Gadoxetate-Enhanced MRI as a Diagnostic Tool in the Management of Hepatocellular Carcinoma: Report from a 2020 Asia-Pacific Multidisciplinary Expert Meeting.

Gadoxetate magnetic resonance imaging (MRI) is widely used in clinical practice for liver imaging. For optimal use, we must understand both its advantages and limitations. This article is the outcome of an online advisory board meeting and subsequent discussions by a multidisciplinary group of experts on liver diseases across the Asia-Pacific region, first held on September 28, 2020. Here, we review the technical considerations for the use of gadoxetate, its current role in the management of patients with hepatocellular carcinoma (HCC), and its relevance in consensus guidelines for HCC imaging diagnosis. In the latter part of this review, we examine recent evidence evaluating the impact of gadoxetate on clinical outcomes on a continuum from diagnosis to treatment decision-making and follow-up. In conclusion, we outline the potential future roles of gadoxetate MRI based on an evolving understanding of the clinical utility of this contrast agent in the management of patients at risk of, or with, HCC.

WHAT TECHNICAL CONSIDERATIONS EXIST FOR THE USE OF GADOXETATE IN MRI?

It is important for us to be familiar with the inherent technical limitations of gadoxetate in magnetic resonance imaging (MRI) of the liver and relevant mitigation techniques before we proceed to examine its current and future utility in clinical practice.

Reduced Arterial Phase Enhancement

Arterial phase images using gadoxetate MRI have been described as unsatisfactory because of weak arterial enhancement, attributed to the smaller administered volume and lower gadolinium content of gadoxetate compared with other gadolinium-based agents [123]. In a retrospective intra-individual study of 49 patients with 56 hepatocellular carcinomas (HCCs), comparing HCC imaging features using gadoxetate MRI, extracellular contrast media (ECCM) MRI with gadopentetate dimeglumine, and dynamic computed tomography (CT), arterial phase hyperenhancement (APHE) on gadoxetate MRI was superior to CT and not inferior to gadopentetate dimeglumine [4]. More recently, a prospective study of dynamic contrast-enhanced (DCE) imaging in 66 patients with 83 HCCs showed similar perfusion parameters and lesion-to-liver contrast, even though liver parenchymal flow and enhancement occurred later with gadoxetate than with gadobenate dimeglumine [5]. However, in clinical practice, where visual evaluation of multiphasic contrast-enhanced MRI is the mainstay, Min et al. [6] showed that ECCM MRI identified APHE in a significantly higher proportion of patients (97.6%) than CT (81.5%; p < 0.001) or gadoxetate MRI (89.5%; p = 0.002). A broader consideration comparing the diagnostic performance of gadoxetate with ECCM and contrast-enhanced CT imaging, which reflects the discussions made by the advisory board, is covered in the ensuing sections of this review.

Several approaches have been proposed to increase arterial enhancement on gadoxetate MRI by achieving a favorable contrast bolus shape [3]. A slower injection rate of 1 mL/s stretches the bolus and increases the magnitude of the peak enhancement [3]. Gadoxetate doses above the approved 0.025 mmol/kg dose may prolong the peak arterial perfusion time and improve liver-to-lesion contrast [3]. Diluting gadoxetate with saline is not practical given the risk of non-sterility.

Subtraction imaging may further enhance the sensitivity of gadoxetate MRI for detecting APHE (Fig. 1). Subtraction images generated from unenhanced and arterial phases significantly increased sensitivity in diagnosing HCCs from 55.9% to 64.1%, with a non-significant decrease in specificity from 94.9% to 92.9%, in a retrospective analysis of 372 malignant nodules in 258 patients at risk of HCC [7].

Fig. 1

58-year-old male chronic hepatitis B carrier with HCC nodule that had APHE better depicted with subtraction imaging.

A. Diffusion-weighted imaging at 1000 s/mm2 shows a nodule in segment 4 with moderate restricted diffusion (arrow) worrisome for malignancy. B. Axial fat-suppressed 3D gradient-recall echo image in the portal venous phase shows the nodule (arrow) demonstrating hypointensity consistent with washout. C. No corresponding APHE is demonstrated at the same region (arrow) where the nodule is expected to be. D. Subtraction imaging, however, reveals the nodule to possess mild non-rim APHE (arrow). Histology revealed poorly differentiated HCC. APHE = arterial phase hyperenhancement, HCC = hepatocellular carcinoma, 3D = three-dimensional

Transient Severe Motion Artifact

Gadoxetate is associated with a very low risk of adverse events [8] and a lower frequency of allergic-like adverse events than macrocyclic ECCM [910].

Although usually a benign event, transient severe motion (TSM) adversely affects the image quality in the arterial phase, which is a critical part of HCC diagnostic assessment. TSM has been reported to be more common with gadoxetate than with other gadolinium-based agents and appears to be more prevalent in the West [1112].

Factors that could influence the occurrence of TSM include previous episodes of TSM, a higher (off-label) gadoxetate volume of 20 mL at a fixed dose of 2 mL/s, a history of allergy to iodinated contrast media for CT, low body weight, chronic obstructive pulmonary disease, oxygen administration, and modified breath-holding [3131415].

Multiple arterial phase image acquisitions after a fixed delay can provide at least one set of well-timed arterial phase images and, in one study, were not compromised by TSM in 81% of patients [1216]. Alternatively, shorter acquisition sequences can be used in single-phase examinations [3]. Motion-insensitive techniques, such as controlled aliasing in parallel imaging results in higher acceleration (dubbed as CAIPIRINHA), volume interpolated breath-hold examination (VIBE), and radial VIBE, are also promising [17].

Reduced Sensitivity for HCC Detection in Hepatic Dysfunction

The major uptake transporter of gadoxetate in the hepatocyte cell membrane is organic anion-transporting polypeptide 1B3 (OATP1B3). For the majority (85%) of HCCs, the expression of OATP1B3 is reduced during progression, which explains the typical hypointensity of HCCs on gadoxetate MRI [181920]. Reduced uptake of gadoxetate in poorly functioning livers can, therefore, reduce gadoxetate sensitivity for diagnosing HCCs [21]. Kim et al. [22] observed that in 189 patients with 240 HCCs, the diagnostic accuracy of gadoxetate MRI decreased with worsening severity of liver cirrhosis: from a mean accuracy of 0.974 in Child-Pugh class A to 0.904 in class B and 0.779 in class C. Concurrently, polymorphisms in OATP genes and concomitant medications that inhibit OATP1B3 may also reduce gadoxetate uptake [2324] and limit diagnostic sensitivity.

WHAT IS THE CURRENT ROLE FOR GADOXETATE MRI IN IMAGING DIAGNOSIS OF HCC?

To comprehend the current role of gadoxetate MRI for HCC diagnosis according to consensus guidelines, we will first need to compare its diagnostic performance against the established modalities of multidetector CT (MDCT) and ECCM MRI, followed by practical approaches to lesion assessment to maintain diagnostic performance.

