An exploratory nomogram model with ¹⁸F-FDG PET/CT metabolic parameters for predicting high-risk status of microvascular invasion in hepatocellular carcinoma patients prior to liver transplantation

Introduction

Hepatocellular carcinoma (HCC) is a malignancy with global prevalence and the third leading cause of cancer-related mortality (1). HCC is strongly influenced by microvascular invasion (MVI), a key determinant of postoperative recurrence and prognosis (2,3). MVI indicates that HCC cells have invaded the microvasculature. Detached tumor cells may disseminate extrahepatically via the bloodstream prior to liver transplantation. In the post-transplant immunosuppressed state, these disseminated foci are prone to proliferate, potentially leading to tumor recurrence and significantly compromising patient survival (4,5). HCC patients with high-risk MVI face poorer outcomes and often require intensified adjuvant therapy (3,4).

Preoperative MVI evaluation has been studied in HCC patients undergoing hepatectomy and fluorine-18-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) has shown its potential as a non-invasive method for screening high-risk MVI prior to or after hepatectomy and the postoperative recurrence of HCC (6-8). Recent studies have also shown the capability of ¹⁸F-FDG PET/CT in predicting pathologic grade, MVI, and HCC recurrence after liver transplantation (9,10).

Liver transplantation remains one of the most effective treatments for HCC, yet the 5-year survival rate post-liver transplantation is <15% (11). In addition, HCC patients with high-risk MVI need to be treated with an accurate therapy treatment plan. Accurate preoperative identification of high-risk MVI in liver transplantation candidates can help improve survival outcomes. Liver transplantation candidates with HCC are more difficult to collect and have more severe conditions and higher requirements for region of interest (ROI) delineation and assessment than patients who undergo hepatectomy. While prior research has focused on MVI prediction for risk stratification of hepatectomy (7-10), strict high-risk MVI definition has not been determined. There are still difficulties in dividing patients into an MVI high-risk and non-high-risk group based on the pathology findings.

This study analyzed the associations between preoperative ¹⁸F-FDG PET/CT parameters (such as TMRmax and tumor diameter), clinical variables, and strict high-risk MVI in liver transplant candidates, aiming to develop an exploratory model to guide individualized treatment strategies for HCC patients. We present this article in accordance with the TRIPOD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1029/rc).

Methods

Participants

A retrospective analysis was performed on the clinical data of HCC patients who underwent liver transplantation at the Third Hospital of Hebei Medical University from November 2021 to November 2024. The patients had undergone ¹⁸F-FDG PET/CT preoperatively. Patient characteristics, including Child-Pugh classification and other relevant clinical data, were summarized and recorded. This study was approved by the Ethics Committee of the Third Hospital of Hebei Medical University (Ethics approval No. W2024-039-1) in compliance with the Declaration of Helsinki and its subsequent amendments. All participants (both donors and recipients) have signed written informed consent for this study. The organs were not procured from prisoners. The organs were achieved from China Organ Transplant Response System.

The inclusion criteria were as follows: (I) HCC confirmed by postoperative pathology; (II) undergoing liver transplantation; and (III) completion of ¹⁸F-FDG PET/CT within 1 month prior to surgery. The exclusion criteria were as follows: (I) concomitant history of other malignancies; (II) receipt of specific interventions between PET/CT and liver transplantation, including chemotherapy, radiotherapy, targeted therapy, or other systemic therapies; (III) active infections or uncontrolled hematologic disorders within the preceding 3 months; and (IV) incomplete clinical medical records.

The image measurement and analysis were completed by the same two skilled radiologists with >10 years of clinical experience each. The radiologists were blinded to the research.

