The diagnostic accuracy of imaging methods in preoperative prediction of vessels encapsulating tumor clusters in hepatocellular carcinoma in East Asian populations: a systematic review and meta-analysis

Introduction

Hepatocellular carcinoma (HCC) is the sixth most common cancer and the third leading cause of cancer-related deaths globally (1-3). Chronic liver diseases, including cirrhosis, are well-established risk factors for HCC development (4). Management strategies for HCC are diverse, comprising surgical resection, locoregional therapies, systemic treatments, and palliative interventions (3,5). Although surgical resection remains the cornerstone for patients with early-stage or selected intermediate-stage HCC, postoperative recurrence is frequent, with reported 5-year recurrence rates approaching 70% (6,7).

Recently, previous studies have described the vessels encapsulating tumor cluster (VETC) pattern and its correlation with adverse outcomes in HCC patients (8-10). In the majority of studies, the VETC pattern is characterized by a continuous lining of sinusoid-like vessels that encase individual tumor clusters, creating a cobweb-like structure (11). Tumor clusters can be directly released into the circulation via anastomoses between the VETC and surrounding blood vessels, promoting rapid tumor spread and heightening the likelihood of recurrence (8-10). The VETC pattern has been identified as a potential prognostic factor for the effectiveness of sorafenib therapy, as well as for adjuvant therapies such as transarterial chemoembolization (TACE) and hepatic arterial infusion chemotherapy (HAIC) in HCC patients (12-14). Additionally, the VETC pattern may act as a reliable biomarker for identifying early-stage HCC patients at heightened risk for recurrence, who might benefit from repeat hepatic resection (RHR) (15). Therefore, the precise assessment and monitoring of VETC status are crucial for optimizing the prognosis of HCC patients.

Currently, the gold standard for diagnosing VETC is pathology. Although needle biopsy can serve as a diagnostic tool for detecting VETC in HCC prior to surgery, it has certain drawbacks. These include the risk of hemorrhage at the biopsy site, the potential for tumor seeding, and the possibility of inaccurate sampling (10). However, with ongoing technological advancements, imaging techniques have increasingly gained prominence as a non-invasive alternative for preoperative tumor assessment. As clinical interest in the VETC characteristics of HCC has grown, numerous individual studies have explored various imaging modalities for its diagnosis (16,17). However, the diagnostic accuracy of these methods varies significantly across different imaging modalities. This variability can be attributed to inconsistencies in discrepancies in imaging protocols, study methodologies, sample sizes, and other factors. Consequently, the reliability of preoperative imaging models for predicting VETC status in HCC remains a significant concern. Considering these challenges, conducting a thorough meta-analysis to assess the diagnostic precision of imaging-based models for predicting VETC in HCC before surgery is both timely and crucial. This research intends to systematically assess and perform a meta-analysis of the efficacy of non-invasive imaging techniques prior to surgery in forecasting VETC status, offering significant insights for the preoperative evaluation of VETC in HCC. We present this article in accordance with the PRISMA reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1534/rc).

Methods

This systematic review and meta-analysis were conducted following the Cochrane Collaboration’s guidelines. The study has been registered in the PROSPERO database (No. CRD42024584454).

Search strategy and study selection

Two researchers conducted a thorough exploration of various databases, including PubMed, Embase, Web of Science, and the Cochrane Library, concentrating on studies that assess the predictive accuracy of VETC before surgery in HCC cases. The search utilized the terms “carcinoma, hepatocellular” and “vessels encapsulating tumor clusters”, along with related synonyms. In addition, reference lists of pertinent publications, review articles, and leading journals were manually examined to discover further research. This search included articles published until August 3, 2024, with no restrictions on either language or publication start date. Comprehensive search strategies are detailed in Table S1.

Studies were eligible for inclusion if they fulfilled the following criteria: (I) HCC diagnosis confirmed by pathological analysis; (II) VETC status determined through pathological examination; (III) inclusion of at least three VETC-positive and three VETC-negative cases; and (IV) preoperative imaging, such as computed tomography (CT), magnetic resonance imaging (MRI), or ultrasound (US), conducted before hepatectomy or liver transplantation.

Exclusion criteria were: (I) case reports, reviews, editorials, letters, comments, or conference proceedings; (II) studies irrelevant to the research scope; (III) duplicate cohorts; (IV) non-human research; and (V) studies lacking sufficient data for constructing a 2×2 diagnostic table or conducting correlation analysis.

