Hepatobiliary agents and their role in LI-RADS

Hepatobiliary agents and their role in LI-RADS

Thomas A. Hope,1 Kathryn J. Fowler,2 Claude B. Sirlin,3 Eduardo A. C. Costa,3

Judy Yee,1 Benjamin M. Yeh,1 Jay P. Heiken2

1Department of Radiology and Biomedical Imaging, University of California San Francisco, 505 Parnassus Avenue – 0628,

San Francisco, CA 94143-0628, USA2Department of Radiology, Washington University School of Medicine, St. Louis, MO, USA3Liver Imaging Group, Department of Radiology, University of California San Diego, San Diego, CA, USA

Abstract

TheLiver ImagingReporting andDataSystem (LI-RADS)was introducedwith the goal of standardizing the diagnosisof hepatocellular carcinoma.The 2014versionofLI-RADSincorporates the use of hepatobiliary contrast agents(HBAs) into the diagnostic algorithm, including gadoxetatedisodium and gadobenate dimeglumine. Three new ancil-lary features are introduced: hepatobiliary phase (HBP)hypointensity andHBP hypointense rim favormalignancy,while HBP isointensity favors benignity. HBP hyperinten-sity favors neithermalignancy nor benignity. In this review,we describe how to use these new features as well asnumerous pitfalls associated with the use ofHBAs, includ-ing hemangiomas, cholangiocarcinomas, and focal conflu-ent fibrosis. Importantly, findings on the HBP are notincluded as major criteria and therefore the criteria for thediagnosis of LI-RADS 5 observations remain unchanged,and so congruencewith theOrganProcurement TransplantNetwork system remains intact. Additionally, we reviewhow the major features in LI-RADS, arterial phase hype-renhancement, threshold growth, and washout and capsuleappearance, may be affected with HBAs. Notably withHBAs, hypointensity on the delayed phase, termed thetransitional phase, does not qualify as washout appearancedue to the possibility of early parenchymal enhancement. Itis hoped that the incorporationofHBAs intoLI-RADSwillhelp create consistency when interpreting HBA enhancedMRIs.

Key words: Hepatocellular carcinoma—Hepatobiliaryagents—Liver—LI-RADS—Gadoxetate—Gadobenatedimeglumine

The Liver Imaging Reporting and Data System (LI-RADS) is an imaging-based diagnostic algorithm aimedto standardize the diagnosis of hepatocellular carci-noma (HCC) in at risk patients [1]. LI-RADS intro-duces and defines terminology to help reduce variabilityin imaging interpretation and reporting. The first ver-sions of LI-RADS focused on traditional extracellularcontrast agents used in MRI and CT. In the 2014 ver-sion of LI-RADS, hepatobiliary contrast agents havebeen incorporated into the diagnostic algorithm. Thepurpose of this article is to review the current status ofhepatobiliary contrast agents (HBAs) in HCC imagingand how their use will be incorporated into LI-RADS(Fig. 1).

Hepatobiliary contrast agents

There are currently two main types of gadolinium-basedcontrast agents (GBCAs) used in liver MRI, extracellularand hepatobiliary agents. Hepatobiliary contrast agentsare those with sufficient hepatobiliary excretion to permithepatobiliary phase imaging in addition to the dynamicpost-contrast phases obtained with extracellular agents.The hepatobiliary phase (HBP) is the temporal windowduring which hepatic parenchymal enhancement occursas a result of hepatocyte uptake of contrast (Fig. 2). Thevariable presence or absence of hepatocytes with func-tioning transporters for HBAs yields additional diag-nostic information for characterization of focal liverobservations.

There are currently two HBAs routinely used inclinical practice, gadoxetate disodium (Eovist, BayerHealthcare) and gadobenate dimeglumine (Multi-Hance, Bracco) (Table 1). Although other agents haveminimal biliary excretion, for example gadofosveset(Ablavar, Lantheus), only these two agents have beenapplied and studied for liver imaging and HCCdiagnosis.

Hepatobiliary Phase LI-RADS Workgroup are given in ‘‘Appendix’’section.

Correspondence to: Thomas A. Hope; email: [email protected]

ª Springer Science+Business Media New York (outside the USA) 2014

AbdominalImaging

Abdom Imaging (2014)

