Distinction Between Hepatocellular Carcinoma and Hypervascular Liver Metastases in Non-cirrhotic Patients Using Gadoxetate Disodium-Enhanced Magnetic Resonance Imaging

Abstract

French

Purpose: This study aims to identify the hallmarks of gadoxetate disodium-enhanced magnetic resonance imaging distinguishing hepatocellular carcinoma (HCC) from hypervascular liver metastases (HLMs). Methods: Between January 2008 and October 2020, among patients who underwent gadoxetate disodium-enhanced MRI, those who met the following criteria were retrospectively included: without chronic hepatitis or liver stiffness ≤ 2.5 kPa on magnetic resonance elastography or F0/F1 on pathological assessment. Two blinded radiologists reviewed the imaging findings to judge the presence or absence of the enhancing capsule, nonperipheral washout, corona enhancement, hypointensity in the transitional/hepatobiliary phase (HBP), hyperintensity on T2-weighted/diffusion-weighted imaging (DWI), mosaic architecture, and blood products/fat in mass. The lesion-to-liver signal intensity ratios in HBP and DWI were also calculated. Univariate and multivariate analyses were performed to identify the imaging hallmarks distinguishing HCC from HLM. Interobserver agreement was calculated using kappa values and intraclass correlation coefficients (ICCs). Results: The final study cohort comprised 72 lesions in 44 patients (mean age, 65.0±11.9 years). Univariate analysis revealed higher frequencies of the following features in HCC than in HLM (P < .10): nonperipheral washout, corona enhancement, transitional phase hypointensity, mosaic architecture, and fat in mass (P = .002-.073). Multivariate analysis revealed that nonperipheral washout and mosaic architecture favored the diagnosis of HCC over that of HLM with odds ratios of 7.66 and 14.6, respectively (P = .038 and .029, respectively). The interobserver agreement for each item was moderate or substantial (kappa or ICC = .447-.792). Conclusion: Peripheral washout and mosaic architecture may be reliable imaging hallmarks for distinguishing HCC from HLM.

Keywords

liver magnetic resonance imaging metastasis hepatocellular carcinoma contrast agent

Introduction

The liver is the most common site of metastatic disease, and the colorectum is the most common primary site of liver metastases.¹ Gadoxetate disodium-enhanced magnetic resonance imaging (MRI) is the most useful imaging modality for the detection of liver metastases, especially owing to diffusion-weighted images (DWI) and hepatobiliary phase (HBP).^2,3 According to a meta-analysis by Vilgrain et al, the combination of DWI and HBP showed very high sensitivity for the detection of liver metastases on a per-lesion basis (95.5%).³ Liver metastases typically demonstrate a rim enhancement on the arterial phase (AP), peripheral washout on the portal venous phase (PVP), delayed central enhancement on PVP and transitional phases (TP), and targetoid diffusion restriction. They appear as hypointense lesions on HBP and may show a hypointense rim on the TP and HBP.⁴ Moreover, gadoxetate disodium-enhanced MRI has been a useful tool for helping differentiate liver metastases from other diseases, including intrahepatic cholangiocarcinoma, hepatic hemangioma, and inflammatory pseudotumor.^5-9 However, liver metastases from hypervascular tumors, including neuroendocrine neoplasm (NEN), gastrointestinal tumor (GIST), renal cell carcinoma (RCC), melanoma, paraganglioma, and thyroid cancer, can be detected as hypervascular lesions.^4,10,11

Most hepatocellular carcinomas (HCCs) are associated with cirrhosis regardless of the etiology.¹² Although rare, HCCs can occur in non-cirrhotic liver; however, there is a lack of significant data on HCC that arises in non-cirrhotic liver.¹³ HCCs in non-cirrhotic livers are detected as hypervascular lesions similar to HCCs in cirrhotic livers.^14-16 Discriminating HCC from hypervascular liver metastases (HLM) in non-cirrhotic patients is often challenging because HLM may exhibit non-rim AP hyperenhancement.¹⁷ It is important to distinguish HCC from HLM because the treatments for these conditions differ.

