Keywords
CT - LI-RADS v2018 - LR-3 - LR-M - LR-TIV - MRI
Introduction
Liver Imaging Reporting and Data System (LI-RADS) v2018 is the newest version of LI-RADS,
representing the unification of the LI-RADS diagnostic criteria with the American
Association for the Study of Liver Diseases (AASLD) guidelines.[1] LI-RADS categorizes observations into ordinal categories based on their probability
of being hepatocellular carcinoma (HCC); LR-1 (definitely benign observation), LR-2
(probably benign), LR-3 (intermediate probability for malignancy), LR-4 (probably
HCC), and LR-5 (definitely HCC), and also differentiates HCC from non-HCC malignancies
(LR-M) and tumor in vein (TIV). LR-3 observations have an intermediate probability
of malignancy, with imaging follow-up preferred over active therapeutic intervention
of these observations. LR-4 observations are considered as probable HCC (not a 100%
certainty); however, the Asian Pacific Association for the Study of the Liver or Korean
Liver Cancer Association–National Cancer Center guidelines consider LR-4 observations
as HCCs,[2]
[3] and a multidisciplinary approach is recommended to manage such lesions, which can
vary based on institutional practices. Radiological differentiation of HCC from non-HCC
malignancies is critical, as the presence of a non-HCC malignancy may preclude the
patient from liver transplantation. LR-M observations represent a heterogeneous milieu,
and comprise of both atypical HCCs and non-HCC malignancies. TIV is a separate category
in the current version of LI-RADS, is an imaging biomarker of biological aggressiveness
and invasiveness, correlates with overall poor clinical outcomes, and is an absolute
contraindication for liver transplantation. Appropriate assignment of LR-3, LR-M,
and LR-TIV categories is critical in a setting of chronic liver disease to minimize
the number of biopsies, and to guide precise patient management. The purpose of this
article is to review the multimodal imaging features of LR-3, LR-M, and LR-TIV observations,
discuss the potential gray zones, and therapeutic implications of these observations.
LR-3
Development of HCC is a multistep carcinogenesis process. Early detection of HCC (<
10 mm) is vital, as these have a limited propensity to be angioinvasive,[4]
[5] and are therefore amenable to early intervention with a curative intent. Focal observations < 10 mm
in size are categorized as either LR-3 (intermediate probability of malignancy) or
LR-4 (probably HCCs; [Fig. 1A–D]). LR-3 category indicates an intermediate probability of malignancy, so as to warrant
routine surveillance with an imaging follow-up rather than active treatment.[1]
[6] The AASLD guidelines of 2018 have proposed a 10-mm threshold (subthreshold HCCs),
as these lesions are challenging to diagnose and characterize reliably due to their
small size, and are less likely to be malignant. In clinical practice, an increasing
number of subcentimeter size hypervascular observations are observed on contrast-enhanced
computed tomography (CECT) and magnetic resonance (MR) imaging (MRI) studies, with
the cumulative risk of HCC progression in these observations being higher than in
hypovascular nodules.[7]
[8] Ranathunga et al[9] showed that the rate of progression from LR-3 to LR-4 was 22.22% ([Fig. 2A–F]) and from LR-3 to LR-5 was 11.1% at least 12 months after the initial observation
was detected. Studies evaluating subcentimeter size observations showing typical imaging
features of HCC on gadoxetic acid-enhanced MRI and diffusion-weighted imaging (DWI)
showed that all nodules (100%) > 5.5 mm progressed to overt HCC within a year in patients
with a history of HCC[10] and 89.9 to 100% of the nodules progress to overt HCCs (≥ 1 cm) within 12 months.[10]
[11] Diagnostic efficacy and sensitivity of CECT and dynamic contrast-enhanced MRI for
detecting early HCCs is based on the observation size, and both modalities have a
low sensitivity in diagnosing subthreshold HCCs.[12] Based on the AASLD LI-RADS v2018 guidelines, current noninvasive diagnostic criteria
do not allow for characterization of a subcentimeter size observation as LR-5. Subset
of subcentimeter size observations demonstrate arterial phase hyperenhancement (APHE),
with either one of the following including nonperipheral washout, enhancing capsule,
or threshold growth and are categorized as LR-4 observations.
Fig. 1 (A–D) LR-3 observation: Axial 3-dimensional T1-weighted (T1-w) fast spoiled gradient-echo
(FSPGR) sequences reveal a 3-mm observation in the subcapsular location of segment
VI of right lobe of liver, which appears isointense to the parenchyma on the precontrast
(A) images, reveals non-rim-like arterial phase hyperenhancement (APHE) (yellow arrow)
on the arterial phase (B), without washout on the portal (C) and venous (D) phase images.
Fig. 2 (A–F) Growth of LR-3 to LR-4 observation in the presence of a treated hepatocellular carcinoma
(HCC): Contrast-enhanced computed tomography (CT) performed on December 20, 2018 reveals
a 6-mm LR-3 observation demonstrating non-rim-like arterial phase hyperenhancement
(APHE) (yellow arrow in A) in segment I, without washout on the portal (B) and venous (C) phase images. Follow-up CT performed on January 2, 2019 shows interval threshold
growth of the LR-3 observation, which presently measures 9.5 mm and reveals washout
(black arrows) on the portal (E) and venous (F) phase images, representing progression to a LR-4 observation. Note a large treated
LR-5 observation with patchy areas of lipiodol deposition (post-transarterial chemoembolization
[TACE]) in right lobe of liver.
