Research progresses of imaging studies on preoperative prediction of microvascular invasion of hepatocellular carcinoma

Abstract

Hepatocellular carcinoma (HCC) is the predominant form of primary liver cancer, accounting for approximately 90% of liver cancer cases. It currently ranks as the fifth most prevalent cancer worldwide and represents the third leading cause of cancer-related mortality. As a malignant disease with surgical resection and ablative therapy being the sole curative options available, it is disheartening that most HCC patients who undergo liver resection experience relapse within five years. Microvascular invasion (MVI), defined as the presence of micrometastatic HCC emboli within liver vessels, serves as an important histopathological feature and indicative factor for both disease-free survival and overall survival in HCC patients. Therefore, achieving accurate preoperative noninvasive prediction of MVI holds vital significance in selecting appropriate clinical treatments and improving patient prognosis. Currently, there are no universally recognized criteria for preoperative diagnosis of MVI in clinical practice. Consequently, extensive research efforts have been directed towards preoperative imaging prediction of MVI to address this problem and the relative research progresses were reviewed in this article to summarize its current limitations and future research prospects.

Keywords

Microvascular invasion hepatocellular carcinoma radiomics texture analysis diagnostic imaging liver

1 Introduction

According to the latest global cancer burden data for 2020 published by the WHO International Agency for Research on Cancer, primary liver cancer accounts for 19.33% of all malignancies and ranks as the sixth most common cancer and the second leading cause of cancer-related deaths [1], with approximately 840,000 new cases and 780,000 deaths every year [2], posing a significant public health challenge. More than 40% of new cases of HCC have occurred in China [3]. The survival rate for HCC within 5 years stands at around 10–20% [4], with a median overall survival time worldwide ranging from only about 20–30 months [5]. One major contributing factor to this low survival rate is the high incidence of tumor recurrence following radical hepatectomy – reaching up to 35% after transplantation and even up to 70% after resection at five years.

At present, a large number of studies have shown that the presence of MVI serves as an independent prognostic factor associated with early recurrence and poor survival in HCC patients who undergo resection surgery [6 –8]. Therefore, accurate prediction of whether MVI is present can assist clinicians choose appropriate surgical programs for patients, such as wide resection margins or anatomic resection [9 –11], which could potentially reduce tumor recurrence rates and improve prognosis [12]. Unfortunately, MVI can only be reliably determined through histopathological examination of surgical specimens obtained during hepatectomy or liver transplantation – equivalent to postoperative diagnosis without providing preoperative guidance in clinical settings. Consequently, there exists an urgent need to investigate effective strategies for predicting MVI presence beforehand which holds immense significance in developing personalized treatment plans for HCC patients’ benefitting purposes.

Numerous reports [13 –15] have attempted to explore the correlation between preoperative serological indicators and MVI. Such as serum alpha-fetoprotein (AFP), des-gamma-carboxyprothrombin (DCP), abnormal prothrombin (PIVKA-II), plasma free micro RNA and others have been widely used in the auxiliary diagnosis and prognosis evaluation of HCC, and its clinical value has gradually been recognized. However, how to determine the critical values still requires further research, and some special biological factors may also have abnormal levels in non-tumor patients with a background of chronic hepatitis or in patients with other tumors, which limits its application value. Hence, there is an unmet clinical need to preoperatively evaluate MVI. Medical imaging evaluation plays an irreplaceable role and can provide valuable information (e.g., position, size, and clinical stage of tumors). But various research based on preoperative images to evaluate MVI showed no consensus. In this study, we aimed to compare the preoperative prediction ability of different imaging methods for MVI to meet the urgent need for individualized diagnosis and treatment of differential MVI status in clinical practice.