Comparison to MDCT

Clinical studies have affirmed the superiority of gadoxetate MRI over MDCT for the sensitivity of diagnosis, particularly for small HCCs. Tsurusaki et al. [25] assessed the performance of combined dynamic phase and hepatobiliary phase (HBP) images on gadoxetate MRI in a prospective intra-individual comparison versus MDCT in 54 patients with 83 histopathologically confirmed HCCs. The combined interpretation of the dynamic phase and HBP of gadoxetate MRI provided a significantly higher sensitivity for lesion detection (83%) than MDCT (70%). Gadoxetate MRI has also been reported to be superior to MDCT in detecting recurrence after radiofrequency ablation (RFA) [26] and hepatectomy [27].

A meta-analysis of 242 studies (15713 patients) published in 2016 reported that gadoxetate MRI provided a significantly higher per-lesion sensitivity and positive predictive value for HCC diagnosis (85.6% and 94.2%, respectively) than contrast-enhanced CT (73.6% and 85.8%) [28]. This conclusion has been corroborated by recent studies [2930] and meta-analyses [31].

Comparison to ECCM MRI

The diagnostic advantages of gadoxetate over other MRI contrast agents are less clear [17]. In a retrospective analysis by Semaan et al. [30], gadoxetate MRI was significantly superior to ECCM MRI in per-patient sensitivity (95.2% vs. 89.5%), but not in per-lesion sensitivity (76.8% vs. 78.5%). A meta-analysis by Feng et al. [32] on eight comparative studies on the diagnosis of HCC also reported similar per-lesion sensitivity rates between ECCM MRI and gadoxetate; however, it also reported that the sensitivity rate of ECCM MRI was significantly lower for HCCs smaller than 2 cm (< 2 cm, 66%; ≥ 2 cm, 87%). As these analyses pooled diagnostic performance, one may argue that there could be publication reporting bias in favor of gadoxetate.

Indeed, in the prospective intra-individual study by Min et al. [33], ECCM MRI showed better sensitivity and accuracy than gadoxetate MRI for the diagnosis of HCC in the Liver Imaging Reporting and Data System (LI-RADS) LR-5 category. Other head-to-head comparison studies seem to suggest that gadoxetate has a lower negative predictive value than ECCM for definite criteria of HCC according to LI-RADS [634]. However, it is interesting to note that the majority of head-to-head intra-individual comparison studies did not demonstrate significant reductions in specificity for diagnosis with the use of gadoxetate, [6303435] except for the study by Paisant et al. [36], where, in nodules 1–2 cm in size, specificity dropped to 66.1% for gadoxetate MRI versus 85.7% for ECCM MRI. However, the sensitivity of gadoxetate has been found to be lower, perhaps attributable to lower per-lesion detection of APHE, as discussed earlier [635]. Depiction of the capsule appearance is poorer in the HBP of gadoxetate MRI [636], as is washout on the portal venous phase (PVP). Inclusion of transitional phase (TP) or delayed phase hypointensity, using gadoxetate as a criterion for washout, increases sensitivity comparable to [36], if not greater than [373839404142], sensitivity of washout on PVP for ECCM.

Few studies have compared gadoxetate with other commercially available hepatocyte-specific agents [4344]. The higher relative uptake of gadoxetate likely contributes to its superior diagnostic performance, particularly where a reduced signal is expected in the background liver tissues, such as in steatosis (Fig. 2) or reduced biliary excretory function. Overall, more head-to-head comparative studies are needed to recommend an MRI contrast agent over other agents.

Fig. 2

56-year-old female with obesity and metabolic syndrome, in whom gadoxetate better depicted an HCC nodule due to higher concentration of physiological contrast uptake by the surrounding non-tumorous liver, in the hepatobiliary phase.

A. Axial dual-gradient echo image in the opposed phase shows a circumscribed nodule (arrow) in subcapsular segment 5/8 of the liver surrounded by severe hepatic steatosis. B. Axial fat-suppressed 3D gradient-recall echo image in the hepatobiliary phase at 90 minutes post intravenous administration of gadobenate. The mass (arrow) appears hyperintense to the surrounding liver, which has a lower signal intensity than expected, due to the marked parenchymal steatosis. C. Axial fat-suppressed 3D gradient-recall echo image in the hepatobiliary phase at 20 minutes post intravenous administration of gadoxetate, performed 6 months following the prior scan (B). The mass (arrow) appears clearly as a hepatobiliary defect, due to significantly larger quantity of gadolinium accumulated in the non-tumorous liver. An estimated 50% of dose of gadolinium is excreted via bile in the case of gadoxetate, as compared to 3%–5% in the case of gadobenate. Histology confirmed well-differentiated HCC. HCC = hepatocellular carcinoma, 3D = three-dimensional

Incorporation into Consensus Guidelines for HCC Management

Both Asian and Western guidelines recommend surveillance for HCC in patients with liver cirrhosis or other risk factors [454647484950]. Asian countries, including China, Japan, and Korea, engage in systematic nationwide surveillance, which is associated with earlier curative treatment and improved survival [5152]. In Western countries, although surveillance is recommended by guidelines, its utilization in real-world practice is limited by physician and patient compliance [53].

Differences between Asian and Western guidelines reflect geographic variations in HCC epidemiology and treatment priorities [5455]. All guidelines include APHE as a major criterion for HCC. In North America and Europe, diagnostic criteria are designed to achieve high specificity in HCC diagnosis to identify candidates for transplantation [4648]. Assessment of washout is restricted to the PVP, which excludes the potential for false-positive results (“pseudo-washout”) in the TP and HBP [55]. Asian guidelines additionally include washout in the TP or HBP of gadoxetate MRI because this criterion improves sensitivity for lesion detection [4549565758] and thus provides opportunities for early interventional treatment [5559], with certain caveats to maintain specificity.

How to Overcome the Lower Specificity of Gadoxetate MRI for HCC Diagnosis

To address the decrease in specificity when incorporating TP/HBP washout into the diagnostic algorithm, findings from other MRI sequences that imply benignity or non-HCC malignancy could reduce the loss of specificity [6061]. Combined APHE plus hypointensity on PVP and/or TP and/or HBP plus non-LR-1/2/M, which reflects the 2018 KLCA-HCC guidelines for HCC diagnosis, showed a sensitivity of 92.5% and specificity of 87.4% [60]. By excluding hemangiomas or cholangiocarcinomas and nodules with a targetoid appearance, the sensitivity of diagnosis using HBP washout was 95.2% and specificity was 82.0% versus 75.3% and 94.1% for washout confined to the PVP; the gain in sensitivity was arguably greater than the loss in specificity [61]. Min et al. [62] found that the addition of at least two ancillary features, from among HBP-hypointense capsule, septum, and T2 spotty hyperintensity, significantly improved diagnostic accuracy (from 79.9% to 88.1%) and sensitivity (79.1% to 88.1%) without changing the specificity of HCC diagnosis (both 87.8%).