Gray Zones in the Assessment of LR-3 Observations
Gray Zones in the Assessment of LR-3 Observations
-
Background liver may limit the detection of LR-3 observations: The presence of transient
hepatic attenuation/intensity differences (THADs/THIDs) in the liver parenchyma due
to inflammation, arterioportal shunts or fistulas, portal vein thrombosis, and posttreatment
sequelae can lead to either a false-positive interpretation or false-negative interpretation
due to obscuration of LR-3 observations. Background iron overload or steatosis may
cause the liver to appear darker than usual, limiting the ability to detect washout
resulting in a false negative. On the contrary, the presence of fibrosis, may lead
to a false impression of washout with a capsule resulting in a false positive and
upgrading of a LR-3 observation to LR-4. Radiologists must note that differentiating
fibrosis from capsule is extremely challenging and may lead to miscategorization.
-
Limited detection and characterization of subcentimeter size observations on CT: Both
LR-3 and LR-4 categories are heterogeneous entities, each assignable by different
combinations of imaging features, which may be further adjusted by the application
of ancillary features (AFs) and tie-breaking rules. Introduction of AFs and tie-breaking
rules introduces interreader variability, which results in lower agreement among radiologists
when assigning LR-3 or LR-4 categories. Due to inherent differences in imaging techniques,
MR tends to detect benign pseudolesions which are categorized as LR-2 or LR-3 more
often, which may remain undetected or uncategorized on CT. Presently, there is no
consensus on a particular choice of modality within the technical recommendations
of LI-RADS. Unlike MR, contrast CT protocols including rate of injection of contrast
are not optimized, which further contributes to the diagnostic dilemma. Limited soft
tissue resolution of CT precludes definitive assessment of washout or enhancing capsule
in subcentimeter size observations. Majority of the AFs are applicable only to MR,
and CT is incapable of assessing these features, which precludes accurate upgrading
or downgrading of these observations. Early HCC is categorized LR-4 more frequently
on extracellular contrast-enhanced MRI than on CT due to the presence of intratumoral
fat[13] and due to the presence of hepatobiliary phase (HBP) hypointensity on gadoxetate-enhanced
MRI.[14]
-
Hypoenhancing observations on the arterial phase (< 20 mm): Nonhypervascular observations
detected on multiphasic CT and MR pose a diagnostic challenge ([Fig. 3A–H]). Very little is known and understood about these nodules. Agnello et al[15] assessed the outcome of LR-3 observations without APHE and showed that 17 out of
55 (31%) LR-3 observations progressed to LR-5 at follow-up. A baseline diameter of
10 mm significantly increased the risk for LR-5 progression of LR-3 observations in
their study. The authors inferred that the hypoenhancing LR-3 observations could represent
precancerous lesions (dysplastic nodules or early HCCs) that develop hypervascularization
and progress to classic HCC during multistep hepatocarcinogenesis process.[16]
[17]
[18] HBP hypointensity is an AF and is related to a decreased or absent expression of
OATP receptors. LR-3 observations with HBP hypointensity and a baseline diameter of
15 mm have an increased risk of developing APHE.[16] However, many observations which are stable or are downgraded also show this feature
and hence category upgrade only based on this AF should be determined with caution.
PEARL: We believe that hypoenhancing LR-3 observations with a baseline diameter of
10 mm or greater should be closely followed-up because of an increased risk of progression
to LR-5. The use of ancillary criteria to upgrade or downgrade a hypoenhancing observation
requires a cautious approach, and more refinement of the guidelines for the use of
AFs in clinical practice is needed.
-
Ancillary features: Changes in hemodynamics of HCC vis-a-vis the background “At-Risk”
liver is the key to making an accurate diagnosis. However, this approach may not be
appropriate in certain populations, inherently reducing the sensitivity for detection
of early HCCs, which is attributable to incomplete neoangiogensis. AFs favoring malignancy
in general include mild to moderate T2 hyperintensity, fat sparing in solid mass,
iron sparing in solid mass, corona enhancement, transitional phase hypointensity,
HBP hypointensity, and restricted diffusion. Of these, HBP hypointensity[16] and DWI[19]
[20] contribute to higher sensitivity for HCCs on MRI. On the contrary, Shropshire et
al[21] assessed 141 LR-3 observations and showed that ancillary criteria did not contribute
to LR-3 observation category changes on follow-up studies. In their study, 40% (57/141)
of baseline LR-3 observations remained LR-3, 8% (11/141) were downgraded to LR-2,
and 42% (59/141) were downgraded to LR-1. Two percent were upgraded to LR-4 (3/141)
or 8% to LR-5 (11/141). In the setting of a new LR-3 observation, AFs such as mild-to-moderate
T2-weighted (T2-w) hyperintensity, restricted diffusion, or HBP hypointensity may
upgrade an observation to LR-4. In our experience, we do not recommend to upstage
observations based on ancillary criteria alone as AFs would need to be weighted appropriately
for each LI-RADS category adjustment, and refinement of the ancillary criteria and
LI-RADS guidelines may be warranted to reduce interobserver variability to improve
categorization of subcentimeter size hypervascular nodules.