Data extraction

Two independent researchers systematically assessed the selected studies and extracted relevant data utilizing standardized extraction forms. The gathered information encompassed: (I) bibliographic details such as author, publication year, study design, and location; (II) participant characteristics, including type of surgery, pathological classification, Barcelona Clinic Liver Cancer (BCLC) stage, etiology, and tumor size; and (III) model-related attributes, including imaging modality, algorithm, and methodology.

Moreover, data regarding the presence or absence of VETC, along with diagnostic outcomes, such as true positives (TPs), false positives (FPs), false negatives (FNs), and true negatives (TNs), were collected. Key diagnostic metrics, including positive likelihood ratio (PLR), negative likelihood ratio (NLR), and area under the curve (AUC), were also extracted. When multiple models were assessed within a single cohort, only the one with the highest diagnostic accuracy was selected for inclusion in the meta-analysis.

Quality assessment

The methodological quality of the studies included was assessed independently by two reviewers, utilizing the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool, which was created by the University of Bristol in the UK. Each reviewer carried out the evaluation separately, and any discrepancies were resolved through discussion or, if necessary, by bringing in a third reviewer.

The assessment concentrated on four principal domains: selection of patients, the index test, the reference standard, and the flow and timing. For every item in the assessment tool, responses were classified as “yes”, “no”, or “unclear” to evaluate potential bias, and categorized as “low risk”, “high risk”, or “unclear risk” to consider concerns regarding applicability.

Statistical analysis

Statistical analyses were conducted utilizing Stata 16.0. A fixed-effects model was utilized to derive sensitivity, specificity, PLR, NLR, and their 95% confidence intervals (CIs) when heterogeneity was minimal (I²<50%), a random-effects model was utilized to estimate these metrics along with their associated CIs. The findings were illustrated as forest plots. The Spearman correlation coefficient was used to assess threshold effects, while the evaluation of publication bias was performed using Deeks’ funnel plot asymmetry test.

The overall diagnostic efficacy was evaluated by creating SROC curves and calculating the AUC. The diagnostic accuracy was classified as low (AUC <0.7), moderate (AUC: 0.7–0.9), or high (AUC >0.9). For cases where significant heterogeneity was present (I²>50%), meta-regression and subgroup analyses were performed to uncover potential sources. Comparisons of sensitivity, specificity, PLR, and NLR among subgroups were executed using the Z test, and the AUC comparisons underwent analysis with the DeLong test. To assess the clinical value of preoperative imaging in predicting VETC status, Fagan’s nomogram was employed.

All statistical tests were two-tailed, with P values deemed statistically significant when <0.05.

Results

Study demographics

The article selection process is detailed in the PRISMA flow diagram. The initial search across PubMed, Embase, Web of Science, and the Cochrane Library identified 2,076 articles. After eliminating duplicates, 1,868 unique records remained. Following a title and abstract screening, this number was reduced to 76. A comprehensive full-text assessment ultimately led to the inclusion of 18 studies, as illustrated in Figure 1.

Figure 1 Flow chart of study selection.

Detailed characteristics of the included studies are provided in Table 1. Among these, 12 were conducted in a single-center setting (18-29), six were multi-center studies (30-35), and only one was prospective (21), while the remaining were retrospective or unclear. A majority of these studies focused on the Chinese population, with only one involving participants from Japan (24). In terms of imaging methods for preoperative diagnosis of VETC, four studies utilized preoperative CT scans (22,26,33,35), 12 applied MRI, and only one included US (27). Additionally, one study combined US and MRI for diagnosis (24). Regarding the type of imaging methods used in conventional radiology (CR) and artificial intelligence (AI), seven studies adopted AI-based approaches (18-20,27,28,34,35), while 11 relied on CR (21-26,29-33). This meta-analysis included a total of 3,615 HCC patients, of whom 1,463 were pathologically confirmed to have VETC after surgery or transplantation, while 2,152 patients did not present with VETC. BCLC staging data were extracted where reported (available in nine of 18 studies). Among these nine studies, eight provided data for patients classified as BCLC stage 0 or A, accounting for 77.0% (1,349/1,751). The remaining study reported patients classified as BCLC stage A or B, accounting for 70.7% (342/484). Details are provided in Table S2. Eight studies reported the distribution of different etiologies among VETC-positive and VETC-negative patients. Among VETC-positive patients, viral etiology accounted for 82.7% (522/631), while other etiologies accounted for 17.3% (109/631). Similarly, among VETC-negative patients, viral etiology accounted for 82.7% (801/968), and other etiologies for 17.3% (167/968). Details are provided in Table S3. The sample sizes and 2×2 table data for these studies can be found in Tables S4,S5.