2 Ultrasound

Ultrasound (US) has the advantages of real-time, non-invasive, non-radiation, and low price which can be used for early screening of HCC. Research [16] found that the irregular shape, unclear boundary, and incomplete capsule of tumors are independent risk factors for predicting MVI. However, traditional ultrasound diagnosis lacks sensitivity and specificity and it needs to be combined with other imaging examinations to improve the prediction performance of MVI. MVI can lead to microcirculatory hemodynamic changes. Contrast-enhanced ultrasound (CEUS) can show the hemodynamic changes of liver tumors and has advantages in evaluating microvascular perfusion of HCC. Previously, a study [17] showed that the mosaic architecture and intratumoral feeding arteries of CEUS are significant for early prognosis, and are an independent risk factors for predicting MVI. Zhu et al. [18] found that 86.7% of MVI patients showed rapid arterial phase washout during CEUS. In combination with tumor number (≥2) and tumor size (≥5 cm), CEUS washout rate can help identify MVI. Someone [19] evaluated the role of quantitative perfusion analysis of 3D CEUS in detecting MVI and showed that tumors with MVI present different hemodynamic parameters, with a larger parameters peak intensity than tumors without MVI. Other scholars [20] analyzed five quantitative perfusion parameters through TIC curves, and the results revealed that the wash-in area under the curve and wash-out area under the curve were higher in MVI-positive group than in MVI-negative group, the use of these techniques allows for the estimation of blood flow parameters of liver cancer lesions. Currently, the evaluation indicators of ultrasonic examination in MVI assessment are controversial which may be due to differences in machine equipment and operator experience levels. Therefore, in-depth research based on standardized collection and establishment of an ultrasound image database is the future development direction.

3 CT

CT has a wide range of clinical applications and plays an important role in the diagnosis of HCC due to its fast scanning speed and high image resolution. In recent years, several studies have reported that the morphological features of CT images are closely correlated with MVI, including tumor size, irregular tumor edge, peritumoral capsule, and so on, but the conclusions of different studies are not consistent [21 –24]. Compared with qualitative diagnosis, quantitative analysis is more objective and it is not easily affected by image quality and observer experience. Wu et al. [25] used a series of quantitative perfusion parameters to identify MVI, a conclusion has been drawn that the tumor PVF (portal vein flow), ΔPVF (difference in PVF between tumor and liver tissue), and the rPVF (ΔPVF/liver PVF ratio) were significantly higher in sHCC with MVI. However, the sample size of this study is small, and different inspection equipment or perfusion schemes may cause differences among individuals, which requires large samples and multi-center verification. Energy Spectrum CT increases the energy and chemical resolution based on retaining the original space, time, and density resolution of conventional CT, which provides more comprehensive image data support for the clinic. Yang et al. [26] found that the iodine concentration (IC) and NIC in the HCC MVI-positive group were significantly higher than those in the MVI-negative group. The ROC curve showed that arterial phase NIC was the most effective in the diagnosis of MVI, and provided more quantitative parameters for MVI prediction than conventional CT. A consistent conclusion is drawn from the study of KIM [27], and it is added that there is a moderate positive correlation between microvessel density (MVD) and NIC-a [28], which indicates the quantitative parameters of iodine in energy spectrum CT reflect angiogenesis and microcirculation perfusion during MVI in HCC, which is expected to be a new method for MVI evaluation. However, the quantitative analysis of iodine concentration in energy spectrum CT to evaluate whether MVI can be applied to other dual-source and dual-energy CT needs to be further discussed.

4 MRI

Magnetic Resonance Imaging (MRI) can display well soft tissues and blood vessels with high resolution because of its unique imaging principle, it has made breakthrough progress in predicting whether MVI in HCC before surgery.

4.1 Dynamic contrast enhanced-MRI, DCE-MRI

The non-invasive dynamic contrast-enhanced MRI (DCE-MRI) method provides valuable insights into tissue perfusion and vascularity. A meta-analysis [29] on DCE-MRI screened 7 MRI features from 235 articles significantly correlated with MVI, among which peritumoral hypointensity on HBP was the most indicative of MVI. The relationship between this feature and MVI can be explained by peritumoral perfusion change resulting from the dysfunction of organic anion-transporting polypeptide transporters in the hepatocytes around the liver cancer [30]. In this regard, Wu et al. [31] also reached a similar conclusion. And Chen et al. [32] found that in patients without peritumoral HBP hypointensity, a radiological capsule is helping in identifying MVI. However, it is less sensitive to rely solely on imaging features to predict MVI. Wang et al. [33] retrospectively analyzed the signal intensity (SI) of normal liver tissue and tumor parenchyma in 113 HCC patients based on Gd-EOB-DTPA enhanced MRI in the arterial and hepatobiliary phases, it is concluded that SI ratio of peritumoral tissue to normal liver in arterial phase (SIAp/Al) is a potential diagnosis marker for MVI, and the AUC value of SIAp/Al is higher than that of peritumoral hypointensity on hepatobiliary phase imaging, which indicates that quantitative analysis and evaluation of MVI may be more convincing. In summary, DCE-MRI can provide a variety of means for the evaluation of MVI and hepatobiliary specific contrast agent have significant advantages in preoperative prediction of MVI, provides more image details than CT.

