44
BENIGN AND MALIGNANT
PRIMARY LIVER NEOPLASMS
Clifford S. Cho • Yuman Fong
BENIGN LIVER TUMORS
Introduction
Benign tumors of the liver include hepatic hemangioma,
hepatocellular adenoma, focal nodular hyperplasia (FNH),
and other less common lesions arising from hepatic epithelial
or mesenchymal tissues (Table 44-1). Benign tumors of the
liver may be found in up to 20% of the population,1 and are
more than twice as common as malignant lesions. As a conse-
quence of increased availability and utilization of abdominal
computed tomography (CT) and magnetic resonance imag-
ing (MRI), they are now being diagnosed with increasing fre-
quency. Hemangiomas and FNH have an entirely benign nat-
ural history and therefore do not warrant resection; adenomas
carry a risk of growth, hemorrhage, or malignant transforma-
tion and should be treated operatively. Correct identification
of these lesions is therefore imperative, and establishment of a
definitive diagnosis is often the primary challenge in manag-
ing this group of patients.
Diagnostic uncertainty is common, and has been reported
to be the indication for operation in as many as 40% of
patients undergoing resection.2,3 Contemporary clinical,
laboratory, and radiographic studies are often incapable
of definitively distinguishing benign from malignant liver
lesions. Symptoms, physical examination findings, and liver
function tests are nonspecific. Tumor markers are normal in
many patients with malignancy, and therefore should not
be relied upon to identify a benign process. Hepatic ultra-
sonography is commonly employed, but often nonspecific.
Currently, the most accurate radiographic modalities are CT
and MRI. These are often complementary tests, and are usu-
ally diagnostic for hemangioma. MRI is becoming the diag-
nostic test of choice, with recent studies demonstrating an
accuracy of 85–95%.4 CT and MRI have traditionally been
less helpful in distinguishing adenoma from FNH,3,5 but
recent advances in MRI contrast agents selectively excreted
by hepatocytes suggest that use of such agents may permit
reasonable differentiation of FNH, which retains contrast
enhancement on delayed imaging, from adenoma, which
does not.6 Table 44-2 presents the MRI and CT characteris-
tics of the more common benign lesions.
A liver biopsy may be indicated when noninvasive tests
are not diagnostic. The precise indications for preopera-
tive liver biopsy remain controversial; in our practice, liver
biopsy is generally reserved for cases in which diagnostic
uncertainty persists after a thorough clinical and radio-
graphic evaluation. Biopsy may be performed by percuta-
neous or laparoscopic fine-needle aspiration (FNA) or core
needle biopsy. The diagnostic accuracy of liver biopsy may
be as low as 40%; therefore, the potential benefits of biopsy
must be balanced against the risk of iatrogenic bleeding and
the potential for tumor seeding of the peritoneal cavity if the
lesion is malignant.7,8 In a recent study from the Memorial
Sloan-Kettering Cancer Center (MSKCC), 30 of 68 patients
(44%) who underwent resection of a benign liver tumor
had a preoperative percutaneous biopsy, but the preopera-
tive biopsy provided the correct diagnosis in only 11 of the
30 cases (37%).9
Indications for resection of benign tumors include diag-
nostic uncertainty with a suspicion of malignancy, severe
mass-related symptoms due to lesion size, and in the case of
hepatic adenoma, the risk of hemorrhage, rupture, or malig-
nant degeneration. The decision to resect is based on patient
history, the radiographic appearance of the tumor, and the
surgeon’s clinical judgment. Because of the predominance
of hemangioma, the majority of patients with benign liver
tumors may be safely observed.2,9 The safety of nonopera-
tive management of asymptomatic patients with FNH or
hemangioma was illustrated in a large study in which 388
patients diagnosed with either hemangioma or FNH by
thorough radiographic evaluation were followed with obser-
vation alone.2 After a median follow-up of 32 months, 87%
of patients demonstrated complete stability in their tumor
size, and no patients developed tumor rupture. Figure 44-1
presents a flow diagram of our basic management approach
for presumed benign tumors of the liver.
When surgical resection is indicated, a variety of approaches
can be considered. These approaches include open or lapa-
roscopic resection or enucleation. When malignancy cannot
be excluded, resection should be performed with a margin of
927
928 Part VII Liver

Solitary hepatic tumor on CT or US

TABLE 44-1: BENIGN TUMORS OF
THE LIVER
MRI
Epithelial tumors Hepatocellular Focal nodular hyperplasia
Hepatocellular adenoma
Nodular regenerative
hyperplasia Uncertain diagnosis Adenoma FNH Hemangioma
Bile duct Bile duct adenoma
Bile duct cystadenoma Consider
Mesenchymal tumors Blood vessel Hemangioma percutaneous Symptomatic Asymptomatic
Hemangioendothelioma biopsy
Adipose Lipoma Observe
Muscle Leiomyoma Uncertain diagnosis
Others Abscess
Focal fatty infiltration Resection

FIGURE 44-1 Treatment algorithm for patient presenting with

presumed solitary benign tumor of the liver.

normal surrounding parenchyma. Resection is also necessary

when the tumor lies deep within the liver, as enucleation is gen-
erally not feasible in this circumstance. Enucleation for benign
tumors is often appropriate, and is our preferred approach for
tumors known to be hemangioma.10 Laparoscopic resection
is appropriate for selected patients with benign tumors, with
retrospective studies verifying the safety of laparoscopic subseg-
mental, segmental, and lobar resections.11,12
Hemangioma
EPIDEMIOLOGY AND ETIOLOGY
Hemangiomas are the most common benign tumors of
the liver; autopsy series suggest that they may be present
in up to 7% of the population.13 Their pathogenesis is
unknown. Hemangiomas of the liver may be associated
with hemangiomas of other organs, and may be multifo-
cal within the liver.14 Recent data suggest that there is no
difference in the incidence of liver hemangioma between
men and women, and sex hormones do not appear to
play a role in the growth of these lesions.15,16 An updated
review from MSKCC of 115 patients with liver heman-
gioma found that 35% of the patients were male.10 A
recent case-control study found no association between
liver hemangioma and the patient’s reproductive history
or use of oral contraceptives.15
PATHOLOGY
The typical cavernous hemangioma is a red-blue, soft, spongy
mass that usually measures between 1 and 2 cm in diameter.
When these lesions are greater than 4 cm in size, they are

TABLE 44-2: RADIOGRAPHIC FEATURES OF BENIGN TUMORS OF THE LIVER

Triphasic Contrast-Enhanced CT MRI

Delayed Delayed
Precontrast Arterial Phase Phase T1 T2 Sequence

Hemangioma Well-defined Peripheral nodular Centripetal Hypointense Hyperintense Centripetal