DOI: 10.1007/s00261-014-0227-5

Adequacy of the hepatobiliary phase

In an ideal setting, gadoxetate disodium undergoes dual-elimination with 50% of the injected dose eliminated inthe bile. Peak parenchymal enhancement occurs atapproximately 20 min following injection and a 20-mindelay acquisition results in an adequate HBP with gad-oxetate disodium for most patients [2]. Gadobenate di-meglumine has lower biliary excretion of approximately2%–4%. While a standard HBP of 1 h in normal patientsis usually adequate, a greater delay of 1–3 h is necessaryfor HBP imaging in patients with underlying cirrhosis. Incirrhotic patients with decreased hepatic function orthose who have an elevated bilirubin, the percentage ofbiliary vs. glomerular excretion may be altered, resultingin delayed or inadequate parenchymal enhancement atthe standard HBP temporal window. Diminished HBPenhancement of the liver parenchyma may relate to de-creased number of functioning hepatocytes or dysfunc-tional hepatocyte transporters [3–5]. An adequate HBP isessential to allow for reliable characterization of obser-vations as hyper/hypo/iso-intense relative to the liverparenchyma. In an adequate HBP, the hepatic paren-chyma should be unequivocally brighter than the hepaticvasculature. If there is persistent contrast within theportal vein or lack of biliary excretion, the HBP may beconsidered sub-optimal [6]. In some instances, a longerdelay may yield improved hepatic parenchymalenhancement; however, predicting which patients willexhibit sub-optimal enhancement and adjusting theprotocol can be challenging in clinical practice.Some authors have found that liver function tests

(hyperbilirubinemia, hypoalbuminemia, prolonged pro-thrombin time, and thombocytopenia) and clinical fac-tors (Child-Pugh class, MELD score, presence of ascites)can be predictive of hepatocyte uptake and adequateHBP enhancement; however, this remains controversialwith conflicting results in the literature [7–9].

LI-RADS: HCC diagnosis

LI-RADS recognizes major features that when present incombination can be used to make the diagnosis of HCC(Fig. 1). While routinely used in practice, HBAs do notcontribute directly to the major features for HCC diag-nosis. This is in part due to the fact that HBAs are notrecognized in the organ procurement and transplantnetwork (OPTN) system of HCC diagnosis [10]. TheOPTN system and HBAs will be discussed in furtherdetail below. Because OPTN defines HCC diagnosis fortransplant priority in patients with liver disease, it isimportant for LI-RADS to maintain congruence. Inpractice, however, HBAs play an important role in thescreening and diagnosis of HCC. LI-RADS v2014introduces several ancillary features related to HBPenhancement. In addition to ancillary features, the con-trast pharmacokinetics of HBAs may impact the appli-cation of the diagnostic major features. The potentialimpact on major feature characterization and the ancil-lary features encountered with HBAs are discussed be-low.

Arterial phase hyperenhancement

Arterial phase hyperenhancement is an essential majorfeature for diagnosis of HCC in LI-RADS. No obser-vation can be categorized as a LI-RADS 5 (DefiniteHCC) observation in the absence of arterial phase hy-perenhancement. Capturing an optimal late arterialphase is critical for nodule to background liver dis-crimination. Gadobenate is dosed identically to that ofa conventional extracellular GBCA with an adminis-tered volume of 0.2 mL/kg. Gadoxetate is formulatedin half the volume (0.1 mL/kg) containing a quarter ofthe concentration of gadolinium. The decreased dose inadministered volume can create difficulty in catchingthe arterial bolus during the optimum late arterialphase. The late arterial phase, when the portal vein isfully enhanced but prior to hepatic vein enhancement,is essential for visualizing HCCs [11]. Many techniqueshave been suggested to improve the capture of thearterial phase including using multiphasic techniques,diluting the bolus, or doubling the size of the bolus[12–16].

In addition to issues related to a smaller bolus, mo-tion during the arterial phase is a commonly encounteredphenomenon with gadoxetate disodium. This has beenlabeled acute transient dyspnea, which appears to result

Fig. 1. LI-RADS algorithm 2014. Major criteria using hepa-tobiliary agents are identical to those used with extracellularagents and includes size, arterial phase hyperintensity,washout appearance, capsule appearance, and thresholdgrowth. Hepatobiliary phase (HBP) imaging adds two ancillaryfeatures suggesting malignancy (HBP hypointensity and aHBP hypointense rim) and one suggesting benignity (HBPisointensity). HBP hyperintensity suggests neither malignancynor benignity.

T. A. Hope et al.: Hepatobiliary agents and their role in LI-RADS

from an inability to adequately breath-hold during thearterial phase after the injection of gadoxetate [17].Altering the injection rate may help to mitigate thisphenomenon; however, it still remains a challenge inclinical practice. LI-RADS v2014, allows the combina-tion of major features between CT and MRI. Hence, ifthe arterial phase on the HBA MRI is inadequate but anobservation demonstrated arterial phase hyperenhance-ment on a recent CT examination, this in combinationwith other major features (washout or capsule appear-ance) observed on the MRI may be sufficient for diag-nosis of a LI-RADS 5 observation.