4.2 Diffusion weighted imaging, DWI

DWI is a functional MRI technique which reflects the degree of diffusivity of water molecules, and its quantitative index apparent diffusion coefficient (ADC) is gradually used in the diagnosis of tumor microvascular lesions which can reflect the number of tumor cells, proliferative activity and capillary perfusion. Zhang et al. [34] found that when the b-value was 1000 s/mm², the ADC value < 1.07×10^–3mm²/s, could be regarded as an independent risk factor for MVI. Another study [35] showed that lower ADC values (<1.227×10^–3 mm²/s) on DWI with b-value of 0.500 s/mm² can be a useful preoperative predictor of MVI for small HCCs. In the case of different b-values, the ADC truncation values are different, so the standardization of parameters is necessary. Different from previous studies on single lesion, Yang et al. [36] retrospectively analyzed 51 patients with bifocal HCC, the results showed that highly similar ADC values for the two HCC lesions can effectively predict MVI in patients with two HCC lesions. However, this hypothesis needs to be verified by whole-genome sequencing before it is widely used in clinic. Using DWI to evaluate MVI has a broad development prospect, but it still has some limitations at present, which needs external verification by big data.

4.3 Intravoxel incoherent motion, IVIM

Intravoxel Incoherent Motion (IVIM) does not require contrast agents and it estimates perfusion in tissues by applying a biexponential signal equation model [37]. Wei et al. [38] prospectively evaluated the potential role of diffusion MRI for preoperative prediction of MVI, found that ADC and D values were significantly higher in MVI-negative. The reason may be the presence of tumor emboli or clusters of cancer cells in branches of hepatic vessels which could restrict the diffusion of water molecules. Meanwhile, this research has found that IVIM model-derived D value (AUC = 0.815) is advantageous over ADC measured with mono-exponential model (AUC = 0.746) for assessing the MVI of HCC. This may be related to the increase of tissue cellularity and pseudodiffusion component fraction during the progression of HCC, and their combined effect may lead to an increase in ADC, while the D is not affected. Howeve, the research on the evaluation of MVI by IVIM is relatively few, and the acquisition time of its model is relatively long, so it is worth thinking about how to shorten the acquisition time on the premise of obtaining sufficient image information currently.

4.4 Diffusional kurtosis imaging, DKI

DKI is a special DWI model that can more accurately quantify the degree of non-Gaussian diffusion of water molecules. Wang et al. [39] firstly suggested that mean kurtosis (MK) was significantly correlated with MVI and the higher MK was a potential predictive biomarker of MVI. Cao et al. [40] prospectively evaluated the diagnostic efficacy of DKI in predicting MVI with comparison to the DWI, concluded only MK differed significantly between the MVI-positive and MVI-negative group which showed moderate diagnostic efficacy (AUC = 0.77), and high-grade HCCs had significantly higher MK values. Among the many dispersion parameters, MK is potentially the only significant prognostic factor that could be attributed to the underlying correlation between MK and tumor-aggressive biological behaviors. There was no obvious correlation between ADC and MVI in this study, this discrepancy is likely due to the selection of b-values. The DKI was obtained under free-breathing, resulting in a decreased signal-to-noise ratio, but, compared to respiratory-triggered and breath-hold imaging, it has good reproducibility and shorter acquisition time. The measurement of DKI was strongly influenced by the ROI selection, how to improve the DKI image quality and standardize the specification parameters need more in-depth research.