In recent years, gadoxetate disodium has become widely used for liver MRI in daily clinical practice because of its high performance in lesion detection and characterization.^18-20 However, few studies have discussed the imaging features of HLM on MRI.^10,21-23 Our aim was to assess that gadoxetate disodium-enhanced MRI can be useful for distinguishing HCC from HLM because of this modality’s multiparametric techniques, including chemical shift imaging, T2-weighted images (T2WI), DWI, dynamic study, and HBP. To the best of our knowledge, a comparison of gadoxetate disodium-enhanced MRI findings between HLM and HCC in non-cirrhotic patients has not been reported to date. Thus, this study aimed to identify the hallmarks of gadoxetate disodium-enhanced MRI in distinguishing HCC from HLM.

Materials and Methods

Patients

This single-center, retrospective, cross-sectional study was approved by the relevant institutional review board, which waived the requirement for obtaining written informed patient consent owing to the retrospective nature of this study. All patients meeting the following inclusion criteria were enrolled between February 2008 and October 2020: (1) without chronic hepatitis or liver stiffness ≤ 2.5 kPa on magnetic resonance elastography or F0/F1 on pathological assessment, (2) presence of hypervascular lesions that show non-rim AP hyperenhancement on gadoxetate disodium-enhanced MRI, (3) presence of HCC with pathological confirmation or HLM with pathological confirmation or clinical diagnosis, (4) lesion size < 30 mm in diameter (to exclude hypervascular lesions with large necrotic area), and (5) older than or equal to age 18. Clinical diagnosis of HLM was defined as the following situations: (a) diagnosed by somatostatin receptor scintigraphy; (b) decreased in tumor size after chemotherapy; and (c) if one liver metastasis was biopsied, additional liver metastases in the same patient were considered to have the same diagnosis.

To evaluate the differential features of HCC from HLM on gadoxetate disodium-enhanced MRI, all patients who met the inclusion criteria were enrolled because of the rarity of HCC in non-cirrhotic patients and HLM. In patients with multiple lesions, the largest 5 lesions were selected for analysis. Therefore, the final study cohort comprised 44 non-consecutive patients (mean age, 65.0 ± 11.9 [range, 35-84] years), with 72 confirmed liver lesions (Figure 1). This group included 37 men (67.1 ± 9.9 [49-82] years) and 7 women (53.6 ± 15.9 [35-84] years).

Figure 1.

Flowchart of patient enrollment.

Magnetic Resonance Imaging Protocols

MRI was performed using a superconducting magnet scanner operated at 1.5 T (Signa EXCITE HD; GE Medical Systems) or 3 T (Discovery 750; GE Medical Systems) with an 8- or a 32-channel phased-array coil, respectively. Dynamic studies were performed at 20–30 s (AP, scan timing was adjusted using the fluoroscopic triggering technique) and 1 (PVP), 2 (TP), and 20 min after administration of gadoxetate disodium. The contrast material (.025 mmol/kg body weight) was administered as an intravenous bolus at a rate of 1 mL/s, followed by flushing with 20 mL saline using a power injector. In- and opposed-phase images were obtained before dynamic study. T2WI and DWI were obtained between TP and HBP. Table 1 presents the MRI parameters in detail. Diagnostic images were obtained in all the sequences in all the patients.

Table 1.

Sequence Parameters for Magnetic Resonance Imaging.