hypodensity enhancement enhancement enhancement
Focal nodular Well-defined Homogeneous Increased scar Hypo/isointense Isointense with Retains gadobenate
hyperplasia hypo/isodensity enhancement uptake possible increased dimeglumine contrast
scar signal enhancement
Hepatocellular Isodensity; fat, Homogeneous Possible Mixed, may be Mixed, may be Does not retain
adenoma hemorrhage or enhancement prolonged hyperintense hyperintense gadobenate
necrosis may be hyperdense dimeglumine contrast
present enhancement enhancement
 Chapter 44
referred to as giant hemangiomas. Histologically, these lesions
are sharply defined but not encapsulated, and are made up
of large, cavernous vascular spaces partly or completely filled
with blood separated by minimal connective tissue stroma.
The amount of stromal tissue is variable, and some lesions
may exhibit extensive areas of necrosis and fibrosis. Throm-
bosis is common, and dystrophic calcification may be present
within the lesion.
The microscopic interface between a hemangioma and
the surrounding liver parenchyma is one that has been well
described, and may influence the ease with which they can be
enucleated.17,18 Zimmermann has described four distinct pat-
terns of tumor-parenchymal interface: (1) a fibrous interface
characterized by relatively avascular capsule-like fibrous lamel-
lae, (2) an interdigitating interface characterized by a mixture
of parenchymal and hemangioma components without a
capsule-like interface, (3) a compression interface character-
ized by direct apposition between hemangioma and paren-
chyma, and (4) an irregular or spongy interface characterized
by a highly irregular border with numerous parenchymal foci
interspersed between dilated blood channels.18 The fibrous
interface is the most common interface in large hemangiomas,
and the presence of this capsule-like avascular fibrous interface
facilitates operative enucleation of these lesions.
DIAGNOSIS
Although cavernous hemangioma can be diagnosed by
biopsy, percutaneous biopsies should be performed with
caution because of their risk of hemorrhage and their subop-
timal diagnostic accuracy. In a recent study from MSKCC,
8 of 55 patients (15%) resected for hemangioma under-
went preoperative percutaneous biopsy.10 The diagnosis of
hemangioma was established by biopsy in only three of 13
patients (23%).
The diagnosis of hemangioma is most often rendered through
the use of high-quality CT and MRI. Noncontrast CT images
will reveal a well-defined hypodense mass that may contain
areas of calcification or central scarring. Contrast-enhanced
images using arterial and portal venous phase series will usually
demonstrate a typical pattern of peripheral nodular enhance-
ment (Fig. 44-2).19,20 In a study of 38 patients with 50 liver
hemangiomas, five specific CT criteria were found to be diag-
nostic of hemangioma: (1) a low-density lesion on an unen-
hanced scan, (2) early peripheral contrast enhancement, (3)
progressive centripetal opacification from the periphery toward
the center, (4) a delay of at least 3 minutes before total opacifi-
cation, and (5) an eventual isodense appearance. In this study,
38 of the 50 lesions (76%) demonstrated all five criteria, and
the presence of criteria 4 and 5 plus any other two criteria was
considered diagnostic.20
When hemangioma is clinically suspected but CT fails to
confirm the diagnosis, MRI should be performed. Hemangi-
omas typically exhibit hypointense signal intensity compared
with the surrounding liver tissue on T1-weighted imaging,
and hyperintense signal intensity on T2-weighted imaging.
The pattern of gadolinium enhancement is similar to that
Benign and Malignant Primary Liver Neoplasms 929
FIGURE 44-2 Arterial phase computed tomographic appearance
of hepatic hemangioma of the right hemiliver. Note the nodular
peripheral contrast enhancement.
seen on contrast-enhanced CT imaging. MRI may be more
sensitive than CT in detecting subtle enhancement, and
recent studies have demonstrated that MRI has a diagnostic
accuracy as high as 96% for hepatic hemangioma.4,21
Several other radiographic tests are available for the diag-
nosis of liver hemangioma, but are generally not indicated.
Because of the accuracy of CT and MRI, Tc-99m-labeled
red blood cell scans are infrequently necessary. However, in
circumstances where both CT and MRI are not diagnostic,
this test may be helpful and is certainly superior to ultra-
sound or even biopsy.22 Use of positron emission tomography
(PET) has also been described in the diagnosis of hepatic and
peripheral hemangiomas.10,23 Limited available data suggest
that hemangiomas are not 18F-fluorodeoxyglucose (FDG)-
avid, suggesting that PET may be helpful in differentiating
hemangioma from metabolically active malignancies.
TREATMENT
Most patients with hepatic hemangiomas should be managed
nonoperatively. The natural history of these lesions is usually
one of stability. Only rarely do patients develop symptoms
attributable to hemangioma, and the onset of spontaneous
rupture is reportable. Only 30 reports of spontaneous intra-
abdominal rupture of hepatic hemangiomas have been cited in
the literature.24 Size alone should not be used as a criterion for
resection, as recent reports have demonstrated that even large
lesions may be safely managed nonoperatively.2,25 In a recent
study from the Netherlands, 38 of 49 patients discovered to
have hepatic hemangiomas were followed nonoperatively. In
this study, the median diameter of nonoperatively managed
lesions was 6 cm. The median follow-up was 52 months,
during which time no patient experienced a complication or
developed symptoms attributable to the lesion.25
Three criteria we use as indications for resection of hem-
angioma are the presence of severe symptoms, an inability
to confidently exclude malignancy, and the development of
930 Part VII Liver
complications. Controversy persists as to whether resection is
advisable for symptom relief. Bismuth’s group reported that
7 of 14 patients (50%) who underwent treatment of heman-
gioma for symptom relief had persistent symptoms following
treatment, prompting them to advise caution against resec-
tion.26 However, it should be noted that 6 of the 14 patients
treated for symptomatic hemangioma underwent embolization
or hepatic arterial ligation rather than surgical excision; of the
seven patients with persistence of symptoms, five were from
the embolization or ligation group. Data from MSKCC and
others suggest that between 75 and 96% of carefully selected
patients who undergo resection or enucleation for symp-
toms will have relief,9,10,14,27,28 but it is imperative that these
patients  be carefully selected. Other pancreaticobiliary and
upper gastrointestinal (GI) causes for the symptoms must be
investigated, and the symptoms should be carefully matched to
the size and nature of the hemangioma. Lesion size appears to
be associated with the likelihood of symptoms. The median
lesion size among patients resected for symptoms at MSKCC
was 14 cm. In our experience, hemangiomas less than 4 cm
in size are almost exclusively asymptomatic, and symptomatic
lesions tend to be larger within the right hemiliver than within
the left.10
Another indication for surgical resection of hepatic
hemangioma is the development of tumor-related compli-
cations. As noted above, intra-abdominal hemorrhage is
extremely uncommon, but when it does occur, it should
be considered a life-threatening emergency treated with a
combination of angiography with embolization and surgery.
Intratumoral bleeding has been reported as a complication
of hemangioma, and is often thought to be associated with
the development of symptoms. Although this complica-
tion is not life threatening, onset of symptoms will often
lead to diagnosis of the lesion and eventual surgical evalu-
ation. Kasabach-Merritt syndrome is another complication
of large hemangiomas characterized by thrombocytope-
nia and consumptive coagulopathy.30 Its pathophysiology
is  thought to involve activation of the clotting cascade
by platelets trapped within the hemangioma. Contro-
versy exists as to the most appropriate intervention for
this condition. Recommended treatments have included
immunosuppressive agents, radiation, surgical resection,
and even transplantation.29,30
When surgical intervention is indicated, both resection
and enucleation have been advocated. Multiple studies
have documented that the majority of cases are treated by
enucleation.10,27,31 In a recent study from MSKCC, enucleation
was performed for 31 of 52 patients (60%) undergoing
operative treatment for hemangioma.10 We believe that this
technique decreases both operative time and operative blood
loss. In a recent analysis of 52 patients undergoing surgical
resection of hemangioma, enucleation was associated with a
lower rate of postoperative complications and a similar rate of
transfusion.31 Hepatic hemangioma is suited for enucleation
because of its benign nature and the fibrous cleavage plane
that often exists between the hemangioma and the surround-
ing hepatic parenchyma.
The technique for enucleation of hepatic hemangio-
mas has previously been described in detail.10,32 In general,
hepatic inflow control is initially achieved by use of the Prin-
gle maneuver. The hepatic artery ipsilateral to the tumor is
identified and dissected proximally to the level of the proper
hepatic artery. For large lesions of the right or left hemiliver,
ligation of the ipsilateral artery can result in some shrinkage
in tumor size. Smaller lesions do not require arterial ligation
and may be adequately devascularized with a more selective
ligation of distal branch vessels. Once arterial inflow has been
controlled, a small hepatotomy should be performed a few
millimeters from the edge of the hemangioma. Division of
this parenchyma will allow entry into the compressed sheath
of liver tissue that usually defines the border between hem-
angioma and normal parenchyma. Gentle dissection within
this sheath is usually possible, and small blood vessels or bile
ducts encountered during parenchymal transaction can be
easily controlled with clips or suture ligatures.
Focal Nodular Hyperplasia
EPIDEMIOLOGY AND ETIOLOGY
Focal nodular hyperplasia (FNH) presents as a nodule
composed of benign-appearing hepatocytes within a liver that
is otherwise histologically normal.33 The second most com-
mon benign tumor of the liver, FNH typically presents as a
solitary lesion and is often discovered incidentally. It occurs
most commonly in young women, with a female-to-male
ratio of 8:1. The average age of patients at the time of diag-
nosis is 35 years.34
The pathogenesis of this lesion is not well understood,
but thought to be a reactive process to a vascular injury or
malformation. The presence of large arteries within a cen-
tral fibrous scar and the absence of portal venous structures
are characteristic of FNH. Some have postulated that this
abnormal vascular anatomy results in chronic malperfusion
and secondary hyperplasia of the surrounding hepatocytes.35
Controversy exists as to whether this lesion, or its natural
history, is associated with the use of oral contraceptives.36,37
Contemporary data suggest that neither the presence, size,
nor number of lesions is influenced by the use of oral con-
traceptives. In an analysis of 216 women with FNH, oral
contraceptive use and pregnancy were not associated with
lesion growth.38 During follow-up, only four lesions (2%)
increased in size, and none of the 12 patients who became
pregnant experienced growth or complications related to
their lesion.
PATHOLOGY
FNH grossly appears as a discrete pale mass with lobula-
tions and abundant septae. A thin capsule often surrounds
the tumor, which is usually free of necrosis or hemorrhage.
A central scar is characteristic of FNH. More than one cen-
tral scar may be apparent, and dilated blood vessels are often
 Chapter 44
evident within the scar. These dilated vessels represent the
large central arteries that are typical of this lesion.
Technically, FNH is not a neoplastic process, but a hyper-
plastic one. FNH is categorized into typical and atypical
types. Typical, or classic FNH, is characterized by a central
stellate fibrous region that contains abnormal arteries but
not portal veins, and is multinodular with nearly normal-
appearing hepatocytes and mildly proliferative bile ducts.37
These bile ducts are generally located at the junction of the
abnormal hepatocytes and fibrous regions. Atypical FNH are
histologically classified as FNH, but do not exhibit one of
the classic features of typical FNH; for example, they may
lack a central scar, or may harbor a portal vein within the
central vascular region. These atypical lesions have been fur-
ther divided into three subtypes: (1) telangiectatic; (2) FNH
with cytologic atypia; and (3) mixed hyperplastic or adeno-
matous FNH.34,37
DIAGNOSIS
FNH does not require surgical resection; therefore, it is criti-
cal to differentiate this lesion from other hypervascular lesions
such as hepatocellular adenoma, hepatocellular carcinoma,
and various metastatic lesions. Radiographic imaging is the
current mainstay for diagnosis; however, percutaneous biopsy
may be warranted in situations where radiographic diagnosis
is not definitive.
Increased availability and recent improvements in con-
trast administration and image detection have resulted in
increased utilization and accuracy of CT imaging for the
diagnosis of FNH. The typical FNH will appear as a well-
demarcated hypointense lesion on noncontrast images. In
the arterial phase, the lesion will become uniformly hyperat-
tenuating because of the homogenous enhancement of the
entire lesion with the exception of the central scar. This pat-
tern is similar to that of hemangioma; however, within an
FNH, the enhancement is uniform throughout the lesion
rather than from the periphery. In the portal phase, the lesion
will become more isointense, and the central scar may show
enhancement as a result of gradual diffusion of the contrast
material into the fibrous scar.39
Currently, the most sensitive test for the diagnosis of FNH
is MRI with gadolinium enhancement, which has reported
FIGURE 44-3 MRI appearance of a large left lateral section focal nodular hyperplasia. Note the gradual contrast washout from the lesion and
enhancement of the central scar on delayed sequences.
Benign and Malignant Primary Liver Neoplasms 931
sensitivity and specificity rates of about 75 and 98%, respec-
tively.6,37,40 In general, FNH is hypointense on T1-weighted
images, and slightly hyperintense on T2-weighted images.
Similar to CT findings, FNH will show brisk and homoge-
nous enhancement with gadolinium. During later phases, the
central scar will enhance and may even become hyperintense
as contrast washes out of the lesion (Fig. 44-3). Other tis-
sue-specific contrast agents have been recently investigated.37
These agents are specific for Kupffer cells and hepatocytes and
may further increase the ability to identify FNH with MRI.6
When imaging studies are equivocal, percutaneous biopsy
may permit diagnosis of FNH in a majority of cases.37 In one
study, a histologic diagnosis of FNH was made by percutane-
ous biopsy in 58% of lesions with nondiagnostic radiographic
imaging characteristics.41 The typical FNH can be diagnosed
with core biopsy when benign-appearing hepatocytes, a prom-
inent arterial supply, absence of a portal vein, and peripheral
bile duct hyperplasia are observed histologically.
TREATMENT
When the diagnosis of FNH is confirmed, treatment should
be nonoperative in the vast majority of patients. FNH is not
a neoplastic process, and cannot be considered premalignant.
As noted above, growth of these lesions is uncommon, and
does not clearly appear to be related to the use of oral con-
traceptives or pregnancy.38 Therefore, resection should not
be recommended as a prophylactic measure against tumor
enlargement and rupture. Patients have on rare occasion been