Combining imaging findings from different imaging modalities allows for the maximization of complementary strengths, such as real-time second-to-second assessment for the presence of APHE in contrast-enhanced ultrasound, as compared to the more conventional “static” evaluation of APHE on CT or MRI [63]. Further research on combined modality assessments, particularly with validation across different healthcare settings, is warranted.

WHAT UNIQUE ENTITIES DOES GADOXETATE ALLOW DEPICTION OF?

The use of gadoxetate revealed entities that would not have been otherwise depicted on ECCM MRI without the TP or HBP phases. These entities are often present in cirrhotic livers and are increasingly recognized to have clinical implications, contributing to the value of gadoxetate MRI.

Non-Hypervascular HBP-Hypointense Nodules

Non-hypervascular HBP-hypointense nodules (NHHNs) represent early HCCs, although approximately one-third are high-grade dysplastic nodules (HGDNs) [64]. Approximately one-third of NHHNs undergo hypervascular transformation to progress to HCC within 3 years [65]. Thus, NHHN may be associated with a higher grade of liver fibrosis. Progression to overt HCC is significantly higher in NHHNs than elsewhere in the hepatic parenchyma [66]. In the setting of hepatitis C virus infection, the risk of progression to HCC appears to be reduced when the viral load is eradicated with direct-acting antivirals (DAA) [67]; de novo HCC occurs more commonly than in patients without NHHN in the sustained viral response state [68].

Ancillary imaging features, such as T1 hyperintensity at baseline and a high growth rate, may be helpful in predicting HCC progression [69]. Concomitant T2-weighted (T2W) mild-to-moderate intensity and restricted diffusion-weighted images (DWIs) also increase the likelihood of malignancy [70]. Elevated serum alpha-fetoprotein (AFP) levels ≥ 100 ng/mL, well-defined margins, and hypointensity on pre-contrast T1W imaging suggest progressed HCC [71].

The significance of NHHNs is not completely understood, given that they can represent a range of lesions, from dysplastic nodules to progressed HCCs [72]. Therefore, the optimal management strategy for borderline lesions remains unclear. However, recent evidence points toward NHHNs as at-risk lesions in pretreatment as well as in post-treatment (RFA, resection, and transplant) settings (Table 1) [737475], although one study did not find a significant difference in survival compared to patients without NHHN [76]. A meta-analysis of eight studies on 842 patients found that the overall pooled hazard ratio (HR) for intrahepatic distant recurrence (IDR) was lower following hepatectomy (2.14) than RFA (3.07), suggesting a role for NHHN in stratifying patients who would benefit from the former [77].

Table 1

NHHN as a Clinical Prognostic Marker

Authors		Patients	Assessments	Key Results	Conclusion(s)
Increased Progression to HCC
	Kim et al. 2016 [69]	60 patients with CLD, with 114 NHHNs	Gadoxetate MRI	• 27 NHHNs in 21 patients transformed to HCC • Hypervascularization associated with T1 hyperintensity (HR 2.69; p = 0.021), history of HCC (HR 2.64; p = 0.021), and initial lesion size (HR 1.09; p = 0.046) • Growth rate of nodules more powerful determinant than baseline clinical and MR features	Careful follow-up should be considered after detection of NHHNs or a higher growth rate
	Hwang et al. 2017 [66]	714 patients with CLD, 120 with NHHNs	Gadoxetate MRI Transient elastography for LS assessment	• NHHNs significantly associated with log LS (OR 1.48; p = 0.002) and hepatitis B infection (OR 3.14; p = 0.017) • 2-year cumulative progression rate of overt HCC: 34.1% from corresponding nodules, 18.3% in other parts of liver (p = 0.071)	NHHNs on gadoxetate MRI frequently progressed to HCC
	Suh et al. 2017 [65]	944 patients with CLD, 1819 NHHNs (16 studies, meta-analysis)	Gadoxetate MRI	• Overall rate of hypervascular transformation: 28.2% • Cumulative incidence of progression: 1 year, 18.3%; 2 years, 25.2%; 3 years, 30.3% • Nodule size (cutoff ≥ 9 mm) was significant predictor for progression	NHHNs detected on gadoxetate MRI carry significant risk of transforming into hypervascular HCCs
	Joo et al. 2020 [71]	298 patients with CLD or cirrhosis, with 334 NHHNs	Gadoxetate MRIPathologic analysis	• Progressed HCCs diagnosed in 44.0%, early HCCs in 20.4%, high-grade DNs in 27.5%, and low-grade DNs or regenerative nodules in 8.1% of NHHNs • Independent predictors for progressed HCC: serum AFP level ≥ 100 ng/mL (OR 2.7; p = 0.01) and MRI features including well-defined margin (OR 5.5; p = 0.003), hypointensity at precontrast T1-weighted imaging (OR 3.2; p < 0.001), intermediate hyperintensity at T2-weighted imaging (OR 3.4; p < 0.001), and restricted diffusion (OR 1.9; p = 0.04)	NHHNs corresponded mainly to progressed HCCs, early HCCs, and high-grade DNs
	Shimizu et al. 2020 [67]	221 patients with HCV infection, 30 with NHHNs	Gadoxetate MRI before and after DAA therapy	• Progression of NHHN to HCC at 2 years significantly reduced after eradication of HCV (p = 0.022) • Hyperintensity on T2-weighted images (RR 14.70; p < 0.001) and achieving SVR (RR 0.29; p = 0.043) were independent factors associated with risk of HCC • During follow-up, 6 (9.2%) NHHNs in the SVR cohort became undetectable on HBP images	Eradication of HCV by DAAs could reduce the hypervascularization rate of NHHNs
	Toyoda et al. 2021 [68]	383 patients with HCV infection, 32 with NHHNs	Gadoxetate MRI before and after DAA therapy		Presence of NHHNs before DAA therapy is a strong risk factor for the development of de novo HCC after SVR
Predicting post-treatment recurrence
	Lee et al. 2015 [73]	139 patients with early-stage HCC, 110 with NHHNs	Gadoxetate MRI pre-RFA	Estimated 5-year RFS rate: 27.9% in patients with NHHNs vs. 71.3% in patients without NHHNs (HR 2.84; p = 0.006) IDR rate at 5 years: 67.5% in patients with NHHNs vs. 17.9% in patients without NHHNs (p < 0.001)	Presence of NHHNs on gadoxetate MRI is a predictive factor of recurrence after RFA of early-stage HCC, particularly IDR
	Song et al. 2017 [74]	141 patients with IDR after RFA for HCC	Gadoxetate MRI pre-RFA	• Precursor nodules (majority NHHNs) present in 46 (32.7%) patients • Time to recurrence was 16.6 months in patients with precursor nodules vs. 24.0 months for patients without precursor nodules (p = 0.011)	Patients with NHHNs had a shorter time to recurrence
	Lee et al. 2019 [75]	345 patients with single nodular HCC ≤ 3 cm, 81 with NHHNs	Gadoxetate MRI prehepatectomy or pre-RFA	• Presence of NHHNs was a significant factor affecting RFS after both hepatic resection (HR 2.75; p = 0.004) and RFA (HR 1.78; p = 0.004) • 5-year RFS after resection: 34.0% with NHHN vs. 65.0% with non-NHHN • 5-year RFS after RFA: 28.0% with NHHN vs. 51.0% with non-NHHN	In non-NHHN, resection offers higher RFS than RFA
	Kim et al. 2020 [77]	842 patients with HCC, 321 with NHHNs (8 studies, meta-analysis)	Gadoxetate MRI prehepatectomy or RFA	• Pooled HR for IDR in NHHN vs. non-NHHN groups = 2.44 (2.14 post-hepatectomy, 3.07 post-RFA)	The presence of NHHN increases risk of IDR and could stratify patients for hepatectomy
	Takeishi et al. 2020 [76]	290 HCC patients, 66 with NHHNs	Gadoxetate MRI prehepatectomy	• Untreated NHHN vs. no NHHN: no significant difference in RFS or OS up to 8 years post resection (p = 0.103 and p = 0.103, respectively) • Treated NHHN and untreated NHHN: no significant difference in RFS or OS (p = 0.158 and p = 0.109, respectively)	NHHN detected on gadoxetate MRI did not reflect prognosis of HCC after hepatectomy