	1.5 T	3T
T1-Weighted Gradient-Echo Image
Sequence	2D fast gradient echo	3D fast gradient echo
Repetition time (ms)	150-170	6.2
Echo time (ms)	2.2 and 4.5	1 and 4.2
Matrix	32-40 × 32-40	35 × 16
Section thickness (mm)	5	3.6
Fat-saturated T2-weighted imaging
Sequence	2D fast spin echo	2D fast spin echo
Repetition time (ms)	Variable (respiratory trigger)	Variable (respiratory trigger)
Echo time (ms)	72	94
Matrix	256 × 192	320 × 320
Section thickness (mm)	6	5
Diffusion-weighted imaging
Sequence	2D spin echo-EPI	2D spin echo-EPI
Repetition time (ms)	Variable (respiratory trigger)	Variable (respiratory trigger)
Echo time (ms)	64.7	59.7
b value (s/mm²)	1000	1000
Matrix	128 × 128	128 × 160
Section thickness (mm)	6	6
Contrast-enhanced MRI
Sequence	3D LAVA	3D LAVA
Repetition time (ms)	3.9	4.8
Echo time (ms)	1.9	2.0
Matrix	320 × 192	320 × 192
Section thickness/intersection gap (mm)	5/−2.5	3.6/−1.8
Flip angle (degree)	12	12
Scan delay after administration	20-30 s, 1, 2, and 20 min	20-30 s, 1, 2, and 20 min

Abbreviations: EPI, echo planar imaging; MRI, magnetic resonance imaging; LAVA, liver acquisition with volume acceleration.

Image Analysis

All anonymized data, including the MRI findings, clinical information, and pathological records, were collected for all patients by the study coordinator (with 13 years of experience in liver imaging), who attempted to determine the size and location of the HCC and HLM lesions on MRI. Subsequently, 2 radiologists, with 13 and 7 years of experience in liver imaging, independently assessed the anonymized MR images. The radiologists were blinded to the clinical information and the final diagnosis. The following Liver Imaging Reporting and Data System (LI-RADS) major and ancillary features were evaluated for their presence or absence: enhancing capsule, nonperipheral washout, corona enhancement, TP hypointensity, HBP hypointensity, DWI hyperintensity, mild-to-moderate T2 hyperintensity, mosaic architecture, blood products in mass, and fat in mass.²⁴ The definitions of these features are as follows.²⁴.

(1) Enhancing capsule, defined as smooth, uniform, sharp border with enhancement around most or all an observed lesion on PVP or TP.

(2) Nonperipheral washout, defined as at least part of the observation showing lower signal intensity than adjacent liver on PVP.

(3) Corona enhancement, defined as peri-observational enhancement in AP or PVP attributable to venous drainage from tumor.

(4) TP hypointensity, defined as lower signal intensity in whole or in part than adjacent liver on TP.

(5) HBP hypointensity, defined as lower signal intensity in whole or in part than adjacent liver on HBP.

(6) DWI hyperintensity, defined as higher signal intensity than adjacent liver on DWI, not attributable solely to T2 shine-through.

(7) Mild-to-moderate T2 hyperintensity, defined as mildly or moderately higher signal intensity than adjacent liver on T2WI and similar to or less than non-iron-overloaded spleen.

(8) Mosaic architecture, defined as the presence of randomly distributed internal nodules or compartments, usually with different imaging features.

(9) Blood products in mass, defined as intralesional or perilesional hemorrhage in the absence of biopsy, trauma, or intervention.

(10) Fat in mass, defined as excess fat within a mass, in whole or in part, relative to adjacent liver.

The signal intensity (SI) of the liver and lesion was measured using HBP and DWI. The SI ratio of the lesion to the liver (SIR_lesion/liver) was calculated as follows

S I R_{\frac{l e s i o n}{i v e r}} = \frac{S I_{l e s i o n}}{S I_{l i v e r}}

where

S I_{l e s i o n}

and

S I_{l i v e r}

are the SI values for the lesion and liver, respectively. For quantitative analysis, the largest possible regions of interest were drawn on the lesion and liver, away from large vessels.

Statistical Analyses

For univariate analysis, categorical variables were compared between HCC and HLM using the chi-squared test, whereas continuous variables were compared using the Wilcoxon test. For multivariate analysis, the odds ratio was estimated by logistic regression analysis using variables that exhibited P values < .10 in the univariate analysis. Cohen’s kappa values (κ) or intraclass correlation coefficients (ICCs, r) were calculated to assess interobserver agreement. Agreement was evaluated as follows: .00-.20, slight; .21-.40, fair; .41-.60, moderate; .61-.80, substantial; and .81-1.00, almost perfect. All statistical analyses were performed using JMP software (version 15.2.0; SAS Institute Inc) and BellCurve for Excel (version 3.21; Social Survey Research Information Co, Ltd). P values < .05 were considered statistically significant.