reported to present with symptoms attributable to FNH.9,42
After careful diagnostic review, resection in such a setting
may be warranted for highly selected cases. Subsegmental
resections are generally preferable for FNH, and minimally
invasive approaches should be considered.
Hepatocellular Adenoma
EPIDEMIOLOGY AND ETIOLOGY
Like FNH, hepatocellular adenoma also occurs predominantly
in young women; however, unlike FNH, it is a neoplastic pro-
cess that is clearly associated with the use of oral contraceptive
932 Part VII Liver
pills (OCPs), as well as with type 1 glycogen storage disease
and diabetes.43 Hepatocellular adenoma was rarely reported
before the introduction of oral contraception, and is believed
to be four times more common in women who use OCPs
compared with those who do not.43 The risk of developing
hepatocellular adenoma is proportional to the duration of
OCP use; patients who have used OCPs for more than 9 years
have a risk of developing hepatocellular adenoma that is
25 times that of the general population. Complications from
hepatocellular adenoma have also been associated with the
use of OCPs. A review of 237 patients found that those tak-
ing OCPs presented with larger tumors (97% >5 cm vs 75%
>5 cm) and were more likely to present with tumor rupture or
hemorrhage than those not taking OCPs (65 vs 25%).44
A majority of hepatocellular adenomas will be symptom-
atic at presentation, and complications associated with these
lesions include tumor growth, rupture with intraperitoneal
hemorrhage, and malignant transformation. The risk of
rupture and intraperitoneal hemorrhage may be as high as
30–50%.45,46 In a review of 54 patients with hepatocellular
adenoma, 21 patients (39%) were diagnosed after the onset
of hemorrhage into the tumor or peritoneal cavity, and only
four patients (7%) had the lesion discovered incidentally.47
In a recent multicenter retrospective analysis of 124 patients
with hepatocellular adenoma, tumor rupture was reported in
31 cases (25%), and the risk of rupture was associated with
increasing tumor size, recent OCP use, or hormonal replace-
ment therapy (within 6 months). Average tumor size for
nonruptured adenoma in this series was 7.2 cm, compared
with 10.5 cm for ruptured adenoma.48 When hemorrhage
does occur, intra-abdominal blood may be observed in up to
60% of patients. Hepatocellular adenoma should be consid-
ered a premalignant condition. Published reports of patients
who have been followed with serial radiographic studies
have described the development of hepatocellular carcinoma
within previous hepatocellular adenoma.45,46,49 In these stud-
ies, an increase in the serum alpha-fetoprotein (AFP) level was
found to be an indicator of malignant transformation.45,46,49
In the aforementioned multicenter analysis of 124 patients
with hepatocellular adenoma, malignancy was detected in
5 cases (4%).48
PATHOLOGY
Hepatocellular adenomas present as solitary lesions in
approximately 75% of cases. Multiple adenomas are common
in patients with glycogen storage disease or hepatic adenoma-
tosis. Adenomas may vary from 1 cm to greater than 20 cm
in size. Grossly, adenomas are sharply demarcated from the
normal parenchyma and appear light in color. Unlike FNH,
hemorrhage is commonly evident on gross examination.
Microscopically, adenomas consist of cords of cells that
closely resemble normal hepatocytes; indeed, histologic dif-
ferentiation between adenoma and normal liver tissue can
be difficult. However, adenoma cells are larger than normal
hepatocytes and may contain large amounts of glycogen
and lipid.50 Adenomas are devoid of bile ducts, and this key
histologic feature helps to distinguish hepatocellular adenoma
from FNH on biopsy. Hepatocytes within an adenoma are
separated by dilated sinusoids that represent the arterial blood
supply of the lesion which, like FNH, typically lacks a portal
venous supply. Adenomas characteristically have little fibrous
connective tissue support and generally lack a tumor capsule.
DIAGNOSIS
The most important diagnosis to exclude in the evaluation of
hepatocellular adenoma is FNH, as their treatments are radi-
cally different. Hepatocellular adenomas are often first detected
by ultrasonography during evaluation for right upper quadrant
abdominal symptoms.50 Adenomas are typically hyperechoic
on ultrasound, which may be a result of their high lipid con-
tent. Other findings may include significant heterogeneity due
to intratumoral hemorrhage, or calcifications due to hemor-
rhage and necrosis. These findings may identify the lesion, but
are not specific enough to diagnose hepatocellular adenoma.
Multiphasic CT is more specific than ultrasonography for
adenoma.51 Nonenhanced images may identify areas of fat
or hemorrhage that are typical of adenoma. The majority of
adenomas will appear sharply demarcated and are hypo- or
isoattenuating. Areas of old hemorrhage and necrosis will
appear as discrete foci of hypoattenuation on nonenhanced
imaging. Arterial phase contrast images may show some
degree of peripheral enhancement due to the larger periph-
eral feeding vessels. However, approximately 80% of cases
will show rapid homogenous enhancement.50 Unlike FNH,
contrast enhancement in adenoma usually does not persist
due to a component of arteriovenous shunting (Fig. 44-4).
Findings on MR imaging have been reported to be
more variable than those of CT.50,52,53 Adenomas have been
described as hyperintense, isointense, and hypointense on
T1-weighted imaging.52,53 Heterogeneity is again common,
with regions of increased signal intensity occurring due to fat,
hemorrhage, and necrosis. Inconsistency is also reported for
T2-weighted imaging, but the majority of adenomas will be
hyperintense relative to the liver on T2 imaging.53 Dynamic
gadolinium-enhanced MRI may also be used to demonstrate
early arterial enhancement. Since adenomas have a scarcity
of Kupffer cells, Kupffer cell–specific agents will result in
decreased signal intensity on T2-weighted imaging.53
If diagnostic uncertainty persists after a thorough imag-
ing evaluation, percutaneous biopsy should be considered,
especially if the possibility of FNH remains within the dif-
ferential diagnosis. Percutaneous FNA and core biopsy have
been shown to be accurate in the diagnosis of hepatocellular
adenoma.43 As noted earlier, the absence of bile ducts within
the lesion is one of the critical elements that differentiates
FNH from hepatocellular adenoma.
TREATMENT
In general, patients with any size hepatocellular adenoma
should undergo surgical therapy in order to prevent the
risk of tumor growth and rupture, and to prevent the risk
 Chapter 44
FIGURE 44-4 Computed tomographic appearance of hepatic adenoma of the right hemiliver. Note the rapid homogenous contrast enhance-
ment seen on arterial phase images and rapid contrast washout seen on delayed images.
of malignant degeneration. Some have recommended initial
discontinuation of OCP use with follow-up imaging to assess
for evidence of regression.9 This approach may be reasonable
in selected patients, but it must be emphasized that regression
is unpredictable, and continued growth, rupture, and malig-
nant transformation even after discontinuation of OCPs have
been reported.43–45
Miscellaneous Benign Tumors
NODULAR REGENERATIVE HYPERPLASIA
Nodular regenerative hyperplasia is an uncommon lesion
associated with conditions of chronic liver disease. Cirrho-
sis and portal hypertension may be present in up to 50%
of cases.54 Radiographically, these lesions appear similar to
FNH; however, portal venous phase images may show almost
complete contrast washout, rendering these lesions almost
imperceptible.36 Given the rarity of this entity, their natural
history is poorly understood, but tumor growth and rupture
are extremely rare. Percutaneous biopsy can be obtained to
confirm the diagnosis of nodular regenerative hyperplasia,
which should be managed nonoperatively.
PELIOSIS HEPATITIS
Hepatic peliosis is an uncommon disorder histologically
characterized by multiple, small, blood-filled sinuses. Peliosis
occurs most commonly in immunocompromised transplant
recipients, AIDS patients, and patients on long-term cortico-
steroid therapy.55,56 Radiographically, they present as diffuse
hypodense areas distributed throughout the liver. CT and
MRI will typically show enhancement on early phase images,
and may progress from central to peripheral enhancement
on delayed imaging.57 Rupture with intraperitoneal hemor-
rhage has been reported from this condition.58,59 The optimal
treatment for bleeding from this lesion has been angiographic
embolization. As this is a diffuse condition of the liver result-
ing from identifiable causes, ultimate treatment should be
directed toward the specific etiology.
Benign and Malignant Primary Liver Neoplasms 933
BILE DUCT ADENOMA
Bile duct adenomas are benign lesions that appear as small
and well-demarcated white subcapsular lesions ranging from
several millimeters to 1 or 2 cm in maximal dimension. These
are typically diagnosed incidentally at the time of operation.
Pathologically, these appear as well-differentiated bile ductular
structures surrounded by a fibrous stroma. These lesions are
typically asymptomatic and appear to be entirely benign.60,61
MALIGNANT LIVER TUMORS
Introduction
Malignant tumors of the liver may arise from the hepatocyte,
the bile duct epithelium, and the endothelial cells within the
liver. The most common primary liver malignancy is hepatocel-
lular carcinoma (HCC), which represents the most common
solid organ cancer worldwide. As a group, malignant tumors of
the liver present major diagnostic and therapeutic challenges.
Although surgery can be potentially curative, most hepatobi-
liary cancers are discovered at a stage too advanced to permit
complete resection.
Over the last two decades, considerable advances have been
made in the diagnosis and treatment of these tumors. Enhance-
ments in imaging have permitted earlier detection and more
accurate staging of disease. The morbidity associated with
surgical resection has decreased, and improving short- and
long-term results are now being achieved with extensive but
rational resections. Contemporary surgical therapy is guided
by improved imaging techniques and a better understanding
of disease biology. Furthermore, novel palliative treatments
such as radiotherapy and ablative techniques have extended
the limits of tumor eradication and treatment.
Hepatocellular Carcinoma
The therapeutic challenge of HCC arises from three factors.
First, it is usually associated with cirrhosis, which limits the
934 Part VII Liver
range of treatment options and increases the morbidity of
any given therapy. Second, HCC is usually asymptomatic at
early stages, during which it has a great propensity for intra-
vascular or intrabiliary extension; as a result, HCC typically
presents at an advanced stage with consequently few effective
therapeutic options. Third, HCC has been resistant to most
conventional forms of cytotoxic chemotherapy, limiting the
array of nonoperative forms of treatment.
EPIDEMIOLOGY AND ETIOLOGY
There are nearly 500,000 new cases of HCC diagnosed yearly
(Table 44-3).62 The incidence of HCC increases with age, and
is four to eight times more common in males than in females.
HCC is strongly associated with chronic liver injury; there-
fore, its geographic distribution closely mirrors that of viral
hepatitis. The etiologic association between hepatitis B (HBV)
infection and HCC is well established. A landmark study
examined the relationship between HBV infection and HCC
among 22,707 male subjects in Taiwan, 15.2% of whom were
chronic HBV carriers as evidenced by detection of HBsAg in
their serum. Of the 116 cases of HCC that developed during
a mean follow-up period of 7 years, 113 occurred in patients
positive for HBsAg. This study demonstrated that HCC was
related not simply to a history of HBV infection but to the
chronic carrier state, and that the relative risk of developing
HCC was 200-fold greater in individuals with evidence of
HBV infection compared with uninfected individuals.63
The hepatitis C virus (HCV) has also been associated with
HCC. Antibodies to HCV have been found in as many as
76% of patients with HCC in Japan, Italy, and Spain64 and
TABLE 44-3: FACTORS ASSOCIATED WITH
THE DEVELOPMENT OF HCC
Infection Hepatitis B virus
Hepatitis C virus
Cirrhosis Laennec’s (alcohol-induced)
Autoimmune hepatitis
Primary biliary cirrhosis
Environmental Aflatoxins
N-nitrosylated compounds
Pyrrolizidine alkaloids
Thorotrast
Metabolic disorders Alpha1-antitrypsin deficiency
Citrullinemia
Familial cholestatic cirrhosis
Galactosemia
Hemochromatosis
Hereditary tyrosinemia
Porphyria cutanea tarda
Type 1 and 3 glycogen storage
disease
Wilson’s disease
in 36% in the United States.65 Unlike HBV-associated HCC,
which rarely occurs before the development of cirrhosis,
HCV-associated HCC does not necessarily arise in the setting
of advanced liver disease. Additionally, the incidence of HCC
in chronic carriers of HCV is estimated to be as high as 5%
per year, compared with 0.5% per year for HBV carriers.66
Chemical carcinogens have also been linked to HCC.
Nitrites, hydrocarbons, solvents, organochlorine pesticides,
primary metals, and polychlorinated biphenyls have been
implicated in development of HCC.67 Colloidal thorium
dioxide (Thorotrast), which emits high level α, β, and γ radi-
ation and was used as an angiographic agent in the 1930s, has
been linked to angiosarcoma, cholangiocarcinoma, and HCC.
Of all the chemicals linked to the development of HCC, the
most important is ethanol. Alcohol abuse has been associated
with the development of HCC, in addition to carcinomas
of the larynx, mouth, and esophagus. Ethanol is believed to
produce HCC through the development of hepatic cirrhosis
or as a cocarcinogen with other agents such as HBV, HCV,
hepatotoxins, and tobacco68–73 rather than through any direct
effect on hepatocytes.
Aflatoxins produced by the fungi Aspergillus flavus and
Aspergillus parasiticus have also been linked to HCC. These
are fungi that grow on grains, peanuts, and other food prod-
ucts, and are the most common cause of food spoilage in the
tropics. These fungi produce aflatoxins designated B1, B2, G1,
and G2. Aflatoxin B1 is the most hepatotoxic, and chronic
exposure to these mycotoxins can promote HCC.74
Congenital conditions may also lead to the development
of HCC. Genetic diseases such as hemochromatosis, Wilson’s
disease, hereditary tyrosinemia, type 1 glycogen storage dis-
ease, hepatic porphyria of both intermittent and cutanea tarda
types, familial polyposis coli, ataxia telangiectasia, familial
cholestatic cirrhosis, biliary atresia, congenital hepatic fibro-
sis, neurofibromatosis, situs inversus, fetal alcohol syndrome,
α-antitrypsin deficiency, and Budd-Chiari syndrome75 have
all been linked to a higher incidence of HCC. Ultimately, the
unifying etiology of HCC for these conditions is chronic liver
injury and inflammation.
PATHOLOGY
HCC has been histologically graded as well differentiated,
moderately differentiated, and poorly differentiated. The well-
differentiated variety may be difficult to distinguish from a
regenerating nodule on biopsy. HCC can also be classified
into three distinct patterns of growth that are associated with
resectability and therefore have a significant influence on long-
term outcome (Fig. 44-5). The hanging type is attached to the
normal liver by a small vascular stalk, even when tumors are
large. Because of this anatomic configuration, these are easily
excised with minimal loss of functional hepatic parenchyma.
The pushing type is generally well demarcated and often encap-
sulated by a fibrous capsule. These displace rather than infil-
trate normal vasculature and typically do not invade the major
vessels. Even when large, they are often resectable. Finally, the
infiltrative type has a very indistinct tumor-to-liver interface,
 Chapter 44 Benign and Malignant Primary Liver Neoplasms 935