AFP = alfa-fetoprotein, CI = confidence interval, CLD = chronic liver disease, DAA = direct-acting antiviral, DN = dysplastic nodule, HBP = hepatobiliary phase, HCC = hepatocellular carcinoma, HCV = hepatitis C virus, HR = hazard ratio, IDR = intrahepatic distant recurrence, LS = liver stiffness, MRI = magnetic resonance imaging, NHHN = non-hypervascular HBP-hypointense nodule, OR = odds ratio, OS = overall survival, RFA = radiofrequency ablation, RFS = recurrence-free survival, RR = relative risk, SVR = sustained virologic response

Subcentimeter Arterially Enhancing and Hepatobiliary Hypointense Lesions

Subcentimeter arterially enhancing and hepatobiliary hypointense lesions (SAELs) are another type of intermediate-probability lesions that do not fit the guideline criteria for definite HCC. They may be at a higher risk than NHHN: the majority of SAELs (57.7%) progressed to overt HCC within 2 years [78]. Venous or late dynamic phase washout was more frequently observed in malignant than in benign SAELs (57.7% vs. 30.6%, respectively) [78]. Interestingly, combining hyperintensity on T2W images or DWI with APHE and washout did not increase the specificity for predicting SAEL progression to hypervascular HCC (both 90.6%) [79].

For SAELs that fulfill the washout criteria for HCC, recurrence-free survival did not differ between early treatment or watchful waiting, even though patients with more SAELs had higher rates of recurrence [80]. Percutaneous ultrasound (US)/MRI fusion-guided RFA is a viable treatment option for subcentimeter recurrent HCCs [81].

Overall, while optimal treatment strategies for NHHN and SAEL are still evolving, the current literature suggests that they warrant close monitoring as they would allude to a higher risk of de novo or metachronous HCC in high-risk patients (Fig. 3).

Fig. 3

80-year-old male with chronic hepatitis B viral infection and known history of HCC, presenting with NHHN that developed definite HCC features within 6 months.

A. Axial fat-suppressed 3D gradient-recall echo image in the HBP (gadoxetate at 20 minutes) shows a subcentimeter hypointense nodule (arrow) in the subcapsular segment 8 near the hepatic dome. B. Corresponding late arterial phase image does not reveal enhancement—this would be in keeping with a NHHN. C. Axial fat-suppressed 3D gradient-recall echo image in the HBP (gadoxetate at 20 minutes) performed 6 months later shows increase in the size of the nodule (arrow). D. Corresponding arterial phase shows non-rim arterial phase hyperenhancement (arrow), in keeping with definite HCC. HBP = hepatobiliary phase, HCC = hepatocellular carcinoma, NHHN = non-hypervascular HBP-hypointense nodule, 3D = three-dimensional

WHAT IS THE IMPACT OF GADOXETATE MRI ON CLINICAL OUTCOMES?

Ultimately, the clinical utility of an imaging modality needs to be assessed beyond its diagnostic performance and measured by its impact on treatment outcomes and cost-effectiveness.

Treatment Decision-Making

Multidisciplinary team (MDT) care, with contributions from oncologists, hepatologists, gastroenterologists, surgeons, and radiologists, is widely regarded as the standard of care in HCC [468283], irrespective of the choice of MRI contrast agent. In several retrospective studies, MDT care was associated with improved survival [8485] and reduced mortality [86], particularly in patients with poor liver function [87].

The diagnostic arm of the Sorafenib and Micro-therapy Guided by Primovist Enhanced MRI in Patients with Inoperable Liver Cancer trial (dubbed as SORAMIC trial) was the first prospective study to investigate the impact of greater diagnostic accuracy on treatment decisions [88]. Gadoxetate MRI using extended criteria for typical, atypical, and early HCC and HGDN [89] identified significantly more patients with more than four lesions than CT and consequent treatment decisions more closely matched definitive treatment decisions: 81.2%–83.3% for gadoxetate MRI versus 70.8%–73.4% for CT. A small retrospective study addressing early HCC reported that the addition of gadoxetate MRI to MDCT detected 18 more (of 82) HCCs and changed therapeutic decisions in 11 more (of 33) patients [90]. Kim et al. [91] found that gadoxetate MRI identified 74 additional nodules in 53 of 323 (16.4%) patients initially investigated by CT, leading to a change in the Barcelona Clinic Liver Cancer stage and treatment plan in 43 (13.3%) patients.

In liver transplantation (LT), accurate lesion detection is critical for allocating recipient candidates according to the Milan criteria. The diagnostic performance of gadoxetate MRI either equals [92] or is superior to MDCT in LT cohorts. The additional use of HBP images significantly improved sensitivity for the detection of small (1–2 cm) HCCs and accuracy of patient allocation (from 88.9% to 92.1%) [73].

Survival and Recurrence Prediction

The higher lesion detection rates and positive impact on treatment decision-making provided by gadoxetate MRI compared to MDCT may translate to higher rates of overall and recurrence-free survival. In a large retrospective study (n = 30023 patients) by Kang et al. [93], all-cause mortality rates per 100 person-years were 36.3 in the CT alone, 15.2 in the CT + gadoxetate MRI (HR 0.64), and 21.7 in the CT + ECCM MRI groups (HR 0.71). For patients with localized disease, mortality was significantly lower in the CT + gadoxetate MRI group than in the CT + ECCM MRI group (HR, 0.89; p = 0.008). A primary limitation of this study is the potential selection bias, even after adjusting for confounders, including centers administering treatment. However, similar findings were reported in a smaller study, where direct comparisons between gadoxetate MRI and CT (thereby reducing confounding factors) were made [91].

