Kuroki H, Nakagawa S, Okabe H, Chikamoto A, Watanabe M, Beppu T, Baba H: Epithelial-mesenchymal transition expression profiles as a prognostic factor for disease-free survival in hepatocellular carcinoma: Clinical significance of transforming growth factor-beta signaling . Oncol Lett 2013, 5 (1):149-154. Zhou YM, Cao L, Li B, Zhang RX, Sui CJ, Yin ZF, Yang JM: Clinicopathological significance of ZEB1 protein in patients with hepatocellular carcinoma . Ann Surg Oncol 2012, 19 (5):1700-1706. Li K, Zhang R, Wen F, Zhao Y, Meng F, Li Q, Hao A, Yang B, Lu ZA-O, Cui YA-O et al : Single-cell dissection of the multicellular ecosystem and molecular features underlying microvascular invasion in HCC.Hepatology. 2023 Nov 16. doi: 10.1097/HEP.0000000000000673. Hepatology (1527-3350 (Electronic)). Han J, Zhang X: Complement Component C3: A Novel Biomarker Participating in the Pathogenesis of Non-alcoholic Fatty Liver Disease . Front Med (Lausanne) 2021, 8 :653293. Zhu C, Song H, Xu F, Yi W, Liu F, Liu X: Hepatitis B virus inhibits the expression of complement C3 and C4, in vitro and in vivo . Oncol Lett 2018, 15 (5):7459-7463. Wang N, Tan H-Y, Lu Y, Chan Y-T, Wang D, Guo W, Xu Y, Zhang C, Chen F, Tang G et al : PIWIL1 governs the crosstalk of cancer cell metabolism and immunosuppressive microenvironment in hepatocellular carcinoma . Signal Transduction and Targeted Therapy 2021, 6 (1):86. Zheng C, Gu X-T, Huang X-L, Wei Y-C, Chen L, Luo N-B, Lin H-S, Jin-Yuan L: Nomogram based on clinical and preoperative CT features for predicting the early recurrence of combined hepatocellular-cholangiocarcinoma: a multicenter study . Radiol Med 2023, 128 (12):1460-1471. Yang J, Zhu S, Yong J, Xia L, Qian X, Yang J, Hu X, Li Y, Wang C, Peng W et al : A Nomogram for Preoperative Estimation of Microvascular Invasion Risk in Hepatocellular Carcinoma: Single-Center Analyses With Internal Validation . Front Oncol 2021, 11 :616976. Pawlik TM, Delman KA, Vauthey JN, Nagorney DM, Ng IO, Ikai I, Yamaoka Y, Belghiti J, Lauwers GY, Poon RT et al : Tumor size predicts vascular invasion and histologic grade: Implications for selection of surgical treatment for hepatocellular carcinoma . Liver Transpl 2005, 11 (9):1086-1092. Xu X, Zhang H-L, Liu Q-P, Sun S-W, Zhang J, Zhu F-P, Yang G, Yan X, Zhang Y-D, Liu X-S: Radiomic analysis of contrast-enhanced CT predicts microvascular invasion and outcome in hepatocellular carcinoma . Journal of Hepatology 2019, 70 (6):1133-1144. Chen S, Wang C, Gu Y, Ruan R, Yu J, Wang S: Prediction of Microvascular Invasion and Its M2 Classification in Hepatocellular Carcinoma Based on Nomogram Analyses . Front Oncol 2021, 11 :774800. Kim SU, Jung Ks Fau - Lee S, Lee S Fau - Park JY, Park Jy Fau - Kim DY, Kim Dy Fau - Ahn SH, Ahn Sh Fau - Choi GH, Choi Gh Fau - Kim KS, Kim Ks Fau - Choi JS, Choi Js Fau - Han K-H, Han Kh Fau - Park YN et al : Histological subclassification of cirrhosis can predict recurrence after curative resection of hepatocellular carcinoma.Liver Int. 2014 Aug;34(7):1008-17. doi: 10.1111/liv.12475. (1478-3231 (Electronic)). Huang J, Li L, Liu FC, Tan BB, Yang Y, Jiang BG, Pan ZY: Prognostic Analysis of Single Large Hepatocellular Carcinoma Following Radical Resection: A Single-Center Study . J Hepatocell Carcinoma 2023, 10 :573-586. Wei X, Jiang Y, Feng S, Lu C, Huo L, Zhou B, Meng Y, Lau WY, Zheng Y, Cheng S: Neoadjuvant intensity modulated radiotherapy for a single and small (≤5 cm) hepatitis B virus-related hepatocellular carcinoma predicted to have high risks of microvascular invasion: a randomized clinical trial . Int J Surg 2023, 109 (10):3052-3060. Xiang C, Shen X, Zeng X, Zhang Y, Ma Z, Zhang G, Song X, Huang T, Yang J: Effect of transarterial chemoembolization as postoperative adjuvant therapy for intermediate-stage hepatocellular carcinoma with microvascular invasion: a multicenter cohort study . Int J Surg 2024, 110 (1):315-323. Ueshima K, Komemushi A, Aramaki T, Iwamoto H, Obi S, Sato Y, Tanaka T, Matsueda K, Moriguchi M, Saito H et al : Clinical Practice Guidelines for Hepatic Arterial Infusion Chemotherapy with a Port System Proposed by the Japanese Society of Interventional Radiology and Japanese Society of Implantable Port Assisted Treatment . Liver Cancer 2022, 11 (5):407-425. Wang K, Xiang YA-O, Yu HM, Cheng YQ, Liu ZH, Qin YY, Shi J, Guo WX, Lu CD, Zheng YX et al : Adjuvant sintilimab in resected high-risk hepatocellular carcinoma: a randomized, controlled, phase 2 trial.Nat Med. 2024 Jan 19. doi: 10.1038/s41591-023-02786-7 . 2024(1546-170X (Electronic)). Xu XF, Diao YK, Zeng YY, Li C, Li FW, Sun LY, Wu H, Lin KY, Yao LQ, Wang MD et al : Association of severity in the grading of microvascular invasion with long-term oncological prognosis after liver resection for early-stage hepatocellular carcinoma: a multicenter retrospective cohort study from a hepatitis B virus-endemic area . Int J Surg 2023, 109 (4):841-849. Zhou ZR, Wang WW, Li Y, Jin KR, Wang XY, Wang ZW, Chen YS, Wang SJ, Hu J, Zhang HN et al : In-depth mining of clinical data: the construction of clinical prediction model with R . Ann Transl Med 2019, 7 (23):796. Wei X, Li N, Li S, Shi J, Guo W, Zheng Y, Cheng S: Hepatitis B virus infection and active replication promote the formation of vascular invasion in hepatocellular carcinoma . BMC Cancer 2017, 17 (1):304. Additional Declarations No competing interests reported. Supplementary Files SupplementaryMaterials.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4410132","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":313865491,"identity":"0c6c3472-94ac-4018-a078-8dd06f231127","order_by":0,"name":"Guoyi Xia","email":"","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Guoyi","middleName":"","lastName":"Xia","suffix":""},{"id":313865493,"identity":"740ff6ab-62ca-460d-8a51-a1ffe6f02be0","order_by":1,"name":"Zeyan Yu","email":"","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Zeyan","middleName":"","lastName":"Yu","suffix":""},{"id":313865495,"identity":"b1df4bec-8175-4ea6-bc6a-30705125d5fb","order_by":2,"name":"Shaolong Lu","email":"","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Shaolong","middleName":"","lastName":"Lu","suffix":""},{"id":313865496,"identity":"dff88d4c-bca2-47dc-83c4-8322c1a98ffd","order_by":3,"name":"Xiaobo Wang","email":"","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital","correspondingAuthor":false,"prefix":"","firstName":"Xiaobo","middleName":"","lastName":"Wang","suffix":""},{"id":313865497,"identity":"b16099ef-52c8-41c0-80d1-df8de4614a16","order_by":4,"name":"Yuanquan Zhao","email":"","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Zhuang Autonomous Region People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yuanquan","middleName":"","lastName":"Zhao","suffix":""},{"id":313865498,"identity":"a580e85d-b97f-4de2-ab9a-880593c41c9d","order_by":5,"name":"Jie Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1ElEQVRIiWNgGAWjYBACxmYGBmYgzcMvf/jAgQ8/SNAiIzmDLfHgzB4ibQJpsTG4wWN8mIONGOXtzAc/F1Tc4ZGc3fPhMAMPgzy/2AFCDmNLlp5x5hkPv8zZDYcLLBgMZ85OIKSFx4yZt+0wj2RD7obDM3gYEgxuE9TC/42Z999hHoMDOQ8O87ARpYWHjZm3AajlRg4DsVrYjKV5jgEd1nPMABjIEoT9Yth/+OFnnprD9vzszY8/fPhhI88vTUhLAypfAr9yEJAnrGQUjIJRMApGPAAA1+1DZov7KkMAAAAASUVORK5CYII=","orcid":"","institution":"Department of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital","correspondingAuthor":true,"prefix":"","firstName":"Jie","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2024-05-13 01:53:45","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4410132/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4410132/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":59872424,"identity":"6dc8007e-8cdb-4278-97a5-2f44b6fa66b9","added_by":"auto","created_at":"2024-07-08 17:13:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":611335,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart of patients screening and grouping.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/0263dbe49cec67a821502afe.png"},{"id":59872425,"identity":"0b1fd931-bad6-4895-9b05-13a98c0c933f","added_by":"auto","created_at":"2024-07-08 17:13:02","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":247767,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan–Meier survival curves analysis of histologic M2 in HCC. A： RFS (\u003cem\u003eP\u003c/em\u003e＜0.0001) (log rank test) B：OS (\u003cem\u003eP\u003c/em\u003e＜0.0001) (log rank test)\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/2d182347e0cc55c4d7d85456.png"},{"id":59873624,"identity":"861dbd83-ce43-4ed8-8a03-4424bb93d855","added_by":"auto","created_at":"2024-07-08 17:29:02","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":472528,"visible":true,"origin":"","legend":"\u003cp\u003eSelection of demographic and clinical features using the least absolute shrinkage and selection operator (LASSO) regression model. Selection of tuning parameter (λ)in the LASSO model by five-fold cross-validation based on 1 standard error of the minimum criteria (1-SE criteria) for M2 grade(A)(B)；Selection of demographic and clinical features using the Boruta method (C)；Using SHAP method to calculate the score of key variables derived from XGboost (D) ；Using BIC values from the best subset to screen variables (E)；Taking the intersection of the selected variables and visualizing it using a Venn diagram(F)\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/ca15cededd8e3114fed24c55.png"},{"id":59872428,"identity":"6d1bd8a1-17f6-47a0-ad65-d70dc2f300cd","added_by":"auto","created_at":"2024-07-08 17:13:02","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":244879,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot of key Variables(A); Nomograms for predicting presence of M2 in HCC patients(B)\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/04d2ea4eed818be52f7c8cbe.png"},{"id":59873291,"identity":"3bd5f957-cc55-4dd5-9218-646f06003f45","added_by":"auto","created_at":"2024-07-08 17:21:02","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":105343,"visible":true,"origin":"","legend":"\u003cp\u003eThe performance of the nomogram for predicting M2 was assessed by the ROC curve in the training cohort (A,) and the validation cohort (B). Calibration curve of nomogram for predicting M2 presence in the training cohort (C) and the validation cohort (D)\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/dec738f0bf8d457bdd5a41f6.png"},{"id":59873293,"identity":"c262e941-20e5-4b34-be9f-6a02c366136b","added_by":"auto","created_at":"2024-07-08 17:21:02","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":163092,"visible":true,"origin":"","legend":"\u003cp\u003eDecision Curve Analysis（DCA） in the training cohort (A) and the validation cohort (B) The Y-axis showed the net benefit. The X-axis represented the threshold probability. The None-line (black line) represented the net benefit when none of the participants were considered to have MVI. The All-line (light gray line) represented the net benefit when all participants were considered to have MVI. The red line represented the net benefit of the nomogram at different threshold probabilities. The area between the “None-line” (black line) and “All-line” (light gray line) in the model curve indicated the clinical utility of the model. Clinical Impact Curve （CIC） in the training cohort (C) and the validation cohort (D). At different threshold probabilities within a given population, the number of high-risk patients and the number of high-risk patients with MVI were shown.