-
Imaging follow-up versus biopsy versus immediate therapy: Management of subcentimeter
size hypervascular nodules is highly debatable and varies between different institutions.
As per the LI-RADS v2018 guidelines, LR-3 observations may undergo follow-up imaging
at 3 to 6 months' interval ([Fig. 4A, B]), whereas LR-4 observations would require a multidisciplinary approach to decide
an individualized management, which may include close follow-up imaging at < 3 months,
biopsy, or in few cases a therapeutic intervention. Biopsy of observations (< 1 cm)
is challenging, may not be technically feasible, and possesses an inherent risk of
sampling bias, with a potential risk of tract seeding. However, if subcentimeter observations
are pathologically confirmed to be HCC on biopsy, immediate treatment is recommended.
Due to these factors, a follow-up strategy based on threshold growth criteria is clinically
more feasible and agreeable to clinicians. Threshold growth is defined as > 50% increase
in diameter in < 6 months, represents a surrogate imaging biomarker for progression
to HCC, is a “major diagnostic criterion” in the LI-RADS v2018 and Organ Procurement
and Transplantation Network-United Network for Organ Sharing guidelines, and, therefore,
allows for the diagnosis of early HCCs (10–19 mm). Lastly, there is no specific recommendation
on prioritizing strategy for the therapeutic management of these indeterminate nodules,
as precise colocalization of such nodules is a technical challenge. There are no studies
comparing the clinical outcomes of subcentimeter nodules which have features suspicious
for HCC between immediate treatment and follow-up imaging, and there is insufficient
clinical evidence to recommend a standardized strategy for monitoring LR-3 observations.
Based on the current evidence, an interval of 6 months is optimal and cost-effective,
as a short-interval follow-up could increase treatment costs, whereas a longer interval
could increase the risk of diagnosing HCC when it is already untreatable.[22]
[23]
[24]
Fig. 3 (A–H) Stability of hypoenhancing (< 20 mm) LR-3 observation: magnetic resonance imaging
(MRI) performed on June 4, 2022 reveals a 14-mm observation (yellow arrows) in segment
VIII of right lobe of liver which reveals mild to moderate T2 hyperintensity (A), hypoenhances on the late arterial phase (B) without nonperipheral washout on the portal venous phase (C) and mild hypointensity on the hepatobiliary phase (D). Follow-up MRI dated October 2, 2022 shows interval stability of the segment VIII
observation (white arrows).
Fig. 4 (A, B) Stability of LR-3 observation: magnetic resonance imaging (MRI) dated March 22,
2022 shows a 3-mm observation demonstrating non-rim-like arterial phase hyperenhancement
(APHE) in the subcapsular location of segment VI of right lobe of liver (yellow arrow),
which has remained stable on the follow-up MRI performed on September 9, 2022.
LR-M
LI-RADS is a probabilistic scale developed primarily for the Western population. The
goal of LI-RADS Category 5 is to have a high specificity and positive predictive value
(PPV) for the diagnosis of HCC, to negate the need for histopathological diagnosis.
The major imaging criteria including observation size, non-rim-like APHE, nonperipheral
washout, capsule appearance, and threshold growth are intended to emphasize this high
specificity.[25]
[26] Majority of malignant observations that arise in an “At-Risk” liver are HCCs, which
could then be graded using the probabilistic scale of LR-1 to LR-5. However, in routine
clinical practice, many HCCs do not meet these stringent criteria. Further, non-HCC
malignant tumors deviate from the standardized probabilistic scale of LR-1 to LR-5,
as they demonstrate imaging features which are indicative of their non-hepatocellular
origin. As a diagnostic algorithm, LI-RADS is unique in that it provides a specific
category, namely, LR-M, for defining observations that are definitely or probably
malignant, but are not specific for the diagnosis of HCC. The imaging differentials
for LR-M observations include intrahepatic cholangiocarcinoma (IH-CCA), combined tumors
such as hepatocholangiocarcinoma (H-ChC), metastasis, lymphoma, and atypical HCC.
Even few benign entities such as atypical abscesses or sclerosing hemangiomas are
also included in the LR-M category. It is critical that radiologists must remember
that LR-M category does not exclude the diagnosis of HCC,[27]
[28]
[29] and the imaging features of such HCCs may have prognostic implication and could
imply poor clinical outcomes.[30] In this regard, though LI-RADS was principally developed as a diagnostic scale,
we do feel it may have prognostic implications which requires validation. LR-M category
observations are subcategorized as a targetoid mass or a non-targetoid mass ([Illustration 1]).
Illustration 1 LR-M in an “at risk liver”.