Washout appearance

Washout appearance is a major feature of HCC thatlikely results from a relative increase in hepatic arterialblood supply as well as decreased interstitial spacecompared to hepatic parenchyma. With extracellularGBCAs, washout appearance is defined as a reduction ofenhancement relative to the background liver on eitherthe portal venous or delayed phases. With gadoxetateadministration, hepatocyte uptake may start as early asthe end of the portal venous phase. With extracellularGBCAs, the post-contrast phases following the portal

Fig. 2. Dynamic and hepatobiliary phase (HBP) images withgadoxetate disodium. Arterial phase is noted with opacifica-tion of the portal vein (A arrow), while the portal venous phasedemonstrates opacification of the hepatic veins (B arrow-heads). On the 5-min delay, note that the vasculature ishypointense relative to the hepatic parenchyma (C). Withextracellular contrast agents, the vasculature will remainhyperintense relative to hepatic parenchyma, but due to

hepatocyte uptake of contrast and increased clearance ofcontrast, the vasculature becomes hypointense; this is termedthe ‘‘transitional phase.’’ On the HBP, typically acquired by20 min after gadoxetate administration, the vasculature ishypointense and biliary excretion is noted in the biliary tree (Ddotted circle). Note the vascular shunt in segment 2/3 (circle)that is hyperintense on the arterial phase and isointense on allother imaging phases.

Table 1. Generic names

Generic name Brand name Dose % biliary excretion HBP timing

Gadoxetate disodium Eovist 0.1 mL/kg, 0.025 mmol/kg 50 ~20 min (15–30 min)Gadobenate dimeglumine MultiHance 0.2 mL/kg, 0.1 mmol/kg 5 ~120 min (1–3 h)


venous phase are called the equilibrium or delayed phases.The term transitional phase has been proposed to describethe dynamic phases after the portal venous phase andbefore the HBP where both hepatocyte uptake andextracellular distribution may contribute to the enhance-ment pattern of an observation (Fig. 2). During thetransitional phase, observations may be hypointense dueto ‘‘washout’’ of contrast material and/or lack of hepa-tocyte uptake relative to that of the surrounding liver.Hence, hypointensity in the transitional phase may notrepresent washout as seen with extracellular agents. Thisphenomenon may be referred to as ‘‘pseudo-washout’’and has been described in the setting of hemangiomas, butcould be seen in any non-hepatocyte containing lesion[18]. Therefore with gadoxetate, washout should only bedescribed if there is unequivocally decreased signalintensity of an observation during the portal venousphase. This issue is not relevant to gadobenate adminis-tration, as hepatocyte uptake is delayed beyond the con-ventional dynamic phases that are acquired.

Threshold growth

Threshold growth is achieved when a lesion grows morethan 50% within 6 months or greater than 100% if over6 months. Dimensions on the HBP may differ fromdimensions measured on other sequences [19], andtherefore it is recommended that lesion diameter be ob-tained on the same phase when comparing two studies tomore accurately determine threshold growth.

Capsule appearance

Capsule appearance describes a peripheral rim of hype-renhancement that surrounds an observation seen on theportal venous or delayed dynamic post-contrast phases.The early parenchymal enhancement during the transi-tional phases after gadoxetate administration may maskthe capsule appearance, as both the tumor capsule/pseudocapsule and enhancing adjacent parenchyma maybe isointense (Fig. 3). Due to the rapid clearance of

Fig. 3. LI-RADS 5 HCC with arterial enhancement, wash-out appearance, and hepatobiliary phase (HBP) hypointen-sity imaged using gadoxetate. The arterial phasedemonstrates a 2.5 cm arterially enhancing lesion (A dottedcircle), which demonstrates washout appearance on the

portal venous phase (B arrows). A faint capsule can be seenon the portal venous phase that becomes isointense on thetransitional phase (B arrows, C). On the HBP, the lesion ishypointense relative to the hepatic parenchyma (D arrow-heads).


gadoxetate from the blood pool, it is possible that cap-sule appearance may be more difficult to characterize. Ifso, this may lower the sensitivity for capsule appearancewhen using gadoxetate.

Ancillary features that favormalignancy

Hepatobiliary phase hypointensity

HBP hypointensity is a new ancillary feature that favorsmalignancy (Fig. 3). As focal observations in the at riskliver progress from dysplastic nodules (DNs) to well-differentiated HCC, uptake of HBAs decreases relativeto the adjacent hepatic parenchyma. Therefore, HCC arecommonly hypointense on the HBP. Observations thatare only seen as hypointense on the HBP have beenshown in retrospective studies to frequently representeither DNs or well-differentiated HCCs; as large sizeobservations have increased risks of malignancy, theseare categorized LI-RADS 3 if less than 2 cm and LI-RADS 4 if greater than 2 cm [20, 21]. The lack ofhepatocyte uptake and resulting hypointense appearancein the HBP may be encountered in any non-hepatocytecontaining observation. Thus, this imaging feature is notspecific for HCC and can be seen in the setting of othermalignancies (LR-M, used to connote non-HCC malig-nancies in the at risk liver) as well as benign observations,hence it is an ancillary feature to be considered in thecontext of assigning HCC probability and is not con-sidered a major feature of HCC.