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/a6bb117239b430dff36ef6bb.png"},{"id":66247421,"identity":"09132860-f08c-474c-ada5-01f89ed391cb","added_by":"auto","created_at":"2024-10-09 08:08:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3929241,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/52530d9c-7868-4557-aede-2a77cebe0d8c.pdf"},{"id":59872423,"identity":"0d1fe785-6a8c-4479-a4fb-597c82186eb7","added_by":"auto","created_at":"2024-07-08 17:13:02","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":14150,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-4410132/v1/14a0698ba5eed52ddc466abc.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Prediction model of M2 with early-stage hepatocellular carcinoma based on multiple machine learning","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHepatocellular carcinoma (HCC) is the most common liver malignancy, ranking sixth and third in terms of global morbidity and mortality, respectively.\u003csup\u003e[\u003c/sup\u003e1\u003csup\u003e,\u003c/sup\u003e 2\u003csup\u003e]\u003c/sup\u003e It is well known that HCC in stage Barcelona Clinic Liver Cancer-A(BCLC-A) is defined as early-stage.\u003csup\u003e[\u003c/sup\u003e3\u003csup\u003e]\u003c/sup\u003e In the 2022 update of the BCLC prognosis prediction and treatment strategy, stage BCLC 0 (normal liver function, single tumor, 2 cm diameter, no vascular invasion or extrahepatic metastasis) was defined as very early-stage. The first-line treatment strategy for HCC in stage BCLC 0/A is surgical resection, ablation and liver transplantation.\u003csup\u003e[\u003c/sup\u003e4\u003csup\u003e]\u003c/sup\u003e Despite some improvements in the diagnosis and treatment of HCC, the proportion of patients with early-stage HCC is \u0026lt;\u0026thinsp;30%;together with the high invasion and heterogeneity of HCC, the overall recurrence rate at 3 years remains\u003c/p\u003e \u003cp\u003eas high as 30\u0026ndash;50%, leading to a poor prognosis.\u003csup\u003e[\u003c/sup\u003e5\u003csup\u003e,\u003c/sup\u003e 6\u003csup\u003e]\u003c/sup\u003e Microvascular invasion (MVI) mainly refers to the cancer cell nests that can only be observed under a microscope in the lumen of the tiny vessels lined with endothelial cells, mostly in the tumor envelope and adjacent liver tissue.\u003csup\u003e[\u003c/sup\u003e7\u003csup\u003e]\u003c/sup\u003e According to the quantity and distribution of cancer cell nests, MVI can be categorized into three levels: M0 level, where no MVI is detected; M1 level, characterized by \u0026le;\u0026thinsp;5 MVIs occurring in the peri-cancerous liver tissue region (\u0026le;\u0026thinsp;1cm); and M2 level, defined by \u0026gt;\u0026thinsp;5 MVIs or MVIs occurring in the distant peri-cancerous liver tissue region (\u0026gt;\u0026thinsp;1cm). \u003csup\u003e[\u003c/sup\u003e8\u003csup\u003e,\u003c/sup\u003e 9\u003csup\u003e]\u003c/sup\u003e MVI is associated with various factors, including the invasive ability of tumor cells, the secretion of angiogenic factors, and immune escape. Tumor cells degrade the vascular basement membrane via secreted proteases, which then invade the vascular lumen to form cancer cell nests. Simultaneously, tumor cells secrete angiogenic factors, promote the formation of neovascularization, and provide conditions for tumor growth and metastasis. Numerous studies have confirmed that MVI is an important factor affecting the recurrence and survival of patients with liver cancer, and personalized treatment based on MVI is of great significance for improve the surgical outcomes of patients with HCC.[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] Studies have shown that patients with M2 have a poorer prognosis after radical resection than those with M0 and M1 \u003csup\u003e[\u003c/sup\u003e8\u003csup\u003e,\u003c/sup\u003e 12\u003csup\u003e]\u003c/sup\u003e, and M2 is a high-risk factor for postoperative residual cancer recurrence and intrahepatic metastasis \u003csup\u003e[\u003c/sup\u003e13\u003csup\u003e,\u003c/sup\u003e 14\u003csup\u003e]\u003c/sup\u003e.Therefore, we should focus not only on the presence or absence of MVI, but also on the M2. If preoperative patients requiring liver resection are judged to have a higher risk of M2, it is recommended to expand the margin or advance the intervention for transformation treatment to prevent early postoperative recurrence and metastasis thus improving patient prognosis.\u003csup\u003e[\u003c/sup\u003e15\u003csup\u003e,\u003c/sup\u003e 16\u003csup\u003e]\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eMachine learning has been widely used in research on HCC. Machine learning is more accurate, more stable, and more fully high-throughput and multi-dimensional clinical data mining, so it can improve the performance of prediction models.\u003csup\u003e[\u003c/sup\u003e17\u003csup\u003e]\u003c/sup\u003e In recent years, with the development of radiomics, an increasing number of radiomic features have been considered to predict MVI and judge the prognosis of HCC.\u003csup\u003e[\u003c/sup\u003e18\u003csup\u003e,\u003c/sup\u003e 19\u003csup\u003e]\u003c/sup\u003eAlthough many prediction models have been used to predict MVI, there is a large heterogeneity in the criteria for HCC inclusion, and the risk prediction studies on M2 remain scarce.\u003csup\u003e[\u003c/sup\u003e20\u0026ndash;22\u003csup\u003e]\u003c/sup\u003e In this study, we screened clinical indicators of HCC through machine learning, established a prediction model for early-stage HCC patients with M2, and drew a nomogram for stratified intervention to reduce postoperative recurrence, which could be beneficial for clinicians.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003ePatients and data collection\u003c/h2\u003e \u003cp\u003eThe clinical data of 888 patients diagnosed with HCC from December 2012 to December 2018 at the Affiliated Cancer Hospital of Guangxi Medical University were collected. The inclusion criteria were as follows: 1) BCLC 0/A stage, 2) hepatitis B surface antigen (HBsAg) (+), and 3) MVI grade M0 and M2.The exclusion criteria were as follows: 1) imaging and blood index examination 1 week before surgery, 2) non-R0 resection, 3) preoperative surgery, ablation, radiofrequency, or neoadjuvant therapy, 4) MVI status not assessed, and 5) incomplete clinical data. Finally, the clinical data of 451 patients were included, and the number of seeds was set using R software. The patients were divided into training and validation groups at a 7:3 ratio. The training group was used to build the prediction model, and the validation group was used for internal validation \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e\u003cb\u003e)\u003c/b\u003e. This was a retrospective and controlled study. This study adhered to the Declaration of Helsinki and was approved by the Ethics Committee of Guangxi Medical University Cancer Hospital(Batch No:KY2024454).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eData collection and processing\u003c/h2\u003e \u003cp\u003eBasic patient information was collected, including data on gender, age, C-reactive protein, D-dimer, complement C3, T helper cells, inhibitory T cells, natural killer cells, B-lymphocytes, total-bilirubin, total-protein, glutamic-pyruvic transaminase, glutamic-oxalacetic transaminase, alpha fetoprotein (AFP), platelet count, absolute neutrophil values, absolute monocyte values, absolute lymphocyte values, prothrombin time, cirrhosis, ascites, tumor size, number of tumors, tumor capsule, BCLC-stages, Edmondson-Steiner classification, recurrence, recurrence-free survival(RFS), survival and overall survival(OS). Postoperative tissue specimens were further examined pathologically to confirm the presence of MVI and subsequently graded. Liver cirrhosis was defined as B-ultrasound liver transient elastography with an elasticity value (or hardness value)\u0026thinsp;\u0026gt;\u0026thinsp;7.1 kPa. RFS as defined as the time interval from the start of surgical treatment to the first recurrence of the tumor. OS was defined as the time from the diagnosis of HCC to death from any cause.\u003c/p\u003e \u003cp\u003e After tumor resection, all patients were followed-up regularly according to clinical guidelines. Follow-ups included laboratory tests (serum AFP level and liver function tests) and imaging studies (ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI)) once every 3 months after surgery. The diagnosis of HCC recurrence was based on two or more examination methods, including ultrasonography, CT, MRI, and hepatic arteriography, with elevated serum AFP levels. The patients with confirmed recurrent HCC underwent further evaluation by a multidisciplinary team. The last follow-up was conducted on 31 December 2020.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eScreening for indicators related to M2\u003c/h2\u003e \u003cp\u003eThe Least Absolute Shrinkage and Selection Operator (LASSO) is a regularization algorithm used for linear regression and related problems. It achieves a sparseness of the model coefficients by adding L1 regular terms to the loss function, thus prompting the model to select relatively few features. The Boruta algorithm is a machine learning algorithm used for feature selection, which is especially suitable for the random forest model. The algorithm determines the threshold by ranking the feature importance. This threshold will be used to distinguish which features are \"important\" and which are \"not\". XGBoost (eXtreme Gradient Boosting) is a machine learning algorithm for gradient-boosting trees that has achieved significant improvements in prediction performance and computing efficiency. This provides a means for assessing the importance of these features. The relative importance of each feature can be obtained by measuring the number of splittings constructing the tree or the contribution of the feature to the target variable during splitting. The Best subset algorithm is a feature-selection method designed to select the best subset from a given set of features to build a linear regression model. The algorithm evaluates the performance of each subset by exhaustively imposing all possible feature combinations and selecting the subset with the most influence on the target.