Results

Lesion Characteristics

The final diagnoses of 72 liver lesions were as follows: HLM (n = 39; mean size, 9.4 ± 5.4 [range, 52-4] mm) and HCC (n = 33, 17.5 ± 6.3 [6-29] mm). Of the 39 HLM lesions, 17 were pathologically confirmed, whereas 22 were clinically diagnosed. In contrast, all HCC lesions were pathologically confirmed. The primary diseases of the 12 patients with HLM were NEN (9 patients), GIST (2 patients), and RCC (1 patient). Other information on patient demographics is presented in Table 2.

Table 2.

Patient Demographics in Each Disease.

Variables	Hypervascular Liver Metastases	Hepatocellular Carcinoma	P Value
Number of patients	12	32	-
Number of lesions	39	33	-
Age (years)	58.5 ± 12.9	67.4 ± 11.1	.033^a
Sex (man: woman)	7:5	30:2	.011^a
Body weight (kg)	56.1 ± 13.6	62.6 ± 7.3	.018^a
Size (mm)	9.4 ± 5.4	17.5 ± 6.3	< .001^a

Continuous variables were analyzed using the Wilcoxon test and are expressed as mean ± standard deviation or median with the range in parentheses. Categorical variables were analyzed using the chi-squared test and are expressed as ratios.

^aP < .05.

Univariate Analysis

There were significant differences between HLM and HCC in terms of nonperipheral washout (21/39 [53.8%] vs 29/33 [87.9%]), mosaic architecture (1/39 [2.6%] vs 9/33 [27.3%]), and fat in mass (1/39 [2.6%] vs 6/33 [18.2%]) (P = .002-.043), (Table 3). Corona enhancement (4/39 [10.3%] vs 9/33 [27.3%]) and TP hypointensity (26/39 [66.7%] vs 29/33 [87.9%]) showed P values < .10 (P = .073 and .051, respectively) (Table 3). There were no significant differences between HLM and HCC in terms of other imaging findings and SIR_lesion/liver on HBP and DWI (P = .310-1.000), (Table 3).

Table 3.

Univariate Analysis and Interobserver Agreement of Hypervascular Liver Metastases vs Hepatocellular Carcinoma.

Variable	Hypervascular Liver Metastases (%)	Hepatocellular Carcinoma (%)	P Value	Interobserver Agreement
Enhancing capsule	25.6 (10/39)	39.4 (13/33)	.310	.531 (.337.728)
Nonperipheral washout	53.8 (21/39)	87.9 (29/33)	.002^b	.530 (.315.745)
Corona enhancement	10.3 (4/39)	27.3 (9/33)	.073^a	.448 (.235.661)
Transitional phase hypointensity	66.7 (26/39)	87.9 (29/33)	.051^a	.646 (.435.858)
Hepatobiliary phase hypointensity	97.4 (38/39)	93.9 (31/33)	.590	.652 (.2021.000)
DWI hyperintensity	94.9 (37/39)	97.0 (32/33)	1.000	.736 (.3891.000)
Mild-moderate T2 hyperintensity	82.1 (32/39)	90.9 (30/33)	.326	.576 (.294.858)
Mosaic architecture	2.6 (1/39)	27.3 (9/33)	.004^b	.486 (.167.805)
Blood products in mass	0 (0/39)	3.0 (1/33)	.458	.660 (.0391.000)
Fat in mass	2.6 (1/39)	18.2 (6/33)	.043^b	.578 (.244.911)
SIR_lesion/liver on HBP	.61 ± .19	.60 ± .19	.705	.780 (.671-.856)
SIR_lesion/liver on DWI	3.05 ± 1.78	2.70 ± .94	.879	.792 (.687.864)

Continuous variables were analyzed by Wilcoxon test and are expressed as mean ± standard deviation. Categorical variables were analyzed by the chi-squared test and are expressed as percentage with numerators and denominators. Interobserver agreement is presented with the 95% confidence interval in parentheses.