4.5 MR elastography, MRE

MR elastography (MRE) is a phase contrast-based MRI technique and a quantitative technique. Previous studies have shown that MRE-measured tumor stiffness (TS) has an emerging role in assessing focal malignancy, and could provide prognostic information. There is a study [41] firstly focused on assessing the diagnostic performance of MRE-based shear strain mapping for predicting MVI in HCC, Li calculated Octahedral shear strain (OSS) maps, recorded the percentage of peritumoral interface length with low shear strain and assessed the diagnostic performance of OSS-pLSL at different frequencies (60,40,30 Hz). The results indicated that shear strain mapping derived from 3D MRE, which was based on the detection of the transition in mechanical properties at the tumor-liver interface, could noninvasively identify the presence of MVI by quantifying the degree of tumor peritumoral tissue adhesion in HCC, especially at 30-Hz, which may further improve the prediction of MVI or prognosis in HCC patients. A recent exploratory study [42] with a big sample size (N = 185) assessed the diagnostic performance of stiffness value quantified by MRE in identifying MVI grade and found the severe-MVI grade had the highest TS, which might be mainly explained by the alterations of the vasculature and extracellular matrix of HCC with MVI. Using MRE features to evaluate MVI is an innovation and a challenge. In the future, it is necessary to develop more sensitive MRE sequences to combine OSS-pLSL with MRE-measured tumor/liver stiffness, imaging-based features, and other indicators to build a more accurate predictor for MVI.

5 PET

Functional imaging of nuclear medicine is a useful molecular imaging tool. One previous study’s [43] multivariate analysis identified 18F-fluorodeoxyglucose positron emission tomography with computed tomography (FDG PET/CT) could be a clinical surrogate marker of microvascular invasion, this reveals that increased 18F-FDG uptake correlates with biologic tumor aggressiveness. The mediators and cellular effectors of inflammation are important components of tumor local environment, pro-inflammatory cells infiltrate into the tumor microenvironment results in increased 18F-FDG uptake at sites of inflammation [44]. Previously, a foreign multicenter study [45] measured the maximum standardized uptake value (SUVmax) in the primary tumor and the mean standardized uptake value of the normal liver (SUVmean) in 158 patients with early HCC of BCLC stage 0 or A, and determined the tumor FDG avidity by calculating the ratio (TLR = SUVmax / SUVmean). The results revealed that TLR, AFP level, and tumor size were significantly associated with the presence of MVI, and the diagnostic efficiency of TLR was the highest (AUC = 0.704). Primary-tumor FDG avidity is an invasive biological index that correlates with tumor recurrence. The analysis of TLR to reflect FDG affinity can reduce the inaccurate measurement of SUV caused by Inter-institutional or inter-observer variabilities, and is more objective. In addition, Celebi et al. [46] found that there is a correlation between tumor SUVmax measured by 18F-FDGPET/MRI imaging and MVI, on multivariate analysis, the SUVmax was the only independent risk factor for the MVI of HCC. But the current research on 18F-FDGPET/MRI evaluation of MVI is rare, and more studies are needed in the future to verify its clinical application value. At the same time, nuclear medical equipment is limited in some remote areas and county-level hospitals, so it is difficult to popularize.

6 Artificial intelligence (AI)

In the past few years, the progress of artificial intelligence (AI) has made it possible to extract useful information from complex clinical data sets. At present, there are two kinds of AI methods widely used in the field of medical imaging, namely, the deep learning method represented by convolution neural network (CNN) and the traditional machine learning method represented by imaging group.