A B

FIGURE 44-5 Growth patterns of hepatocellular carcinoma. A. “Hanging” type. MRI demonstrating a mass (arrow) that is attached by a stalk to
the right lobe of the liver. B. “Pushing” type. These are generally encapsulated, well-circumscribed tumors (arrow). C. “Infiltrating/invading” type.
This lesion has diffusely infiltrated the entire left two-thirds of the liver. Note the left biliary ductal dilation (arrow).

and tends to have a much greater degree of vascular infiltration
and invasion even when the tumor is relatively small. Because
of the indistinct interface, resection of infiltrative-type HCC
is occasionally complicated by positive margins. The practical
nature of this gross pathologic classification scheme is rein-
forced by the distinctive radiologic appearance of these three
different growth patterns on imaging and the resulting ability
to categorize HCC preoperatively.76
An important pathologic determination involves the
distinct appearance and clinical behavior of the fibrolamel-
lar variant of HCC (Table 44-4). On gross and radiologic
inspection, fibrolamellar HCC is generally well demarcated,
936 Part VII Liver
TABLE 44-4: COMPARISON OF STANDARD
HCC WITH FIBROLAMELLAR VARIANT HCC
Fibrolamellar
Characteristic HCC HCC
Male:Female 4:1–8:1 1:01
Median age 55 25
Tumor morphology Invasive Well-circumscribed
Resectability <25% 50–75%
Cirrhosis 90% 5%
AFP elevation 80% 5%
HBV positivity 65% 5%
often encapsulated, and contains a central fibrotic area that
can sometimes make differentiation of fibrolamellar HCC
and FNH difficult. This variant generally occurs in young
patients without underlying cirrhosis.77 AFP, which is com-
monly elevated in usual cases of HCC, is not elevated in
fibrolamellar HCC. Other serum markers that are often
elevated in fibrolamellar HCC include neurotensin and vita-
min B12-binding protein.78 The fibrolamellar variant of HCC
is associated with prolonged survival compared with typical
HCC, which is likely due to the well-demarcated nature of
their tumors and the greater range in treatment options for
patients without underlying cirrhosis.79
CLINICAL PRESENTATION
Despite the fact that they are generally slow-growing tumors,
most cases of HCC present at an advanced stage that is no
longer amenable to potentially curative therapies. Because
the liver is relatively hidden behind the right costal cartilages,
tumors can reach a substantial size before becoming palpable.
Furthermore, the large functional reserve of the liver generally
masks any small impairment resulting from local parenchy-
mal disturbances by tumor. Small HCC tumors are usually
therefore asymptomatic, and are discovered by screening pro-
grams80–82 or incidentally during imaging performed for other
abdominal conditions.
As a result, the majority of HCC cases present with local
symptoms resulting from large tumor mass. Patients usu-
ally complain of a dull and vague right upper quadrant
pain that is sometimes attributable to the shoulder. Hepa-
tomegaly is a frequent accompanying finding. The liver edge
is often hard and irregular because of both the tumor and
the usual accompanying cirrhosis. A vascular bruit can be
heard over the liver in about 25% of cases.83 General symp-
toms of malignancy, including anorexia, nausea, lethargy,
and weight loss, are common. The most common clinical
presentation is the triad of right upper quadrant pain, a pal-
pable mass, and weight loss.84–86 Central necrosis of large
tumors can lead to fever, and HCC can occasionally present
as pyrexia of unknown origin. For the majority of patients,
the presentation of HCC will also be the first presentation
of their underlying cirrhosis. In one study population in
which 90% of patients were eventually found to have cir-
rhosis, fewer than 10% of patients were thought at first eval-
uation for HCC to have chronic liver disease on the basis of
their history and clinical examination.85
Hepatic decompensation is another common presenta-
tion for HCC, with patients seeking medical attention as a
result of symptoms of liver failure like ascites, jaundice, or
encephalopathy. This acute decompensation of liver function
is usually a consequence of replacement of functional paren-
chyma by tumor in a patient with previously compensated
cirrhosis. HCC also has a great propensity for vascular inva-
sion and intravascular growth; hepatic failure can also result
from portal vein occlusion secondary to intravascular tumor
thrombus.87–89 A rare cause of liver failure is Budd-Chiari
syndrome resulting from direct invasion and occlusion of
the hepatic vein and inferior vena cava by tumor and tumor
thrombus. Gastrointestinal hemorrhage often complicates
the clinical course of patients with HCC and is the presenting
finding in 10% of cases.89 In approximately half of these cases,
bleeding is from esophageal varices89 that can result from por-
tal hypertension due to cirrhosis alone or cirrhosis with the
added contribution of intraportal tumor thrombus. Patients
with bleeding esophageal varices have an extraordinarily poor
prognosis, with a median survival measurable in weeks.88 In
one study, nearly a quarter of patients with HCC died of
massive variceal hemorrhage.87 Gastrointestinal bleeding can
also occur from other causes, such as a benign peptic ulcer or
direct invasion of the alimentary tract by tumor.89 The most
dramatic presentation of HCC is tumor rupture, which is the
initial presenting event in 2–5% of HCC patients (Fig. 44-6).
Symptoms of tumor rupture include acute abdominal pain
and swelling, and signs include abdominal distension, guard-
ing, rebound tenderness, and ileus. Patients also commonly
have signs of hemodynamic instability or overt hypovolemic
shock. Diagnosis is confirmed by findings of tumor mass and
peritoneal blood through imaging, laparotomy, or paracen-
tesis.90–94
Up to half of HCC patients initially present with jaundice,
which is most commonly due to hepatic insufficiency.84,95–97
On rare occasions (fewer than 10% of jaundiced patients),
jaundice associated with HCC is a result of biliary obstruc-
tion by extraluminal or intraluminal tumor or tumor-induced
hemobilia.84,98–102 In the clinical evaluation of jaundice for
patients with HCC, it is critically important to distinguish
hepatocellular failure from obstruction. The former etiology
usually indicates that the patient is beyond any therapy, while
the latter can often be treated with effective palliation and
possible cure.99,101,103–107
In rare cases (<5%), HCC can present with paraneoplas-
tic syndromes attributable to hormonal or immune effects
of tumor.108 The most important of these are hypoglycemia,
erythrocytosis, hypercalcemia, and hypercholesterolemia.
Porphyria cutanea tarda, virilization and feminization syn-
dromes, carcinoid syndrome, hypertrophic osteoarthropathy,
hyperthyroidism, and osteoporosis can also occur.109–111
 Chapter 44
FIGURE 44-6 Computed tomographic appearance of ruptured hepatocellular carcinoma. Note the acute extravasation of contrast seen on early
arterial phase images with pooling of blood seen on delayed images.
DIAGNOSIS
For patients with suspected HCC, the aims of diagnostic inves-
tigations are (1) verification of diagnosis; (2) determination of
extent of disease; and (3) estimation of functional liver reserve.
Verification of Diagnosis. Diagnosis of HCC can usually
be established noninvasively with a combination of history,
physical examination, imaging, and laboratory blood tests.
There is little diagnostic doubt in a patient with a liver mass
consistent with a HCC on CT or MRI and a serum AFP
greater than 500 ng/dL. This combination of findings is diag-
nostic, and treatment can be initiated without the need for a
tissue diagnosis. The presence of cirrhosis or hepatitis infec-
tion (as documented by presence of HBsAg or HCV in the
blood) serves to further confirm the diagnosis.
In the patient with a space-occupying lesion on ultraso-
nography or CT with a nondiagnostic AFP level, the role of
percutaneous needle biopsy has been debated. There is little
doubt that needle biopsy can be diagnostic for HCC; how-
ever, iatrogenic complications of biopsy are not infrequent.
Hemorrhage or tumor rupture can occur, and there is also
a risk of tumor spillage and seeding of the needle biopsy
tract.112 In cases of potentially resectable HCC where the
diagnostic certainty is high, we typically proceed to surgical
exploration without a biopsy. Indeed, in this clinical scenario,
the histologic appearance of the non-neoplastic liver may
have a greater impact on surgical planning; if the presence of
advanced cirrhosis will preclude safe resection, we will biopsy
the nontumoral liver for histologic evaluation.
In patients with nondiagnostic AFP who are not surgical
candidates (and therefore not candidates for curative therapy),
tumor biopsy is performed for those who may be candidates
for palliative therapy. In such cases, fine-needle aspirate for
cytologic evaluation has been shown to yield a higher percent-
age of correct diagnoses compared with microhistology of core
needle biopsy specimens (86 vs 66%).113 Patients who are not
Benign and Malignant Primary Liver Neoplasms 937
candidates for palliative therapy do not need a definitive diag-
nosis, and biopsy is discouraged for this cohort.
Extent of Disease Evaluation. The two issues that must be
resolved by the extent of disease evaluation are (1) whether the
disease is isolated to the liver, and (2) whether the distribution
of tumor within the liver is amenable to surgical resection.
The most common sites of HCC metastases include lung,
peritoneum, adrenal, and bone. Chest radiography is there-
fore mandatory. Cross-sectional imaging such as CT or MRI
of the abdomen and pelvis should be scrutinized for perito-
neal and adrenal sites of disease. Many centers consider bone
scans to be mandatory prior to liver resection; this should
certainly be performed for patients with pain that could be
attributable to bony metastases. The presence of extrahepatic
disease gravely alters the prognosis, as it eliminates the poten-
tially curative treatment option of hepatectomy.
The extent of liver involvement is usually determined by
CT. In the evaluation of HCC, cross-sectional imaging is used
to determine the number and distribution of liver tumors
and to identify radiographic evidence of vascular invasion. In
this regard, triple phase (noncontrast, arterial phase, and por-
tal phase) CT images should be obtained. HCC tumors are
generally highly vascular; however, some tumors may become
isodense with the surrounding liver on contrast-enhanced
images, and some tumors are only visible during the noncon-
trast-enhanced phase. HCC has a great propensity for vascu-
lar invasion and extension, and tumor thrombus in the portal
vein, hepatic vein, or vena cava is not unusual. Scans should
therefore be scrutinized for evidence of vascular invasion, since
therapy and prognosis can be significantly altered by the pres-
ence of such findings. If vascular invasion is suspected but not
proven by CT, Doppler ultrasonography or MRI is indicated.
At some centers, hepatic angiography is a standard compo-
nent of the diagnostic evaluation of HCC.114,115 Some have even
advocated routine use of lipiodol injected angiographically to
938 Part VII Liver
further delineate hepatic extent of disease.116 This lipid is prefer-
entially retained in HCC because of its particle size. There is no
doubt that these angiographic methods are highly sensitive in
detecting the presence of tumor. However, with current helical
CT or MRI, the incremental yield of this invasive diagnostic
modality is minor. We rely on angiography for select circum-
stances in which small tumors are suspected but not visible by
conventional cross-sectional imaging (eg, a patient with a very
high AFP level with only minimal disease seen on CT).
Determination of Functional Liver Reserve. Various liver
function tests, alone or in combination, have been touted as
useful for predicting risks of liver resection and other treat-
ments for HCC. Various single serum measures of liver func-
tion have been suggested to be useful predictors of periopera-
tive outcome, including serum bilirubin117 and serum alanine
aminotransferase (ALT).118 A doubling of bilirubin has been
suggested as a contraindication for liver resection.119 Others
have used a platelet count less than 50,000, or a prolonged
prothrombin time greater than 4 seconds over control, to
be relative contraindications for hepatic resection.119 Most
investigators have not relied on a single parameter, but use a
combination of clinical and biochemical parameters to gauge
safety of hepatectomy and other liver-directed treatments.
The most clinically useful system is the Child-Turcotte-Pugh
(Child) classification, which is a point scoring system for
evaluation of liver function based on serum bilirubin, coagu-
lation profile, serum albumin, presence or absence of ascites
and encephalopathy, and nutritional status (Table 44-5).120,121
Functionally, well-compensated cirrhosis is classified as Child
classification grade A; decompensating cirrhosis is grade B;
decompensated cirrhosis is grade C. Generally, partial hepa-
tectomy is offered only to patients who are Child A and the
most favorable class B patients.122 In general, class C patients
are only offered supportive care, since even nonsurgical abla-
tive methods such as transarterial embolization are associated
with procedure-related mortality in one-third of patients.123
Many sophisticated dynamic measures of liver function
have also been used in attempts to quantitate hepatic function.
Investigators have used the elimination of certain dyes that are
exclusively cleared by the liver, such as bromosulphthalein or
TABLE 44-5: CHILD-TURCOTTE-PUGH
GRADING SYSTEM FOR CIRRHOSIS
Criterion 1 Point 2 Points 3 Points
Bilirubin (mg/dL) <2.0 2.0–2.9 >2.9
INR <1.7 1.7–2.3 >2.3
Albumin (g/dL) >3.5 2.8–3.4 <2.8
Ascites Absent Mild Moderate–severe
Encephalopathy Absent Mild Moderate–severe
(grade 1–2) (grade 3–4)
Child-Turcotte-Pugh class A: 5–6 points; B: 7–9 points; C: 10–15 points.
indocyanine green, as measures of hepatic function. Galactose
clearance and [14C] aminopyrine clearance have also been used
to evaluate the metabolic capacity of the liver. Of these, the
most commonly utilized evaluations in clinical practice are
indocyanine green retention at 15 minutes124 and the [14C]
aminopyrine breath test,125 although controversy still exists
concerning their utility.126 We do not use these tests on a rou-
tine basis in our care of patients with HCC, and have found
clinical Child-Turcotte-Pugh classification sufficiently discrim-
inatory for selecting patients for therapies.
Another relatively simple test that may be predictive of peri-
operative outcome, and which we use on occasion, is hepatic
venous wedge pressure. By passing a venous catheter through
the vena cava into the hepatic vein, the hepatic venous pres-
sure can be directly ascertained. With balloon occlusion of the
hepatic vein, the hepatic venous wedge pressure, which is a
reflection of portal pressure, can be determined. These mea-
surements have been touted as useful in segregating Child-
Turcotte-Pugh B patients who may have favorable results from
resection versus those likely to have major complications.127
POTENTIALLY CURATIVE TREATMENT
Therapies for HCC can be separated into resection, ablation,
radiation therapy, systemic chemotherapy or immunother-
apy, and supportive care. Resectional therapy represents the
only potentially curative option. We will begin with a discus-
sion of these, particularly emphasizing recent advances and
comparison of partial hepatectomy with total hepatectomy
or liver transplantation.
Partial Hepatectomy. Partial hepatectomy represents the
most common procedure for HCC performed with curative
intent (Table 44-6). The liver is normally a very resilient organ
with remarkable regenerative capacity. In a noncirrhotic liver,
routine recovery can be expected even after resection of over
two-thirds of functional parenchyma.128 In the United States,
nearly half of the patients with HCC will have no associated
cirrhosis.129 Operative mortality at most major centers is gen-
erally less than 5%, and very extensive procedures are justified
by the low risk and the potential for long-term survival and
cure. Resection is associated with a 5-year survival estimate of
nearly 40% (Fig. 44-7).126–130 For patients without cirrhosis,
partial hepatectomy is a relatively safe procedure, and is the
treatment of choice for HCC.
In contrast, most cases of HCC worldwide are associated
with cirrhosis. The associated cirrhosis greatly increases the
risk of partial hepatectomy. This increase in risk is partly a
result of intraoperative factors. These patients will usually
have rigid and hard hepatic parenchyma and established vari-
ces that are difficult to manipulate and prone to hemorrhage.
Additionally, these patients will have thrombocytopenia and
coagulation defects that further exacerbate their risk of bleed-
ing. Postoperatively, the liver may not regenerate, resulting in
liver failure. Furthermore, postoperative exaggeration of por-
tal hypertension may lead to ascites and variceal bleeding. It
is understandable; therefore, that resection is associated with
 Chapter 44 Benign and Malignant Primary Liver Neoplasms 939