Abbreviations: DWI, diffusion-weighed image; SIR_lesion/liver, signal intensity ratio of lesion to liver; HBP, hepatobiliary phase.

^aP < .10.

^bP < .05.

Interobserver Agreement

The interobserver agreement of all items was moderate to substantial (κ or r = .448-.792) (Table 3).

Multivariate Analysis

Multivariate analysis revealed that nonperipheral washout (odds ratio [95% confidence interval] for HCC vs HLM, 7.66 [1.11-52.6]) and mosaic architecture (14.6 [1.11-1.31–1.62 × 10²]) favored HCC over HLM (P = .038 and .029, respectively) (Table 4). Figures 2 and 3 present the case examples. Figure 2 shows a typical case of HCC with nonperipheral washout and mosaic architecture, whereas Figure 3 shows a typical case of HLM without nonperipheral washout and mosaic architecture.

Table 4.

Multivariate Analysis for Distinguishing Hepatocellular Carcinoma From Hypervascular Liver Metastases.

Variable	Odds Ratio (Hepatocellular Carcinoma vs Hypervascular Liver Metastases)	P Value
Nonperipheral washout	7.66 (1.11-52.6)	.038^a
Corona enhancement	1.34 (.30-5.94)	.699
Transitional phase hypointensity	.79 (.11-5.98)	.823
Mosaic architecture	14.6 (1.31-1.62 × 10²)	.029^a
Fat in mass	6.83 (.49-95.3)	.153

Data are presented with the 95% confidence interval in parentheses.

^aP < .05.

Figure 2.

A case with hepatocellular carcinoma. A 54-year-old man with hepatocellular carcinoma (16 mm) at S5. The hypervascular lesion shows mosaic architecture on fat-saturated T2-weighted image (arrow) and nonperipheral washout on portal venous phase image (arrowhead).

Figure 3.

A case with hypervascular liver metastases from gastrointestinal stromal tumor. A 57-year-old woman with hypervascular liver metastasis (15 mm) at S7. The hypervascular lesion does not show mosaic architecture on fat-saturated T2-weighted image and nonperipheral washout on portal venous phase image (dotted arrow).

Discussion

In our study, HCC had certain characteristic imaging features on MRI, including nonperipheral washout and mosaic architecture, compared to HLM. The main mechanism of nonperipheral washout in HCC is due to the venous return to the peripheral hepatic sinusoids.²⁵ Reduced intranodular portal venous blood supply, tumoral hypercellularity with corresponding reduction in extracellular volume, and intrinsic hypointensity may also be associated with nonperipheral washout in HCC.²⁵ More than half of the HLMs showed nonperipheral washout despite venous return of HLM to the hepatic veins. This may be caused by the evident enhancement of the background liver on PVP due to the high contrast on MRI. Mosaic architecture is the specific finding of HCC, which refers to the presence within a mass of internal nodules or compartments differing in enhancement or intensity separated by fibrous septations.²⁵ In contrast, fibrous septations are rare in HLMs. The frequency of mosaic architecture was not high in this study (27.3% [9/33]) because the size of the included HCCs was relatively small (mean 17.5 ± 6.3 mm) in this study.