The HBP features depicted by gadoxetate may be contributory. In a study by Kim et al. [94], the median disease-free survival (DFS) in the ECCM MRI group (35.8 months) was longer than that in the gadoxetate MRI group with NHHN (25.8 months) but shorter than that in the gadoxetate MRI group without NHHN (48.6 months). Another study reported that the presence of satellite nodules and peritumoral hypointensity in HBP were independent factors associated with tumor recurrence, regardless of the Milan criteria [95]. In patients who underwent LT, the presence of NHHN predicted DFS within and outside the Milan criteria [96].

A large retrospective study of 549 patients with HCC within the Milan criteria showed that a predictive model built on preoperative imaging and laboratory factors (aspartate aminotransferase:platelet ratio index, tumor size, arterial rim enhancement, and presence of NHHNs) was non-inferior to combined preoperative and postoperative histopathologic factors in predicting early recurrence after curative resection [97]. A local tumor progression prediction model following RFA for HCC has also been reported [98]. These multimodal prediction models require testing and validation in different practical settings.

Cost-Effectiveness Analyses of Gadoxetate MRI

Despite its limited sensitivity, US is widely used for HCC surveillance [99]. For new nodules > 1 cm detected on US in high-risk patients, further investigation using MDCT or MRI is mandatory. Gadoxetate MRI provides higher HCC detection rates and fewer false-negative findings than US [100] and could be more cost-effective than US for surveillance of high-risk populations [101102]. However, large prospective studies confirming the survival benefit of gadoxetate MRI over US are necessary for it to be considered a population screening tool.

Cost-effectiveness analyses are influenced by region-specific healthcare costs, reimbursement models, and many variables other than diagnostic accuracy alone. A study from Japan showed that gadoxetate MRI was associated with lower direct costs (US$18642; 2017 data) and generated a greater number of quality-adjusted life years (QALYs) (9.502) than ECCM MRI ($20274, 9.303 QALYs) or MDCT ($21279, 9.215 QALYs) [103]. An interesting health economic assessment of two different health systems (Korea and Thailand) drew similar conclusions [104]. In Korea, from the payer’s perspective, the total cost to reach a confirmed treatment decision was US$3087/patient using gadoxetate MRI, versus $3205 for MDCT and $3403 for ECCM MRI. In Thailand, gadoxetate MRI was the least costly option for the payer ($702/patient vs. $931 for MDCT, $873 for ECCM MRI), though less so from the provider’s (hospital’s) perspective. Another study based only in Korea [105] on early-stage HCC reported that QALYs were higher for gadoxetate (5.52) than for MDCT (5.08). In China, He et al. [106] found that the total diagnostic and treatment costs per patient after initial gadoxetate MRI evaluation were similar to MDCT (US$4586 vs. $4653; 2018 data) and lower than that for ECCM MRI ($4753).

Overall, despite higher upfront costs, gadoxetate MRI appears to be as, if not more, cost-effective than other advanced imaging modalities in the management of HCCs, after considering the reduced need for confirmatory diagnostic procedures and unnecessary treatment. A consensus on treatment protocols and transitional probabilities would facilitate the direct comparison of health systems, which is particularly relevant in Asia as national guidelines move toward adopting HBP hypointensity as a major criterion for definite HCC diagnosis.

PROMISING ROLES OF GADOXETATE MRI IN PATIENTS WITH HCC BESIDES THE HCC DIAGNOSIS

Here, we review emerging evidence for the future utility of gadoxetate in two promising domains: identifying HCC with pathologic features associated with poorer prognosis and one-stop pre-treatment assessment of regional liver function. A summary of the key references for each is provided in Table 2 [107108109110111112113114115116117118119120] and Table 3 [121122123124125126127128129130131132].