Hepatobiliary phase hypointense rim

HBP hypointense rim is a new ancillary feature that ismore frequently seen with hyperintense HCCs than be-nign hyperintense observations (Fig. 4) [22]. The hypo-intense rim is thought to be a correlate to the fibrouscapsule that surrounds HCCs, corresponding to thecapsule appearance on portal venous and delayed phaseimages with extracellular contrast agents.

Ancillary features that favor benignity

Hepatobiliary phase isointensity

HBP isointensity is a new ancillary feature that suggestsbenignity. This feature is most useful for characterizingnodular vascular shunts seen primarily in the arterialphase as benign or probably benign perfusion alterations(Fig. 5) [23].

Hepatobiliary phase hyperintensity

Hyperintensity seen on the HBP neither connotesmalignancy or benignity. Benign hepatocellular nodulessuch as focal nodular hyperplasia and some DNs maybe mildly hyperintense in the HBP. A small minority ofHCCs, between 5% and 10%, can be hyperintense inthe HBP as well, often times markedly hyperintense.One way to distinguish between a hyperintense HCCand a benign observation is the presence of a hypoin-tense rim on the HBP, which would suggest HCC(Fig. 4) [22].

Fig. 4. Hyperintense HCC with hypointense rim. On the 3-min delay coronal, the inferior component demonstrateswashout and capsule appearance (A arrowhead). The supe-rior component already is accumulating contrast withinhepatocytes (B circle). Note how the capsule appearance isseen only on the observations that are hypointense relative to

the hepatic parenchyma (A). On the coronal HBP phase im-age, the mass is hyperintense relative to the surroundinghepatic parenchyma with a surrounding hypointense rim (Barrows) representing a corollary to the capsule appearanceseen with extracellular contrast agents. This hypointense rimsuggests malignancy.


OPTN and hepatobiliary phase agents

Current Organ Procurement Transplantation Network(OPTN) guidelines do not include HBAs in their criteria[24]. The focus of OPTN guidelines is to determine whichlesions are definite HCC (LI-RADS/OPTN 5) in thecontext of organ allocation for liver transplantation.Patients with imaging findings of T2 stage (within Milancriteria) HCC by OPTN criteria may be eligible forautomatic exception points and improved wait list status.The intended context of LI-RADS extends beyond thescope of the transplantation setting; however, congru-ence between the two systems is essential for integrationin the clinical setting. To maintain congruence, HBAfeatures do not contribute to the diagnosis of LR-5observations in LI-RADS. The actual impact of HBAMRI on the assignment of major features for HCC usingLI-RADS and OPTN system is not known. It is possiblethat the sensitivity for LI-RADS 5 observations will belower with HBAs due to the exclusion of delayed phasehypointensity from the definition of washout appearance

as well as the possible decreased sensitivity for capsuleappearance. Alternatively, there may be improved sen-sitivity due to hypointense HBP observations that canhelp direct evaluation in the arterial phase and detectionof subtle arterial enhancement and washout that mayhave been missed otherwise. Studies are needed to di-rectly compare gadoxetate with conventional extracellu-lar contrast agents for assignment of LI-RADScategories and for characterization of major and ancil-lary LI-RADS features.

Pitfalls (Table 2)

Hemangioma

In cirrhotic livers, hemangiomas are frequently scleroticand rapidly enhancing. These lesions often enhance in thearterial phase and become isointense in the portal venousand delayed phases (referred to as pseudo-washout) [18].With gadoxetate administration, hemangiomas may be-come hypointense in the transitional phase [18]. Imaging

Fig. 5. Vascular shunts (Case 1) are typically seen asgeographic arterial enhancing regions (A circle) that becomeisointense on the hepatobiliary phase (HBP) (B). The findingof HBP isointensity can be helpful in demonstrating benignity,

particularly for nodular appearing vascular shunts. In somecases (Case 2), the perfusional abnormality (C) can result indecreased hepatocyte function that can be hypointense onthe HBP (D arrow).

Table 2. Summary of pitfalls when using hepatobiliary agents, and associated imaging characteristics

Pitfall Imaging characteristics

Hemangioma May mimic HCC as are hypointense in the transitional phase and HBP. Look for marked hyperintensity inT2-weighted imaging

Cholangiocarcinoma Target appearance, ductal dilation, and lobulated shape suggests cholangiocarcinoma over HCCCirrhotic nodules Typically demonstrates isointensity to mild hyperintensity in the HBP, and are less than 1 cmDysplastic nodules May be hypointense in the HBP overlapping with HCC. Lack of arterial enhancement and mild T2 hypointensity

suggests dysplastic nodulesFocal fat Loss of signal on out-of-phase imaging and the absence of suspicious enhancement patterns suggests focal fat,

although microscopic intra-lesion fat can be seen in HCCConfluent fibrosis Lacks early contrast enhancement, has T2 hyperintensity and is characteristically in segments 4, 5, and 8Perfusional abnormalities Typically only seen in the arterial phase, and are isointense in the HBP


with gadobenate mirrors extracellular GBCAs during thedynamic phases, and therefore enhancement on delayedimages mimicking the blood pool is diagnostic of a hem-angioma [25]. On HBP imaging, hemangiomas will behypointense mimicking a suspicious lesion [26]. The key indiagnosing hemangiomas is the utilization of T2-weightedimaging, where hemangiomas are characteristically bright(Fig. 6).