\u003c/p\u003e \u003cp\u003eSHAP (SHapley Additive exPlanations) is a model used to interpret machine learning model prediction results. The SHAP values are based on the Shapley-value concept in game theory, which provides a fair, consistent, and efficient way to assign contributing values to each feature, thereby explaining how the model's output for each sample is formed. For each feature, the SHAP value considers all possible subset combinations that the feature may have formed with other features and calculates the average contribution of the feature to the output in these combinations.\u003c/p\u003e \u003cp\u003eThe clinical characteristic factors derived from the four machine algorithms were intersected, and a Venn diagram was created. Ultimately, the analysis identified the definitive predictive factors associated with M2. These factors were used to construct the nomogram. The nomogram measures each regression coefficient in the logistic regression on a scale of 0\u0026ndash;100 points. The points for each independent variable were summed, and the predicted probabilities were obtained from the total points. The predictive performance and accuracy of the nomograms were evaluated using the area under the receiver operating characteristic curve(AUROC) and calibration curves, respectively. Decision curve analysis (DCA) and clinical impact curve (CIC) were used to assess the clinical utility of the nomogram by calculating the net benefit of different threshold probability points.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eNormally distributed continuous variable data were presented as the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD), and comparisons between the two groups were performed using the Student\u0026rsquo;s t-test. For non-normally distributed data, the median and interquartile range (M (Q1, Q3)) was used to present the data, and the Mann\u0026ndash;Whitney U test was used for comparisons between the two groups. Counts were described as numbers and percentages (%), and comparisons between the two groups were performed using the χ2 test. Logistic regression was used to conduct the univariate and multivariate analyses. RFS and OS were calculated using Kaplan\u0026ndash;Meier curves, which were compared using the log-rank test. All statistical tests were two-tailed, and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant. The data were processed using version SPSS 26.0 (IBM, New York, USA) and version R software (version 4.3.2, R Project for Statistical Computing).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\"\u003e\n \u003ch2\u003eBaseline characteristics\u003c/h2\u003e\n \u003cp\u003eIn total, 451 patients with HCC were retrospectively included in this study. There were 363 (80.49%) M0 patients with a mean age of 51.1 years, and 88 (19.51%) M2 patients with a mean age of 52.0 years. Detailed clinical case characteristics are shown in (Table\u0026nbsp;\u003cspan\u003e1\u003c/span\u003e\u003cstrong\u003e)\u003c/strong\u003e. A baseline comparison of the clinical case characteristics between the training group (n\u0026thinsp;=\u0026thinsp;317) and validation group (n\u0026thinsp;=\u0026thinsp;134) showed that the differences in all metrics were not statistically significant (P\u0026thinsp;\u0026gt;\u0026thinsp;0.05) \u003cstrong\u003e(\u003c/strong\u003eTable\u0026nbsp;\u003cspan\u003e2\u003c/span\u003e\u003cstrong\u003e)\u003c/strong\u003e.\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 1\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eComparison of baseline data among M0-M2 stage HCC patients\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLevels\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eM0 (N\u0026thinsp;=\u0026thinsp;363)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eM2 (N\u0026thinsp;=\u0026thinsp;88)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge(year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.1\u0026thinsp;\u0026plusmn;\u0026thinsp;11.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e52.0\u0026thinsp;\u0026plusmn;\u0026thinsp;10.4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCRP(mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.2\u0026thinsp;\u0026plusmn;\u0026thinsp;14.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.8\u0026thinsp;\u0026plusmn;\u0026thinsp;12.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eD.dimer(\u0026micro;g/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1\u0026thinsp;\u0026plusmn;\u0026thinsp;1.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1\u0026thinsp;\u0026plusmn;\u0026thinsp;1.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eC3(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.0\u0026thinsp;\u0026plusmn;\u0026thinsp;0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.9\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTh Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.9\u0026thinsp;\u0026plusmn;\u0026thinsp;8.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.2\u0026thinsp;\u0026plusmn;\u0026thinsp;7.9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTreg Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.0\u0026thinsp;\u0026plusmn;\u0026thinsp;6.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20.4\u0026thinsp;\u0026plusmn;\u0026thinsp;5.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNK Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.1\u0026thinsp;\u0026plusmn;\u0026thinsp;7.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.5\u0026thinsp;\u0026plusmn;\u0026thinsp;8.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB.Lymphocytes(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.8\u0026thinsp;\u0026plusmn;\u0026thinsp;5.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.6\u0026thinsp;\u0026plusmn;\u0026thinsp;5.1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTBIL(\u0026micro;mol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.8\u0026thinsp;\u0026plusmn;\u0026thinsp;8.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.4\u0026thinsp;\u0026plusmn;\u0026thinsp;9.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALB(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.7\u0026thinsp;\u0026plusmn;\u0026thinsp;4.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37.7\u0026thinsp;\u0026plusmn;\u0026thinsp;3.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALT(U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e46.0\u0026thinsp;\u0026plusmn;\u0026thinsp;50.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37.3\u0026thinsp;\u0026plusmn;\u0026thinsp;19.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAST(U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e46.5\u0026thinsp;\u0026plusmn;\u0026thinsp;38.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45.8\u0026thinsp;\u0026plusmn;\u0026thinsp;22.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePlatelet(*10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e196.8\u0026thinsp;\u0026plusmn;\u0026thinsp;75.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e203.0\u0026thinsp;\u0026plusmn;\u0026thinsp;70.8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNeutrophil(*10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.6\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.8\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMonocyte(*10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.5\u0026thinsp;\u0026plusmn;\u0026thinsp;0.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.5\u0026thinsp;\u0026plusmn;\u0026thinsp;0.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLymphocytes (*10\u003csup\u003e9\u003c/sup\u003e/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.8\u0026thinsp;\u0026plusmn;\u0026thinsp;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.7\u0026thinsp;\u0026plusmn;\u0026thinsp;0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAFP(ng/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e264 (72.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e49 (55.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e99 (27.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e39 (44.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHBV-DNA(IU/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;10\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e179 (49.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (48.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;10\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e184 (50.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45 (51.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e61 (16.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e302 (83.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e81 (92%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePT(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e230 (63.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e62 (70.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e133 (36.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 (29.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAscites(ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e322 (88.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e70 (79.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41 (11.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (20.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Size(cm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e220 (60.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30 (34.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e143 (39.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e58 (65.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCirrhosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86 (23.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11 (12.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e277 (76.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e77 (87.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Number\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSingle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e340 (93.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e87 (98.