Risk of bias assessment

The methodological quality assessment findings of the studies incorporated in this review are illustrated in Figure 2. Concerning patient selection, eight studies (20,21,24-26,29,31,32) were classified with an unclear risk of bias, primarily due to ambiguous patient recruitment protocols or possible complications tied to unsuitable exclusion criteria. In relation to the index test domain, three studies (20,21,29) were evaluated as having a high risk of bias, stemming from uncertainty about whether the clinical data were blinded from the test outcomes. Furthermore, four studies (18,24,28,34) were rated as having an unclear risk of bias. Within the reference standard domain, eight studies (19,21,24-28,33) were rated as having an unclear risk of bias. Overall, the studies demonstrated a low to moderate risk of bias, with few concerns regarding their applicability. According to the quality assessment criteria, none of the studies were excluded. From the 18 studies included, the Deeks’ funnel plot asymmetry test indicated no signs of publication bias (P=0.43). Moreover, the threshold effect analysis also showed no significant impact, with a Spearman correlation coefficient of 0.131 and P=0.483. The pooled diagnostic performance metrics from the meta-analysis are presented in the forest plot (Figure 3): sensitivity was 0.79 (95% CI: 0.73–0.84), specificity was 0.83 (95% CI: 0.78–0.87), the PLR was 4.64 (95% CI: 3.44–6.26), and the NLR was 0.25 (95% CI: 0.19–0.33). Heterogeneity was significant, as indicated by Higgins’ I² values for sensitivity (73.18%), specificity (83.56%), PLR (70.26%), and NLR (69.49%), necessitating the use of a random-effects model. The summary receiver operating characteristic (SROC) curve yielded an AUC of 0.88 (95% CI: 0.85–0.91; Figure 4A), indicating moderate diagnostic accuracy.

Figure 2 Stacked bar charts of QUADAS-2 scores of methodologic study quality that provide an overview of study quality in the whole meta-analysis. QUADAS-2, Quality Assessment of Diagnostic Accuracy Studies-2.

Figure 3 Forest plots of tests. The accuracy estimates of the included studies about (A) sensitivity, (B) specificity, (C) PLR, and (D) NLR. CI, confidence interval; df, degrees of freedom; NLR, negative likelihood ratio; PLR, positive likelihood ratio; t, testing cohort; T, training cohort; V, validation cohort.

Figure 4 SROC plot (A) and Fagan’s nomogram (B) of the included studies. AUC, area under the curve; LR, likelihood ratio; neg, negative; pos, positive; prob, probability; SENS, sensitivity; SPEC, specificity; SROC, summary receiver operating characteristic.

The Fagan nomogram suggested that, with a pre-test probability of 20%, a PLR of 5 increased the post-test probability of VETC to 54%, while an NLR of 0.25 decreased it to 6% (Figure 4B).

To minimize overfitting during model training, a subset meta-analysis was conducted using only validation set data, excluding training set information. This subset included 12 studies involving 863 HCC patients, of whom 326 had VETC, and 537 did not. The Deeks’ funnel plot asymmetry test for this subset again showed no publication bias (P=0.58; Figure S1). Threshold effect analysis remained nonsignificant, with a Spearman correlation coefficient of 0.252 and P=0.430.

Meta‑regression and subgroup analysis

The forest plots for the subset analysis (Figures S2,S3) showed pooled sensitivity of 0.79 (95% CI: 0.70–0.85), specificity of 0.77 (95% CI: 0.68–0.84), PLR of 3.40 (95% CI: 2.47–4.69), and NLR of 0.28 (95% CI: 0.20–0.38). The SROC curve for this subset had an AUC of 0.85 (95% CI: 0.81–0.88; Figure S4), again reflecting moderate diagnostic accuracy. According to the Fagan nomogram, a positive test result increased the post-test probability of detecting VETC to 46%, while a negative result decreased it to 6% (Figure S5).