6.1 Machine learning (ML)

As an emerging non-invasive image analysis and data extraction technology, radiomics has been used to predict MVI in HCC and has shown the potential value. Previous studies have found that radiomics based on CT, MRI, 18F-FDGPET/CT, and ultrasonography have shown great potential in predicting MVI. Xia et al [47] developed a radiomics approach to predict MVI based on CT in patients with HCC from four medical centers, and combined tumor characteristics and radiomics score to predict MVI in internal (AUC = 0.86) and external test sets (AUC = 0.84). Radiomic texture analysis quantifies tumor heterogeneity. Li et al. [48] found that texture analysis based on multi-phase CT images can be used to predict MVI of HCC, and the most valuable parameter comes from the gray-level run-length matrix. But at present, how to standardize feature extraction/selection algorithms of texture analysis is a difficult problem, and which feature extraction/selection algorithms can predict MVI more accurately need extensive clinical research and verification. A study [49] used MRI arterial phase images to analyze the peritumoral and intratumoral histogram features, the results showed that histogram defined peritumoral (peri-) mean, median, kurtosis, and radiological features were associated with MVI in sHCCs (≤3 cm) and none of the features extracted from the intratumor area were significantly different between the MVI-positive and MVI– negative groups. However, this study does not import other multi-parameters and only analyzes the maximal cross-sectional areas of the tumors, may lead to deviation in the results. A study [50] showed that nomogram of [18F]FDG PET/CT combined with clinical risk factors can be used to predict MVI status and disease-free survival (DFS) in patients with very-early-and early-stage (BCLC 0, BCLC A) HCC, which means that it is helpful to guide clinical treatment and prognosis to a great extent. Hu et al. [51] retrospectively reviewed 482 HCC patients who underwent CEUS, and the results showed that the radiomics nomogram (based on the three factors: the radiomics score, AFP, and tumor size) showed better performance for MVI detection (AUC = 0.731) than the clinical nomogram (based on AFP and tumor size)(AUC = 0.634), indicating that imaging score improved the predictive efficiency of MVI. In a multicenter study [52] based on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI, eight imaging models were established according to different VOIs (VOIintratumor and VOIintratumor + peritumor) and three MRI sequences (T2-weighted, AP, and HBP) and fused sequences (combined of three sequences), and clinical-radiological (CR) model and clinical-radiological-radiomics (CRR) model were constructed. The results show that the prediction efficiency of the fused sequences-based VOIintratumor + peritumor radiomics model is the best, and the CRR model improved evaluation efficacy over the CR model. In addition, a study [53] has found that the radiographic-radiomic (RR) model based on contrast-enhanced CT demonstrates good performance for predicting MVI and clinical outcomes. This suggests that the combination of clinical features, radiographic features and radiomic features has significant advantages in MVI evaluation, and is the research trend in the future. But how to create a standard definition and a normalized pipeline for radiomics studies is a question worth pondering.

6.2 Deep learning (DL)

Deep learning (DL) is a particular form of machine learning based on artificial neural networks (ANNs). With the continuous progress of development, DL has triggered some researchers to explore the area of combining DL with MVI and achieve some progress. A study [54] firstly developed an MVI predictive model based on image analysis using convolutional neural network and showed that the 3D-CNN models based on CT images had high diagnostic efficiency in the validation set (AUC = 0.91), indicating that the 3DCNN model has broad development prospects in predicting MVI. Zhang et al. [55] used a three-dimensional convolutional neural network (3D CNN) to develop four deep-learning models based on MRI images, including three single-layer models based on single-sequence, and fusion model combining three sequences. The results suggested that the fusion model combining T2WI, T2-SPIR and PVP had better diagnostic performance in predicting MVI which achieved an AUC of 0.81, sensitivity of 69%, and specificity of 79%. These machine learning models may facilitate decision-making in HCC treatment but requires further validation. Even though AI carries much promise for changing future clinical practice, lots of issues must be addressed before broad implementation is possible, such as data standardization and explainability of ML/DL algorithms.

In summary, the current research status shows that preoperative imaging assessment of MVI has made some progress, but it is still challenging. Future research on preoperative imaging assessment of MVI may need to shift from independent single imaging modalities to multi-modal imaging studies, fully combining clinical, imaging and radiomics features, as well as deep learning, to establish more accurate predictions model to improve precision medical care for HCC patients.

Ethics approval

This is a review article and thus no institutional ethics approval was required.

Funding

Funding from the Clinical Foundation of The First Hospital of Lanzhou University, (No. Ldyyyn2020-14) is gratefully acknowledged.