LR-M: Targetoid Appearance
LR-M: Targetoid Appearance
Imaging appearance of a targetoid mass is closely associated with the imaging appearance
of IH-CCA, as this is the most common non-hepatocellular primary malignant hepatic
neoplasm. Histologically, IH-CCA reveals a peripheral hypercellular rim (representing
viable tumor) and a central core of desmoplasia or ischemia. Temporal enhancement
features ([Fig. 5A–F]) of these targetoid observations mirror tumoral histology and include rim-like APHE
(enhancement in the observation periphery), peripheral washout on the venous phase
(washout is most pronounced in the observation periphery), and delayed central enhancement
(core of the observation enhances the late venous phases). Observations which demonstrate
rim-like APHE with or without progressive concentric enhancement are reported to be
biologically aggressive and have worse prognosis.[31]
[32]
[33] On the transitional phase or HBP, these observations reveal moderate to marked hypointense
periphery surrounding a milder hypointense core ([Fig. 6A–C]). Hepatobiliary-specific gadolinium agents are absorbed by hepatocytes via the OATP1
transported mechanism, including gadobenate disodium (Gd-BOPTA) and gadoxetate disodium
(Gd-EOB-DTPA), of which only gadobenate disodium is available for clinical use in
India. Radiologists must note that only 2 to 4% of gadoxetate disodium is absorbed
by hepatocytes and excreted in the biliary tree between 60 and 120 minutes' interval.
Quantum of biliary excretion is dependent on the liver function status, with poor
or nonexcretion seen in severe hepatocellular dysfunction or obstructive biliopathy
with elevated bilirubin levels, and therefore impacts the appearance of the observation
on the HBP. In our experience, a majority of these targetoid observations on HBP reveal
a relatively mild hyperintense core, the intensity of which is less or nearly isointense
to the background hepatic parenchyma but more than the intensity of the spleen, reflecting
contrast retention within the extracellular desmoplastic core. Appearance of these
targetoid observations on DW and T2-w sequences parallel the temporal enhancement
features and reveal a concentric morphology, comprising of moderate to marked hypointense
(restricted diffusivity) periphery with milder hypointensity (facilitated diffusivity)
in the core ([Fig. 6A–C]).
Fig. 5 (A–F) LR-M observation. A 63-year old male with chronic liver disease and portal hypertension
with elevated CA 19–9 (67.52 U/mL). A 34-mm observation (yellow arrows) in segment
VIII of right lobe of liver which appears hypointense to the background parenchyma
on the precontrast (A) three-dimensional T1-weighted fast spoiled gradient-echo (FSPGR) sequence, and demonstrates
rim-like arterial phase hyperenhancement (APHE) (B) with peripheral washout and progressive enhancement of the core on the venous phases
(C–F).
Fig. 6 (A–C) Targetoid morphology of LR-M observation in a 63-year-old male with chronic liver
disease and portal hypertension with elevated CA 19–9 (67.52 U/mL). On the hepatobiliary
phase (A), there is retention of contrast within its core (red arrow) surrounded by a thin
hypointense rim (dotted yellow arrows). On the T2-weighted (T2-w) image (B), the observation reveals a hyperintense core (red arrow) surrounded by a thick irregular
intermediate signal outer rind (dotted yellow arrows). On the diffusion-weighted (DW)
image (C), the core reveals facilitated diffusivity (red arrow) whereas the rind reveals restricted
diffusion (dotted yellow arrows).
LR-M: Nontargetoid Appearance
LR-M: Nontargetoid Appearance
Less common clinical scenarios which would necessitate the radiologist in applying
the LR-M category would include infiltrative masses without TIV, marked diffusion
restriction, necrosis, or severe ischemia, each in the absence of other features of
an LR-5 observation. Few AFs which could be suggestive of a non-HCC malignancy include
biliary dilatation, capsular retraction, and multiplicity. There is a relative paucity
of data on these imaging criteria and would require validation both in the research
and clinical setting.
Gray Zones in the Assessment of LR-M Observation
Gray Zones in the Assessment of LR-M Observation
-
Observation Size
-
Unlike LR-4 and LR-5 observations, there are no specific size criteria defined for
LR-M observations and smaller observations are known to be diagnostically challenging.
Small IH-CCAs in an “at-risk” liver may show features that are indistinguishable from
HCCs, including non-rim APHE with nonperipheral washout,[34] and may lack the typical targetoid morphology on the AFs on the T2-w and DW, and
HBP.[35] In our experience, such observations are discussed in the multidisciplinary tumor
boards and we recommend either close follow-up imaging at 1 to 2 months' interval
or alternatively a percutaneous biopsy if technically feasible, an approach that is
similar to LR-4 observations.
-
Combined Neoplasm (HepatoCholangiocarcinoma)
-
H-ChC are biphenotypic tumors which reveal overlapping features of HCC and IH-CCA.[36]
[37] These observations pose a diagnostic challenge and in a small minority of cases
may be classified as LR-5. The presence of even one targetoid LR-M feature (rim-like
APHE or peripheral washout on the venous phase or delayed central enhancement) is
sufficient to classify the observation as LR-M. If unequivocally present, rim-like
APHE in only a part of the observation would require the observation to be categorized
as LR-M. If there is a diagnostic dilemma between rim APHE and non-rim APHE, the radiologist
should err on the side of reporting rim APHE and assign an LR-M category to the observation.