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMultiple\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (6.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1 (1.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Capsule\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eComplete\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e326 (89.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e65 (73.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIncomplete\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37 (10.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (26.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBCLC-stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0-Stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29 (8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1 (1.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1-Stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e334 (92%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e87 (98.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEdmondson-stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eI-II\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e204 (56.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23 (26.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIII-IV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e159 (43.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e65 (73.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRecurrence\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e227 (62.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e40 (45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e136 (37.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e48 (54.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRFS(mon)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16.0\u0026thinsp;\u0026plusmn;\u0026thinsp;14.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10.4\u0026thinsp;\u0026plusmn;\u0026thinsp;11.2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSurvival\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLive\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e304 (83.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e50 (56.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDead\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e59 (16.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38 (43.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOS(mon)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.4\u0026thinsp;\u0026plusmn;\u0026thinsp;15.7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25.8\u0026thinsp;\u0026plusmn;\u0026thinsp;12.7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"4\"\u003eAbbreviations: CRP, C-reactive protein;C3, Complement C3; Th Cells, CD3\u0026thinsp;+\u0026thinsp;CD4\u0026thinsp;+\u0026thinsp;T Cells; Treg Cells, CD3\u0026thinsp;+\u0026thinsp;CD8\u0026thinsp;+\u0026thinsp;T Cells; NK Cells, Natural Killer Cells; AFP, alpha fetoprotein; Edmondson-stage, Edmondson-Steiner stage;RFS, recurrence-free survival;OS, overall survival.\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 2\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eComparison of baseline data between training group and validation group for HCC patients\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eNumber(%)/Mean(SD)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003ep\u003c/em\u003e-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTraining-Group\u003c/p\u003e\n \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;317)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eValidation-Group\u003c/p\u003e\n \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;134)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge(year)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.2 (11.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.4 (11.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCRP(mg/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.38 (13.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.80 (14.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.779\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eD.dimer(\u0026micro;g/mL)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.13 (1.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.89 (1.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.081\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eC3(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.94 (0.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.95 (0.27)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.821\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTh Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.9 (7.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.5 (8.09)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.642\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTreg Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.2 (6.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20.3 (6.31)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.224\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNK Cells(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.9 (7.68)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.4 (8.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.576\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB.Lymphocytes(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.9 (5.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.6 (5.08)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.640\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTBIL(\u0026micro;mol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.6 (7.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15.9 (10.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.170\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALB(g/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.5 (4.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e38.4 (4.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.732\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eALT(U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45.1 (52.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42.3 (28.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.457\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAST(U/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e47.7 (40.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43.4 (20.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.137\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePlatelet(*10\u003csup\u003e9\u003c/sup\u003e/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e201 (78.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e191 (63.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.169\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNeutrophil(*10\u003csup\u003e9\u003c/sup\u003e/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.68 (1.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.49 (1.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.176\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMonocyte(*10\u003csup\u003e9\u003c/sup\u003e/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.47 (0.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.51 (0.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.385\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLymphocytes(*10\u003csup\u003e9\u003c/sup\u003e/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.82 (0.64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.79 (0.62)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.562\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAFP(ng/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.738\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e222 (70.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e91 (67.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95 (30.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43 (32.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHBV-DNA(IU/ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le; 10\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e159 (50.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e63 (47.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;10\u003csup\u003e3\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e158 (49.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e71 (53.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGender\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e48 (15.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 (14.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e269 (84.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e114 (85.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePT(s)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.786\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le; 13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e207 (65.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e85 (63.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e110 (34.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e49 (36.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAscites(ml)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.535\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le; 20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e273 (86.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e119 (88.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44 (13.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e15 (11.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Size(cm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.382\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le; 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e171 (53.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e79 (59.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026gt;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e146 (46.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55 (41.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCirrhosis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.502\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e65 (20.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 (23.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e252 (79.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e102 (76.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Number\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSingle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e300 (94.