Cholangiocarcinoma and other malignancies(LR-M)

Any non-hepatocellular lesion will be hypointense in theHBP, including metastasis and primary hepatic malig-nancies such as cholangiocarcinoma (Fig. 7). HBP hy-pointensity is a common finding in intrahepaticcholangiocarcinoma [27, 28]. Key distinguishing features

Fig. 6. Hemangioma imaged with both extracellular and he-patobiliary agents. With extracellular agents, the observationremains mildlyhyperintense to the adjacent hepatic paren-chyma on the 5-min delay image (E arrow). Note that the ves-sels within the liver remain hyperintense relative to the hepaticparenchyma on the 5-min delay with extracellular agents. With

gadoxetate, the hemangioma becomes hypointense on the 5-min delay (H arrowhead), termed the transitional phase, whichbecomes more pronounced on the hepatobiliary phase (B solidcircle). The key to differentiating malignant observations fromhemangiomas with gadoxetate is the hyperintensity seen onT2-weighted imaging (A dashed circle).


between intrahepatic cholangiocarcinoma and HCCs areassociated ductal dilation, target appearance (arterial rimenhancement with peripheral washout and progressiveconcentric enhancement), and lobulated shape (Fig. 8).Features that favor HCC over cholangiocarcinoma in-clude diffuse arterial phase hyperenhancement, diffusewashout appearance, intra-lesional fat, nodule-in-noduleappearance, or mosaic architecture.

Cirrhotic nodules

Cirrhotic nodules are small nodules typically less than1 cm that make up the background of the hepaticparenchyma in patients with cirrhosis. Frequently thesenodules can be variably hyperintense in the HBP (Fig. 9).

Dysplastic nodules

DNs are nodules that differ from the background cir-rhotic nodules in enhancement pattern, fat or iron con-tent, or size. DNs are typically iso- to hyperintense in theHBP although they can be hypointense in the HBP,

particularly high-grade DNs (Fig. 9) [29]. DNs lack thearterial hyperenhancement and T2 hyperintensity seenwith hypervascular HCCs [30].

Focal fat

Focal fat can mimic malignant lesions on HBP images ifthey are performed with fat saturation (Fig. 10) [31].Drop of signal on in- and out-of-phase images as well astypical locations are helpful for diagnosis as well as theabsence of suspicious enhancement patterns. Care mustbe made not to confuse intra-lesional fat seen in HCC forbenign focal fat.

Confluent fibrosis

Due to the increased blood pool clearance of gadoxetatedisodium, there is no retention of contrast in fibroustissue as seen with extracellular GBCAs. Therefore,bands of fibrous tissue are hypointense relative to hepaticparenchyma in the HBP, and mass-like fibrous tissue asseen with focal confluent fibrosis can mimic malignant

Fig. 7. Cholangiocarcinoma. This segment 4 observationdemonstrates arterial enhancement (A arrowheads) and wash-out appearance on the portal venous phase (B). On the hepa-

tobiliary phase, the lesion is hypointense (C circle), but hascharacteristic ductal dilation seen best on the T2-weightedimaging (D arrow) suggesting cholangiocarcinoma (LI-RADSM).


Fig. 9. HCC with nodule-in-nodule appearance within adysplastic nodule. There is a small focus of arterialenhancement within a dysplastic nodule (A arrow). Withextracellular gadolinium, the dysplastic nodule remains iso-intense to the liver parenchyma on the delayed phase (Barrow), while with gadoxetate the dysplastic nodule be-comes hypointense relative to liver parenchyma on thedelayed phase, termed the transitional phase (C arrow-

head). The dysplastic nodule and focus of HCC are hypo-intense to the liver parenchyma on the hepatobiliary phase(D circle), with numerous mildly hyperintense cirrhotic nod-ules in the surrounding parenchyma. This lesion would becharacterized as LI-RADS 4 on both studies as the lesiondoes not demonstrate washout or capsule appearance.Hypointensity on the transitional phase does not qualify aswashout appearance.

Fig. 8. Cholangiocarcinoma with target appearance. On theportal venous phase, peripheral enhancement of the observationis noted (A),withprogressive central enhancement so that on thehepatobiliary phase the peripheral cellular component is hypo-intense due to the lack of functional hepatocytes (B arrows) while

portions of the fibrotic core retain contrast in the interstitial space(B arrowheads). On diffusion-weighted imaging, the peripheralcellular component can be seen as low signal intensity on theapparent diffusion coefficient (ADC) map (C arrows) due to therestricted diffusion relative to the fibrotic core.