-
Multifocality
-
LR-M observations may be multifocal in an “at-risk” liver. Multifocal LR-Ms ([Fig. 7A–D]) generally would be expected to demonstrate identical morphology and temporal enhancement
features, which would imply similar histology and biological aggressiveness. Occasionally,
LR-Ms may occur synchronously with observations which may be subclassified either
as LR-3, LR-4, or LR-5, and this is where the diagnostic conundrum lies ([Fig. 8A–D]). In such a clinical context, these observations are reviewed in a multidisciplinary
tumor board. Based on the multidisciplinary team (MDT) recommendations, a fluorodeoxyglucose
(FDG) positron emission tomography (PET)/CT is usually performed to assess FDG uptake
in the observations, the presence of extrahepatic disease, and aid in tissue sampling.
Alternatively, due to cost constraints, a percutaneous biopsy may be performed without
a FDG PET/CT to obtain histological proof of either a non-HCC or HCC malignancy in
view of the impact of tissue diagnosis on therapy and long-term outcomes.
Fig. 7 (A–D) Multifocal LR-M observations in a 63-year-old male with chronic liver disease and
portal hypertension with elevated CA 19–9 (67.52 U/mL). On the three-dimensional T1-weighted
fast spoiled gradient-echo (FSPGR) sequence acquired during the late arterial phase
(A–D), five discrete observations demonstrating rim-like arterial phase hyperenhancement
(APHE) are scattered in both lobes of liver. Note few small subcentimeter size (<
10 mm) LR-3 observations in segment VIII (A; dotted white arrows), and, two discrete enhancing skeletal lesions in a right-sided
rib and vertebral body.
Fig. 8 (A–D) LR-M observations in association with LR-3 and LR-5 observations. Treated c/o hepatocellular
carcinoma (HCC) in a hepatitis B virus (HBV)-induced chronic liver disease. Underwent
liver donor liver transplant (LDLT). On routine ultrasonographic surveillance, detected
to have few focal liver lesions in the transplant graft. On the axial three-dimensional
T1-weighted fast spoiled gradient-echo (FSPGR) sequence acquired during the late arterial
phase (A–D), two discrete observations demonstrating rim-like arterial phase hyperenhancement
(APHE) (A, B; yellow arrows) are seen in the transplant graft liver, the larger observation in
panel A revealing capsular retraction. Note a concomitant LR-5 observation (> 10 mm)
revealing a nodule-in-nodule appearance (C; red arrow) and a subcentimeter size (< 10 mm) LR-3 observation in the graft liver.
-
Role of Positron Emission Tomography/Computed Tomography in LR-M observations
-
FDG PET/CT is not recommended for the detection of HCC due to its low sensitivity.
Increased uptake of FDG in HCC has been reported to be a surrogate imaging biomarker
of biological aggressiveness and has also been associated with poor outcomes ([Fig. 9A–C]). Newer PET tracers such as 11C-acetate have a higher sensitivity for HCC detection and monitoring the response
to locoregional therapies. PET/CT with 11C-choline overcomes the limitations of 11C-acetate, and is useful in the therapeutic management of patients with HCC. Further
studies are required to clearly define the clinical impact of dual-tracer PET/CT in
patients with HCC. In our setting of live liver donor transplants, PET/CT may be used
as single imaging tool to exclude extrahepatic disease (nodes, adrenal, pulmonary,
or skeletal metastases) and to assess the quantum of FDG uptake in patients with HCC
beyond the University of California, San Francisco criteria or with a suspected TIV
prior to downstaging. Anecdotally, we have seen that PET/CT may be useful in determining
biologic aggressiveness of LR-Ms, which would allow the interventional radiologist
to biopsy the observation with the highest maximum standardized uptake value (SUVmax)
to obtain maximum diagnostic yield.
Fig. 9 (A–C) LR-M observation. LR-M observation (yellow arrow) in the posterior subcapsular aspect
of the transplant graft liver demonstrates rim-like arterial phase hyperenhancement
(APHE) (A), and increased metabolic activity (black arrows) on the 18-fluorodeoxyglucose (FDG)
positron emission tomography (PET)/computed tomography (CT) (B, C) images.
-
Biopsy
-
LR-M observations invariably undergo targeted percutaneous biopsy ([Fig. 10A–D]), at our institution, unless the biopsy is not technically feasible, or there are
specific contraindications to performing a biopsy (coagulopathy, patient is on blood
thinners, observations are in close proximity to critical vascular structures). Biopsy
of LR-Ms is imperative, as the presence of pathologically proven non-HCC malignancy
precludes the patient from transplantation ([Fig. 11A–F]). Alternatively, if an LR-M observation is sampled and pathologically proven to
be an HCC, there is a growing need for molecular and immune classification of HCCs.
In our experience, the cutting edge of the biopsy needle must traverse the peripheral
enhancing rim of the observation to obtain an optimum diagnostic yield. Follow-up
imaging is not recommended, unless the observation cannot be biopsied or the biopsy
is negative. Duration between the interval follow-up is decided as per MDT recommendations
and is individualized.
Fig. 10 (A–D) Positron emission tomography (PET)/computed tomography (CT) and biopsy of LR-M observation.