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e127 (94.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMultiple\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (5.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 (5.22%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTumor-Capsule\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.838\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eComplete\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e276 (87.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e115 (85.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIncomplete\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41 (12.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 (14.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBCLC-stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.138\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (5.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13 (9.70%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e300 (94.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e121 (90.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEdmondson-stage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.416\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eI-II\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e164 (51.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e63 (47.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIII-IV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e153 (48.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e71 (53.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cstrong\u003eComparison of RFS and OS M0-M2 stage HCC patients.\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003eThe median follow-up time was 42.0 months (range: 2\u0026ndash;63 months). The median RFS in the M0 group was 30.8 months, and the median OS was not achieved. The median RFS in the M2 group was 10.4 months and the median OS was 43.1 months. The hazard ratio (HR) of M0 RFS compared to M2 was 2.149 (95% confidence interval (CI): 1.542\u0026ndash;2.996, P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001), and the HR of M0 OS compared to M2 was 4.348 (95% CI: 2.852\u0026ndash;6.63, P\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e2\u003c/span\u003e\u003cstrong\u003e)\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\"\u003e\n \u003ch2\u003eScreening feature variables based on machine learning\u003c/h2\u003e\n \u003cp\u003eThe LASSO algorithm screening variables \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e3\u003c/span\u003eA, \u003cspan\u003e3\u003c/span\u003eB\u003cstrong\u003e)\u003c/strong\u003e, Boruta algorithm screening variables \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e3\u003c/span\u003eC\u003cstrong\u003e)\u003c/strong\u003e, XGboost algorithm screening variables \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e3\u003c/span\u003eD\u003cstrong\u003e)\u003c/strong\u003e, and Best_subset algorithm screening variables \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e3\u003c/span\u003eE\u003cstrong\u003e)\u003c/strong\u003e were used to finally determine the C3 level, tumor size, tumor membrane integrity, and Edmondson\u0026ndash;Steiner classification as the four key variables \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e3\u003c/span\u003eF\u003cstrong\u003e)\u003c/strong\u003e.\u003c/p\u003e\n \u003cdiv id=\"Sec10\"\u003e\n \u003ch2\u003eConstruction and validation of model\u003c/h2\u003e\n \u003cp\u003eThe four key variables screened using the training set data were included in the multivariate logistic regression analysis and prediction model based on the odds ratio (OR) value and 95% CI \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e4\u003c/span\u003eA\u003cstrong\u003e)\u003c/strong\u003e, and a prediction model based on the multivariate logistic regression analysis \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e4\u003c/span\u003eB\u003cstrong\u003e)\u003c/strong\u003e. The AUC in the training set was 0.765 (95% CI: 0.696\u0026ndash;0.843) \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e5\u003c/span\u003eA\u003cstrong\u003e)\u003c/strong\u003e and that in the validation set was 0.807 (95% CI: 0.712\u0026ndash;0.903) \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e5\u003c/span\u003eB\u003cstrong\u003e)\u003c/strong\u003e. The prediction results of the calibration curve were close to the actual results in both the training and validation sets, with good consistency (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.220 and 0.292, respectively; P\u0026thinsp;=\u0026thinsp;0.864 and 0.590, respectively) \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e5\u003c/span\u003eC, \u003cspan\u003e5\u003c/span\u003eD\u003cstrong\u003e)\u003c/strong\u003e. The best cut-off probability obtained using the ROC curve was 0.185, with MVI corresponding to the optimal critical probability value and a total score of 120 points.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec11\"\u003e\n \u003ch2\u003eEvaluation and application of the model\u003c/h2\u003e\n \u003cp\u003eFinally, the clinical utility of the nomogram model was assessed using DCA and CIC. The results showed that use of a nomogram prediction model within the 0.10\u0026ndash;0.60 probability threshold in the training set was beneficial for patients. Furthermore, use of the nomogram prediction model within the 0.05\u0026ndash;0.85 probability threshold in the validation set was beneficial for patients \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e6\u003c/span\u003eA, \u003cspan\u003e6\u003c/span\u003eB\u003cstrong\u003e)\u003c/strong\u003e. The CIC found that the predictive power of the nomogram model was significant \u003cstrong\u003e(\u003c/strong\u003eFig.\u0026nbsp;\u003cspan\u003e6\u003c/span\u003eC, \u003cspan\u003e6\u003c/span\u003eD\u003cstrong\u003e)\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur results indicated that M2 stage in HCC may be associated with lower complement C3 levels, tumor size\u0026thinsp;\u0026gt;\u0026thinsp;5 cm, incomplete tumor capsules, and late Edmondson\u0026ndash;Steiner grade. Using these influencing factors, we developed a nomogram model that could effectively distinguish the presence of M2 in patients with HCC.\u003c/p\u003e \u003cp\u003eThe long-term prognosis of patients with HCC remains poor in the early to middle stages after treatment, which is mainly due to the high recurrence rate after primary resection caused by de novo tumors resulting from intrahepatic metastasis or underlying liver pathologies.\u003csup\u003e[\u003c/sup\u003e2\u003csup\u003e,\u003c/sup\u003e 6\u003csup\u003e]\u003c/sup\u003e MVI, considered an important marker of aggressive behavior in HCC, can significantly affect the intrahepatic metastasis of tumor cells through the portal circulation, leading to tumor recurrence after curative surgery.\u003c/p\u003e \u003cp\u003e \u003csup\u003e[\u003c/sup\u003e23\u003csup\u003e]\u003c/sup\u003e For MVI to occur, the cancer cells must acquire the ability to disrupt ligation/adhesion, decompose the extracellular matrix, and migrate to and utilize alternative energy sources. These changes are commonly observed during epithelial-to-mesenchymal transition (EMT), in which malignant cells de-differentiate from a polar, adherent phenotype to a mobile mesenchymal state with a more aggressive and resilient biology.\u003csup\u003e[\u003c/sup\u003e24\u003csup\u003e]\u003c/sup\u003e Increasing evidence suggests that the presence of MVI is associated with detection of the EMT phenotype in primary tumors. In MVI-positive cases, M2 grade is an obvious indicator of poor prognosis in HCC.\u003csup\u003e[\u003c/sup\u003e25\u003csup\u003e]\u003c/sup\u003e Myeloid cells and macrophages in the tumor microenvironment of HCC promote MVI and tumor progression through midkine proteins, providing favorable conditions for rapid tumor growth and invasion.\u003csup\u003e[\u003c/sup\u003e26\u003csup\u003e]\u003c/sup\u003e Unfortunately, M2 classification can only be diagnosed by histopathological examination after surgical resection; therefore, we need to look for important risk factors for MVI and its M2 grade and develop predictive models to make optimal clinical decisions, which may have important implications for postoperative recurrence detection and anti-recurrence strategies in patients with HCC.\u003c/p\u003e \u003cp\u003eIn this study, approximately 19.51% of patients with HCC (88/451) were classified as M2 grade. Our results suggesting significant differences in OS and RFS between M0 and M2 patients, with a shorter RFS and OS observed in M2 patients. These results are comparable with Yao et al. \u003csup\u003e[\u003c/sup\u003e9\u003csup\u003e]\u003c/sup\u003e who reported similar findings, which may be related to the similar baseline settings in both studies. We used LASSO, Boruta, XGboost, and Best_subset to screen the four machine learning methods by five-fold cross-validation and used the intersection to derive the most important factors related to M2.\u003c/p\u003e \u003cp\u003eComplement C3 is a plasma protein synthesized by the liver and macrophages and is a key component of the complement system. C3 has the highest complement content and plays a central role. The complement system is an important component of the immune system and can be activated through multiple pathways. C3b binds to the surface of the pathogen and facilitates its clearance through multiple pathways. In addition, complement C3 participates in processes such as cytotoxicity, the inflammatory response, and the removal of pathogenic microorganisms to enhance the body's ability to resist infection. Moreover, complement C3 increases vascular permeability, virus neutralization, and cytolysis. C3 is an important component of the immune system and plays an important role in maintaining immune balance and resisting infection. Studies have shown that serum fatty acids, adipose tissue-derived cytokines, and gut-derived endotoxins participate in complement activation. Following complement activation, C3 interacts with different types of hepatic innate immune cells, ultimately participating in the pathogenesis of non-alcoholic fatty liver disease (NAFLD).\u003csup\u003e[\u003c/sup\u003e27\u003csup\u003e]\u003c/sup\u003e Other studies have shown that the liver synthesizes most blood proteins, except γ -globin, liver damage reduces the synthesis of C3 and C4, and that hepatitis B virus (HBV) infection may induce the formation of various antigen\u0026ndash;antibody complexes, which activate the complement system, leading to excessive consumption of complement components including C3 and C4. It is well known that HBV infection, as well as NAFLD, are important factors in the development of HCC.