TABLE 44-6: OUTCOMES AFTER LIVER RESECTION FOR HCC

Operative Survival Survival Survival

Author/Year n Mortality 1-y 5-y 10-y Comments

Okuda et al/1984269 98 NR 62% – – –

Nagao et al/1987142 94 19% 58% 20% – –
Kanematsu et al/1988270 107 NR 83% 26% – –
Yamanaka et al/1990271 295 NR 76% 31% – –
Ringe et al/1991272 131 NR 68% 54% – –
Sasaki et al/1992273 186 NR – 44% – Cirrhotics
57 NR – 68% – Noncirrhotics
Nagasue et al/1993134 229 7% 80% 26% 19%
Takenaka et al/1994274 229 1% 89% 76% – <70 y old
39 5% 87% 52% 0 >70 y old
Suenaga et al/1994275 134 NR 100% 68% – –
Bismuth et al/1995276 68 NR 74% 40% 26% Noncirrhotics
Lai et al/1995277 343 5% 60% 24% 1987–1991
Vauthey et al/1995130 106 6% – 41% – –
Takenaka et al/1996278 280 2% 88% 50% – –
Fong et al/1999129 154 5% 80% 39% – 67% cirrhosis
Poon RT et al/2001279 230 NR – 37% – –
Belghiti et al/2002280 328 6% 81% 37% 13% 50% cirrhosis
Esnaola et al/2003281 586 5% 36% 14% 47% cirrhosis
Ng et al/2005282 404 2% 88% 58% – HCC <5 cm
380 3% 74% 39% – HCC >5 cm
Cho et al/2008135 184 5% – 38% – 35% cirrhosis

NR, not reported.

increased morbidity and mortality in these patients. Even

for a cirrhotic patient with well-compensated liver function,
we are reluctant to remove more than 20–25% functional
parenchyma.126,136–139 Until recently, even at centers with a
low mortality for partial hepatectomy in the noncirrhotic
1.00 population, partial hepatectomy for patients with cirrhosis
Resection was associated with a 10% mortality or more.134,140–144 This
Ablation
Chemo/supportive explains the nihilistic view adopted by some for this disease,
0.75 and explains the interest in treating this disease by total hepa-
tectomy and liver transplantation. Nevertheless, even in this
period of time, cirrhotic patients who survive the operation
have a 5-year survival of approximately 30%.137–141 Over the
Survival

0.50
last decade, a number of series have demonstrated increas-
ing safety of partial hepatectomy in cirrhotic patients. Due
to improvements in patient selection, surgical technique, and
0.25
perioperative support, the mortality at most major centers
treating HCC has been reduced to less than 5%.129,131,132,135
Patient selection for surgery is primarily driven by
0 hepatic function. As discussed above, the most commonly
0 12 24 36 48 60 used clinical selection criteria for a patient’s fitness for sur-
Months gery relies on the Child-Turcotte-Pugh score. Few surgeons
FIGURE 44-7 Survival curve for 154 patients undergoing hepa- are willing to perform hepatic resection for patients with
tectomy for hepatocellular carcinoma at Memorial Sloan-Kettering Child C cirrhosis. Most surgeons will only consider resec-
Cancer Center. (Used with permission from Fong Y, et al. An analysis of 412 tion for patients with class A liver functional reserve and the
cases of hepatocellular carcinoma at a Western center. Ann Surg. 1999;229[6]:792.) best class B patients.
940 Part VII Liver

The major changes in operative conduct that have

improved perioperative outcome include the willingness to
use inflow occlusion during resection and the willingness to
accept nonanatomic resection. Temporary occlusion of the
hepatic artery and portal vein during liver resection by clamp-
ing the gastrohepatic ligament has been a useful technique
for reducing blood loss during hepatectomy for patients
without cirrhosis.145 In the past, surgeons have been reluc-
tant to use such inflow occlusion (the Pringle maneuver) in
cirrhotic patients because of fears that cirrhotic parenchyma
will not tolerate even transient ischemia. Recent studies have
indicated that the reluctance to use this technique is largely
unfounded, and that the cirrhotic liver can tolerate a Pringle
maneuver for well over 30 minutes.146,147 The most impor-
tant change in operative technique has been the willingness
to use limited, nonanatomic resections. For patients without
cirrhosis, most major centers adhere to the anatomic bound-
aries of the various segments during liver resection for cancer.
Hemihepatectomies, sectionetomies, and segmentectomies
are preferred over wedge and other nonanatomic resections
because limited resections are more likely to result in a posi-
tive microscopic margin.148 In the cirrhotic liver, however,
a smaller resection margin is acceptable if it will reduce the
chance of postoperative liver failure. The smallest resection
that will remove all gross tumors is generally used at most
centers. FIGURE 44-8 Vascular invasion of hepatocellular carcinoma. These
tumors have propensity for intravascular extension. Note the tumor
As the safety of resections has improved, increasing thrombus in the portal vein (arrow).
experience in the treatment of HCC has resulted in long-
term results that allow for the analysis of prognostic factors.
Many factors that in years past were thought to be contra-
indications to surgical resection have not been substanti-
ated by these data. It is now clear that multiple lesions do we developed a nomogram that appears to permit far more
not preclude surgical resection.132,134,135,138 Presentation with accurate prediction of individual recurrence-free and overall
intraductal tumor and obstructive jaundice does not pre- survival outcomes than is permissible using conventional
clude long-term survival after surgical resection. Therefore, staging systems (Fig. 44-9).135
it is very important in a patient who presents with HCC
and jaundice to distinguish biliary obstruction from hepatic Neoadjuvant Therapy. Many groups have attempted to
insufficiency as the cause for the jaundice. Synchronous treat HCC with local or systemic therapies prior to surgi-
direct invasion of adjacent organs such as the diaphragm by cal resection. The rationale for such neoadjuvant therapies
HCC is also not an absolute contraindication to resectional is that (1) large primary tumors may be sufficiently reduced
surgery.149,150 in bulk to make resection safer, and (2) local and systemic
One cohort with a particularly poor prognosis is com- microscopic disease may be reduced or eradicated to improve
prised of those with major intravascular extension of tumor long-term outcomes. In this regard, methods that have been
(Fig. 44-8). Even though tumor thrombus can be treated employed to achieve these goals include transarterial chemo-
with liver resection and thrombus extraction, the risk of dis- embolization,159,160 combined chemotherapy (doxorubicin
seminated disease is extremely high.151 If the tumor thrombus and 5-fluorouracil) and radiation therapy (2100 cGy), a com-
involves the vena cava or main portal vein, liver resections bination of hepatic artery ligation, hepatic artery infusion of
accompanied by venous tumor thrombectomies are unlikely chemotherapeutic agents, radioimmunotherapy, and frac-
to yield long-term survival. tionated regional radiotherapy,161 and transarterial yttrium
A number of staging systems have been proposed for pur- 90 microspheres.162,163
poses of postoperative risk stratification.152–158 In a recent In a French study, preoperative administration of lipiodol/
analysis of the MSKCC experience, the median disease- doxorubicin-based transarterial chemoembolization (TACE)
specific survival of 184 patients with HCC treated with partial prior to partial hepatectomy in 49 patients resulted in tumor
hepatectomy was 54 months. Incorporating the prognosti- downstaging in 42% and total tumoral necrosis in 50%, with
cally informative variables of patient age, operative blood five patients being converted from unresectable to resect-
loss, microscopic resection margin status, presence of satellite able disease. A trend toward improved survival was observed
lesions, presence of vascular invasion, tumor size, and AFP, among patients demonstrating a partial or complete response
 Chapter 44 Benign and Malignant Primary Liver Neoplasms 941

Points 0 10 20 30 40 50 60 70 80 90 100

Age
25 70 80 90

EBL
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

P
Margin
N
Y
Satellites
N

Y
Vascular invasion
N
>5
Size (cm)
<5

log (AFP) 0 3 6 9 13

Total points 0 10 20 30 40 50 60 70 80 90 100

3 year
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

5 year
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1

FIGURE 44-9 Prognostic nomogram for prediction of overall survival following resection of hepatocellular carcinoma. The individual patient’s
values are located on each variable axis and a line is drawn upward to determine the number of points received for each variable. A line is drawn
downward from the sum of these numbers on the “Total Points” axis to obtain the likelihood of 3-year and 5-year survival. (Used with permission
from Cho CS, Gonen M, Shia J, et al. A novel prognostic nomogram is more accurate than conventional staging systems for predicting survival after resection of hepatocellular
carcinoma. J Am Coll Surg. 2008;206[2]:288.)

to neoadjuvant TACE compared with those demonstrating
no response or those not undergoing resection alone, suggest-
ing that response to preoperative therapy may enhance proper
selection of patients for operative intervention.164 Another
investigative form of preoperative treatment is immunoemboli-
zation. In a comparison of 39 patients treated with neoadjuvant
transarterial immunoembolization using the immunogenic
Streptococcus pyogenes preparation OK-432 versus neoadju-
vant TACE, enhanced tumoral necrosis was observed among
patients treated with transarterial immunoembolization. Two-
and 3-year disease-free survival after resection were 85 and 51%
after transarterial immunoembolization and 56 and 47% after
TACE, respectively.165 In a randomized analysis of 91 patients
with HCC assigned to resection alone versus neoadjuvant che-
moembolization or immunotherapy followed by resection,
overall survival was 18 months after resection alone versus 36.3
months after neoadjuvant therapy followed by resection.166
Although these strategies are promising, their analyses have
involved relatively small numbers of patients, and the use of
neoadjuvant therapy remains investigative.
An alternative neoadjuvant strategy to improve out-
comes after resection that has been adopted more widely is
preoperative portal vein embolization (PVE). Embolization
of the portal vein nourishing the portion of liver in which
the tumor resides results in compensatory hypertrophy of the
contralateral hemiliver. Single-institution analyses have identi-
fied improved postoperative mortality and morbidity among
patients treated with preoperative PVE with no compromise
in oncologic outcomes.167–169
Adjuvant Therapy. Approximately one-third of patients
can expect to remain durably free of disease after partial hepa-
tectomy for HCC; the fact that the majority of patients will
recur indicates the prevalence of microscopic residual disease
at the time of liver resection.170,171 This has motivated keen
interest in developing effective adjuvant therapy directed at
microscopic residual disease.
In a Chinese study, 61 patients with resected HCC were
randomized to no further therapy or postoperative hepatic infu-
sion of lipiodol and cisplatin with systemic epirubicin. Inter-
estingly, patients receiving adjuvant therapy appeared to have
a higher extrahepatic recurrence rate and worse outcomes.172
Another study of 57 patients with resected HCC randomized
to hepatic arterial infusional and systemic epirubicin versus no
further treatment demonstrated no differences in overall and
disease-free survival.173
942 Part VII Liver