PET/CT imaging

Imaging was performed using a Philips Vereos PET/CT scanner (Philips Healthcare, Best, The Netherlands) with ¹⁸F-FDG (Atomic High-Tech Co., Ltd., Shijiazhuang, China; radiochemical purity >95%) by the two abovementioned radiologists. Patients fasted for 6–8 h before intravenous administration of ¹⁸F-FDG (0.10–0.12 mCi/kg). Whole-body PET/CT (skull-to-mid-thigh) commenced 60 min post-injection. The CT parameters were as follows: 120 kV; 180 mAs; 3.0 mm slice thickness/interval; and a 512×512 matrix. PET data were acquired in 3D mode (7–8 bed positions, 2 min/bed). Iterative CT-based attenuation correction and reconstruction generated fused axial, coronal, and sagittal images.

In this study, all patients followed the standard clinical PET/CT examination protocol: at least 6 hours of fasting was required before the scan, and blood glucose levels were routinely measured before the injection of FDG. Our inclusion criteria required a blood glucose level of ≤11.1 mmol/L (200 mg/dL), and no patients were excluded or re-assigned due to high blood glucose.

Image analysis

The ROI for HCC lesions on the Philips IntelliSpace Portal workstation was delineated on PET/CT images. The maximum standardized uptake value (SUVmax) of the tumor lesions was recorded. Using 40% of the tumor SUVmax as the threshold, the mean standardized uptake value (SUVmean), and metabolic and tumor volume (MTV) were measured. The total lesion glycolysis (TLG) was calculated using the following formula: TLG = MTV × SUVmean. In addition, the SUVmax and SUVmean of a 1-cm diameter region at the level of the aortic arch in the mediastinum (measured at the center of the aorta and avoiding the aortic wall) were recorded. The lesion SUVmax-to-mediastinal SUVmax ration was calculated and denoted as the TMRmax, while the lesion SUVmean-to-mediastinal SUVmean ration was denoted as the TMRmean. Furthermore, the SUVmax and SUVmean of normal liver tissue were measured and recorded using a 1-cm diameter ROI. The lesion SUVmax-to-normal liver tissue SUVmax ratio was calculated and denoted as the TLRmax and the lesion SUVmean-to-normal liver tissue SUVmean ratio was denoted as the TLRmean.

During the PET/CT scan, all ROI of the lesions were manually delineated and confirmed under the direct guidance and fusion of the same-machine enhanced CT or MRI images, rather than solely relying on threshold segmentation. This was performed to ensure anatomical accuracy even when the contrast was low. At the same time, we conducted strict internal quality control, where two physicians independently delineated the key parameters (such as TMRmax) and calculated the intra-group correlation coefficient (ICC =0.82), indicating good consistency.

Tumor number and diameter

Tumor diameter was measured based on cross-referenced PET/CT and contrast-enhanced CT (CE-CT) or MRI findings. The larger measurements from CE-CT or MRI were prioritized for cases in which the PET/CT-derived diameter was smaller than the diameter from CE-CT or MRI. Tumor number and long-axis diameter were measured and recorded. Tumor number was categorized using a binary classification method with ≥3 lesions defined as multiple lesions. The sum of the longest diameters of the three largest lesions was recorded as the maximum tumor long-axis diameter for multiple lesions.

Laboratory parameters

Laboratory parameters were retrospectively selected from tests performed 1 d prior to surgery, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), the AST/ALT ratio, alkaline phosphatase (ALP), and gamma-glutamyl transferase (GGT). The most recent preoperative levels of alpha-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist-II (PIVKA-II) were recorded with the initial AFP and PIVKA-II levels obtained during admission when HCC was suspected. The difference values and ratios between the initial and final measurements of AFP and PIVKA-II were calculated.

Pathologic assessment of MVI status

Two attending pathologists independently assessed MVI using hematoxylin-eosin staining and immunohistochemically analyzed specimens. Both pathologists were blinded to patients’ clinical data and ancillary test results before evaluation. Due to the small sample size of the retrospective study, the group M1 consisted of only 15 cases. To avoid statistical bias, we adopted a dichotomous approach of “non-high-risk group (M0 + M1)” and “high-risk group (M2)”. Lesions were categorized as follows: (I) in the non-high-risk group, the postoperative pathologic examination revealed no MVI or low-risk MVI, which was defined as the presence of ≤5 carcinoma cell clusters involving endothelial cells observable under the microscope; (II) in the high-risk group, the postoperative pathologic examination revealed high-risk MVI, which was defined as the presence of >5 carcinoma cell clusters involving endothelial cells observable under the microscope.