\u003csup\u003e[\u003c/sup\u003e28\u003csup\u003e]\u003c/sup\u003e Other studies have shown that PIWIL1 induces HCC cells to secrete complement C3, which mediates HCC through the interaction of cells and myeloid-derived suppressor cells (MDSCs), initiating the expression of the immunosuppressive cytokine IL-10. Neutralizing IL-10 secretion reduces the immunosuppressive activity of MDSCs in the HCC microenvironment with PIWIL1 expression, thereby promoting the initiation, development, and progression of HCC.\u003csup\u003e[\u003c/sup\u003e29\u003csup\u003e]\u003c/sup\u003e However, few studies have investigated the association between complement C3 and MVI. In our study, the OR for complement C3 was \u0026lt;\u0026thinsp;1, indicating that complement C3 is a protective factor.\u003c/p\u003e \u003cp\u003eStudies have shown that tumor aggressiveness increases with an increase in tumor size, and that tumor size is correlated with MVI.\u003csup\u003e[\u003c/sup\u003e30\u003csup\u003e]\u003c/sup\u003e A study from an international multicentre database showed that the incidence of MVI increased with the size of the resected HCC tumors.\u003csup\u003e[\u003c/sup\u003e31\u003csup\u003e,\u003c/sup\u003e 32\u003csup\u003e]\u003c/sup\u003e Histological examination revealed that venous invasion-positive tumors had a strong tendency for invasion, resulting in an irregular tumor margin irregular and incomplete capsule. A radiomics-based study demonstrated that tumor size and incomplete capsules were highly reliable predictors of MVI.\u003csup\u003e[\u003c/sup\u003e33\u003csup\u003e]\u003c/sup\u003e In a prediction model, Chen et al.\u003csup\u003e[\u003c/sup\u003e34\u003csup\u003e]\u003c/sup\u003e showed that an incomplete tumor capsule is an important factor for predicting M2. Similarly, in our study, an incomplete tumor capsule was shown to be an independent risk factor for M2.\u003c/p\u003e \u003cp\u003eThe Edmondson-Steiner grade is used to evaluate the malignancy of HCC based on the morphological and histological characteristics of tumor cells, and grades I and II represent high levels of differentiation, while grades III and IV represent poorly differentiated and undifferentiated tumors, respectively. The Edmondson-Steiner grade has been shown to be associated with the prognosis of HCC.\u003csup\u003e[\u003c/sup\u003e35\u003csup\u003e]\u003c/sup\u003e MVI and Edmondson-Steiner grades are important components of the pathological grade of HCC. Several predictive models have linked the Edmondson\u0026ndash;Steiner classification to MVI.\u003csup\u003e[\u003c/sup\u003e36\u003csup\u003e]\u003c/sup\u003e Our study found that clinical Edmondson\u0026ndash;Steiner type was significantly associated with M2 and was an independent predictor of M2 grade, implying that a Edmondson\u0026ndash;Steiner classification indicating poor cellular differentiation corresponds to a worse MVI grade.\u003c/p\u003e \u003cp\u003eIn a recent randomized controlled study of BCLC-A stage, a tumor diameter of 5 cm was associated with high-risk MVI. The disease-free survival rates at 1, 2, 3, and 5 years were 86.7%, 76.7%, 60.0%, and 56.3%, respectively. In contrast, in the intention-to-treat population, the survival rates in the surgery-alone group were 90.0%, 66.7%, 52.8%, and 45.7%, respectively (P\u0026thinsp;=\u0026thinsp;0.448). No statistically significant difference in the survival outcome was observed between neoadjuvant radiotherapy and upfront surgery in patients with early HCC and those with a high MVI risk.\u003csup\u003e[\u003c/sup\u003e37\u003csup\u003e]\u003c/sup\u003e Studies have also shown that postoperative adjuvant transcatheter arterial chemoembolization(TACE) is beneficial for patients with middle- and advanced-stage HCC and MVI.\u003csup\u003e[\u003c/sup\u003e38\u003csup\u003e]\u003c/sup\u003e Ueshima K et al. \u003csup\u003e[\u003c/sup\u003e39\u003csup\u003e]\u003c/sup\u003e showed that hepatic arterial infusion chemotherapy(HAIC) is a potential first-line treatment option for patients with advanced HCC and MVI but without distant metastasis. In a recent study, Wang K et al. \u003csup\u003e[\u003c/sup\u003e40\u003csup\u003e]\u003c/sup\u003e showed significantly prolonged RFS with MVI (median RFS: 27.7 months vs 15.5 months; HR: 0.534, 95% CI: 0.360\u0026ndash;0.792; P\u0026thinsp;=\u0026thinsp;0.002). Many studies have shown that for MVI and high-risk, resectable HCC patients, active preoperative neoadjuvant therapy other than radiotherapy may improve RFS and OS; therefore, more aggressive treatment options should be adopted for patients with M2 grade HCC.\u003csup\u003e[\u003c/sup\u003e41\u003csup\u003e]\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eThe nomogram is considered a user-friendly and practical prediction tool with high accuracy and good discriminative power and is widely used for the evaluation of prognostic or prognostic events.\u003csup\u003e[\u003c/sup\u003e42\u003csup\u003e]\u003c/sup\u003e Therefore, we developed a nomogram combining C3 level, tumor size, tumor envelope integrity, and Edmondson\u0026ndash;Steiner grade to predict M2. According to the optimal calibration curve, the predicted probabilities were in good agreement with the actual observed values. In training set, the AUC was 0.765 (95% CI: 0.696\u0026ndash;0.843), and in validation set, it was 0.807 (95% CI: 0.712\u0026ndash;0.903). The optimal cut-off probability obtained using the ROC curve was 0.185, and the M2 total score was calculated according to the nomogram. The DCA results showed that the use of a nomogram to predict M2 may add more benefit than treating all patients, with or without any patients in the training and validation sets. The CIC results indicated that the predictive power of the nomogram model was significant.\u003c/p\u003e \u003cp\u003eIt is important to note that this study had several limitations. First, all patients with HCC in our study tested positive for HBsAg. Although the established nomogram showed satisfactory discriminatory performance, our prediction model may only be applicable to HCC caused by HBV and does not consider HCC caused by hepatitis C virus or non-alcoholic/alcoholic hepatitis.\u003csup\u003e[\u003c/sup\u003e43\u003csup\u003e]\u003c/sup\u003e Second, this study had a small sample size, was retrospective, and inevitably had case\u0026ndash;selection bias. Therefore, a prospective study with a balanced population and a large sample size is required to confirm the reliability of our nomogram. Third, our study was conducted in one study unit without any validation; it is necessary to validate our results using data from multiple centres. Finally, because the model was based on clinicopathological data, the mechanisms leading to early relapse and metastasis in M2 patients remain unclear, and the accuracy may be further improved if specific markers of M2 are added.\u003c/p\u003e \u003cp\u003eIn conclusion, complement C3, tumor size\u0026thinsp;\u0026gt;\u0026thinsp;5 cm, incomplete tumor capsule, and Edmondson\u0026ndash;Steiner stages III\u0026ndash;IV were identified as important predictors for the development of M2 in patients with HCC. We then combined the above variables to create a practical nomogram to make the individualized prediction of the M2 rank more objective and accurate. According to the nomogram scoring system, if patients are considered to be at a high risk of M2, more aggressive and precise treatment is tailored to reduce the potential risk of recurrence. Finally, our nomogram could improve individualized treatment designs and facilitate the selection of surveillance plans to develop more effective treatment options for patients with HCC.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAFP:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eAlpha fetoprotein\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eALP:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eAlkaline phosphatase\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eALT\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003eAlanine aminotransferase\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAST\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003eAspartate aminotransferase\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAUROC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eArea Under the Receiver Operating Characteristic Curve\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eBCLC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003eBarcelona Clinic Liver Cancer\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCI\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Confidence interval\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCIC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Clinical Impact Curve\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCT\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Computed tomography\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDCA\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Decisive cure analysis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHAIC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Hepatic Artery Infusion Chemotherapy\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHCC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Hepetacullar carcinoma\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHR\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Hazard Ratio\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eMRI\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Magnetic resonance imaging\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eMVI\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Microvascular invasion\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eOR\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Odds ratio\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRBC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Red blood cells\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eROC\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u003cem\u003e：\u003c/em\u003e\u003c/strong\u003e Receiver operating characteristic\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eTACE:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eTranscatheter arterial chemoembolization\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eWBC:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eWhite blood cells\u003cstrong\u003e\u003c/strong\u003e\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to acknowledge the statistical assistance offered by Qingyuan Zhang，Jie Lin，Peng Zhu，Hualin Wu.