Fig. 11. Focal confluent fibrosis. 20 min hepatobiliaryphase (HBP) after the injection of gadoxetate demonstratesa nodular liver with innumerable background cirrhotic nod-ules (A). In the segments 5 and 8, there is a wedge-shapedhypointense lesion (A arrows) with associated T2 hyperin-tensity seen on low b value diffusion-weighted imaging (C

arrows) consistent with focal confluent fibrosis. With gad-oxetate, there is decreased retention of contrast on theHBP in fibrous tissue due to the increased blood poolclearance. To distinguish confluent fibrosis from infiltrativeHCC, there is no evidence of contrast enhancementabnormality (B).

Fig. 10. Focal fat. On hepatobiliary phase imaging focal fatcan appear hypointense relative to adjacent hepatic paren-chyma (B circle), similar to the appearance on portal venousphase imaging (A circle). Focal fat will characteristically lose

signal on out-of-phase gradient echo imaging compared to in-phase imaging (C circle, D). It is important to differentiate dif-fuse fatty deposition seen in HCCs and benign appearing focalfat.


lesions (Fig. 11) [32]. The characteristic location in seg-ments 4, 5, and 8, morphology, lack of early contrastenhancement, and T2 signal intensity can be used to

make the diagnosis [33]. Portal vein thrombosis can alsomimic focal confluent fibrosis in the HBP due to wedge-shaped reduction in portal flow (Fig. 12).

Fig. 12. HCC with portal vein invasion and hepatobiliaryphase hypointensity related to perfusional alteration. 6 cmmass in the right lobe of the liver (A–C white arrows) invading

the feeding portal vein (A black arrow). The resulting perfu-sional alteration results in decreased hepatocyte uptake ofcontrast peripheral to the lesion (A arrowheads).

Fig. 13. Large infiltrative HCC in the right hepatic lobeinvading the portal vein (LI-RADS 5V). On the portal venousphase, there is no differential enhancement to allow for thedelineation of the hepatic tumor due to occlusion of the portal

vein from tumor thrombus (A). On the hepatobiliary phaseimages, the delineation of the tumor margins is betterappreciated (B arrows), although it can be difficult to distin-guish tumor from perfusional alterations.


Infiltrative HCC and portal vein thrombosis

Infiltrative HCC may be difficult to discern from thebackground liver due to poorly defined margins and poorlesion to parenchymal contrast (Fig. 13). Diffusion-weighted imaging can be useful when infiltrative tumor issuspected. Portal vein thrombosis may be masked byrapid portal venous clearance and diminished portal veinto tumor contrast (i.e., both the thrombus and the vesselwill be dark during later transitional and HBP images).Findings that suggest infiltrative HCC over focal con-fluent fibrosis include bulging liver contour, portal veinthrombosis, suspicious enhancement, and location [33].

Perfusional abnormalities

Alterations in vascular flow within the liver can result inaltered hepatic function. Although arterial shunts oftenhave no correlate on HBP images, portal venous occlu-sion can result in HBP hypointensity in the vasculardistribution of the occlusion. This finding can complicatethe delineation of HCCs that occlude the portal vein(Figs. 12, 13). Rarely vascular shunts may result in he-patobiliary phase hypointensity (5) [34].

Conclusion

In the 2014 version of LI-RADS, HBP findings will beincorporated into the diagnostic algorithm. HBP find-ings will not contribute to major features and LI-RADS 5 categorization. HBP isointensity and hypoin-tense rim are new ancillary features that suggestmalignancy. HBP isointensity is a new ancillary featurethat suggests benignity. While HBAs provide new in-sight into hepatic observations and may improve sen-sitivity of HCC diagnosis, there are many pitfalls andchallenges associated with their use. Knowledge of theimpact of HBAs on HCC major diagnostic features aswell as the role of HBAs in the imaging algorithms isessential for optimal interpretation of findings in atrisk patients.

Appendix: Hepatobiliary Phase LI-RADS Workgroup Members

Carlo Bartolozzi, Mustafa Bashir, Giuseppe Brancatelli,Victoria Chernyak, Jin-Young Choi, Kathryn Fowler,Masoom Haider, Jay Heiken (workgroup chair), Tho-mas Hope, Keyanoosh Hosseinzadeh, Hero Hussain,Kartik Jhaveri, Masayuki Kanematsu, Jeong Min Lee,John Leyendecker, Osamu Matsui, Don Mitchell, ElmarMerkle, Giovanni Morana, Takamichi Murakami, ScottReeder, Jens Ricke, Alla Rozenblit, Wolfgang Schima,Claude Sirlin, Jaap Stoker, Janio Szklaruk, ChristophZech.