Model construction

A least absolute shrinkage and selection operator (LASSO)-optimized logistic regression model incorporating TMRmax and maximum tumor diameter was developed to predict high-risk MVI in HCC. A nomogram was constructed to provide a more intuitive representation of the exploratory model.

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics 26.0 and R 4.3.2 software. The “glmnet” package in R was applied for LASSO regression analysis and the “rms” package was used for constructing nomograms and calibration curves. Statistical significance was defined as a P value <0.05. Normally distributed continuous variables are presented as the means ± standard deviation and were compared between groups using an independent sample t-test. Non-normally distributed continuous variables are presented as median (Q1, Q3) and were compared between groups using the Mann-Whitney U test. Categorical variables were compared between groups using the Chi-squared (χ²) test. Factors that demonstrated significant differences between the groups were then incorporated into a binary logistic regression model for univariate analysis, LASSO regression, and multivariate analysis to construct the predictive model and generate an exploratory nomogram based on this model. The exploratory model was assessed using the C-index, ROC curve, and calibration curve.

Results

Patient characteristics

During the study period, 134 patients were screened, and 55 cases were excluded for the following reasons: (I) missing key clinical or pathological data (n=12 cases); (II) poor image quality of PET/CT making analysis impossible (n=18 cases); (III) received neoadjuvant therapy before transplantation (n=25 cases). Therefore, a total of 79 patients (70 males and 9 females) were included in this study. The age range for these patients was 31–71 years with a median age of 53 years. All patients recovered well after liver transplantation with no significant adverse reactions observed. The statistical results of routinely measuring blood glucose levels before the FDG injection were (5.8±1.3 mmol/L).

Based on the postoperative histopathologic analysis, 27 cases had no endothelial tumor clusters under microscopic examination and 15 exhibited ≤5 endothelial tumor clusters, and were categorized as the non-high-risk MVI group. The remaining 37 cases demonstrated >5 endothelial tumor clusters and were categorized as the high-risk MVI group. The distribution of TMRmax was 2.1 (1.5–3.4), the TMRmean was 1.7 (1.3–2.6), the TLRmax was 1.3 (1.0–2.1), and the TLRmean was 1.3 (1.0–1.8). The interval between the initial and final AFP measurements was 42 [20–121] d and the interval for PIVKA-II was 42 [18–118] d. The baseline clinical characteristics (except the tumor number) of the high-risk and non-high-risk groups were comparable, as summarized in Table 1 (P>0.05).

Univariate analysis of predictive factors for the MVI high-risk group in HCC patients

Univariate analysis showed that SUVmax, SUVmean, TMRmax, TMRmean, maximum tumor diameter, TLG, AST/ALT ratio, GGT, initial AFP, final AFP, initial PIVKA-II, final PIVKA-II, and the number of lesions were statistically significant predictors of MVI (P<0.05). However, the other factors were not statistically significant predictors of MVI (Table 2). LASSO analysis and 10-fold cross-validation were performed to exclude relatively less important variables to select variables and avoid overfitting the model (Figure 1). The results showed that with an increasing regularization parameter (λ value), only TMRmax and maximum tumor diameter coefficients remained non-zero, while all other parameters were excluded (Figures 1,2). Further exploratory analysis of the optimal cut-off values for final AFP and PIVKA-II showed that the best cut-off value for the final AFP was 10.4 ng/mL with a specificity of 73.81% and a sensitivity of 72.97%. The best cut-off value for the final PIVKA-II under the exploratory analysis was 45.6 mAU/mL with a specificity of 47.62% and a sensitivity of 91.89%.

Figure 1 The LASSO coefficient path diagram of 13 risk factors. LASSO, least absolute shrinkage and selection operator.