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by grants from the National Natural Science Foundation of China (82260345), the Key Research and Development Program of Guangxi (AB22080066), the \u0026quot;139\u0026quot; Plan for Training High-level Backbone Medical Talents in Guangxi (G202003008), \u0026nbsp;the Guangxi Medical University Outstanding Young Talents Training Program , Scientific Research Project of Hunan Provincial Health Commission(No: 202204013436),and the ScientificResearch Project of shaoyang city Science and Technology Bureau(NO:2022GZ4161)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGuoyi Xia and Zeyan Yu have contributed equally to this work.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Authors and Affiliations\u003c/p\u003e\n\u003cp\u003eDepartment of Hepatobiliary Surgery, Guangxi Medical University Cancer Hospital, Nanning, 530021 Guangxi, China.\u003c/p\u003e\n\u003cp\u003eGuoyi Xia,Zeyan Yu,Shaolong Lu,Xiaobo Wang\u0026amp;Jie Chen\u003c/p\u003e\n\u003cp\u003eDepartment of Hepatobiliary Surgery, The Central Hospital of Shaoyang,\u003c/p\u003e\n\u003cp\u003eShaoyang,422000 Hunan,China.\u003c/p\u003e\n\u003cp\u003eGuoyi Xia\u003c/p\u003e\n\u003cp\u003eDepartment of Hepatobiliary Surgery, Guangxi Zhuang Autonomous Region People\u0026apos;s Hospital, Nanning, 530021 Guangxi, China.\u003c/p\u003e\n\u003cp\u003eYuanquan Zhao\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eConsent to Participate declaration\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGuoyi Xia，Zeyan Yu and Jie Chen contributed to the conception and design of the study and drafted the manuscript; Shaolong lu participated in data collection and literature research; and Xiaobo Wang ，Yuanquan Zhao contributed to data analysis and interpretation. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Corresponding author\u003c/p\u003e\n\u003cp\u003eCorrespondence to\u0026nbsp;Jie Chen.\u003c/p\u003e\n\u003cp\u003eEmail: [email protected] \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuman Ethics and Consent to Participate declarations\u003c/p\u003e\n\u003cp\u003eThis was a retrospective and controlled study. This study adhered to the Declaration of Helsinki and was approved by the Ethics Committee of Guangxi Medical University Cancer Hospital (Clinical Trial Number:KY2024454).\u003c/p\u003e\n\u003cp\u003eConflicting \u0026nbsp;interests\u003c/p\u003e\n\u003cp\u003eThe author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eYang JD, Hainaut P, Gores GJ, Amadou A, Plymoth A, Roberts LR: \u003cstrong\u003eA global view of hepatocellular carcinoma: trends, risk, prevention and management\u003c/strong\u003e. \u003cem\u003eNat Rev Gastroenterol Hepatol \u003c/em\u003e2019, \u003cstrong\u003e16\u003c/strong\u003e(10):589-604.\u003c/li\u003e\n\u003cli\u003eLlovet JM, Kelley RK, Villanueva A, Singal AG, Pikarsky E, Roayaie S, Lencioni R, Koike K, Zucman-Rossi J, Finn RS: \u003cstrong\u003eHepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eNat Rev Dis Primers \u003c/em\u003e2021, \u003cstrong\u003e7\u003c/strong\u003e(1):6.\u003c/li\u003e\n\u003cli\u003eLlovet JM BC, Bruix J.: \u003cstrong\u003ePrognosis of hepatocellular carcinoma: the BCLC staging classification\u003c/strong\u003e. \u003cem\u003eSemin Liver Dis \u003c/em\u003e1999, \u003cstrong\u003e19\u003c/strong\u003e(3):329-338.\u003c/li\u003e\n\u003cli\u003eReig M, Forner A, Rimola J, Ferrer-F\u0026agrave;brega J, Burrel M, Garcia-Criado \u0026Aacute;, Kelley RK, Galle PR, Mazzaferro V, Salem R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eBCLC strategy for prognosis prediction and treatment recommendation: The 2022 update\u003c/strong\u003e. \u003cem\u003eJ Hepatol\u003c/em\u003e, \u003cstrong\u003e76\u003c/strong\u003e(3):681-693.\u003c/li\u003e\n\u003cli\u003eBrown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM: \u003cstrong\u003eManagement of Hepatocellular Carcinoma: A Review\u003c/strong\u003e. \u003cem\u003eJAMA Surg \u003c/em\u003e2023, \u003cstrong\u003e158\u003c/strong\u003e(4):410-420.\u003c/li\u003e\n\u003cli\u003eLlovet JM, Pinyol R, Kelley RK, El-Khoueiry A, Reeves HL, Wang XW, Gores GJ, Villanueva A: \u003cstrong\u003eMolecular pathogenesis and systemic therapies for hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eNat Cancer \u003c/em\u003e2022, \u003cstrong\u003e3\u003c/strong\u003e(4):386-401.\u003c/li\u003e\n\u003cli\u003eErstad DJ, Tanabe KK: \u003cstrong\u003ePrognostic and Therapeutic Implications of Microvascular Invasion in Hepatocellular Carcinoma\u003c/strong\u003e. \u003cem\u003eAnn Surg Oncol \u003c/em\u003e2019, \u003cstrong\u003e26\u003c/strong\u003e(5):1474-1493.\u003c/li\u003e\n\u003cli\u003eCong WM, Bu H, Chen J, Dong H, Zhu YY, Feng LH, Chen J, Guideline C: \u003cstrong\u003ePractice guidelines for the pathological diagnosis of primary liver cancer: 2015 update\u003c/strong\u003e. \u003cem\u003eWorld J Gastroenterol \u003c/em\u003e2016, \u003cstrong\u003e22\u003c/strong\u003e(42):9279-9287.\u003c/li\u003e\n\u003cli\u003eYao LQ, Li C, Diao YK, Liang L, Jia HD, Tang SC, Zeng YY, Wu H, Wang MD, Gu LH\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGrading severity of microscopic vascular invasion was independently associated with recurrence and survival following hepatectomy for solitary hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eHepatobiliary Surg Nutr \u003c/em\u003e2024, \u003cstrong\u003e13\u003c/strong\u003e(1):16-28.\u003c/li\u003e\n\u003cli\u003eHuo T-I, Ho S-Y, Liao J-I: \u003cstrong\u003ePredicting post-resection early recurrence of hepatocellular carcinoma: Defining the role of microvascular invasion\u003c/strong\u003e. \u003cem\u003eLiver Int \u003c/em\u003e2023, \u003cstrong\u003e43\u003c/strong\u003e(12):2826-2827.\u003c/li\u003e\n\u003cli\u003eYao L-Q, Chen Z-L, Feng Z-H, Diao Y-K, Li C, Sun H-Y, Zhong J-H, Chen T-H, Gu W-M, Zhou Y-H\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eClinical Features of Recurrence After Hepatic Resection for Early-Stage Hepatocellular Carcinoma and Long-Term Survival Outcomes of Patients with Recurrence: A Multi-institutional Analysis\u003c/strong\u003e. \u003cem\u003eAnn Surg Oncol \u003c/em\u003e2022.\u003c/li\u003e\n\u003cli\u003eWang H, Liu R, Mo H, Li R, Lian J, Liu Q, Han S: \u003cstrong\u003eA novel nomogram predicting the early recurrence of hepatocellular carcinoma patients after R0 resection\u003c/strong\u003e. \u003cem\u003eFront Oncol \u003c/em\u003e2023, \u003cstrong\u003e13\u003c/strong\u003e:1133807.\u003c/li\u003e\n\u003cli\u003eWang K, Xiang Y, Yan J, Zhu Y, Chen H, Yu H, Cheng Y, Li X, Dong W, Ji Y\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA deep learning model with incorporation of microvascular invasion area as a factor in predicting prognosis of hepatocellular carcinoma after R0 hepatectomy\u003c/strong\u003e. \u003cem\u003eHepatol Int \u003c/em\u003e2022, \u003cstrong\u003e16\u003c/strong\u003e(5):1188-1198.\u003c/li\u003e\n\u003cli\u003eSheng X, Ji Y, Ren G-P, Lu C-L, Yun J-P, Chen L-H, Meng B, Qu L-J, Duan G-J, Sun Q\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA standardized pathological proposal for evaluating microvascular invasion of hepatocellular carcinoma: a multicenter study by LCPGC\u003c/strong\u003e. \u003cem\u003eHepatology International \u003c/em\u003e2020, \u003cstrong\u003e14\u003c/strong\u003e(6):1034-1047.\u003c/li\u003e\n\u003cli\u003eYang P, Si A, Yang J, Cheng Z, Wang K, Li J, Xia Y, Zhang B, Pawlik TM, Lau WY\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA wide-margin liver resection improves long-term outcomes for patients with HBV-related hepatocellular carcinoma with microvascular invasion\u003c/strong\u003e. \u003cem\u003eSurgery \u003c/em\u003e2019, \u003cstrong\u003e165\u003c/strong\u003e(4):721-730.\u003c/li\u003e\n\u003cli\u003eWang H, Du P-C, Wu M-C, Cong W-M: \u003cstrong\u003ePostoperative adjuvant transarterial chemoembolization for multinodular hepatocellular carcinoma within the Barcelona Clinic Liver Cancer early stage and microvascular invasion\u003c/strong\u003e. \u003cem\u003eHepatobiliary Surgery and Nutrition \u003c/em\u003e2018, \u003cstrong\u003e7\u003c/strong\u003e(6):418-428.\u003c/li\u003e\n\u003cli\u003eLee HW, Kim H, Park T, Park SY, Chon YE, Seo YS, Lee JS, Park JY, Kim DY, Ahn SH\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA machine learning model for predicting hepatocellular carcinoma risk in patients with chronic hepatitis B\u003c/strong\u003e. \u003cem\u003eLiver Int \u003c/em\u003e2023, \u003cstrong\u003e43\u003c/strong\u003e(8):1813-1821.\u003c/li\u003e\n\u003cli\u003eHwang YJ, Bae JS, Lee Y, Hur BY, Lee DH, Kim H: \u003cstrong\u003eClassification of microvascular invasion of hepatocellular carcinoma: correlation with prognosis and magnetic resonance imaging\u003c/strong\u003e. \u003cem\u003eClin Mol Hepatol \u003c/em\u003e2023, \u003cstrong\u003e29\u003c/strong\u003e(3):733-746.\u003c/li\u003e\n\u003cli\u003eXu Y, Chen Y, Wu J, Pan J, Liao C, Su H: \u003cstrong\u003eThe utility of Vision Transformer in preoperatively predicting microvascular invasion status of hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eHpb \u003c/em\u003e2023, \u003cstrong\u003e25\u003c/strong\u003e(5):533-542.\u003c/li\u003e\n\u003cli\u003eZhang J, Zeng F, Jiang S, Tang H, Zhang J: \u003cstrong\u003ePreoperative prediction model of microvascular invasion in patients with hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eHPB (Oxford) \u003c/em\u003e2023, \u003cstrong\u003e25\u003c/strong\u003e(1):45-53.\u003c/li\u003e\n\u003cli\u003eWang H, Lu Y, Liu R, Wang L, Liu Q, Han S: \u003cstrong\u003eA Non-Invasive Nomogram for Preoperative Prediction of Microvascular Invasion Risk in Hepatocellular Carcinoma\u003c/strong\u003e. \u003cem\u003eFront Oncol \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e:745085.\u003c/li\u003e\n\u003cli\u003eLi J, Su X, Xu X, Zhao C, Liu A, Yang L, Song B, Song H, Li Z, Hao X: \u003cstrong\u003ePreoperative prediction and risk assessment of microvascular invasion in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eCrit Rev Oncol Hematol \u003c/em\u003e2023, \u003cstrong\u003e190\u003c/strong\u003e:104107.\u003c/li\u003e\n\u003cli\u003eRodriguez-Peralvarez M, Luong TV, Andreana L, Meyer T, Dhillon AP, Burroughs AK: \u003cstrong\u003eA systematic review of microvascular invasion in hepatocellular carcinoma: diagnostic and prognostic variability\u003c/strong\u003e. \u003cem\u003eAnn Surg Oncol \u003c/em\u003e2013, \u003cstrong\u003e20\u003c/strong\u003e(1):325-339.