The forest plots for sensitivity and specificity revealed significant heterogeneity, with I² values of 73.18% and 83.56%, respectively. To explore potential sources of heterogeneity, meta-regression analysis was performed, evaluating seven covariates outlined in Table S6. The analysis identified study design, study location, and MRI methodology as significant contributors to sensitivity heterogeneity (P<0.05). Similarly, study design and MRI methodology were found to significantly influence specificity (P<0.05).

Subgroup analyses were conducted to evaluate variations in diagnostic accuracy for predicting VETC status across different preoperative subgroups. As shown in Table 2, studies employing gadolinium-ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI demonstrated significantly higher AUC values compared to those using gadopentetate dimeglumine (Gd-DTPA)-enhanced MRI (Z=7.30, P<0.001). Additionally, studies defining VETC diagnostic thresholds at 1% or 5% showed significantly better AUC values than those using a threshold of 55% (Z=2.31, P=0.021; Z=3.73, P<0.001, respectively). Studies integrating AI-based models achieved significantly higher AUCs compared to those relying on conventional imaging approaches (Z=0.95, P<0.001). However, no significant differences in AUC were observed between various study designs or imaging modalities. Notably, no significant differences in AUC were observed between the CT and MRI subgroups (P>0.99), suggesting comparable diagnostic performance across these imaging modalities, as detailed in Table 2.

Comparison of diagnostic performance between CR and AI

Table S7 provides a pairwise comparison of diagnostic performance metrics, including sensitivity, specificity, PLR, NLR, and AUC, between AI and CR. The analysis demonstrates that AI consistently outperforms CR across all metrics.

To address potential overfitting in the training set, an additional analysis was performed using validation group data. Table 3 highlights the diagnostic performance comparison in the validation group, showing that the AI subgroup achieved significantly higher specificity and AUC than the CR subgroup (Z=−2.55, P=0.01; Z=−2.02, P=0.043, respectively).

Discussion

Beyond its prognostic utility, VETC profoundly alters hepatic hemodynamics by promoting a hyperpermeable tumor neovasculature that increases blood flow and facilitates the cluster embolization of tumor cells into portal veins, thereby accelerating intrahepatic dissemination and MVI—key contributors to the approximately 70% 5-year recurrence rate after surgical resection. This hemodynamic dysregulation not only hampers effective drug delivery in treatments such as sorafenib or TACE but also manifests as heterogeneous enhancement on multiphasic imaging.

This study conducted a systematic review and meta-analysis to assess the diagnostic accuracy of various imaging modalities in the preoperative prediction of VETC presence in HCC patients. By synthesizing data from 18 studies, the findings revealed that imaging modalities demonstrate moderate diagnostic performance in non-invasively predicting the presence of VETC. The PLR for imaging methods was 4.64, indicating that a positive imaging result is approximately 4.64 times more likely to reflect the presence of VETC than its absence. These results underscore the potential value of imaging methods as a reliable tool for non-invasive preoperative assessment of VETC status.

Subgroup analyses indicated that imaging techniques integrated with AI exhibited significantly superior diagnostic performance, including enhanced AUC and specificity, compared to traditional imaging methods. AI-assisted imaging notably enhanced diagnostic accuracy. Traditional imaging modalities used for identifying VETC patterns in HCC have face inherent limitations, primarily due to their reliance on subjective interpretation by radiologists, which can reduce repeatability and affect the robustness of the model. In contrast, AI offers distinct advantages, including the ability to extract detailed microscopic quantitative information applicable across various imaging diagnostic processes, as well as being unaffected by human biases or ingrained cognitive patterns, thereby broadening their applicability. In the validation cohort, AI achieved an AUC of 0.88 (95% CI: 0.85–0.91) and specificity of 0.84 (95% CI: 0.70–0.92), further demonstrating its promising potential (33,36-38). In conclusion, AI-based imaging approaches exhibit superior diagnostic performance over traditional imaging methods in identifying VETC patterns in HCC.

In the subgroup analysis, we observed that the diagnostic accuracy of MRI using Gd-EOB-DTPA was significantly superior to that of Gd-DTPA [AUC (95% CI): 0.90 (0.87–0.92) vs. 0.74 (0.70–0.77)]. This finding highlights the enhanced diagnostic performance of Gd-EOB-DTPA in the MRI evaluation of VETC in HCC. We hypothesize that the hepatobiliary phase (HBP) contains critical information related to the microvascular invasion (MVI) status of HCC (39-41), such as the low signal observed around tumors in the HBP, which is an independent predictor of MVI. Given that VETC is significantly associated with frequent MVI (39), the HBP may also provide valuable information regarding VETC.