Figure 2 The 10-fold cross-validation curve.

Multivariate analysis of predictive factors for the MVI high-risk group in HCC patients

Variables selected by univariate analysis with statistical significance and filtered through LASSO regression (TMRmax and maximum tumor diameter) were included in multivariate binary logistic regression analysis. TMRmax [odds ratio (OR) =1.757, 95% confidence interval (CI): 1.124–2.746; P=0.01] and maximum tumor diameter (OR = 1.247, 95% CI: 1.067–1.456; P=0.005) were independent factors for predicting MVI (Table 2). For every unit increase in TMRmax, the risk of MVI expression increased by 1.757 times. For every 1 cm increase in maximum tumor diameter, the risk of MVI expression increased by 1.247 times, indicating that TMRmax and maximum tumor diameter have significant predictive value for MVI expression. An ROC curve analysis of TMRmax and maximum tumor diameter was performed (Figure 3). The exploratory analysis demonstrated an area under the ROC curve (AUC) of 0.716 for TMRmax with an optimal cut-off value of 2.1 (specificity, 64.29%; sensitivity, 70.27%). The AUC was 0.729 at a cut-off threshold of 8.3 cm (specificity, 90.48%; sensitivity, 51.35%) for maximum tumor diameter.

Figure 3 Factors and model prediction of high-risk MVI ROC curve. MVI, microvascular invasion; ROC, receiver operating characteristic; SUVmax, maximum standardized uptake value; TMRmax, lesion SUVmax-to-mediastinal SUVmax ratio.

Model construction

A LASSO-optimized logistic regression model incorporating TMRmax and maximum tumor diameter was developed to predict high-risk MVI in HCC. Internal validation has been performed via 1,000 bootstrap resamples. In each bootstrap repetition, the entire process from LASSO variable selection to final model fitting is fully repeated. The average optimistic bias calculated based on 1,000 bootstrap samples has been subtracted from the initial C-index to obtain this correction value. The exploratory model demonstrated a good performance with a C-index: 0.779 (95% CI: 0.672–0.865) with 90.48% specificity and 59.46% sensitivity, after the internal validation of bootstrap method. The calibration performance of the model was also evaluated through the bootstrap method. Calibration plots showed excellent agreement between predicted and observed probabilities with the bias-corrected curve closely aligned with the ideal line (Figure 4). The logistic equation is as follows:

logit(p)=−2.890+0.564×TMRmax+0.220×maximumtumordiameter[1]

Figure 4 Calibration curve of the predicted high-risk MVI column chart for HCC patients. HCC, hepatocellular carcinoma; MVI, microvascular invasion.

An exploratory nomogram was constructed to provide a more intuitive representation of the predictive model (Figure 5).

Figure 5 Nomogram predicting high-risk MVI status in HCC patients. d, maximum tumor diameter; HCC, hepatocellular carcinoma; MVI, microvascular invasion; SUVmax, maximum standardized uptake value; TMRmax, lesion SUVmax-to-mediastinal SUVmax ratio.

Representative ¹⁸F-FDG PET/CT images illustrating MVI status in HCC patients are shown in Figures 6,7.

Figure 6 A 51-year-old male patient with HCC. PET/CT imaging was performed 24 d prior to liver transplantation. A hypermetabolic focus was identified at the junction between liver segments S4/S8 (marked by the red arrows) with an SUVmax of 6.4, SUVmean of 3.6, TMRmax of 4.9, TMRmean of 3.7, and a long axis diameter of 2.9 cm. Using the model formula, the patient’s MVI risk was classified as high (0.5116= −2.890 + 0.564×4.9 + 0.220×2.9). Postoperative histopathologic analysis confirmed the diagnosis of hepatocellular carcinoma in the left liver lobe with MVI grading of M2, indicating a high-risk group. The model predicted high-risk MVI status aligns with the pathologic findings. HCC, hepatocellular carcinoma; MVI, microvascular invasion; PET/CT, positron emission tomography/computed tomography; SUVmax, maximum standardized uptake value; SUVmean, the mean standardized uptake value; TMRmax, lesion SUVmax-to-mediastinal SUVmax ratio; TMRmean, lesion SUVmean-to-mediastinal SUVmean ratio.