Table 2

Gadoxetate as a Measure of Tumor Aggressiveness

Authors		Patients	Assessments	Key Results	Conclusion(s)
Predicting MVI
	An et al. 2015 [107]	268 patients with single HCC	• Gadoxetate MRI prehepatectomy	• MRI features associated with early recurrence (< 2 years): rim enhancement (OR 3.83), peritumoral parenchymal enhancement in arterial phase (OR 2.64), satellite nodule (OR 4.07), and tumor size (OR 1.66) • Model derived from these variables had AUC of 0.788 in prediction of risk of early recurrence	Prediction model derived from gadoxetate MRI variables preoperatively can estimate risk of early recurrence
	Lee et al. 2017 [108]	197 patients with HCC ≤ 5 cm	• Gadoxetate MRI prehepatectomy	• MRI features associated with MVI < 2 years: peritumoral APHE (OR 5.184), peritumoral hypointensity on HBP (OR 4.705), and non-smooth margins (OR 3.555) • Early recurrence rates higher in patients with two or three significant MRI findings vs. none (27.9% vs. 12.6%; p = 0.030)	Combination of ≥ 2 gadoxetate MRI findings can be used as a preoperative imaging biomarker for predicting MVI, with specificity > 90%
	Hu et al. 2018 [111]	1163 patients (10 studies, meta-analysis)	• MRI presurgery • Histopathology	• MVI associated with peritumoral enhancement (OR 4.04; p < 0.05) and peritumoral hypointensity on HBP (OR 10.62; p < 0.05) • Diagnostic accuracy analysis revealed high specificity (0.90–0.94) but low sensitivity (0.29–0.40) for both features to assess MVI	Two peritumoral imaging features are significantly associated with MVI. These features highly suggest MVI only when present with a high false-negative rate
	Ahn et al. 2019 [112]	179 patients with single HCC	• Gadoxetate MRI prehepatectomy, including texture analysis of tissue heterogeneity	• MRI features associated with early recurrence (< 1 year), MVI, and tumor grade: satellite nodules and peritumoral HBP (p < 0.05) • Texture analysis added to MRI findings increased diagnostic performance for predicting early recurrence from 0.70 to 0.83	Gadoxetate MRI findings with texture parameters are useful to predict early recurrence, MVI, and higher grade
	Kim et al. 2019 [114]	167 patients with single HCC 2–5 cm	• Gadoxetate MRI radiomic model (3 or 5 mm peritumoral border extension) • Postoperative clinicopathological model	• Prognostic performance of gadoxetate MRI radiomic model (3 mm peritumoral border extension) was comparable to clinicopathological model (c-index difference −0.021; p = 0.758) for DFS • A combined gadoxetate MRI radiomic and clinicopathological model was not significantly different from the clinicopathological model	These findings suggest the importance of including peritumoral changes in the radiomic analysis of HCC
	Min et al. 2020 [116]	100 patients with single HCC ≤ 5 cm	• Gadoxetate MRI • Histopathology	• Based on four imaging features (non-smooth tumor margin, irregular rim-like enhancement in arterial phase, peritumoral APHE, peritumoral HBP hypointensity), overall inter-observer agreement was fair to moderate for MVI probability (κ = 0.41) • AUCs for diagnosis of MVI were lower for HCCs > 3 cm (range, 0.55–0.69) than for ≤ 3 cm (range, 0.59–0.75)	Considerable inter-observer variability exists in the assessment of MVI using MRI
	Hong et al. 2021 [117]	4274 patients with HCC (36 studies, meta-analysis)	• MRI presurgery • Histopathology	• MVI was associated with larger tumor size (> 5 cm; diagnostic OR 5.2), rim arterial enhancement (OR 4.2), arterial peritumoral enhancement (4.4), peritumoral hypointensity on HBP (8.2), non-smooth tumor margin (3.2), multifocality (7.1), and hypointensity T1WI (4.9) • Peritumoral hypointensity on HBP and multifocality showed very high meta-analytic pooled specificities for diagnosing MVI (91.1% and 93.3%, respectively)	MRI features predictive of MVI can be useful in the management of HCC
	Lee et al. 2021 [118]	516 patients with single HCC ≤ 3 cm	• MRI before surgical resection and RFA	• MVI was associated with AFP (≥ 15 ng/mL), protein induced by vitamin K absence-II (≥ 48 mAU/mL), arterial peritumoral enhancement, and hepatobiliary peritumoral hypointensity • AUC of the predictive model incorporating these factors was 0.87 and 0.82 in derivation and validation cohorts, respectively • Resection was associated with lower rate of early recurrence (< 2 years) than RFA based on risk of MVI after propensity score matching (p < 0.05)	The model predicted risk of MVI with high accuracy in patients with a small HCC
	Song et al. 2021 [120]	1618 patients (11 studies, meta-analysis)	• MRI presurgery • Histopathology	• MVI associated with non-smooth tumor margin (diagnostic OR 4.62) and incomplete tumor capsule (OR 2.25) • Sensitivity and specificity of these 2 parameters were 0.757 and 0.597 vs. 0.646 and 0.552, respectively • AUC of the non-smooth tumor margin variable for predicting MVI was 0.72 and AUC of the incomplete tumor capsule variable for predicting MVI was 0.62	Non-smooth tumor margins and incomplete tumor capsules on MRI are important for preoperative prediction of MVI
Predicting other aggressive histological features
	Chen et al. 2018 [109]	34 patients with HCC	• Free-breathing DCE-MRI using gadoxetate • Histology	• DCE-MRI-derived Ktrans was negatively correlated with Ki-67 indices (rho −0.408, p = 0.017) and histological grades (rho −0.444, p = 0.009) of HCC • Kep and Ve were related with tumor MVD (rho −0.405, p = 0.017; rho 0.385, p = 0.024, respectively) • Ktrans, Kep, and iAUC demonstrated moderate diagnostic performance for discriminating high-grade from low-grade HCC without significant differences	Free-breathing DCE-MRI is technically feasible and offers a potential avenue to differentiate between low-grade and high-grade HCC
	Choi et al. 2018 [110]	242 patients with HCC (204 CK19-, 38 CK19+)	• Gadoxetate and diffusion-weighted MRI presurgical resection	• Predictors of CK19+ HCC: irregular tumor margin (p = 0.024), arterial rim enhancement (p < 0.001), lower tumor-to-liver SI ratio on HBP images (≤ 0.522; p = 0.01), and lower tumor-to-liver ADC ratio (≤ 0.820; p < 0.001) • When all 4 criteria were satisfied, specificity of CK19+ HCC diagnosis was 99.5% • RFS rates were lower in patients with CK19+ vs. CK19− HCCs after resection (63.9% vs. 79.9% at 2 years; p = 0.001)	Gadoxetate and diffusion-weighted MRI features may be helpful to predict CK19+ HCC with early recurrence after curative resection
	Dong et al. 2019 [113]	91 patients with solitary HCC	• Gadoxetate MRI pre-resection • Histopathology	• Tumor incomplete capsules or non-capsules (p = 0.001) and intratumoral vessels (p = 0.002) were associated with BRAF expression • Tumor incomplete capsules or non-capsules (p = 0.001) and intratumoral vessels (p = 0.013) were associated with RAF1 expression • No MRI features were associated with expression of VEGFR2 or VEGFR3	The presence of incomplete capsule or intratumoral vessels and the absence of capsule are potential indicators of high BRAF and RAF1 expression. Gadoxetate MRI may facilitate choice of gene therapy in HCC
	Chen et al. 2020 [115]	115 patients with surgery-proven HCC	• Gadoxetate MRI presurgery • Progenitor cell markers: CK19 or EpCAM expression	• Predictors of HCCs expressing progenitor cell markers were AFP ≥ 155.25 ng/mL (p < 0.001), skewness on T2WI ≤ 1.10 (p = 0.024), uniformity on pre-T1WI ≤ 0.91 (p = 0.024), irregular tumor margin (p = 0.006), targetoid appearance (p = 0.001), and absence of mosaic architecture (p = 0.014) • Combining any three variables provided diagnostic accuracy of 0.86, sensitivity of 0.97, and specificity of 0.74	Noninvasive prediction of HCCs with progenitor phenotype can be achieved with high accuracy by integrated interpretation of biochemical and radiological information
	Rhee et al. 2021 [119]	476 patients with single HCCs, including 84 with MTM-HCC	• Gadoxetate MRI pre-resection • Histopathology	• MRIC for MTM-HCC: • MRIC-1, ≥ 20% arterial phase hypovascular component. MRIC-1 showed high sensitivity and negative predictive value (88% and 95% in training cohort, and 88% and 97% in validation cohorts, respectively) • MRIC-2, ≥ 50% hypovascular component and ≥ 2 ancillary findings (intratumoral artery, peritumoral APHE, and non-smooth tumor margin) MRIC-2 demonstrated moderate sensitivity and high specificity (47% and 94% in training cohort, and 46% and 96% in pooled validation cohorts, respectively). MRIC-2 was an independent poor prognostic factor for OS in both training and pooled validation cohorts

ADC = apparent diffusion coefficient, AFP = alfa-fetoprotein, APHE = arterial phase hyperenhancement, AUC = area under the receiver operating characteristic curve, CK19 = cytokeratin 19, DCE = dynamic contrast-enhanced, DFS = disease-free survival, EpCAM = epithelial cell adhesion molecule, HBP = hepatobiliary phase, HCC = hepatocellular carcinoma, iAUC = area under the gadolinium concentration–time curve, Kep = reflux rate constant, Ktrans = volume transfer constant, MRI = magnetic resonance imaging, MRIC = MRI criteria, MTM = macrotrabecular massive, MVD = microvascular density, MVI = microvascular invasion, OR = odds ratio, OS = overall survival, RFA = radiofrequency ablation, RFS = recurrence-free survival, SI = signal intensity, T1WI = T1-weighted imaging, T2WI = T2-weighted imaging, Ve = volume fraction