According to a previous study, the combined set of DWI and HBP showed high sensitivity and specificity for the detection of liver metastases from NENs.¹⁰ More liver metastases were detected via gadoxetate disodium-enhanced MRI than by conventional extracellular gadolinium-enhanced MRI.²⁶ Therefore, gadoxetate disodium-enhanced MRI is useful for screening liver metastases from NENs. These studies included heterogeneously enhanced or hypovascular liver metastases, whereas we included only hypervascular metastases to compare with hypervascular HCC. In contrast, there seems to be little evidence of the imaging features of other HLMs. To the best of our knowledge, to date, a comparison of gadoxetate disodium-enhanced MRI findings between HLM and HCC in non-cirrhotic patients has not been reported. Therefore, the results of this study can be clinically important because treatments for HLM and HCC differ. In addition, we found good interobserver agreement of all imaging findings; thus, gadoxetate disodium-enhanced MRI for distinguishing HCC from HLM appears to be highly accurate and can be readily reproduced in further studies and clinical practice. Conversely, it is premature to mention whether HCC can be differentiated from HLM by imaging alone based on the results of this study. Further prospective studies with larger sample sizes are required. In addition, although we used a part of LI-RADS major and ancillary features because they are widely used imaging methods for evaluating liver lesions, the applicability of LI-RADS features to non-cirrhotic patients needs to be studied in the future.

Our study has some limitations. First, we included patients with clinically diagnosed HLMs. In daily practice, it is difficult to obtain pathological confirmation for all HLMs; therefore, our findings are clinically significant. Second, the primary diseases of HLM are not uniform, including NEN, GIST, and RCC. Characteristic imaging findings may vary among different diseases. Third, the retrospective nature of the study and the relatively small number of HLM and HCC lesions in non-cirrhotic patients were also limitations. The greater number of HCC patients in our study could potentially lead to problems with reproducibility of the results.

In conclusion, as we hypothesized, gadoxetate disodium-enhanced MRI is a useful tool for distinguishing HCC from HLM. Specifically, peripheral washout and mosaic architecture may be reliable imaging hallmarks to distinguish HCC from HLM.

Footnotes

Acknowledgments

We would like to thank Editage () for English language editing.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Shintaro Ichikawa

References

de Ridder

de Wilt

JHW

Simmer

Lemmens

Nagtegaal

. Incidence and origin of histologically confirmed liver metastases: An explorative case-study of 23,154 patients. Oncotarget. 2016;7:55368-55376.

Choi

Kim

Park

, et al. Diagnostic performance of CT, gadoxetate disodium-enhanced MRI, and PET/CT for the diagnosis of colorectal liver metastasis: Systematic review and meta-analysis. J Magn Reson Imaging. 2018;47:1237-1250. doi: 10.1002/jmri.25852

Vilgrain

Esvan

Ronot

Aubé

Chatellier

. A meta-analysis of diffusion-weighted and gadoxetic acid-enhanced MR imaging for the detection of liver metastases. Eur Radiol. 2016;26:4595-4615.

Karaosmanoglu

Onur

Ozmen

Akata

Karcaaltincaba

. Magnetic resonance imaging of liver metastasis. Semin Ultrasound CT MR. 2016;37:533-548.

Kovač

Galun

Đurić-Stefanović

et al. Intrahepatic mass-forming cholangiocarcinoma and solitary hypovascular liver metastases: Is the differential diagnosis using diffusion-weighted MRI possible? Acta Radiol. 2017;58:1417-1426.

Goshima

Kanematsu

Watanabe

et al. Hepatic hemangioma and metastasis: differentiation with gadoxetate disodium-enhanced 3-T MRI. AJR Am J Roentgenol. 2010;195:941-946.

Motosugi

Ichikawa

Onohara

et al. Distinguishing hepatic metastasis from hemangioma using gadoxetic acid-enhanced magnetic resonance imaging. Invest Radiol. 2011;46:359-365.

Hardie

Egbert

Rissing

. Improved differentiation between hepatic hemangioma and metastases on diffusion-weighted MRI by measurement of standard deviation of apparent diffusion coefficient. Clin Imaging. 2015;39:654-658.

Ichikawa

Motosugi

Suzuki

Shimizu

Onishi

. Imaging features of hepatic inflammatory pseudotumor: Distinction from colorectal liver metastasis using gadoxetate disodium-enhanced magnetic resonance imaging. Abdom Radiol (NY). 2020;45:2400-2408.

10.