\u003c/li\u003e\n\u003cli\u003eMima K, Hayashi H, Kuroki H, Nakagawa S, Okabe H, Chikamoto A, Watanabe M, Beppu T, Baba H: \u003cstrong\u003eEpithelial-mesenchymal transition expression profiles as a prognostic factor for disease-free survival in hepatocellular carcinoma: Clinical significance of transforming growth factor-beta signaling\u003c/strong\u003e. \u003cem\u003eOncol Lett \u003c/em\u003e2013, \u003cstrong\u003e5\u003c/strong\u003e(1):149-154.\u003c/li\u003e\n\u003cli\u003eZhou YM, Cao L, Li B, Zhang RX, Sui CJ, Yin ZF, Yang JM: \u003cstrong\u003eClinicopathological significance of ZEB1 protein in patients with hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eAnn Surg Oncol \u003c/em\u003e2012, \u003cstrong\u003e19\u003c/strong\u003e(5):1700-1706.\u003c/li\u003e\n\u003cli\u003eLi K, Zhang R, Wen F, Zhao Y, Meng F, Li Q, Hao A, Yang B, Lu ZA-O, Cui YA-O\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSingle-cell dissection of the multicellular ecosystem and molecular features underlying microvascular invasion in HCC.Hepatology. 2023 Nov 16. doi: 10.1097/HEP.0000000000000673.\u003c/strong\u003e\u003cem\u003e Hepatology \u003c/em\u003e(1527-3350 (Electronic)).\u003c/li\u003e\n\u003cli\u003eHan J, Zhang X: \u003cstrong\u003eComplement Component C3: A Novel Biomarker Participating in the Pathogenesis of Non-alcoholic Fatty Liver Disease\u003c/strong\u003e. \u003cem\u003eFront Med (Lausanne) \u003c/em\u003e2021, \u003cstrong\u003e8\u003c/strong\u003e:653293.\u003c/li\u003e\n\u003cli\u003eZhu C, Song H, Xu F, Yi W, Liu F, Liu X: \u003cstrong\u003eHepatitis B virus inhibits the expression of complement C3 and C4, in vitro and in vivo\u003c/strong\u003e. \u003cem\u003eOncol Lett \u003c/em\u003e2018, \u003cstrong\u003e15\u003c/strong\u003e(5):7459-7463.\u003c/li\u003e\n\u003cli\u003eWang N, Tan H-Y, Lu Y, Chan Y-T, Wang D, Guo W, Xu Y, Zhang C, Chen F, Tang G\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePIWIL1 governs the crosstalk of cancer cell metabolism and immunosuppressive microenvironment in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eSignal Transduction and Targeted Therapy \u003c/em\u003e2021, \u003cstrong\u003e6\u003c/strong\u003e(1):86.\u003c/li\u003e\n\u003cli\u003eZheng C, Gu X-T, Huang X-L, Wei Y-C, Chen L, Luo N-B, Lin H-S, Jin-Yuan L: \u003cstrong\u003eNomogram based on clinical and preoperative CT features for predicting the early recurrence of combined hepatocellular-cholangiocarcinoma: a multicenter study\u003c/strong\u003e. \u003cem\u003eRadiol Med \u003c/em\u003e2023, \u003cstrong\u003e128\u003c/strong\u003e(12):1460-1471.\u003c/li\u003e\n\u003cli\u003eYang J, Zhu S, Yong J, Xia L, Qian X, Yang J, Hu X, Li Y, Wang C, Peng W\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA Nomogram for Preoperative Estimation of Microvascular Invasion Risk in Hepatocellular Carcinoma: Single-Center Analyses With Internal Validation\u003c/strong\u003e. \u003cem\u003eFront Oncol \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e:616976.\u003c/li\u003e\n\u003cli\u003ePawlik TM, Delman KA, Vauthey JN, Nagorney DM, Ng IO, Ikai I, Yamaoka Y, Belghiti J, Lauwers GY, Poon RT\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eTumor size predicts vascular invasion and histologic grade: Implications for selection of surgical treatment for hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eLiver Transpl \u003c/em\u003e2005, \u003cstrong\u003e11\u003c/strong\u003e(9):1086-1092.\u003c/li\u003e\n\u003cli\u003eXu X, Zhang H-L, Liu Q-P, Sun S-W, Zhang J, Zhu F-P, Yang G, Yan X, Zhang Y-D, Liu X-S: \u003cstrong\u003eRadiomic analysis of contrast-enhanced CT predicts microvascular invasion and outcome in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eJournal of Hepatology \u003c/em\u003e2019, \u003cstrong\u003e70\u003c/strong\u003e(6):1133-1144.\u003c/li\u003e\n\u003cli\u003eChen S, Wang C, Gu Y, Ruan R, Yu J, Wang S: \u003cstrong\u003ePrediction of Microvascular Invasion and Its M2 Classification in Hepatocellular Carcinoma Based on Nomogram Analyses\u003c/strong\u003e. \u003cem\u003eFront Oncol \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e:774800.\u003c/li\u003e\n\u003cli\u003eKim SU, Jung Ks Fau - Lee S, Lee S Fau - Park JY, Park Jy Fau - Kim DY, Kim Dy Fau - Ahn SH, Ahn Sh Fau - Choi GH, Choi Gh Fau - Kim KS, Kim Ks Fau - Choi JS, Choi Js Fau - Han K-H, Han Kh Fau - Park YN\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eHistological subclassification of cirrhosis can predict recurrence after curative resection of hepatocellular carcinoma.Liver Int. 2014 Aug;34(7):1008-17. doi: 10.1111/liv.12475.\u003c/strong\u003e (1478-3231 (Electronic)).\u003c/li\u003e\n\u003cli\u003eHuang J, Li L, Liu FC, Tan BB, Yang Y, Jiang BG, Pan ZY: \u003cstrong\u003ePrognostic Analysis of Single Large Hepatocellular Carcinoma Following Radical Resection: A Single-Center Study\u003c/strong\u003e. \u003cem\u003eJ Hepatocell Carcinoma \u003c/em\u003e2023, \u003cstrong\u003e10\u003c/strong\u003e:573-586.\u003c/li\u003e\n\u003cli\u003eWei X, Jiang Y, Feng S, Lu C, Huo L, Zhou B, Meng Y, Lau WY, Zheng Y, Cheng S: \u003cstrong\u003eNeoadjuvant intensity modulated radiotherapy for a single and small (\u0026le;5 cm) hepatitis B virus-related hepatocellular carcinoma predicted to have high risks of microvascular invasion: a randomized clinical trial\u003c/strong\u003e. \u003cem\u003eInt J Surg \u003c/em\u003e2023, \u003cstrong\u003e109\u003c/strong\u003e(10):3052-3060.\u003c/li\u003e\n\u003cli\u003eXiang C, Shen X, Zeng X, Zhang Y, Ma Z, Zhang G, Song X, Huang T, Yang J: \u003cstrong\u003eEffect of transarterial chemoembolization as postoperative adjuvant therapy for intermediate-stage hepatocellular carcinoma with microvascular invasion: a multicenter cohort study\u003c/strong\u003e. \u003cem\u003eInt J Surg \u003c/em\u003e2024, \u003cstrong\u003e110\u003c/strong\u003e(1):315-323.\u003c/li\u003e\n\u003cli\u003eUeshima K, Komemushi A, Aramaki T, Iwamoto H, Obi S, Sato Y, Tanaka T, Matsueda K, Moriguchi M, Saito H\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eClinical Practice Guidelines for Hepatic Arterial Infusion Chemotherapy with a Port System Proposed by the Japanese Society of Interventional Radiology and Japanese Society of Implantable Port Assisted Treatment\u003c/strong\u003e. \u003cem\u003eLiver Cancer \u003c/em\u003e2022, \u003cstrong\u003e11\u003c/strong\u003e(5):407-425.\u003c/li\u003e\n\u003cli\u003eWang K, Xiang YA-O, Yu HM, Cheng YQ, Liu ZH, Qin YY, Shi J, Guo WX, Lu CD, Zheng YX\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAdjuvant sintilimab in resected high-risk hepatocellular carcinoma: a randomized, controlled, phase 2 trial.Nat Med. 2024 Jan 19. doi: 10.1038/s41591-023-02786-7\u003c/strong\u003e. 2024(1546-170X (Electronic)).\u003c/li\u003e\n\u003cli\u003eXu XF, Diao YK, Zeng YY, Li C, Li FW, Sun LY, Wu H, Lin KY, Yao LQ, Wang MD\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAssociation of severity in the grading of microvascular invasion with long-term oncological prognosis after liver resection for early-stage hepatocellular carcinoma: a multicenter retrospective cohort study from a hepatitis B virus-endemic area\u003c/strong\u003e. \u003cem\u003eInt J Surg \u003c/em\u003e2023, \u003cstrong\u003e109\u003c/strong\u003e(4):841-849.\u003c/li\u003e\n\u003cli\u003eZhou ZR, Wang WW, Li Y, Jin KR, Wang XY, Wang ZW, Chen YS, Wang SJ, Hu J, Zhang HN\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIn-depth mining of clinical data: the construction of clinical prediction model with R\u003c/strong\u003e. \u003cem\u003eAnn Transl Med \u003c/em\u003e2019, \u003cstrong\u003e7\u003c/strong\u003e(23):796.\u003c/li\u003e\n\u003cli\u003eWei X, Li N, Li S, Shi J, Guo W, Zheng Y, Cheng S: \u003cstrong\u003eHepatitis B virus infection and active replication promote the formation of vascular invasion in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eBMC Cancer \u003c/em\u003e2017, \u003cstrong\u003e17\u003c/strong\u003e(1):304.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hepatocellular carcinoma (HCC), early-stage, machine learning, nomogram","lastPublishedDoi":"10.21203/rs.3.rs-4410132/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4410132/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Microvascular invasion (MVI) is a crucial factor for early recurrence and poor outcomes in hepatocellular carcinoma (HCC). However, there are few studies on M2 classification. We aimed to build a predictive model for M2 in early-stage HCC, assisting clinical decision-making.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e We retrospectively enrolled 451 patients with early-stage HCC and employed multiple machine learning algorithms to identify the risk factors influencing the robustness of M2. \u0026nbsp;Model performance was evaluated using receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA)，and clinical impact curve (CIC).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e There were 363 M0 and 88 M2 cases. Differences in recurrence-free survival (RFS) and overall survival(OS) between the M0 and M2 groups were statistically significant \u0026nbsp;(\u003cem\u003eP\u003c/em\u003e \u0026lt;0.0001). Complement C3, tumor size\u0026gt; 5cm, incomplete tumor capsule, and Edmondson-Steiner stage III-IV were independent risk factors for M2.The prediction model showed an area under the receiver operating characteristic curve(AUROC) of 0.765 and 0.807 in the training and validation groups, respectively. Calibration curves showed good agreement between actual and predicted M2 risks, and the DCA and CIC showed a significant clinical efficacy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThe nomogram-based model had a good predictive effect for M2 in patients with early-stage HCC ,providing guidance for treatment decisions.\u003c/p\u003e","manuscriptTitle":"Prediction model of M2 with early-stage hepatocellular carcinoma based on multiple machine learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-07-08 17:12:57","doi":"10.21203/rs.3.rs-4410132/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4b60834c-abe6-4a99-9965-cc013da1ad42","owner":[],"postedDate":"July 8th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-10-09T08:08:12+00:00","versionOfRecord":[],"versionCreatedAt":"2024-07-08 17:12:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4410132","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4410132","identity":"rs-4410132","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}