Table 3

Gadoxetate for Determination of Regional Liver Function

Authors		Patients	Assessments	Key Results	Conclusion(s)
To estimate liver function
	Ryeom et al. 2004 [125]	10 rabbits with CC14-induced liver injury	• Gadoxetate MRI-derived HEF, calculated from deconvolution analysis of aortic and hepatic parenchymal time–intensity curves • ICG retention at 15 minutes	• HEF correlated with change in ICG R15 (Pearson r = −0.965, p = 0.000)	Gadoxetate MRI HEF correlates with ICG R15 and represents a direct, noninvasive technique for quantitative evaluation of liver function
	Yamada et al. 2011 [126]	23 patients undergoing preoperative evaluation	• HUI derived from VL and mean SI on gadoxetate MRI • VS • ICG-PDR	• HUI and VS correlated significantly with ICG-PDR (R = 0.87)	Liver function can be estimated quantitatively from signal intensities and volumes of liver and spleen on gadoxetate MRI, which may improve estimation of segmental liver function
	Geisel et al. 2017 [121]	37 patients undergoing right PVE and extended right hemihepatectomy	• Lobar volume, KGR, RE, HUI, and FSF for each lobe, derived from gadoxetate MRI	• RE of LLL increased after PVE and decreased to 0.48 at 10 days after surgery • KGR was 14.06% ± 9.82%/week from PVE to 14 days post PVE • HUI of LLL increased after PVE (p < 0.05 at 14 and 28 days) vs. pre-PVE • HUI of residual liver decreased after surgery vs. presurgery
	Haimerl et al. 2017 [122]	131 patients with normal liver, Child-Pugh A and B	• Gadoxetate-enhanced T1 relaxometry-based indices and SI-based indices, including HUI • ICG-PDR	• All gadoxetate MRI-based liver function indices correlated with ICG-PDR. Relaxometry-based indices provided better correlation than SI-based indices with ICG-PDR • Taking account of liver volume provided stronger correlations for both SI-based and T1 relaxometry-based indices with ICG-PDR	Gadoxetate-enhanced T1 relaxometry, in combination with liver volume, is a potential tool for monitoring liver function
	Jung et al. 2018 [123]	29 patients undergoing image-guided SBRT	• SBRT targeting accuracy assessed by parenchymal changes on HBP gadoxetate MRI at 2–4 months • ICD between treated HCC and parenchymal changes	• Median ICD in 3D direction was 6.81 mm (IQR 4.27–9.61 mm) • No significant difference in ICD between intrahepatic marker and diaphragm guidance	Hepatic parenchymal changes on gadoxetate MRI can be used to assess targeting accuracy on SBRT
	Rassam et al. 2019 [124]	20 patients undergoing preoperative assessment of FRL function	• 99mTc-metbromin HBS • Dynamic gadoxetate MRI	• Gadoxetate MRI-derived mean Ki correlated with HBS-derived MUR for total and FRL function (Pearson r = 0.70, p = 0.001 and r = 0.89, p < 0.001, respectively)	Dynamic gadoxetate MRI is comparable to HBS for liver function assessment, with the potential to avoid PHLF
To predict postoperative hepatic dysfunction
	Asenbaum et al. 2018 [127]	62 patients undergoing major liver resection (≥ 4 segments)	• RLE for each FLR (remnantRLE), derived from mean SI on preop gadoxetate MRI • ICG-PDR • functFLR, calculated as FLR x remnantRLE ÷ body weight	• Probability of PHLF in 16 patients related to FLR (p = 0.015), proportion of FLR (p = 0.004), weight-adapted FLR (p = 0.003), remnant RLE (p = 0.002), and functFLR (p = 0.002) in univariate analyses and to functFLR in multivariate analyses (0.561; p = 0.002)	functFLR was superior to established variables in predicting PHLF
	Kim et al. 2018 [128]	73 patients undergoing liver resection	• RLE, FRLV, rHUI, rHUI-BW • ICG-PDR	• RLE, FRLV, rHUI, rHUI-BW, and ICG-PDR were independent predictors of PHLF in 18 patients (p = 0.011, p = 0.034, p < 0.001, p = 0.001, p = 0.003, respectively) on multivariate analyses • AUCs were larger for rHUI and rHUI-BW than other predictors, including significant differences vs. ICG-PDR (p = 0.016 and p = 0.0007, respectively)	Gadoxetate MRI predicted post-hepatectomy PHLF better than ICG-PDR
	Bastati et al. 2020 [129]	265 with CLD	• FLIS, derived from gadoxetate MRI HBP features: hepatic enhancement, biliary excretion, and SI in portal vein		FLIS identified patients with advanced CLD at increased risk for first hepatic decompensation and mortality
	Tsujita et al. 2020 [130]	41 patients with HCC	• Gadoxetate MRI prehepatectomy for HCC with portal vein invasion	• 9 patients developed severe PHLF. LSR of the remnant liver was significantly higher than that of the resected liver (p < 0.001) • 9 patients with severe PHLF demonstrated significantly lower rHUI (p < 0.001) and rHUI/HUI (p < 0.001) vs. no/mild PHLF • Decreased rHUI (p = 0.012, AUC = 0.885) and rHUI/HUI (p = 0.002, AUC = 0.852) were independent predictors of severe PHLF	Gadoxetate MRI may be a promising noninvasive examination for assessing global and regional liver function, allowing estimation of FLR and prediction of severe PHLF
	Huang et al. 2021 [131]	133 patients with HCC (45 patients received major hepatectomy)	• Gadoxetate MRI with computer-aided virtual hepatectomy • ICG test	• T1 relaxation time reduction rate (T1ratio) and FV correlated with ICG test (rho −0.304 and −0.449; p < 0.05) • Low rT1ratio (< 66.5%) and high rT1pos (> 217.5 ms) may predict major complications (AUC 0.831 and 0.756, respectively; p < 0.05) • rT1ratio was an independent risk factor for postoperative major complications (OR 0.845; p < 0.05)	Quantitative regional liver function assessed by gadoxetate MRI can predict short-term outcomes after major hepatectomy in patients with HCC
	Notake et al. 2021 [132]	67 patients undergoing major hepatectomy for biliary malignancy	• rHUI, derived from gadoxetate MRI • Total bilirubin • Prothrombin time • Preoperative ICG • ICGK-F • FLRV • FLR proportion (FLR:total VL)	• rHUI predicted grade B or C PHLF in 8 patients (AUC 0.896) • rHUI < 0.410 was predictive of increased risk of PHLF (OR 2.0 × 103) No other investigated indices were predictive of PHLF	rHUI is a potentially useful predictor of PHLF after major hepatectomy for biliary malignancy

AUC = area under the curve, CLD = chronic liver disease, FLIS = functional liver imaging score, FLR = future liver remnant, FLRV = FLR volume, FRL = future remnant liver, FRLV = FRL volume, FSF = fat signal fraction, FV = functional liver volume, HBP = hepatobiliary phase, HBS = hepatobiliary scintigraphy, HEF = hepatic extraction fraction, HR = hazard ratio, HUI = hepatocellular uptake index, ICD = intercenter discrepancy, ICG = indocyanine green, ICGK = ICG plasma clearance rate, ICGK-F = ICGK × FLR proportion, IQR = interquartile range, KGR = kinetic growth rate, LLL = left liver lobe, LSR = liver-spleen ratio, MRI = magnetic resonance imaging, MUR = mebrofenin uptake ratio, OR = odds ratio, PDR = plasma disappearance rate, PHLF = post-hepatectomy liver failure, PVE = portal vein embolization, RE = relative enhancement, rHUI = remnant HUI, rHUI-BW = remnant HUI corrected for body weight, RLE = relative liver enhancement, SBRT = stereotactic body radiation therapy, SI = signal intensity, VL = liver volume, VS = spleen volume

Measure of Tumor Aggressiveness

OATP1B3 expression is regulated in part by β-catenin signaling [133]. Beta-catenin-activated HCCs are associated with reduced levels of markers of aggressive biological behavior (AFP and AFP-L3 fractions) [134] and HBP uptake of gadoxetate [135136]. Incomplete capsules or non-capsules and intratumoral vessels have been reported to be associated with high BRAF and RAF1 expression [113], with implications for the use of targeted therapies.