Figure 7 A 60-year-old male patient with HCC. PET/CT was performed 20 d prior to liver transplantation. A mildly hypermetabolic lesion in the liver segment V (indicated by the red arrows) was observed with an SUVmax of 3.1, an SUVmean of 2.1, a TMRmax of 1.8, a TMRmean of 1.4, and a longest diameter of 3.3 cm. Based on the model formula, the patient’s MVI status was predicted to be non-high-risk (−1.3744= −2.890 + 0.564×1.4 + 0.220×3.3). Postoperative histopathologic examination revealed well-differentiated HCC in the right liver lobe with an MVI grade of M0, corresponding to the non-high-risk group. The model prediction aligned with the histopathologic findings. HCC, hepatocellular carcinoma; MVI, microvascular invasion; PET/CT, positron emission tomography/computed tomography; SUVmax, maximum standardized uptake value; SUVmean, the mean standardized uptake value; TMRmax, lesion SUVmax-to-mediastinal SUVmax ratio; TMRmean, lesion SUVmean-to-mediastinal SUVmean ratio.

Discussion

The current study showed that TMRmax and maximum tumor diameter were independent factors for predicting MVI. ROC curve analysis demonstrated an AUC of 0.716 for TMRmax with an exploratory optimal cut-off value of 2.1 and an AUC of 0.729 for maximum tumor diameter with an exploratory cut-off threshold of 8.3 cm. The LASSO-optimized logistic regression model incorporating TMRmax and maximum tumor diameter demonstrated a moderate good C-index of 0.779 (95% CI: 0.672–0.865), with a specificity of 90.48% and a sensitivity of 59.46%.

While SUVmax had limited MVI predictive reproducibility due to PET/CT protocol variability, lesion-to-reference ratios showed superior clinical utility. Previous studies established T-SUVmax/L-SUVmean thresholds (≥1.2, 64.3% sensitivity/86.7% specificity; ≥1.69 correlating with vascular invasion; P=0.04) (12,13). Our analysis revealed that a TMRmax ≥2.1 greatly related with high-risk MVI (C-index =0.716), whereas TLR metrics lacked significance. Discrepancies originated from prior cohorts, as follows (12,13): (I) advanced HCC staging (larger tumors and severe cirrhosis) in transplant candidates versus prior cohorts; and (II) cirrhosis-induced heterogeneous tracer distribution complicating metabolic quantification. TMRmax outperforms absolute SUV parameters by circumventing cirrhotic reference tissue limitations, establishing that TMRmax may act as a useful biomarker for preoperative MVI risk stratification in decompensated HCC patients.

TMRmax was highly associated with high-risk MVI with an exploratory cut-off threshold of 8.3 cm, which demonstrated the tumor diameter was useful for predicting the prognosis of HCC as previously reported (14). A tumor diameter ≥8.3 cm may be a key factor for screening high-risk MVI, potentially mediated by tumor vascularization and immune microenvironment remodeling. The elevated threshold versus prior studies reflects the following: (I) stringent high-risk MVI criteria beyond mere presence/absence assessment; (II) inclusion of transplant candidates with larger tumor burdens; and (III) multifocal measurement protocols (summing the three largest foci). This methodologic distinction aligns with advanced HCC pathobiology in cirrhotic livers.

Jiang et al. (15) identified MTV/TLG thresholds (109.45 mL/932.6 g) predicting MVI in transplant recipients (P=0.003/0.001). Another study showed that AFP, number of tumors, CT Dmax, and tumor-to-normal liver uptake ratio (TLR) were all predictors of MVI, which can combine a nomogram with a close agreement between predicted and actual MVI probabilities and an AUC of 0.965 (15). While TLG showed marginal predictive value for high-risk MVI in our cohort, MTV lacked significance, suggesting metabolic activity (TLG) outweighs pure volumetry (MTV). Current metabolic volume parameters demonstrate limited consensus for high-risk MVI prediction in HCC, necessitating multicenter validation to establish standardized criteria.