Previous studies have reported conflicting results regarding the cutoff values for positive VETC patterns. Renne et al. (39) conducted a large-scale, global multicenter study using the K-means adaptive segmentation algorithm, which established 55% as the optimal cutoff. They also found that VETC patterns defined by 1% and 5% (11) thresholds were significantly associated with poorer prognosis in HCC. In our study, we found that the diagnostic performance of subgroups defined by 1% and 5% VETC-positive thresholds was significantly superior to that of the subgroup defined by a 55% threshold. This suggests that pathological thresholds of 1% and 5% may be more advantageous for preoperative imaging model prediction.

Assessing heterogeneity is essential in meta-analyses to ensure the reliability of pooled effect sizes. This study revealed considerable heterogeneity within the validation group (I²=86%) (42). Meta-regression analysis identified study design and MRI modality as primary contributors to this variability. Based on data from 18 studies, we recommend stratifying future preoperative VETC imaging models for HCC by study design and MRI modality to minimize heterogeneity, enhance comparability, and facilitate the identification of optimal predictive models. Some of the included studies provided information on BCLC stage and underlying etiology, showing a consistent pattern: most patients were classified as BCLC stage 0 or A, and viral infection was the predominant cause. However, detailed information on non-viral etiologies—such as alcohol-related or Nonalcoholic fatty liver disease -related HCC—was not available, which may partly explain the observed heterogeneity. Furthermore, variations in imaging devices, settings, and image quality could introduce heterogeneity. Standardizing image normalization and preprocessing before modeling could help address these variations (43). However, due to limitations in available data, these factors were not examined in our meta-regression or subgroup analyses. To investigate these influences more comprehensively, future research should consider individual participant meta-analyses.

There are several limitations in our study that should be acknowledged. Firstly, the retrospective nature of the studies included in this analysis could introduce selection bias. Secondly, the majority of studies were conducted in China, with only one originating from Japan, which may limit the generalizability of the findings to a global population. Nonetheless, the analysis of specificity in the validation group still confirmed the stability of the findings.

Conclusions

In summary, our meta-analysis indicates that preoperative imaging methods provide moderate level of accuracy for the non-invasive assessment of VETC. AI-based models outperformed conventional CR methods in terms of diagnostic accuracy. Nonetheless, the possibility of publication bias necessitates careful interpretation of these findings. Consequently, large-scale prospective studies, together with independent validation cohorts, are critical for establishing the clinical utility and reliability of AI-based approaches in clinical practice.

Acknowledgments

We thank Yuren 930 for allowing the authors to collaborate on this study.

Footnote

Reporting Checklist: The authors have completed the PRISMA reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1534/rc