A 34-mm observation (yellow arrow) in segment VIII of right lobe of liver which demonstrates
rim-like arterial phase hyperenhancement (APHE) (A), and a targetoid morphology on the PET/CT (B, C) images comprising of a metabolically active outer rind (black arrows) surrounding
a relatively nonactive core (mimicking the appearance on the T2-weighted [T2-w] and
diffusion-weighted [DW] images). Axial CT (D) image shows a Trucut biopsy needle traversing the segment VIII observation.
Fig. 11 (A–F) Histopathology and immunohistochemistry (IHC) of the biopsied LR-M observation in
Fig. 10 revealed an intrahepatic cholangiocarcinoma (IH-CCA). Hematoxylin and eosin
(H&E) (A) shows an infiltrative neoplasm composed of atypical cells arranged in nests and
irregular anastomosing glands set in a fibrotic stroma. Tumor cells are diffusely
and strongly positive for PanCK (B), CK7 (D), and CK19 (E) and negative for Hepar1 (C) and synaptophysin (F), chromogranin and arginase (not shown). Images contributed by Dr. Toshi Mishra.
-
ExtraHepatic Findings in LR-M
-
As LR-M observations are considered to be biologically more aggressive, it would be
pertinent to look for extrahepatic metastatic disease. The common sites would include
regional (porta, periportal, common hepatic, retroportal-precaval), retroperitoneal,
or lower thoracic adenopathy, or organ-based disease such as adrenal, pulmonary, or
skeletal metastases ([Fig. 12A–C]). At our institution, either a CT scan of the chest, abdomen, and pelvis along with
a bone scan may be performed, or alternatively a whole body FDG PET/CT scan may be
performed to assess for extrahepatic disease. If extrahepatic disease is detected,
tissue sampling is pertinent prior to strategizing therapeutic options ([Fig. 13 A–F]).
Fig. 12 (A–C) A 63-year-old male with chronic liver disease and portal hypertension with elevated
CA 19–9 (67.52 U/mL) with multifocal LR-M observations along with nodal and skeletal
lesions detected on a fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed
tomography (CT). 18-FDG PET/CT images (A, B) reveal a metabolically active lesion in the left sacral ala (black arrows). Axial
CT (D) image shows a Trucut biopsy needle traversing the left sacral lesion.
Fig. 13 (A–F) Histopathology and immunohistochemistry (IHC) of the biopsied left sacral lesion
in Fig. 12 revealed metastatic cholangiocarcinoma (IH-CCA). Hematoxylin and eosin
(H&E) (A) shows fragments of bony trabeculae separated by anatomizing cords and trabeculae
of atypical cell in a dense fibrotic stroma. Tumor cells are diffusely and strongly
positive for PanCK (B), CK7 (D), and CK19 (E) and negative for Hepar1 (C) and synaptophysin (F), chromogranin and arginase (not shown). Images contributed by Dr. Toshi Mishra.
-
Other Causes of LR-M
-
Occasionally, an “at-risk” liver may develop hypervascular metastases from an extrahepatic
primary, which could be erroneously categorized as LR-4 or LR-5 observations. Presently,
there is no clinical data available which discusses the incidence of APHE metastases
in an “at-risk” liver and the frequency of misinterpretation using the LI-RADS probabilistic
categorization scale. LR-M category also includes few benign lesions such as abscesses
or sclerosis hemangiomas. Categorization of benign lesions as LR-M by the radiologist
is a major clinical problem, especially if these patients undergo surgical intervention
without prior histological proof. Noninvasive imaging biomarkers have yet to be devised
and validated to clearly differentiate probably or definitely malignant LR-M observations.
Further, when in doubt, these observations would always need to be biopsied for histological
proof.
LR-TIV
Vascular invasion is defined as the presence of tumor within the branches of the portal
or hepatic veins, and can be subclassified as either microvascular or macrovascular.
Microvascular invasion is observed on histological examination, the current gold standard.
Macrovascular invasion is observed on gross examination or imaging and is referred
to as TIV in LI-RADS. In a setting of HCC, the reported prevalence of TIV is approximately
6.5 to 44.0%.[38]
[39]
[40]
[41]
[42] Portal vein tumor thrombus in HCC indicates the invasive nature and biological aggressiveness
of the neoplasm, corresponds to reduced tolerance to chemotherapy and rapid deterioration
of liver function due to decreased reserve, and is associated with poor clinical outcomes.[43] Presence of TIV is an absolute contraindication for liver transplantation,[44]
[45] and hence accurate diagnosis of this entity and understanding the pitfalls in assessment
of TIV are critical to patient management.
Absence of a parenchymal observation does not preclude the LR-TIV category. However,
as a standalone imaging feature, the presence of TIV is not sufficient to classify
an observation as an HCC with vascular invasion, as other non-HCC malignancies may
also present with TIV. A systematic review and meta-analysis of 17 studies by Van
der Pol et al reported pooled percentages of HCC for all observations categorized
as TIV of 79%.[46] In another meta-analytic study, Kim et al[47] observed the pooled percentages of HCC and non-HCC in LR-TIV were 70.9% (95% confidence
interval [CI], 55.7–82.5%; I
2 = 59%) and 29.2% (95% CI, 17.5–44.4%; I
2 = 59%), respectively.