Although TACE is used extensively for the treatment of
unresectable disease, randomized studies have not supported
the use of this modality in the adjuvant setting. Indeed, three
studies have demonstrated worse survival for those treated
after resection with chemoembolization.174–176 There have
been two positive randomized trials of adjuvant therapy after
resection for HCC. The first involves the use of the retinoid-
derivative polyprenoic acid, which had been shown to inhibit
hepatocarcinogenesis in rodents.177 In a study randomizing
patients after curative resection or percutaneous ethanol injec-
tion for HCC to polyprenoic acid or placebo, significantly
higher numbers of patients receiving placebo developed addi-
tional HCC. This compound is not currently available in the
United States, but these data encourage further study of this
and other retinoid derivatives in adjuvant treatment for HCC
and in chemoprevention of patients at high risk for develop-
ment of HCC. Another positive adjuvant study involved the
use of radioembolization employing transarterial delivery of
131
I-lipiodol. This compound has demonstrated significant
activity against small HCC, but problems with dosimetry have
limited its use for patients with bulky unresectable disease. In
a prospective, randomized study, 21 patients who received 50
mCi of transarterial131 I-lipiodol within 6 weeks of liver resec-
tion were compared with 22 patients receiving no adjuvant
therapy. Three-year survival rates for the treated group and the
control group were 85 and 46%, respectively.178 These findings
await larger multicenter studies to confirm long-term survival
improvement and to demonstrate the feasibility of delivering
radioembolization in other centers.
Until recently, no systemic chemotherapy regimen had
demonstrated efficacy for HCC. However, a large prospec-
tive multicenter randomized trial involving 602 patients
confirmed a survival advantage associated with the use of the
multikinase inhibitor sorafenib for patients with advanced
unresectable HCC.179 Median survival and time to radio-
graphic progression were 10.7 and 5.5 months, respectively,
in the sorafenib group compared with 7.9 and 2.8 months,
respectively in the placebo control group. Whether sorafenib
may offer benefit in an adjuvant setting for patients having
undergone resection of HCC is presently under investigation.
Total Hepatectomy and Liver Transplantation. From
a theoretical standpoint, total hepatectomy with liver trans-
plantation is the most attractive therapeutic option for HCC
(Table 44-7). This treatment allows for tumor resection with
the widest possible margins, and permits removal of diseased
and tumorigenic parenchyma that may contain microscopic
metastatic disease and be predisposed to the formation of
additional primary tumors. Initial experience with liver trans-
plantation for HCC was disappointing. Several studies dem-
onstrated that large tumors, multiple tumors, and the presence
of microscopic and/or macroscopic vascular invasion were
associated with poor outcomes.180–183 Indeed, early compara-
tive analyses did not uniformly identify improved outcomes
associated with transplantation as compared to resection. A
landmark study defining the appropriate role for transplan-
tation described 3-year post-transplant survival estimates of
85% for highly selected patients with small tumor burden.184

TABLE 44-7: OUTCOMES AFTER LIVER TRANSPLANTATION FOR HCC

Operative
Author/Year n Mortality Survival 1-y Survival 3-y Survival 5-y

O’Grady et al/1988283 50 23% 40% – –

Ringe et al/1989284 52 15% – 37% –
Yokoyama et al/1990182 80 13% 64% 45% 45%
Iwatsuki et al/1991285 71 NR – 43% –
Pichlmayr et al/1992286 87 24% – – 20%
Bismuth et al/1993123 60 5% – 49% –
Selby et al/1995181 105 NR 66% 39% 36%
Pichlmayr et al/1995287 36 18% 57% 31% 27%
Schwartz et al/1995288 57 0% 72% 57% –
Mazzaferro et al/1996184 48 5% – – 75% (4 y)
Llovet et al/1998289 58 14% 84% 74% 74%
Hemming et al/2001290 112 13% 78% 63% 57%
Yao et al 2001291 70 NR 91% – 72%
Duffy et al/2007292 467 NR 82% 65% 52%
Sotiropoulos et al/2007293 100 14% 76% – 60%
Marelli et al/2008294 100 16% 74% 62% 45%
Onaca et al/2009295 587 NR 85% 74% 68%
Halazun et al/2009296 150 3% 85% 68% 60%

NR, not reported.