Although AFP/PIVKA-II demonstrated MVI predictive potential, the thresholds remained inconsistent. Mao et al. (16) reported that an AFP >400 ng/mL is an independent MVI risk factor, while Herrero et al. (17) established AFP-based tumor scoring models. Another study showed that high AFP levels and active tumor metabolism can predict MVI occurrence in HCC patients (18). Our univariate analysis showed transient AFP/PIVKA-II significance (initial/final measurements) but excluded post-LASSO regression. Ratio/difference analyses revealed no significance, potentially due to the following: (I) cohort limitations (non-elevated biomarkers in 38% patients); (II) temporal assay variability; and (III) restricted sample size (n=79). These findings suggested serum biomarkers require standardized temporal sampling protocols and larger multi-center validation to determine clinical utility in preoperative MVI risk stratification.

There are some limitations in this study. First, this was a single-center retrospective study with a limited cohort (n=79) that risks sampling bias and precludes MVI sub-stratification. Second, some other clinical parameters, like CT Dmax and tumor staging grades, were not included. Third, the study merges MVI-negative and low-risk MVI into a single “non-high-risk” group due to limited sample size, which may lead to some bias to the results. Fourth, the exploratory nomogram model hasn’t been validated in an external dataset. Fifth, the current study lacks the data of the recurrence and/or survival rate. Sixth, the authors used LASSO for variable selection and then reported odds ratios, confidence intervals, and P values from a standard multivariable logistic regression using the selected predictors. This “select-then-infer” approach does not account for uncertainty introduced by the variable selection step and can lead to overly optimistic estimates and potentially misleading inference. Prospective multi-center validation with expanded samples is required to mitigate biomarker temporal variability, enable MVI sub-stratification (between MVI-negative, low-risk MVI and high-risk MVI groups), and optimize clinical applicability. The post-transplant recurrence and/or survival rate should also be collected in further studies. Strict statistical analysis should be performed to avoid statistical bias. These steps will address current limitations in HCC metabolic parameter standardization for preoperative risk assessment.

Conclusions

Preoperative ¹⁸F-FDG PET/CT demonstrated clinical utility in preoperative high-risk MVI screening for HCC liver transplant candidates. TMRmax (optimal cut-off, 2.1) and tumor diameter (≥8.3 cm) may be closely associated with high-risk MVI. The exploratory nomogram model integrating TMRmax and tumor diameter achieved a moderate good accuracy with a C-index of 0.779, which can help on high-risk MVI assessment. This tool may facilitate refined preoperative risk stratification for the HCC liver transplant candidates.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1029/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1029/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1029/prf

Funding: This study was supported by the Hebei Provincial Health Commission Research Project (No. 20240288).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1029/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. This study was approved by the Ethics Committee of the Third Hospital of Hebei Medical University (Ethics approval No. W2024-039-1) in compliance with the Declaration of Helsinki and its subsequent amendments. All participants (both donors and recipients) have signed written informed consent for this study.