As per LI-RADS v2018, the LR-TIV category ([Table 1]) can only be applied in the unequivocal presence of enhancing soft tissue within
the vein, regardless of the presence of a parenchymal observation.[48] Presence of enhancement within the thrombus is a definitive feature ([Fig. 14A–E]), and is aimed to attain a high specificity toward the diagnosis of TIV which would
imply a 100% accuracy in detecting the presence of vascular invasion.
Fig. 14 (A–E) LR-TIV (tumor in vein) due to LR-5 observation. Note an observation in segment VIII
of right lobe of liver demonstrating nonperipheral washout with an irregular enhancing
capsule on the portal phase (A), contiguous with an enhancing soft tissue expanding the anterior division of the
right portal vein, with propagation of the TIV into main portal vein (black arrows).
Note the nonenhancing bland thrombus attached to the TIV in the main portal vein (white
arrows).
Table 1
Definitive and suggestive features of LR-TIV as per CT/MRI LI-RADS v2018
|
Imaging features diagnostic of TIV
|
Imaging features suggestive of TIV
|
|
Definition
|
Unequivocal presence of enhancing soft tissue in vein, regardless of presence of parenchymal
mass
|
• Occluded vein with ill-defined walls
• Occluded vein with restricted diffusion
• Occluded or obscured vein in contiguity with malignant parenchymal mass
• Heterogeneous vein enhancement
|
|
Applicable modalities
|
CT, MRI, CEUS
|
CT, MRI
|
|
Categorization
|
LR-TIV
|
Presence of these features prompt careful scrutiny for the presence of enhancing soft
tissue in the vein
|
|
Comments
|
CT/MRI - characterize TIV on any phase
CEUS–early arrival (same as hepatic arterial opacification) helps differentiate from
bland thrombus
|
|
Abbreviations: CEUS, contrast-enhanced ultrasound; CT, computed tomography; LI-RADS,
Liver Imaging Reporting and Data System; MRI, magnetic resonance imaging; TIV, tumor
in vein.
There are few imaging features which are suggestive but not definitive for the diagnosis
of TIV on CT/MRI, and are hence not specific enough to classify an observation as
LR-TIV. These include an occluded vein with indistinct walls, occluded vein with restricted
diffusion, occluded or obscured vein in contiguity with malignant parenchymal mass,
and heterogeneous enhancement ([Fig. 15A–C]). Thompson et al[49] reported that obscuration or poor delineation of the walls of a vein are considered
a feature associated with TIV rather than a bland thrombus. DW images may aid in differentiating
bland thrombus from TIV. LI-RADS v2018 recommends that DWI sequences should be acquired
with ≥ 2 b values (including b = 50 and b = 400–1000 second mm2). Based on these parameters, an occluded vein has a higher likelihood of representing
TIV if it reveals true restricted diffusion unequivocally higher DWI signal compared
with hepatic parenchyma and/or apparent diffusion coefficient (ADC) value unequivocally
lower than parenchyma. Though studies[50] have reported high sensitivities and specificities of DW-ADC for the detection of
TIV, there is considerable overlap between the ADC values of TIV and bland thrombus,[51] and hence DW is not that accurate to differentiate the two entities ([Fig. 16A–F]). Heterogeneous enhancement within an occluded vein is highly suspicious for vascular
invasion; however, the presence of collaterals in a thrombus may mimic enhancement.
Further, the readers must be aware of flow-related artifacts which may mimic enhancement
and represents a potential pitfall. In this context, CE ultrasound has a higher sensitivity
in detecting TIV as compared with CT/MRI, and may be used as a problem-solving tool
to differentiate TIV from bland thrombus.
Fig. 15 (A–C) Suggestive but not definitive features of LR-TIV (tumor in vein): Occluded vein
with indistinct walls (A), occluded vein with restricted diffusion (B), and occluded or obscured vein in contiguity with malignant parenchymal mass (C).
Fig. 16 (A–F) LR-TIV (tumor in vein) versus bland thrombus: Axial T2-weighted (T2-w) (A), diffusion-weighted (DW) (B; b-value 1,000), and portal venous phase (C) images acquired at the level of the porta reveal an expansile enhancing soft tissue
(yellow arrows; C) within the right portal vein and its anterior and posterior divisions extending
into the main portal vein, which reveals mild to moderate hyperintense signal on the
T2-w images (A) and restricted diffusivity (B). Axial T2-w (A), DW (B; b-value 1,000), and portal venous phase (C) images acquired at the level of portomesenteric confluence reveal a nonenhancing
intraluminal filling defect within the confluence (F) representing a bland thrombus (white arrows) which reveals moderate to marked hyperintense
signal on the T2-w images (D) and restricted diffusivity (E), suggesting considerable overlap in the T2-w and DW signal changes in TIV and bland
thrombus. Enhancing soft tissue within a vein is the ONLY unequivocal finding of LR-TIV.