Hayoz

Vietti-Violi

Duran

et al. The combination of hepatobiliary phase with Gd-EOB-DTPA and DWI is highly accurate for the detection and characterization of liver metastases from neuroendocrine tumor. Eur Radiol. 2020;30:6593-6602.

11.

Nakayama

Yoshimitsu

Irie

, et al. Fat in liver metastasis from renal cell carcinoma detected by chemical shift MR imaging. Abdom Imaging. 2003;28:657-659.

12.

Kulik

El-Serag

. Epidemiology and management of hepatocellular carcinoma. Gastroenterology. 2019;156:477-491.e1.

13.

Desai

Sandhu

Lai

J-P

Sandhu

. Hepatocellular carcinoma in non-cirrhotic liver: A comprehensive review. World J Hepatol. 2019;11:1-18.

14.

Jamwal

Krishnan

Kushwaha

Khurana

. Hepatocellular carcinoma in non-cirrhotic versus cirrhotic liver: A clinico-radiological comparative analysis. Abdom Radiol (NY). 2020;45:2378-2387.

15.

Lafitte

Laurent

Soyer

, et al. MDCT features of hepatocellular carcinoma (HCC) in non-cirrhotic liver. Diagn Interv Imaging. 2016;97:355-360.

16.

Di Martino

Saba

Bosco

, et al. Hepatocellular carcinoma (HCC) in non-cirrhotic liver: Clinical, radiological and pathological findings. Eur Radiol. 2014;24:1446-1454.

17.

Laroia

Sasturkar

Rastogi

Pamecha

. Solitary hypervascular liver metastasis from neuroendocrine tumor mimicking hepatocellular cancer: All that glitters is not gold. Indian J Nucl Med. 2015;30:42-46.

18.

Ichikawa

Isoda

Shimizu

, et al. Distinguishing intrahepatic mass-forming biliary carcinomas from hepatocellular carcinoma by computed tomography and magnetic resonance imaging using the Bayesian method: A bi-center study. Eur Radiol. 2020;30:5992-6002.

19.

Ichikawa

Motosugi

Morisaka

Kozaka

Goshima

Ichikawa

. Optimal combination of features on gadoxetate disodium-enhanced MR imaging for non-invasive differential diagnosis of hepatocellular carcinoma: The JAMP-HCC study. Magn Reson Med Sci. 2021;20:47-59.

20.

Murakami

Sofue

Hori

. Diagnosis of hepatocellular carcinoma using Gd-EOB-DTPA MR imaging Online ahead of print Aug 21, 2021. Magn Reson Med Sci. doi:10.2463/mrms.rev.2021-0031.

21.

Besa

Ward

Cui

Jajamovich

Kim

Taouli

. Neuroendocrine liver metastases: Value of apparent diffusion coefficient and enhancement ratios for characterization of histopathologic grade. J Magn Reson Imaging. 2016;44:1432-1441.

22.

Gultekin

Turk

Yurtsever

, et al. Apparent diffusion coefficient values for neuroendocrine liver metastases. Acad RadiolSuppl. 2021;28:S81-S86.

23.

Schmid-Tannwald

Thomas

Ivancevic

, et al. Diffusion-weighted MRI of metastatic liver lesions: Is there a difference between hypervascular and hypovascular metastases?. Acta Radiol. 2014;55:515-523.

24.

CT/MRI LI-RADS® v2018. https://www.acr.org/Clinical-Resources/Reporting-and-Data-Systems/LI-RADS/CT-MRI-LI-RADS-v2018. (accessed January 28, 2022).

25.

Choi

Lee

Sirlin

. CT and MR imaging diagnosis and staging of hepatocellular carcinoma: Part II. Extracellular agents, hepatobiliary agents, and ancillary imaging features. Radiology. 2014;273:30-50.

26.

Morin

Drolet

Daigle

, et al. Additional value of gadoxetic acid-enhanced MRI to conventional extracellular gadolinium-enhanced MRI for the surgical management of colorectal and neuroendocrine liver metastases. HPB (Oxford). 2020;22:710-715.