 Chapter 44
This work motivated the introduction of the so-called “Milan
criteria” (single tumors ≤5 cm in maximal dimension or no
more than three tumors each ≤3 cm in maximal dimension),
which have been adopted by the United Network of Organ
Sharing (UNOS) as a system of selecting HCC patients eli-
gible for liver transplantation. As a result of these defined
criteria, there is also accumulating experience with the use
of neoadjuvant locoregional therapies such as TACE and per-
cutaneous ablation as means to downstage patients to meet
transplantation eligibility.185–187
Attempts to compare partial hepatectomy and liver trans-
plantation for HCC have been challenging because of the
fundamental differences in patient populations selected for
each treatment modality. Patients selected for partial hepatec-
tomy generally have good liver function and can have tumors
of enormous size.188–190 Partial hepatectomy for patients with
small tumors also results in very favorable outcomes. For a
patient with a tumor that is less than 5 cm in size, the 5-year
overall survival is 45–57%.129,191,192 In light of organ short-
ages and costs of liver transplantation, partial hepatectomy
should be regarded as the curative treatment of choice for
patients without cirrhosis or with Child’s A classification cir-
rhosis. Indeed, survival outcomes after partial hepatectomy
among patients with preserved hepatic function with HCC
meeting Milan criteria are comparable to those observed after
transplantation.193 For patients with severe liver dysfunction,
total hepatectomy and transplantation is the better option
and may indeed be the only viable option.
Palliative Therapy. The majority of patients presenting with
HCC will not have disease amenable to potentially curative
surgical intervention. The presence of underlying liver disease
often renders them not treatable by partial hepatectomy, and
most patients present with disease burden beyond accepted
eligibility criteria for total hepatectomy and orthotopic liver
transplantation. If the disease is nevertheless completely or
largely confined to the liver, local tumor ablative therapies
(including percutaneous ethanol injection, radiofrequency
ablation, and cryoablation) and embolization (chemoembo-
lization or radioembolization) can result in reasonable local
control of disease. Two randomized trials comparing chemo-
embolization to symptom control have demonstrated a sig-
nificant survival benefit associated with the use of palliative
chemoembolization.194,195 Nonrandomized studies have sug-
gested that a similar survival benefit may be expected with the
use of radiofrequency ablation for properly selected cases.196
As outlined earlier, systemic sorafenib has also been associ-
ated with an incremental survival benefit for patients with
unresectable HCC.179
Intrahepatic Cholangiocarcinoma
EPIDEMIOLOGY AND ETIOLOGY
Cancers of the bile duct are uncommon, with approximately
4000 cases diagnosed in the United States annually.197 Bile
Benign and Malignant Primary Liver Neoplasms 943
duct carcinoma may occur anywhere along the biliary tree,
and is commonly divided into distal, proximal, and intrahe-
patic varieties. It is a disease of the elderly, with the major-
ity of patients diagnosed older than 65 years of age, and the
peak incidence occurs in the eighth decade of life.198 In the
absence of therapeutic intervention, bile duct cancers are
rapidly fatal, and the majority of patients will die within a
year of diagnosis. Death usually results from hepatic failure or
biliary sepsis.197,199,200 Long-term survival is highly dependent
on the efficacy of surgical therapy. Indeed, it has been shown
that location within the biliary tree has no impact on survival,
provided that complete resection is achieved.201 However, it
is more likely that a patient with distal bile duct cancer can
be resected with curative intent, which explains the relatively
favorable prognosis of distal tumors.
Conditions resulting in chronic biliary inflammation
have been associated with an increased incidence of cholan-
giocarcinomas. These conditions include primary sclerosing
cholangitis, choledochal cysts, and chronic biliary infections.
Primary Sclerosing Cholangitis. In Western nations, the
disease most often associated with development of cholan-
giocarcinoma is primary sclerosing cholangitis (PSC). This
is an autoimmune disease characterized by inflammation
of the periductal tissues. In advanced cases, it is character-
ized by multifocal strictures of the intrahepatic and extra-
hepatic bile ducts.185–187 The majority (70–80%) of patients
with PSC also have associated inflammatory bowel disease,
typically in the form of ulcerative colitis.185 In a longitudinal
study of patients with PSC, 8% of patients developed clini-
cally apparent cholangiocarcinoma over a 5-year period.185
Indeed, a high incidence (30–40%) of occult cholangio-
carcinoma has been found in autopsy or explant specimens
from patients with PSC.202,203 Cholangiocarcinoma pre-
senting in patients with PSC is often multifocal and prone
to recurrence, and typically not amenable to treatment by
partial hepatectomy. Liver transplantation is often the only
effective treatment for these patients, not only because of
the likelihood of multifocal cancer, but also because of the
baseline hepatic insufficiency that often results from the
underlying inflammatory disease.204,205
Choledochal Cysts or Caroli’s Disease. The increased
risk of cholangiocarcinoma in patients with congenital cystic
disease of the biliary tree is well recognized.206,207 The reason
for the malignant transformation is thought to be related to
chronic inflammation and bacterial contamination within
areas of cystic dilatation.206,208–210 Early excision of the chole-
dochal cyst significantly reduces the risk of cancer.206,208 Fifteen
percent to 20% of adult patients with unexcised choledochal
cysts, or cysts previously treated with bypass, will be found to
harbor cholangiocarcinoma.206,208
Pyogenic Cholangiohepatitis and Other Hepatic
Infections. In Asia, chronic infections of the liver can pre-
dispose patients to the development of cholangiocarcinoma.
Pyogenic cholangiohepatitis or oriental cholangiohepatitis
944 Part VII Liver
results from chronic portal bacteremia and portal phlebitis,
which gives rise to intrahepatic pigment stone formation.
This hepatolithiasis then leads to recurrent episodes of chol-
angitis and biliary stricture formation.211,212 Those who do
not succumb to sepsis will have an estimated 10% risk of
developing cholangiocarcinoma.212 In southeast Asia, biliary
parasites (Clonorchis sinensis, Opisthorchis viverrini) are also
associated with an increased risk of cholangiocarcinoma.213
In areas where these parasites are endemic, the incidence of
cholangiocarcinoma is as high as 87 per 100,000.213
Environmental Toxins. Several radionuclides and chemical
carcinogens, including thorium, radon, nitrosamines, dioxin,
and asbestos, have also been implicated in the genesis of
cholangiocarcinomas.
PATHOLOGY AND CLASSIFICATION
Cholangiocarcinoma can arise anywhere within the biliary
tree. Approximately 10% of cholangiocarcinoma cases arise
within the intrahepatic bile ducts.214–217 Extrahepatic chol-
angiocarcinomas are more common and can occur along
the entire length of the extrahepatic biliary system from the
confluence of the hepatic ducts to the ampulla. Some have
classified these extrahepatic tumor into proximal (hilar), mid,
and distal bile duct tumors. We agree with the convention
of dividing cholangiocarcinomas into intrahepatic, perihilar,
and distal subgroups, thus eliminating the mid-duct group.218
Peripheral or intrahepatic cholangiocarcinoma is diag-
nosed in 1000–2000 patients in the United States annually.219
Clinical presentation is similar to that for HCC, with the
most common symptoms being right upper quadrant pain,
epigastric pain, and weight loss.214,219 In fact, difficulty can
be encountered in distinguishing peripheral cholangiocarci-
nomas from HCC or metastatic tumors from unknown ori-
gin. Jaundice occurs in only 24% of patients with peripheral
cholangiocarcinoma compared with 71% of patients with
hilar tumors.219 Because the tumor is usually asymptomatic
in early stages, most patients have advanced disease at presen-
tation. On cross-sectional imaging by CT or MRI, periph-
eral cholangiocarcinoma is often confused with HCC or
metastatic tumor from unknown primary. Unlike HCC, AFP
levels will be normal. Search for alternative primary cancers
that may have produced liver metastases will not be fruit-
ful. A solitary lesion not associated with the gallbladder, in
a patient with no cirrhosis and no other primary cancer, and
with a normal serum AFP, should raise suspicion of a periph-
eral cholangiocarcinoma. Intrahepatic metastases and tumor
growth along the biliary tract frequently occur, and can make
it even more difficult to distinguish these tumors from meta-
static disease originating from a distant site.
The segregation of extrahepatic cholangiocarcinomas
into perihilar and distal subtypes is practical for purposes of
operative planning. Tumors that are proximal to the cystic
duct–common duct junction typically require a liver resec-
tion for extirpation; these represent approximately 40–60%
of cases of cholangiocarcinoma, and include the hilar
cholangiocarcinomas or so-called “Klatskin tumors.”200,201,218–221
Tumors that are distal to the cystic duct typically require pan-
creaticoduodenectomy for extirpation. Fewer than 10% of
patients will present with multifocal or diffuse involvement
of the biliary tree.222
Lymph node metastases are common with peripheral
cholangiocarcinoma, and the assessment of hilar lymph node
appears to provide useful prognostic information.223,224 The
(primary) tumor, (regional lymph) node, (remote) metastases
(TNM) staging of intrahepatic or peripheral cholangiocarci-
noma is the same as that for HCC.
TREATMENT
Partial Hepatectomy. Whenever possible, surgical resection
is the treatment of choice. In a series of 42 patients with intra-
hepatic cholangiocarcinoma, survival was indistinguishable
from that of 70 patients with hilar cholangiocarcinomas.225
Median survival was 12 months, and no patient survived more
than 42 months. Others have reported more favorable results.
In a series of 20 patients with peripheral cholangiocarcinoma
undergoing surgery over a 10-year period, median survival was
21 months.214 Four patients lived more than 3 years, and a sin-
gle patient was alive 5 years after resection. In our own report
of 32 cases of resected peripheral cholangiocarcinoma, median
survival was 59 months with an actuarial 5-year survival of
42%. Vascular invasion and intrahepatic satellite lesions were
predictors of worse survival (p < .05).214 In a recent update of
the MSKCC experience, the median disease-specific survival
for 82 patients with intrahepatic cholangiocarcinoma treated
with surgical resection was 36 months. This stands in con-
trast to a median disease-specific survival of only 9 months
for patients with disease not amenable to partial hepatectomy;
unfortunately, only a third of patients with intrahepatic chol-
angiocarcinoma evaluated at MSKCC were ultimately found
to have resectable disease. The presence of multiple tumors,
tumors greater than or equal to 5 cm, and nodal metastases
was found to be associated with worse survival after resection
on multivariate analysis.224
Total Hepatectomy and Liver Transplantation.
Historically, outcomes after liver transplantation for patients
with cholangiocarcinoma have been suboptimal; in 1991, an
actuarial 5-year survival rate of 17% was reported for 109
intrahepatic and extrahepatic cholangiocarcinoma patients
transplanted at various centers throughout the world. In this
series, there were no significant differences in recurrence rates
for hilar and peripheral tumors.226 More recently, encourag-
ing outcomes have been observed with the use of liver trans-
plantation for selected patients with hilar cholangiocarcinoma
following a course of neoadjuvant chemoradiation.227,228 The
Mayo Clinic has utilized a protocol of external beam radiation
therapy, chemosensitizing 5-FU, and capecitabine chemother-
apy followed by staging laparotomy and, for patients without
evidence of hilar nodal or distant metastases, liver transplanta-
tion for patients with unresectable hilar cholangiocarcinoma
or hilar cholangiocarcinoma arising in a setting of primary
 Chapter 44
sclerosing cholangitis.229,230 Estimated actuarial 5-year survival
of patients completing this therapy was 82%, which compared
favorably to the 21% 5-year survival observed among a cohort
of 26 patients with hilar cholangiocarcinoma who under-
went conventional surgical resection at the same institution.
Whether this promising strategy could be applied for patients
with peripheral cholangiocarcinoma remains unknown.
Chemotherapy. Data for chemotherapy or radiation in
treatment of this disease is not encouraging. Response rates
with 5-fluorouracil (5-FU) and 5-FU-based systemic chemo-
therapy regimens have been generally poor.231 A 5% com-
plete response and 46% partial response for the treatment of
peripheral cholangiocarcinoma was reported with a regimen
of initial whole-liver irradiation to 2100 cGy in seven frac-
tions, doxorubicin, cisplatin, and 131I anticarcinoembryonic
antigen (CEA) antibody. Although the median survival was
14 months from diagnosis and 10 months from treatment,
no patient survived more than 2 years from the onset of diag-
nosis.232 A recent phase II clinical trial examining the use of
hepatic arterial floxuridine and dexamethasone for patients
with unresectable intrahepatic cholangiocarcinoma and HCC
identified a radiographic response rate of 47.1% with 2-year
survival estimates of 67%, suggesting that liver-directed che-
motherapeutic regimens may hold some palliative promise for
patients with unresectable intrahepatic cholangiocarcinoma.233
Other Primary Malignancies of the Liver
HEPATOBLASTOMA
Hepatoblastoma affects approximately 1 in 100,000 children
and is the most common primary malignant liver tumor in
children.234,235 It is usually diagnosed before the age of 3 years,
with a 2:1 male predominance. Patients usually present with
abdominal swelling,234,235 and the serum AFP is elevated in
over 75% of cases. CT scans will reveal a vascular mass that is
often (50%) speckled with calcifications.236 Overall long-term
survival varies between 15 and 37%.236–239 Poor prognosis is
associated with unresectable tumors and tumors demonstrat-
ing aneuploidy and anaplastic characteristics.237,240,241
Complete resection is possible in 50–65% of children with
hepatoblastoma, and is associated with cure rates between 30
and 70%.240,241 Unlike adult primary liver tumors, chemother-
apy may produce a response in a significant number of patients
with hepatoblastoma. Preoperative chemotherapy has been
used with some success in converting unresectable tumors to
resectability.242,243 Adjuvant chemotherapy has also been used
following resection of hepatoblastoma. In one report, 20%
of 24 patients with hepatoblastoma were relapse-free 8–42
months after surgical resection coupled with adjuvant vincris-
tine, doxorubicin, 5-FU, and cyclophosphamide.244 Radiation
therapy has been used in the treatment of unresectable hepato-
blastomas, but its utility is yet to be proven.242,245 Orthotopic
liver transplantation should be considered for children with
unresectable hepatoblastoma if the tumor does not become
Benign and Malignant Primary Liver Neoplasms 945
resectable after preoperative chemotherapy. In a report of 18
patients undergoing liver transplantation for unresectable
hepatoblastoma, tumors recurred in six patients, but five have
survived disease-free for more than 2 years with actuarial sur-
vival rates of approximately 50%.226
ANGIOSARCOMA
These malignant mesenchymal tumors of the liver are also
referred to as hemangiosarcomas. Approximately 25 cases
occur in the United States each year.246 Peak incidence is in
the sixth and seventh decades, with 85% of cases occurring
in males.247 Presenting symptoms are as for any liver tumor,
and most commonly include abdominal pain, abdominal
swelling (usually due to hepatomegaly), liver failure, nausea,
anorexia, emesis, and jaundice. These malignant tumors have
been associated with exposure to Thorotrast, arsenic, or vinyl
chloride.
Angiosarcomas are aggressive neoplasms; partial hepa-
tectomy can result in long-term survival, but most patients
present with advanced tumors not amenable to complete
resection. Distant metastases are found at initial presenta-
tion in half of patients. Most patients die within 6 months
of diagnosis. Even with resectable tumors, few patients sur-
vive more than 1–3 years after complete resection due to
the onset of metastatic disease. Results of radiation therapy
and chemotherapy or both have been disappointing.247
Results of orthotopic liver transplantation for treatment of
angiosarcoma have also been poor, with disease recurrences
reported in 9 of 14 transplant patients with tumors classi-
fied as either angiosarcomas or epithelioid tissue sarcomas.
The 2-year survival rate was 15%, with no patient surviving
more than 28 months after transplantation.226
The liver can occasionally be the primary site for rhabdo-
myosarcoma,248 although this is more common in children
than adults. Hepatic metastases from a gastrointestinal or
uterine primary need to be ruled out before the diagnosis of
primary leiomyosarcoma of the liver can be made. Surgical
resection is the treatment of choice for these primary hepatic
sarcomas,248 and unresectable disease typically portends an
unfavorable prognosis. Undifferentiated sarcomas of the liver
are very rare and usually occur in children between the ages
of 6 and 15 years.249,250 Most undifferentiated sarcomas of
the liver present at an advanced stage, when surgical resec-
tion is not possible. These patients rapidly succumb to their
disease, as they are generally not responsive to radiotherapy
or chemotherapy.250
EPITHELIOID HEMANGIOENDOTHELIOMA
Epithelioid hemangioendothelioma is another malignant
soft tissue tumor of endothelial cell origin.246,248,251 Factor VII
immunohistochemical staining differentiates hemangioen-
dothelioma from other nonvascular tumors. Unlike infantile
hemangioendothelioma, which is benign, the adult variety is
malignant and highly aggressive. Average age at presentation
is 50 years. Afflicted patients usually present with nonspecific
946 Part VII Liver
complaints that include pain and an abdominal mass. In
contrast to angiosarcoma, there is a female predominance
(63% of patients).248 Vinyl chloride exposure has also been
implicated as a possible etiology of epithelioid hemangioen-
dothelioma in some patients.251
Radical surgery has been advocated for cases of resect-
able epithelioid hemangioendothelioma.246 Unfortunately,
these tumors are almost always diffuse and multifocal,
and therefore unlikely to be cured by partial hepatectomy.
Percutaneous biopsy is undertaken when this diagnosis is
suspected. Intraoperative frozen section analysis is not typ-
ically helpful, as special immunostaining is often needed
for definitive diagnosis. Patients with hemangioendothe-
liomas should be considered for total hepatectomy and
liver transplantation. In a series of patients who underwent
orthotopic liver transplantation for epithelioid hemangio-
endothelioma, 7 of 21 patients developed disease recur-
rence.226 The actuarial survival rate was 82% at 2 years and
43% at 5 years.
Technical Considerations in
Hepatic Resection
The past two decades have seen a dramatic improvement in
perioperative outcome after hepatic resection. High-volume
centers now routinely report operative mortality rates of less
than 5%, and often as low as 2–3%.129,232–234 A recent review
of over 1800 resections from MSKCC also documented a
significant improvement in blood loss, transfusion require-
ments, and postoperative length of stay in patients undergo-
ing hepatic resection over a 10-year time period.254 There is
no single factor solely responsible for this reduction in mor-
bidity and mortality; better patient selection, the evolution
of hepatobiliary surgery as an area of specialization, advances
in anesthetic technique, and optimization of operative tech-
nique have all contributed to these improved results.
A better appreciation of the segmental nature of hepatic
anatomy has resulted in an increasing use of anatomically
based resections and, more importantly, an increased use of
parenchymal-sparing segmental resections. In a study from
MSKCC, the number of hepatic segments resected was a
strong predictor of outcome and, along with operative blood
loss, was an independent predictor of both the morbidity and
mortality of hepatic resection.234,255 As the number of resected
segments increased, there was an almost linear rise in the rate
of complications and postoperative mortality (Fig. 44-10).
We have observed a significant increase in the proportion of
segmental resections performed over the last decade, resulting
in a decline in the number of segments resected. Both of these
factors correlated closely with the observed reduction in mor-
tality and the overall improvement in perioperative outcome.
Data from this and other studies suggest that measures
aimed at preserving hepatic parenchyma without compromis-
ing the oncologic integrity of the resection have a significant
influence on operative morbidity and mortality. In selected
90
Complications
75 p < 0.001
60
45
Percent
30
n = 344 n = 345 n = 150 n = 381 n = 519 n = 64
9
Mortality
p < 0.001
6
3
0
≤1 2 3 4 5 6
Number of hepatic segments resected
FIGURE 44-10 Perioperative complications and mortality stratified
by the number of hepatic segments resected. (Used with permission from
Jarnagin WR, Gonen M, Fong Y, et al. Improvement in perioperative outcome after
hepatic resection: analysis of 1803 consecutive cases over the past decade. Ann Surg.
2002;236[4]:402.)
patients, parenchymal-sparing segmental resections offer the
same benefit as more extensive lobar resections with less risk.
The use of segmental resections allows a complete but less
extensive resection to be performed in patients with limited
disease, and permits greater flexibility for those with more
extensive disease or decreased hepatic functional reserve.
Additionally, anatomically based segmental resections have
been shown to be superior to nonanatomic wedge resections
with respect to blood loss and tumor clearance.256–258
For major hepatic resections, we typically begin by estab-
lishing control of the vascular inflow. Several different tech-
niques for vascular inflow control during hepatic resection
have been described.259 In the 1950s, the technique of extra-
hepatic portal dissection and transection, prior to parenchy-
mal division, became a common practice.260 This technique
consists of individual dissection and ligation of the ipsilateral
hepatic artery and portal vein within the hilus of the liver.
This extrahepatic technique is still commonly employed
today, with division of the inflow being performed with the
use of stapling devices or suture ligation.261,262 Some have
noted, however, that this extrahepatic method for inflow con-
trol is time consuming, and can result in inadvertent injury to
aberrant vascular or biliary structures.257
The technique of intrahepatic vascular inflow control was
first reported by Couinaud and Launois.259,263,264 This tech-
nique is based on the anatomic observation that the structures
of the portal triad enter the liver together as a pedicle, carrying
the enveloping Glisson’s capsule with them into the hepatic
parenchyma. Thus, within the liver, all three structures of the
porta are contained within a very strong and well-formed
sheath (pedicle), which can be isolated and divided en masse
within the liver. No such well-characterized sheath exists out-
side the liver; therefore, the extrahepatic portal structures are
 Chapter 44 Benign and Malignant Primary Liver Neoplasms 947