Funding: This study received funding from the National Key Clinical Specialist Construction Programs.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1534/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Dopazo C, Søreide K, Rangelova E, Mieog S, Carrion-Alvarez L, Diaz-Nieto R, Primavesi F, Stättner S. Hepatocellular carcinoma. Eur J Surg Oncol 2024;50:107313. [Crossref] [PubMed]
3. Yang C, Zhang H, Zhang L, Zhu AX, Bernards R, Qin W, Wang C. Evolving therapeutic landscape of advanced hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol 2023;20:203-22. [Crossref] [PubMed]
4. Toh MR, Wong EYT, Wong SH, Ng AWT, Loo LH, Chow PK, Ngeow J. Global Epidemiology and Genetics of Hepatocellular Carcinoma. Gastroenterology 2023;164:766-82. [Crossref] [PubMed]
5. Xie DY, Zhu K, Ren ZG, Zhou J, Fan J, Gao Q. A review of 2022 Chinese clinical guidelines on the management of hepatocellular carcinoma: updates and insights. Hepatobiliary Surg Nutr 2023;12:216-28. [Crossref] [PubMed]
6. Wang DY, Liu L, Qi XS, Su CP, Chen X, Liu X, Chen J, Li HY, Guo XZ. Hepatic Re-resection Versus Transarterial Chemoembolization for the Treatment of Recurrent Hepatocellular Carcinoma after Initial Resection: a Systematic Review and Meta-analysis. Asian Pac J Cancer Prev 2015;16:5573-8. [Crossref] [PubMed]
7. Brown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM. Management of Hepatocellular Carcinoma: A Review. JAMA Surg 2023;158:410-20. [Crossref] [PubMed]
8. Hanley KL, Feng GS. A new VETC in hepatocellular carcinoma metastasis. Hepatology 2015;62:343-5. [Crossref] [PubMed]
9. He C, Zhou Z, Jiang H, Yin Z, Meng S, Zhang J, Huang P, Xu K, Bian L, Xiao Z, Wang J. Epithelial-Mesenchymal Transition is Superior to Vessels-Encapsulate Tumor Cluster in Promoting Metastasis of Hepatocellular Carcinoma: a Morphological Evidence. J Cancer 2017;8:39-47. [Crossref] [PubMed]
10. Lu L, Wei W, Huang C, Li S, Zhong C, Wang J, Yu W, Zhang Y, Chen M, Ling Y, Guo R. A new horizon in risk stratification of hepatocellular carcinoma by integrating vessels that encapsulate tumor clusters and microvascular invasion. Hepatol Int 2021;15:651-62. [Crossref] [PubMed]
11. Fang JH, Zhou HC, Zhang C, Shang LR, Zhang L, Xu J, Zheng L, Yuan Y, Guo RP, Jia WH, Yun JP, Chen MS, Zhang Y, Zhuang SM. A novel vascular pattern promotes metastasis of hepatocellular carcinoma in an epithelial-mesenchymal transition-independent manner. Hepatology 2015;62:452-65. [Crossref] [PubMed]
12. Fang JH, Xu L, Shang LR, Pan CZ, Ding J, Tang YQ, Liu H, Liu CX, Zheng JL, Zhang YJ, Zhou ZG, Xu J, Zheng L, Chen MS, Zhuang SM. Vessels That Encapsulate Tumor Clusters (VETC) Pattern Is a Predictor of Sorafenib Benefit in Patients with Hepatocellular Carcinoma. Hepatology 2019;70:824-39. [Crossref] [PubMed]
13. Wang JH, Li XS, Tang HS, Fang RY, Song JJ, Feng YL, Guan TP, Ruan Q, Wang J, Cui SZ. Vessels that encapsulate tumor clusters (VETC) pattern predicts the efficacy of adjuvant TACE in hepatocellular carcinoma. J Cancer Res Clin Oncol 2023;149:4163-72. [Crossref] [PubMed]
14. Lin C, He Y, Liu M, Wu A, Zhang J, Li S, Li S, Cao Q, Liu F. Vessels That Encapsulate Tumor Clusters (VETC) Predict cTACE Response in Hepatocellular Carcinoma. J Hepatocell Carcinoma 2023;10:383-97. [Crossref] [PubMed]
15. Chen ZY, Guo ZX, Lu LH, Mei J, Lin WP, Li SH, Wei W, Guo RP. The predictive value of vessels encapsulating tumor clusters in treatment optimization for recurrent early-stage hepatocellular carcinoma. Cancer Med 2021;10:5466-74. [Crossref] [PubMed]
16. Li Z, Song W, Zhang J, Li Q, Song Z, Ren X, Wen Y, Li X, Yao H, Gao Y, Tang Z. Identification of vessels encapsulating tumor clusters in solitary hepatocellular carcinoma via imaging biomarkers in preoperative contrast-enhanced magnetic resonance imaging. Quant Imaging Med Surg 2024;14:8586-600. [Crossref] [PubMed]