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author contributions

CRediT authorship contribution statement

Yi-xiang Li: Writing – original draft, Data curation, Conceptualization. Meng-meng Qu: Data curation, Conceptualization. Wei-long Lv: Data curation. Li-li Wang: Data curation. Xiao-yu Liu: Data curation. Ying Zhao: Data curation. Junqiang Lei: Writing – review & editing, Conceptualization.

References

Sung

, Ferlay

, Siegel

, 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.

Bray

, Ferlay

, Soerjomataram

, et al. GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424.

Zheng

, Qu

, Zhang

, et al. Liver cancer incidence and mortality in China: Temporal trends and projections to 2030. Chin J Cancer Res. 2018;30(6):571–9.

Zhou

, Sun

, Wang

, et al. Guidelines for Diagnosis and Treatment of Primary Liver Cancer in China (2017 Edition). Liver Cancer. 2018;7(3):235–26.

Yang

, Hainaut

, Gores

, et al. A global view of hepatocellular carcinoma: Trends, risk, prevention and management. Nat Rev Gastroenterol Hepatol. 2019;16:589–604.

Shah

, Cleary

, Wei

, et al. Recurrence after liver resection for hepatocellular carcinoma: Risk factors, treatment, and outcomes. Surgery. 2007;141(3):330–9.

Wang

, Wu

, Cong

, et al. Microvascular invasion predicts a poor prognosis of solitary hepatocellular carcinoma up to 2 Cm based on propensity score matching analysis. Hepatol Res. 2019;49:344–54.

Zhang

, Wang

, Wei

, et al. An eastern hepatobiliary surgery hospital microvascular invasion scoring system in predicting prognosis of patients with hepatocellular carcinoma and microvascular invasion after R0 liver resection: A large-scale, multicenter study. Oncologist. 2019;24:e1476–88.

Zhou

, Sun

, Wang

, et al. Guidelines for the diagnosis and treatment of hepatocellular carcinoma (2019 edition). Liver Cancer. 2020;9:682–720.

10.

Yang

, Si

, Yang

, et al. A wide-margin liver resection improves longterm outcomes for patients with HBV-related hepatocellular carcinoma with microvascular invasion. Surgery. 2019;165:721–30.

11.

Sun

, Li

, Shi

, et al. Anatomic versus non-anatomic resection of hepatocellular carcinoma with microvascular invasion: A systematic review and meta-analysis. Asian J Surg. 2021;44:1143–50.

12.

Erstad

, Tanabe

, et al. Prognostic and therapeutic implications of microvascular invasion in hepatocellular carcinoma. Ann Surg Oncol. 2019;26:1474–93.

13.

Xiao

, Chen

Z-B

, Jin

H-L

, et al. Treatment selection of recurrent hepatocellular carcinoma with microvascular invasion at the initial hepatectomy[J]. Am J Transl Res. 2019;11(3):1864–75.

14.

Cucchetti

, Piscaglia

, Grigioni

, et al. Preoperative prediction of hepatocellular carcinoma tumour grade and micro-vascular invasion by means of artificial neural network: A pilot study. J Hepatol. 2010;52(6):880–8.

15.

Liu

, Wang

, Zhu

, et al. A preoperative measurement of serum microRNA-125b may predict the presence of microvascular invasion in hepatocellular carcinomas patients. Transl Oncol. 2016;9(3):167–72.

16.

Wang

, Nie

, Dong

, et al. Study on correlation between two-dimensional ultrasound, contrast-enhanced ultrasound and microvascular invasion in hepatocellular carcinoma. Clin Hemorheol Microcirc. 2022;80(2):97–106.

17.

Wang

, Chai

, Cai

, et al. Prognostic and pathological implications of contrast-enhanced ultrasound features in hepatocellular carcinoma. J Cancer Res Ther. 2023;19(4):1040–7.

18.

Zhu

, Qing

, Yan

, et al.

Can the contrast-enhanced ultrasound washout rate be used to predict microvascular invasion in hepatocellular carcinoma?

Ultrasound Med Biol 2017;43(8):1571–80.

19.