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, et al. 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. Sun J, Xia Y, Shen F, et al. Chinese expert consensus on the diagnosis and treatment of hepatocellular carcinoma with microvascular invasion (2024 edition). Hepatobiliary Surg Nutr 2025;14:246-66. [Crossref] [PubMed]
3. Li Z, Xu L, Zhu S, et al. Current Advances in Classification, Prediction and Management of Microvascular Invasion in Hepatocellular Carcinoma. J Cell Mol Med 2025;29:e70746. [Crossref] [PubMed]
4. Lu JP, Feng JK, Zhao Y, et al. Grading risk of microvascular invasion impacts survival in hepatocellular carcinoma patients undergoing adjuvant transarterial chemoembolization: A multicenter study. Eur J Surg Oncol 2025;51:110102. [Crossref] [PubMed]
5. Yang S, Ni H, Zhang A, et al. Grading severity of MVI impacts long-term outcomes after laparoscopic liver resection for early-stage hepatocellular carcinoma: A multicenter study. Am J Surg 2024;238:115988. [Crossref] [PubMed]
6. Xiang C, Shen X, Zeng X, et al. Effect of transarterial chemoembolization as postoperative adjuvant therapy for intermediate-stage hepatocellular carcinoma with microvascular invasion: a multicenter cohort study. Int J Surg 2024;110:315-23. [Crossref] [PubMed]
7. Lee SW, Jeong SY, Kim SJ. Diagnostic performance of FDG PET/CT radiomics in predicting microvascular invasion in hepatocellular carcinoma compared to conventional metabolic parameters: a systematic review and meta-analysis. Ann Nucl Med 2025;39:1146-56. [Crossref] [PubMed]
8. Jiang C, Ma G, Liu Q, et al. The value of preoperative 18F-FDG PET metabolic and volumetric parameters in predicting microvascular invasion and postoperative recurrence of hepatocellular carcinoma. Nucl Med Commun 2022;43:100-7. [Crossref] [PubMed]
9. Li Y, Liu J, Lu D, et al. The predictive value of (18)F-FDG PET/CT metabolic parameters and radiomics characteristics for microvascular invasion of hepatocellular carcinoma. Radiography (Lond) 2025;31:103263. [Crossref] [PubMed]
10. Wu F, Cao G, Lu J, et al. Correlation between 18 F-FDG PET/CT metabolic parameters and microvascular invasion before liver transplantation in patients with hepatocellular carcinoma. Nucl Med Commun 2024;45:1033-8. [Crossref] [PubMed]
11. Erstad DJ, Tanabe KK. Prognostic and Therapeutic Implications of Microvascular Invasion in Hepatocellular Carcinoma. Ann Surg Oncol 2019;26:1474-93. [Crossref] [PubMed]
12. Ahn SY, Lee JM, Joo I, et al. Prediction of microvascular invasion of hepatocellular carcinoma using gadoxetic acid-enhanced MR and (18)F-FDG PET/CT. Abdom Imaging 2015;40:843-51. [Crossref] [PubMed]
13. Lin CY, Liao CW, Chu LY, et al. Predictive Value of 18F-FDG PET/CT for Vascular Invasion in Patients With Hepatocellular Carcinoma Before Liver Transplantation. Clin Nucl Med 2017;42:e183-7. [Crossref] [PubMed]
14. Wang Y, Luo S, Jin G, et al. Preoperative clinical-radiomics nomogram for microvascular invasion prediction in hepatocellular carcinoma using 18F-FDG PET/CT. BMC Med Imaging 2022;22:70. [Crossref] [PubMed]
15. Jiang S, Gao X, Tian Y, et al. The potential of (18)F-FDG PET/CT metabolic parameter-based nomogram in predicting the microvascular invasion of hepatocellular carcinoma before liver transplantation. Abdom Radiol (NY) 2024;49:1444-55. [Crossref] [PubMed]
16. Mao S, Yu X, Yang Y, et al. Preoperative nomogram for microvascular invasion prediction based on clinical database in hepatocellular carcinoma. Sci Rep 2021;11:13999. [Crossref] [PubMed]
17. Herrero A, Boivineau L, Cassese G, et al. Progression of AFP SCORE is a Preoperative Predictive Factor of Microvascular Invasion in Selected Patients Meeting Liver Transplantation Criteria for Hepatocellular Carcinoma. Transpl Int 2022;35:10412. [Crossref] [PubMed]
18. Wang T, Chen X, Huang H, et al. Early prediction of microvascular invasion (MVI) occurrence in hepatocellular carcinoma (HCC) by (18)F-FDG PET/CT and laboratory data. Eur J Med Res 2024;29:395. [Crossref] [PubMed]