17. Hu S, Zou Q, Shen Z, Xie Y, Li S, Jiao J, Zhang H, Cheng M, Zhang Y. Preliminary study on the ability of (18)F-fluorodeoxyglucose positron emission tomography/computed tomography radiomics to predict vessels that encapsulate tumor clusters and prognosis in hepatocellular carcinoma. Quant Imaging Med Surg 2025;15:6217-33. [Crossref] [PubMed]
18. Chu T, Zhao C, Zhang J, Duan K, Li M, Zhang T, Lv S, Liu H, Wei F. Application of a Convolutional Neural Network for Multitask Learning to Simultaneously Predict Microvascular Invasion and Vessels that Encapsulate Tumor Clusters in Hepatocellular Carcinoma. Ann Surg Oncol 2022;29:6774-83. [Crossref] [PubMed]
19. Dong X, Yang J, Zhang B, Li Y, Wang G, Chen J, Wei Y, Zhang H, Chen Q, Jin S, Wang L, He H, Gan M, Ji W. Deep Learning Radiomics Model of Dynamic Contrast-Enhanced MRI for Evaluating Vessels Encapsulating Tumor Clusters and Prognosis in Hepatocellular Carcinoma. J Magn Reson Imaging 2024;59:108-19. [Crossref] [PubMed]
20. Fan Y, Yu Y, Wang X, Hu M, Du M, Guo L, Sun S, Hu C. Texture Analysis Based on Gd-EOB-DTPA-Enhanced MRI for Identifying Vessels Encapsulating Tumor Clusters (VETC)-Positive Hepatocellular Carcinoma. J Hepatocell Carcinoma 2021;8:349-59. [Crossref] [PubMed]
21. Li C, Wen Y, Xie J, Chen Q, Dang Y, Zhang H, Guo H, Long L. Preoperative prediction of VETC in hepatocellular carcinoma using non-Gaussian diffusion-weighted imaging at high b values: a pilot study. Front Oncol 2023;13:1167209. [Crossref] [PubMed]
22. Pan J, Huang H, Zhang S, Zhu Y, Zhang Y, Wang M, Zhang C, Zhao YC, Chen F. Intraindividual comparison of CT and MRI for predicting vessels encapsulating tumor clusters in hepatocellular carcinoma. Eur Radiol 2025;35:61-72. [Crossref] [PubMed]
23. Qu Q, Liu Z, Lu M, Xu L, Zhang J, Liu M, Jiang J, Gu C, Ma Q, Huang A, Zhang X, Zhang T. Preoperative Gadoxetic Acid-Enhanced MRI Features for Evaluation of Vessels Encapsulating Tumor Clusters and Microvascular Invasion in Hepatocellular Carcinoma: Creating Nomograms for Risk Assessment. J Magn Reson Imaging 2024;60:1094-110. [Crossref] [PubMed]
24. Wang F, Numata K, Funaoka A, Liu X, Kumamoto T, Takeda K, Chuma M, Nozaki A, Ruan L, Maeda S. Establishment of nomogram prediction model of contrast-enhanced ultrasound and Gd-EOB-DTPA-enhanced MRI for vessels encapsulating tumor clusters pattern of hepatocellular carcinoma. Biosci Trends 2024;18:277-88. [Crossref] [PubMed]
25. Wang M, Cao L, Wang Y, Huang H, Cao S, Tian X, Lei J. Prediction of vessels encapsulating tumor clusters pattern and prognosis of hepatocellular carcinoma based on preoperative gadolinium-ethoxybenzyl-diethylenetriaminepentaacetic acid magnetic resonance imaging. J Gastrointest Surg 2024;28:442-50. [Crossref] [PubMed]
26. Wang Y, Wang M, Cao L, Huang H, Cao S, Tian X, Lei J. A nomogram for preoperative prediction of vessels encapsulating tumor clusters (VETC) pattern and prognosis of hepatocellular carcinoma. Am J Surg 2024;234:172-8. [Crossref] [PubMed]
27. Xu W, Zhang H, Zhang R, Zhong X, Li X, Zhou W, Xie X, Wang K, Xu M. Deep learning model based on contrast-enhanced ultrasound for predicting vessels encapsulating tumor clusters in hepatocellular carcinoma. Eur Radiol 2025;35:989-1000. [Crossref] [PubMed]
28. Yu Y, Fan Y, Wang X, Zhu M, Hu M, Shi C, Hu C. Gd-EOB-DTPA-enhanced MRI radiomics to predict vessels encapsulating tumor clusters (VETC) and patient prognosis in hepatocellular carcinoma. Eur Radiol 2022;32:959-70. [Crossref] [PubMed]