References

1. American College of Radiology (2013) Liver imaging reporting anddata system (LI-RADS). American College of Radiology, Wash-ington, DC. http://www.acr.org/Quality-Safety/Resources/LIRADS. Accessed 25 May 2014

2. Bashir MR, Breault SR, Braun R, Do RK, et al. (2014) Optimaltiming and diagnostic adequacy of hepatocyte phase imaging withgadoxetate-enhanced liver MRI. Acad Radiol 21(6):726–732

3. Tajima T, Takao H, Akai H, Kiryu S, et al. (2010) Relationshipbetween liver function and liver signal intensity in hepatobiliaryphase of gadolinium ethoxybenzyl diethylenetriamine pentaaceticacid-enhanced magnetic resonance imaging. J Comput Assist To-mogr 34(3):362–366

4. Tamada T, Ito K, Higaki A, Yoshida K, et al. (2011) Gd-EOB-DTPA-enhanced MR imaging: evaluation of hepatic enhancementeffects in normal and cirrhotic livers. Eur J Radiol 80(3):e311–e316

5. Motosugi U, Ichikawa T, Sou H, Sano K, et al. (2009) Liverparenchymal enhancement of hepatocyte-phase images in Gd-EOB-DTPA-enhanced MR imaging: which biological markers ofthe liver function affect the enhancement? J Magn Reson Imaging30(5):1042–1046

6. Lee NK, Kim S, Kim GH, Heo J, et al. (2012) Significance of the‘‘delayed hyperintense portal vein sign’’ in the hepatobiliary phaseMRI obtained with Gd-EOB-DTPA. J Magn Reson Imaging36(3):678–685

7. Kim JY, Lee SS, Byun JH, Kim SY, et al. (2013) Biologic factorsaffecting HCC conspicuity in hepatobiliary phase imaging with li-ver-specific contrast agents. Am J Roentgenol 201(2):322–331

8. Kim H, Kim M-J, Park M-S, Cha S-W, et al. (2010) Potentialconditions causing impairment of selective hepatobiliary enhance-ment of gadobenate dimeglumine-enhanced delayed magnetic res-onance imaging. J Comput Assist Tomogr 34(1):113–120

9. Chernyak V, Kim J, Rozenblit AM, Mazzoriol F, et al. (2011) He-patic enhancement during the hepatobiliary phase after gadoxetatedisodium administration in patients with chronic liver disease: therole of laboratory factors. J Magn Reson Imaging 34(2):301–309

10. Wald C, Russo MW, Heimbach JK, Hussain HK, et al. (2013) NewOPTN/UNOS policy for liver transplant allocation: standardiza-tion of liver imaging, diagnosis, classification, and reporting ofhepatocellular carcinoma. Radiology 266(2):376–382

11. Mori K, Yoshioka H, Takahashi N, Yamaguchi M, et al. (2005)Triple arterial phase dynamic MRI with sensitivity encoding forhypervascular hepatocellular carcinoma: comparison of the diag-nostic accuracy among the early, middle, late, and whole triplearterial phase imaging. Am J Roentgenol 184(1):63–69

12. Hope TA, Saranathan M, Petkovska I, Hargreaves BA, et al.(2013) Improvement of gadoxetate arterial phase capture with ahigh spatio-temporal resolution multiphase three-dimensionalSPGR-dixon sequence. J Magn Reson Imaging 38:938–945

13. Zech CJ, Vos B, Nordell A, Urich M, et al. (2009) Vascularenhancement in early dynamic liver MR imaging in an animalmodel: comparison of two injection regimen and two differentdoses Gd-EOB-DTPA (gadoxetic acid) with standard Gd-DTPA.Invest Radiol 44(6):305–310

14. Tamada T, Ito K, Yoshida K, Kanki A, et al. (2011) Comparison ofthree different injection methods for arterial phase of Gd-EOB-DTPA enhanced MR imaging of the liver. Eur J Radiol 80(3):e284–e288

15. Motosugi U, Ichikawa T, Sano K, Sou H, et al. (2011) Double-dosegadoxetic acid-enhanced magnetic resonance imaging in patientswith chronic liver disease. Invest Radiol 46(2):141–145

16. Motosugi U, Ichikawa T, Sou H, Sano K, et al. (2009) Dilutionmethod of gadolinium ethoxybenzyl diethylenetriaminepentaaceticacid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging(MRI). J Magn Reson Imaging 30(4):849–854

17. Davenport MS, Viglianti BL, Al-Hawary MM, Caoili EM, et al.(2013) Comparison of acute transient dyspnea after intravenousadministration of gadoxetate disodium and gadobenate dimeglumine:effect on arterial phase image quality. Radiology 266(2):452–461