Gray Zones in the Assessment of LR-TIV Observation
Gray Zones in the Assessment of LR-TIV Observation
-
High Specificity and Lower Sensitivity
-
Goal of LI-RADS category LR-TIV is to have a high specificity and PPV toward the diagnosis
of TIV. However, this results in an inherently lower sensitivity, and not all cases
of TIV would be accurately classified as LR-TIV. This is especially true when the
observations reveal an infiltrative or permeative growth pattern, such as those observed
with diffuse infiltrative HCCs. Infiltrative HCCs often do not reveal the characteristic
non-rim-like APHE and the arterial enhancement may be very heterogeneous, patchy,
nodular, or miliary, and washout is challenging to detect as it may mimic the background
fibrosis ([Fig. 17A–D]). TIV is often associated with diffuse infiltrative HCC.[52] In equivocal cases, subtraction images may help detect subtle enhancement in cases
of TIV.[53] Occasionally, TIV can be present in the absence of a definitive parenchymal mass
lesion ([Fig. 18A–F]).
Fig. 17 (A–D) A 70-year-old man with an infiltrative LR-M observation. Axial three-dimensional
T1-weighted fast spoiled gradient-echo (FSPGR) sequences acquired during the late
arterial (B) phase reveal a heterogeneously enhancing observation demonstrating nodular or miliary
morphology (black arrows in B) in the right lobe of liver. Washout is challenging to detect on the portal (C) and hepatic venous (D) phases images as it mimics the background fibrosis. Few discrete subcentimeter size
observations with rim-like arterial phase hyperenhancement (APHE) 9yellow arrows in
B–D) are seen within this large observation, which persist on the portal and hepatic
venous phases.
Fig. 18 (A–H) A 53-year-old man with cirrhosis. Tumor in vein (TIV) without a definitive parenchymal
mass. Axial three-dimensional T1-weighted fast spoiled gradient-echo (FSPGR) sequences
acquired during the late arterial (A–D) and portal venous (E–H) reveal enhancing TIV in bilobar portal venous segmental and subsegmental branches,
propagating into the right and left portal veins and the main portal vein. On the
portal venous phase, numerous linear or branching hypoenhancing observations in the
right lobe of liver represent TIV within the segmental portal venous branches. No
definitive parenchymal mass was observed.
-
LR-TIV in Non-HCC Malignancies
-
Non-HCC malignancies (LR-M) may present with TIV in cirrhotic patients, which can
easily be misinterpreted as HCC. Van der Pol et al observed that among the non-HCC
LR-TIV observations, 37% were combined HCC-CCA, 25% were IH-CCA, 6% were sarcomatoid
carcinoma, and 32% remained undetermined.[46] The presence of imaging clues such as rim-like APHE or peripheral washout should
alert the reader to the presence of a LR-M observation ([Fig. 19A, B]).
Fig. 19 (A, B) A 60-year-old male with chronic liver disease with elevated CA 19–9 to 1384 U/mL,
alpha-fetoprotein (AFP) 2.47 ng/mL, and PIVKA 10.2 mAU/mL. LR-M observation with tumor
in vein (TIV): Axial contrast-enhanced computed tomography (CT) images acquired in
the late arterial (A) and portal (B) venous phases reveal a large observation demonstrating rim-like arterial phase hyperenhancement
(APHE) in the right lobe of liver (LR-M), contiguously invading and expanding the
anterior division of the right portal vein and right portal vein, with enhancing TIV
seen (dotted yellow arrows).
-
Expansion of an Occluded Vein
-
Band thrombosis of the portosplanchnic veins occurs in cirrhotic patients due to portal
hypertension, reduced/sluggish portal flow, or the presence of a malignancy. Expansion
of a vein is considered a sign of vascular invasion; however, this sign is also seen
during acute thrombosis of a vein.
-
T1 Hyperintense Signal of a Thrombus
-
Occasionally, the presence of hemorrhagic component within a thrombus may appear hyperintense
on the unenhanced T1-w images, and mimic enhancement ([Fig. 20A–C]). In such cases, subtraction images will differentiate pseudoenhancement from true
enhancement.
Fig. 20 (A–C) Hemorrhage within a thrombus: Unenhanced axial three-dimensional T1-weighted fast
spoiled gradient-echo (FSPGR) sequence reveals intraluminal T1 hyperintense signal
within the left portal vein (dotted yellow arrow). Note an observation [M] in segment
VIII of right lobe of liver. On the arterial phase, a serpiginous enhancing structure
is seen anterior to the intraluminal T1 hyperintense signal in the left portal vein
(white arrow). Subtraction images reveal no enhancement within the left portal vein.
Persistent enhancement of the serpiginous structure confirms it to be a branch of
the left hepatic artery.
-
Enhancement within a Thrombus
-
If the clot is not dissolved, numerous collateral vessels develop around an occluded
vein which may mimic enhancement. Further, serpiginous vessels do develop in cases
of recanalization which also mimic enhancement and may lead to a spurious diagnosis
of LR-TIV.
-
Necrosis with the TIV
-
Necrosis of the intraluminal tumor within an expanded and occluded vein, may lead
to absence of intraluminal enhancement; however, this would not rule out the presence
of TIV.[52]
Conclusion
This article reviews the potential gray zones in applying major features and AFs in
assigning LR-3, LR-4, and LR-M categories on CT and MRI in routine clinical practice,
with potential solutions to these diagnostic challenges. In-depth understanding and
knowledge of the diagnostic algorithm, its gray zones and potential pitfalls, and
regional practice variations are pertinent for the precise diagnosis of these entities
and to guide patient management.