The established imaging features associated with prognosis in staging systems for HCC include tumor size, number, and location [46484950]. Less established are features that can signal microvascular invasion (MVI), a known pathologic marker for poor prognosis, including non-smooth tumor margins, irregular rim-like APHE, and HBP hypointensity (Fig. 4) [107108116120137]. Ahn et al. [112] showed that peritumoral hypointensity on HBP was a significant independent predictor of early recurrence, MVI, and tumor grade. This finding was corroborated by two meta-analyses [111117]. Combined laboratory and imaging predictive models could predict MVI in single, small (≤ 3 cm) HCCs [118].

Fig. 4

55-year-old male with Child’s B liver cirrhosis secondary to viral hepatitis B found to have HCC with MRI features that depict microvascular invasion.

A. Axial fat-suppressed 3D gradient-recall echo image in the arterial phase showing a mass in segment 7, with peritumoral arterial phase hyperenhancement (arrow), rim-like enhancement, and non-smooth tumor margin. B. Axial image of the same lesion in the portal-venous phase shows corresponding peritumoral hypointensity (arrow). Histology confirmed HCC with microvascular invasion; the patient developed early recurrence of tumor 6.5 months following hepatic resection. HCC = hepatocellular carcinoma, 3D = three-dimensional

HCCs that express a progenitor phenotype, defined as positive for cytokeratin-19 (CK19) or epithelial cell adhesion molecule expression, are associated with poorer outcomes. Chen et al. [115] showed that in addition to serum AFP levels, skewness on T2W imaging, uniformity on pre-T1W imaging, irregular tumor margins, targetoid appearance, and absence of mosaic architecture were associated with the progenitor phenotype. In another study, Choi et al. [110] observed that irregular tumor margin, arterial rim enhancement, lower HBP tumor:liver signal intensity (SI) ratio, and lower tumor:liver apparent diffusion coefficient ratio were significant.

Quantitative parameters on DCE imaging of gadoxetate MRI can additionally provide information on the histologic grade and prognosis. The DCE-MRI-derived volume transfer constant correlated significantly with Ki-67 proliferation status (a measure of tumor growth rate) and histologic grade of HCC, and the reflux rate constant and volume fraction of the extracellular space correlated with tumor microvessel density [109]. Rhee et al. [119] found that gadoxetate MRI criteria based on the arterial phase hypovascular component were capable of stratifying the probability of macrotrabecular-massive HCC, a subtype associated with aggressive behavior and poor prognosis.

Radiomic models can reduce inter-observer variability [112], but their performance remains limited. Kim et al. [114] compared radiomics with a postoperative clinicopathologic and combined clinicopathologic-radiomic (CCR) model for the prediction of early (≤ 2 years) and late (> 2 years) DFS. The radiomic model showed lower prognostic performance than the CCR and clinicopathological models. As with the other aforementioned predictive models, more extensive testing and validation, along with the use of standardized imaging and pathological criteria, are necessary.

Assessment of Liver Function

99mTc-mebrofenin and indocyanine green (ICG) are recognized markers of liver function and are substrates of OATP1B1/B3 and OATP1B3, respectively [138]. Although the 99mTc-mebrofenin test is not commonly performed, ICG remains a common test for preoperative decision-making. The extent of gadoxetate enhancement has been shown to correlate with 99mTc-mebrofenin scintigraphy and ICG clearance and may therefore play a role in the assessment of liver functional reserve [124125126139], with the distinct added advantage of segmental liver function evaluation [127].

Bastati et al. [129] developed a functional liver imaging score (FLIS) from three HBP features of gadoxetate MRI: hepatic enhancement, biliary contrast excretion, and persistence of SI in the portal vein, predictive of first hepatic decompensation in patients with compensated advanced chronic liver disease (CLD), but not of decompensated advanced CLD. FLIS was an independent risk factor for mortality. Quantitative DCE assessment techniques have also been employed [124125].

For the accurate evaluation of quantitative liver function using gadoxetate MRI, it is important to correct for liver volume and the contrast enhancement effect of gadoxetate in the extracellular fluid space [126]. The hepatocellular uptake index (HUI) has been retrospectively validated as the most reliable indicator of quantitative liver function corresponding to the ICG clearance test, with T1-relaxometry-based indices providing a better correlation than SI-based indices [122]. HUI has also been prospectively validated in patients with portal venous embolization for monitoring segmental liver function [121] and for identifying participants contraindicated for major hepatectomy (i.e., ICG R15 > 20%) [139]. Gadoxetate MRI demonstrates post-stereotactic body radiation therapy hepatic parenchymal changes as a relatively sharp SI gradient and could potentially determine the targeting accuracy [123].

Liver parenchymal SI on HBP gadoxetate MRI can also be used to predict post-hepatectomy liver failure (PHLF) [140]. Studies in this area are limited by small sample sizes due to the relatively low proportion of patients suffering from PHLF. Yamada et al. [126] were among the first to correlate HUI with the plasma disappearance rate of ICG (ICG-PDR). Remnant HUI (rHUI) and rHUI corrected for body weight have subsequently been shown to predict PHLF [128130132]. T1 relaxometry-based indices have also been shown to predict major complications after hepatectomy [131]. Prospective validation of the proposed models and development of automated techniques are desirable.

CONCLUSIONS

Gadoxetate is a widely employed hepatocyte-specific MRI contrast agent with a favorable safety profile that combines the advantages of multiphasic MRI with functional uptake. The current literature alludes to the higher diagnostic accuracy that gadoxetate MRI confers over CT for the diagnosis of HCC; its advantage over ECCM is less certain.

Evolving evidence suggests that gadoxetate may improve treatment decision-making, thereby reducing disease recurrence and increasing patient survival. Limited data indicate some degree of cost-effectiveness, despite the higher cost of gadoxetate compared with ECCM in several Asian health systems. Newer applications of gadoxetate MRI are promising for determining HCC tumor biology and assessing non-tumorous liver function. Large prospective studies are necessary to address specific areas that would optimally define the use of gadoxetate MRI for the standard-of-care management of patients with HCC.