18. Doo KW, Lee CH, Choi JW, Lee J, et al. (2009) ‘‘Pseudo washout’’sign in high-flow hepatic hemangioma on gadoxetic acid contrast-enhanced MRI mimicking hypervascular tumor. Am J Roentgenol193(6):W490–W496


http://www.acr.org/Quality-Safety/Resources/LIRADS

http://www.acr.org/Quality-Safety/Resources/LIRADS

19. Nakamura Y, Toyota N, Date S, Oda S, et al. (2011) Clinicalsignificance of the transitional phase at gadoxetate disodium-en-hanced hepatic MRI for the diagnosis of hepatocellular carcinoma:preliminary results. J Comput Assist Tomogr 35(6):723–727

20. Motosugi U, Ichikawa T, Sano K, Sou H, et al. (2011) Outcome ofhypovascular hepatic nodules revealing no gadoxetic acid uptake inpatients with chronic liver disease. J Magn Reson Imaging34(1):88–94

21. Kumada T, Toyoda H, Tada T, Sone Y, et al. (2011) Evolution ofhypointense hepatocellular nodules observed only in the hepatob-iliary phase of gadoxetate disodium-enhanced MRI. Am J Roent-genol 197(1):58–63

22. Suh YJ, Kim M-J, Choi J-Y, Park YN, et al. (2011) Differentiationof hepatic hyperintense lesions seen on gadoxetic acid-enhancedhepatobiliary phase MRI. Am J Roentgenol 197(1):W44–W52

23. ParkY,Kim SH,KimSH, JeonYH, et al. (2010)Gadoxetic acid (Gd-EOB-DTPA)-enhanced MRI versus gadobenate dimeglumine (Gd-BOPTA)-enhanced MRI for preoperatively detecting hepatocellularcarcinoma: an initial experience. Korean J Radiol 11(4):433–440

24. UNOS (2013) OPTN policy 3.6: allocation of livers. UNOS.http://optn.transplant.hrsa.gov/PoliciesandBylaws2/policies/pdfs/policy_8.pdf. Accessed May 25 2014.

25. Gupta RT, Marin D, Boll DT, Husarik DB, et al. (2012) Hepatichemangiomas: difference in enhancement pattern on 3T MRimaging with gadobenate dimeglumine versus gadoxetate disodium.Eur J Radiol 81(10):2457–2462

26. Tamada T, Ito K, Yamamoto A, Sone T, et al. (2011) Hepatichemangiomas: evaluation of enhancement patterns at dynamicMRI with gadoxetate disodium. Am J Roentgenol 196(4):824–830

27. Peporte ARJ, Sommer WH, Nikolaou K, Reiser MF, et al. (2013)Imaging features of intrahepatic cholangiocarcinoma in Gd-EOB-DTPA-enhanced MRI. Eur J Radiol 82(3):e101–e106

28. Kang Y, Lee JM, Kim SH, Han JK, et al. (2012) Intrahepatic mass-forming cholangiocarcinoma: enhancement patterns on gadoxeticacid-enhanced MR images. Radiology 264(3):751–760

29. Kogita S, Imai Y, Okada M, Kim T, et al. (2010) Gd-EOB-DTPA-enhanced magnetic resonance images of hepatocellular carcinoma:correlation with histological grading and portal blood flow. EurRadiol 20(10):2405–2413

30. Sano K, Ichikawa T, Motosugi U, Sou H, et al. (2011) Imagingstudy of early hepatocellular carcinoma: usefulness of gadoxeticacid-enhanced MR imaging. Radiology 261(3):834–844

31. Yeom SK, Byun JH, Kim HJ, Park SH, et al. (2013) Focal fatdeposition at liver MRI with gadobenate dimeglumine and gad-oxetic acid: quantitative and qualitative analysis. Magn ResonImaging 31(6):911–917

32. Husarik DB, Gupta RT, Ringe KI, Boll DT, et al. (2011) Contrastenhanced liver MRI in patients with primary sclerosing cholangitis:inverse appearance of focal confluent fibrosis on delayed phase MRimages with hepatocyte specific versus extracellular gadoliniumbased contrast agents. Acad Radiol 18(12):1549–1554

33. Park YS, Lee CH, Kim BH, Lee J, et al. (2013) Using Gd-EOB-DTPA-enhanced 3-T MRI for the differentiation of infiltrativehepatocellular carcinoma and focal confluent fibrosis in liver cir-rhosis. Magn Reson Imaging 31(7):1137–1142

34. Sun H, Lee J, Shin C, Lee D, et al. (2010) Gadoxetic acid-enhancedmagnetic resonance imaging for differentiating small hepatocellularcarcinomas. Invest Radiol 45(2):96–103


http://optn.transplant.hrsa.gov/PoliciesandBylaws2/policies/pdfs/policy_8.pdf

http://optn.transplant.hrsa.gov/PoliciesandBylaws2/policies/pdfs/policy_8.pdf

Documents

Hepatobiliary agents and their role in LI-RADS