29. Fan Y, Yu Y, Hu M, Wang X, Du M, Guo L, Hu C. Imaging features based on Gd-EOB-DTPA-enhanced MRI for predicting vessels encapsulating tumor clusters (VETC) in patients with hepatocellular carcinoma. Br J Radiol 2021;94:20200950. [Crossref] [PubMed]
30. Chen FM, Du M, Qi X, Bian L, Wu D, Zhang SL, Wang J, Zhou Y, Zhu X. Nomogram Estimating Vessels Encapsulating Tumor Clusters in Hepatocellular Carcinoma From Preoperative Gadoxetate Disodium-Enhanced MRI. J Magn Reson Imaging 2023;57:1893-905. [Crossref] [PubMed]
31. Chen H, Dong H, He R, Gu M, Zhao X, Song K, Zou W, Jia N, Liu W. Optimizing predictions: improved performance of preoperative gadobenate-enhanced MRI hepatobiliary phase features in predicting vessels encapsulating tumor clusters in hepatocellular carcinoma-a multicenter study. Abdom Radiol (NY) 2024;49:3412-26. [Crossref] [PubMed]
32. Chen HL, He RL, Gu MT, Zhao XY, Song KR, Zou WJ, Jia NY, Liu WM. Nomogram prediction of vessels encapsulating tumor clusters in small hepatocellular carcinoma ≤ 3 cm based on enhanced magnetic resonance imaging. World J Gastrointest Oncol 2024;16:1808-20. [Crossref] [PubMed]
33. Feng Z, Li H, Zhao H, Jiang Y, Liu Q, Chen Q, Wang W, Rong P. Preoperative CT for Characterization of Aggressive Macrotrabecular-Massive Subtype and Vessels That Encapsulate Tumor Clusters Pattern in Hepatocellular Carcinoma. Radiology 2021;300:219-29. [Crossref] [PubMed]
34. Yang J, Dong X, Wang F, Jin S, Zhang B, Zhang H, Pan W, Gan M, Duan S, Zhang L, Hu H, Ji W. A deep learning model based on MRI for prediction of vessels encapsulating tumour clusters and prognosis in hepatocellular carcinoma. Abdom Radiol (NY) 2024;49:1074-83. [Crossref] [PubMed]
35. Zhang C, Zhong H, Zhao F, Ma ZY, Dai ZJ, Pang GD. Preoperatively predicting vessels encapsulating tumor clusters in hepatocellular carcinoma: Machine learning model based on contrast-enhanced computed tomography. World J Gastrointest Oncol 2024;16:857-74. [Crossref] [PubMed]
36. Bo Z, Song J, He Q, Chen B, Chen Z, Xie X, Shu D, Chen K, Wang Y, Chen G. Application of artificial intelligence radiomics in the diagnosis, treatment, and prognosis of hepatocellular carcinoma. Comput Biol Med 2024;173:108337. [Crossref] [PubMed]
37. Calderaro J, Seraphin TP, Luedde T, Simon TG. Artificial intelligence for the prevention and clinical management of hepatocellular carcinoma. J Hepatol 2022;76:1348-61. [Crossref] [PubMed]
38. Nam D, Chapiro J, Paradis V, Seraphin TP, Kather JN. Artificial intelligence in liver diseases: Improving diagnostics, prognostics and response prediction. JHEP Rep 2022;4:100443. [Crossref] [PubMed]
39. Renne SL, Woo HY, Allegra S, Rudini N, Yano H, Donadon M, Viganò L, Akiba J, Lee HS, Rhee H, Park YN, Roncalli M, Di Tommaso L. Vessels Encapsulating Tumor Clusters (VETC) Is a Powerful Predictor of Aggressive Hepatocellular Carcinoma. Hepatology 2020;71:183-95. [Crossref] [PubMed]
40. Vernuccio F, Gagliano DS, Cannella R, Ba-Ssalamah A, Tang A, Brancatelli G. Spectrum of liver lesions hyperintense on hepatobiliary phase: an approach by clinical setting. Insights Imaging 2021;12:8. [Crossref] [PubMed]
41. Hong SB, Choi SH, Kim SY, Shim JH, Lee SS, Byun JH, Park SH, Kim KW, Kim S, Lee NK. MRI Features for Predicting Microvascular Invasion of Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis. Liver Cancer 2021;10:94-106. [Crossref] [PubMed]
42. Lijmer JG, Bossuyt PM, Heisterkamp SH. Exploring sources of heterogeneity in systematic reviews of diagnostic tests. Stat Med 2002;21:1525-37. [Crossref] [PubMed]
43. Park J, Lee JM, Kim TH, Yoon JH. Imaging diagnosis of hepatocellular carcinoma: Future directions with special emphasis on hepatobiliary magnetic resonance imaging and contrast-enhanced ultrasound. Clin Mol Hepatol 2022;28:362-79. [Crossref] [PubMed]