Wang

, Wang

, Liu

, et al. The role of quantitation of real-time 3-dimensional contrast-enhanced ultrasound in detecting microvascular invasion: An in vivo study. Abdom Radiol (NY). 2016;41(10):1973–9.

20.

Dong

, Qiu

, Yang

, et al. Potential application of dynamic contrast enhanced ultrasound in predicting microvascular invasion of hepatocellular carcinoma. Clin Hemorheol Microcirc. 2021;77(4):461–9.

21.

Renzulli

, Brocchi

, Cucchetti

, et al.

Can current preoperative imaging be used to detect microvascular invasion of hepatocellular carcinoma?

Radiology. 2016;279(2):432–42.

22.

Song

, Li

, Luo

, et al. The importance of a nonsmooth tumor margin and incomplete tumor capsule in predicting HCC microvascular invasion on preoperative imaging examination: A systematic review and meta-analysis. Clin Imaging. 2021;76:77–82.

23.

Zhu

, Yang

, Li

, et al. Incomplete tumor capsule on preoperative imaging reveals microvascular invasion in hepatocellular carcinoma: A systematic review and meta-analysis. Abdom Radiol (NY). 2019;44(9):3049–57.

24.

, Xu

, Weng

, et al. CT image-based texture analysis to predict microvascular invasion in primary hepatocellular carcinoma. J Digit Imaging. 2020;33(6):1365–75.

25.

, Tan

, Zhou

, et al. Liver computed tomographic perfusion in the assessment of microvascular invasion in patients with small hepatocellular carcinoma. Invest Radiol. 2015;50(4):188–94.

26.

Yang

, Zhang

, Jia

, et al. Dual energy spectral CT imaging for the evaluation of small hepatocellular carcinoma microvascular invasion. Eur J Radiol. 2017;95:222–7.

27.

Kim

, Lee

, Yoon

, et al. Prediction of microvascular invasion of hepatocellular carcinoma: Value of volumetric iodine quantification using preoperative dual-energy computed tomography. Cancer Imaging. 2020;20(1):60.

28.

Luo

, Li

, Xie

, et al. Preoperative normalized iodine concentration derived from spectral CT is correlated with early recurrence of hepatocellular carcinoma after curative resection. Eur Radiol. 2021;31(4):1872–82.

29.

Hong

, Choi

, Kim

, et al. MRI features for predicting microvascular invasion of hepatocellular carcinoma: A systematic review and meta-analysis. Liver Cancer. 2021;10(2):94–106.

30.

Lee

, Kim

, Lee

, et al. Preoperative gadoxetic acid-enhanced MRI for predicting microvascular invasion in patients with single hepatocellular carcinoma. J Hepatol. 2017;67(3):526–34.

31.

, Zhu

, Liu

, et al. Peritumoral imaging manifestations on Gd-EOB-DTPA-Enhanced MRI for preoperative prediction of microvascular invasion in hepatocellular carcinoma: A systematic review and meta-analysis. Front Oncol. 2022;12:907076.

32.

Chen

, Li

, Zhang

, et al. MRI features for predicting microvascular invasion and postoperative recurrence in hepatocellular carcinoma without peritumoral hypointensity. J Hepatocell Carcinoma. 2023;10:1595–608.

33.

Wang

, Li

, Lei

, et al. The value of the signal intensity of peritumoral tissue on Gd-EOB-DTPA dynamic enhanced MRI in assessment of microvascular invasion and pathological grade of hepatocellular carcinoma. Medicine (Baltimore). 2021;100(20):e25804.

34.

Zhang

, Xie

, Li

, et al. Prediction of microvascular invasion in HCC by a scoring model combining Gd-EOB-DTPA MRI and biochemical indicators. Eur Radiol. 2022;32(6):4186–97.

35.

, Zeng

, Liu

, et al.

Microvascular invasion in small hepatocellular carcinoma: Is it predictable with preoperative diffusion-weighted imaging?

J Gastroenterol Hepatol. 2014;29(2):330–6.

36.

Yang

, Wang

, Tang

, et al. ADC similarity predicts microvascular invasion of bifocal hepatocellular carcinoma. Abdom Radiol (NY). 2018;43(9):2295–302.

37.

Le Bihan

, Breton

, Lallemand

, et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology. 1988;168(2):497–505.

38.

Wei

, Huang

, Tang

, et al. IVIM improves preoperative assessment of microvascular invasion in HCC. Eur Radiol. 2019;29(10):5403–14.

39.

Wang

, Yang

, et al. Assessment of microvascular invasion of hepatocellular carcinoma with diffusion kurtosis imaging. Radiology. 2018;286(2):571–80.

40.

Cao

, Chen

, Duan

, et al. Diffusion kurtosis imaging (DKI) of hepatocellular carcinoma: Correlation with microvascular invasion and histologic grade. Quant Imaging Med Surg. 2019;9(4):590–602.

41.

, Yin

, Hu

, et al. MR elastography-based shear strain mapping for assessment of microvascular invasion in hepatocellular carcinoma. Eur Radiol. 2022;32(7):5024–32.

42.

Zhang

, Li

, Zhu

, et al. The value of quantitative MR elastography-based stiffness for assessing the microvascular invasion grade in hepatocellular carcinoma. Eur Radiol. 2023;33(6):4103–14.

43.

Lim

, Salloum

, Chalaye

, et al. 18F-FDG PET/CT predicts microvascular invasion and early recurrence after liver resection for hepatocellular carcinoma: A prospective observational study. HPB (Oxford). 2019;21(6):739–47.

44.

Mantovani

, Allavena

, Sica

, et al. Cancer-related inflammation. Nature. 2008;454(7203):436–44.

45.

Hyun

, Eo

, Song

, et al. Preoperative prediction of microvascular invasion of hepatocellular carcinoma using 18F-FDG PET/CT: A multicenter retrospective cohort study. Eur J Nucl Med Mol Imaging. 2018;45(5):720–6.

46.

Çelebi

, Görmez

, Ilgun

, et al. The value of 18F-FDG PET/MRI in prediction of microvascular invasion in hepatocellular carcinoma. Eur J Radiol. 2022;149:110196.

47.

Xia

, Zhou

, Meng

, et al. Predicting microvascular invasion in hepatocellular carcinoma using CT-based radiomics model. Radiology. 2023;307(4):e222729.

48.

, Xu

, Weng

, et al. CT image-based texture analysis to predict microvascular invasion in primary hepatocellular carcinoma. J Digit Imaging. 2020;33(6):1365–75.

49.

Wang

, Sun

, Zhou

, et al. Histogram peritumoral enhanced features on MRI arterial phase with extracellular contrast agent can improve prediction of microvascular invasion of hepatocellular carcinoma. Quant Imaging Med Surg. 2022;12(2):1372–84.

50.

, Zhang

, Fang

, Zhang

, et al. Radiomics analysis of [18F]FDG PET/CT for microvascular invasion and prognosis prediction in very-early- and early-stage hepatocellular carcinoma. Eur J Nucl Med Mol Imaging. 2021;48(8):2599–614.

51.

, Wang

, Huang

, et al. Ultrasound-based radiomics score: A potential biomarker for the prediction of microvascular invasion in hepatocellular carcinoma. Eur Radiol. 2019;29(6):2890–901.

52.

Tian

, Hua

, Peng

, et al. Preoperative evaluation of Gd-EOB-DTPA-Enhanced MRI radiomics-based nomogram in small solitary hepatocellular carcinoma (≤3 cm) with microvascular invasion: A two-center study. J Magn Reson Imaging. 2022;56(5):1459–72.

53.

, Zhang

, Liu

, et al. Radiomic analysis of contrast-enhanced CT predicts microvascular invasion and outcome in hepatocellular carcinoma. J Hepatol. 2019;70(6):1133–44.

54.

Jiang

, Cao

, et al. Preoperative identification of microvascular invasion in hepatocellular carcinoma by XGBoost and deep learning. J Cancer Res Clin Oncol. 2021;147(3):821–33.

55.

Zhang

, Lv

, Qiu

, et al. Deep learning with 3D convolutional neural network for noninvasive prediction of microvascular invasion in hepatocellular carcinoma. J Magn Reson Imaging. 2021;54(1):134–43.