isolated and divided individually. The technique of pedicle

ligation is ideally suited for right-sided tumors situated away
from the hilus and requiring a right hemihepatectomy. For
tumors close to the hilus, pedicle ligation may compromise
the resection margin and is therefore inappropriate. For
major left hemihepatectomies, pedicle ligation may be used
but extrahepatic control is typically employed, as the longer
extrahepatic course of the left hepatic inflow allows for rela-
tively easy extrahepatic dissection.
Two different methods have been reported for intrahepatic
ligation of the portal pedicle during hepatic hemihepatec-
tomy.263,264 These methods differ with respect to the direction
from which the pedicle is approached, and the sequence for
division of the parenchyma and the pedicle. In the posterior
approach (or hepatotomy method), hepatotomies are cre-
ated both anterior and posterior to the hilum on the side of
resection (Fig. 44-11). A clamp or other instrument is used
to isolate the pedicle sheath, and the pedicle is then tran-
sected with a vascular stapler prior to division of the hepatic
parenchyma (Fig. 44-12). In the anterior approach (or tran-
section method), the hepatic parenchyma is initially divided Clamp Stapler
along the line of the main hepatic fissure under Pringle inflow
occlusion until the pedicle is identified. Once identified, the FIGURE 44-12 Placement of stapling device after isolation of the
pedicle is isolated from within the liver and divided. Regard- right portal pedicle.
less of approach, both techniques obviate the need for extra-
hepatic dissection within the hilus of the liver.
With either approach, the technique of pedicle ligation is made in the caudate process just posterior to the hilum.
for right hemihepatectomy first involves mobilization of the The hepatic parenchyma is further dissected and a finger or
right liver off the inferior vena cava. Initial division of the clamp passed to connect the two hepatotomies to encircle the
lowermost retrohepatic veins draining the caudate process is portal pedicle with a tape or vessel loop. With traction, the
essential. Failure to do this can result in tearing of these veins right and left main pedicles can be exteriorized. After con-
during pedicle isolation.265 Once the liver has been mobilized, firming adequate and appropriate isolation with clamping,
the gallbladder is removed and the hilar plate lowered. A hep- the pedicle is divided with either an Endo GIA (Covidien,
atotomy is created anterior to the hilum, extending from the Mansfield, MA) or TA stapler (Ethicon Surgical, Cincinnati,
right side of the gallbladder fossa. A second parallel incision OH).257 This technique may also be used for segmental resec-
tions, as further dissection of the hepatic parenchyma after
initial isolation of the main pedicle permits isolation of pedi-
cle branches to individual hepatic sections or segments.
Once inflow control has been achieved, control of the hepatic
venous outflow is undertaken; this may again be achieved intra-
hepatically during parenchymal transection, or outside of the
IV liver. In general, we prefer to obtain extrahepatic control except
for bulky tumors that involve the junction of the hepatic vein
V
with the inferior vena cava, in which case hepatic vascular isola-
D tion may be appropriate. In almost all cases, with satisfactory
A control of central venous pressure and careful dissection of the
E major veins, extrahepatic outflow control can be achieved.
B
Once the vascular inflow and outflow have been
C controlled, the liver parenchyma may be transected with any
VI
variety of techniques or instruments. Our preferred approach
is to crush the parenchyma with a clamp to expose the intra-
hepatic bile ducts and vessels, which are then individually
ligated and divided. Over the past decade, a number of
devices have been described for parenchymal transection.
FIGURE 44-11 Hepatotomy placement for pedicle ligation. A and The ultrasonic dissector, water-jet dissector, harmonic
C. Right hepatectomy. A and B. Right anterior sectorectomy. B and scalpel, stapling devices, and most recently radiofrequency
C. Right posterior sectorectomy. D and E. Left hepatectomy. coagulators have all been used for parenchymal transection.
948 Part VII Liver
The majority of reports regarding the use of these instru-
ments are descriptive, and little data exist to suggest that
one technique is better than another with respect to intra-
operative blood loss.32,257,266–268 In the setting of cirrhosis and
significant steatosis, the crushing technique for parenchy-
mal transection is probably not ideal because the liver tissue
tends to fracture and small vascular or biliary structures are
more easily torn. In these situations, the use of noncrushing
instruments such as the harmonic scalpel, which simultane-
ously coapt and coagulate, may be beneficial.
REFERENCES
1. Karhunen PJ. Benign hepatic tumours and tumour like conditions in
men. J Clin Pathol. 1986;39(2):183–188.
2. Weimann A, Ringe B, Klempnauer J, et al. Benign liver tumors: differential
diagnosis and indications for surgery. World J Surg. 1997;21(9):983–990.
3. Whitney WS, Herfkens RJ, Jeffrey RB, et al. Dynamic breath-hold
multiplanar spoiled gradient-recalled MR imaging with gadolinium
enhancement for differentiating hepatic hemangiomas from malignan-
cies at 1.5 T. Radiology. 1993;189(3):863–870.
4. Mitchell DG, Saini S, Weinreb J, et al. Hepatic metastases and cavernous
hemangiomas: distinction with standard- and triple-dose gadoteridol-
enhanced MR imaging. Radiology. 1994;193(1):49–57.
5. Soyer P, Gueye C, Somveille E, et al. MR diagnosis of hepatic metasta-
ses from neuroendocrine tumors versus hemangiomas: relative merits of
dynamic gadolinium chelate-enhanced gradient-recalled echo and unen-
hanced spin-echo images. AJR Am J Roentgenol. 1995;165(6):1407–1413.
6. Grazioli L, Morana G, Kirchin MA, et al. Accurate differentiation
of focal nodular hyperplasia from hepatic adenoma at gadobenate
dimeglumine-enhanced MR imaging: a prospective study. Radiology.
2005;236(1):166–177.
7. Abdelli N, Bouche O, Thiefin G, et al. Subcutaneous seeding on the
tract of percutaneous cytologic puncture with a fine needle of a hepatic
metastasis from colonic adenocarcinoma. Gastroenterol Clin Biol.
1994;18(6–7):652–656.
8. Vergara V, Garripoli A, Marucci MM, et al. Colon cancer seeding
after percutaneous fine needle aspiration of liver metastasis. J Hepatol.
1993;18(3):276–278.
9. Charny CK, Jarnagin WR, Schwartz LH, et al. Management of 155
patients with benign liver tumours. Br J Surg. 2001;88(6):808–813.
10. Yoon SS, Charny CK, Fong Y, et al. Diagnosis, management, and
outcomes of 115 patients with hepatic hemangioma. J Am Coll Surg.
2003;197(3):392–402.
11. Descottes B, Glineur D, Lachachi F, et al. Laparoscopic liver resection of
benign liver tumors. Surg Endosc. 2003;17(1):23–30.
12. Rogula T, Gagner M. Current status of the laparoscopic approach to liver
resection. J Long Term Eff Med Implants. 2004;14(1):23–31.
13. Ishak KG, Rabin L. Benign tumors of the liver. Med Clin North Am.
1975;59(4):995–1013.
14. Metry DW, Hawrot A, Altman C, et al. Association of solitary, segmental
hemangiomas of the skin with visceral hemangiomatosis. Arch Dermatol.
2004;140(5):591–596.
15. Gemer O, Moscovici O, Ben Horin CL, et al. Oral contraceptives and
liver hemangioma: a case-control study. Acta Obstet Gynecol Scand.
2004;83(12):1199–1201.
16. Reddy KR, Kligerman S, Levi J, et al. Benign and solid tumors of the
liver: relationship to sex, age, size of tumors, and outcome. Am Surg.
2001;67(2):173–178.
17. Zimmermann A, Baer HU. Fibrous tumor-liver interface in large hepatic
neoplasms: its significance for tumor resection and enucleation. Liver
Transpl Surg. 1996;2(3):192–199.
18. Zimmermann A. Tumors of the liver: pathological aspects. In: Blumgart
LH, Fong Y, eds. Surgery of the Liver and Biliary Tract. 3rd ed. New York,
NY: W.B. Saunders Co.;2003;1343–1396.
19. Quinn SF, Benjamin GG. Hepatic cavernous hemangiomas: simple
diagnostic sign with dynamic bolus CT. Radiology. 1992;182(2):
545–548.
20. Ashida C, Fishman EK, Zerhouni EA, et al. Computed tomogra-
phy of hepatic cavernous hemangioma. J Comput Assist Tomogr. 1987;
11(3):455–460.
21. Itai Y, Ohtomo K, Furui S, et al. Noninvasive diagnosis of small cavern-
ous hemangioma of the liver: advantage of MRI. AJR Am J Roentgenol.
1985;145(6):1195–1199.
22. Tsai CC, Yen TC, Tzen KY. The value of Tc-99m red blood cell SPECT
in differentiating giant cavernous hemangioma of the liver from other
liver solid masses. Clin Nucl Med. 2002;27(8):578–581.
23. Hatayama K, Watanabe H, Ahmed AR, et al. Evaluation of hemangioma
by positron emission tomography: role in a multimodality approach.
J Comput Assist Tomogr. 2003;27(1):70–77.
24. Cappellani A, Zanghi A, Di Vita M, et al. Spontaneous rupture of a giant
hemangioma of the liver. Ann Ital Chir. 2000;71(3):379–383.
25. Terkivatan T, Vrijland WW, Den Hoed PT, et al. Size of lesion is not a
criterion for resection during management of giant liver haemangioma.
Br J Surg. 2002;89(10):1240–1244.
26. Farges O, Daradkeh S, Bismuth H. Cavernous hemangiomas of the liver:
are there any indications for resection? World J Surg. 1995;19(1):19–24.
27. Kammula US, Buell JF, Labow DM, et al. Surgical management of benign
tumors of the liver. Int J Gastrointest Cancer. 2001;30(3):141–146.
28. Fioole B, Kokke M, Van Hillegersberg R, et al. Adequate symptom relief
justifies hepatic resection for benign disease. BMC Surg. 2005;5(1):7.
29. Hall GW. Kasabach-Merritt syndrome: pathogenesis and management.
Br J Haematol 2001;112(4):851–862.
30. Hochwald SN, Blumgart LH. Giant hepatic hemangioma with Kasa-
bach-Merritt syndrome: is the appropriate treatment enucleation or liver
transplantation? HPB Surg. 2000;11(6):413–419.
31. Lerner SM, Hiatt JR, Salamandra J, et al. Giant cavernous liver
hemangiomas: effect of operative approach on outcome. Arch Surg.
2004;139(8):818–821.
32. Baer HU, Dennison AR, Mouton W, et al. Enucleation of giant hem-
angiomas of the liver. Technical and pathologic aspects of a neglected
procedure. Ann Surg. 1992;216(6):673–676.
33. Terminology of nodular hepatocellular lesions. International Working
Party. Hepatology. 1995;22(3):983–993.
34. Nguyen BN, Flejou JF, Terris B, et al. Focal nodular hyperplasia of the
liver: a comprehensive pathologic study of 305 lesions and recognition of
new histologic forms. Am J Surg Pathol. 1999;23(12):1441–1454.
35. Bioulac-Sage P, Balabaud C, Wanless IR. Diagnosis of focal nodular
hyperplasia: not so easy. Am J Surg Pathol. 2001;25(10):1322–1325.
36. Gibbs JF, Litwin AM, Kahlenberg MS. Contemporary management of
benign liver tumors. Surg Clin North Am. 2004;84(2):463–480.
37. Hussain SM, Terkivatan T, Zondervan PE, et al. Focal nodular hyper-
plasia: findings at state-of-the-art MR imaging, US, CT, and pathologic
analysis. Radiographics. 2004;24(1):3–17.
38. Mathieu D, Kobeiter H, Cherqui D, et al. Oral contraceptive intake in
women with focal nodular hyperplasia of the liver. Lancet. 1998;352(9141):
1679–1680.
39. Mortele KJ, Praet M, Van Vlierberghe H, et al. CT and MR imaging
findings in focal nodular hyperplasia of the liver: radiologic-pathologic
correlation. AJR Am J Roentgenol. 2000;175(3):687–692.
40. Mortele KJ, Praet M, Van Vlierberghe H, et al. Focal nodular hyperplasia of
the liver: detection and characterization with plain and dynamic-enhanced
MRI. Abdom Imaging. 2002;27(6):700–707.
41. Fabre A, Audet P, Vilgrain V, et al. Histologic scoring of liver biopsy
in focal nodular hyperplasia with atypical presentation. Hepatology.
2002;35(2):414–420.
42. Herman P, Pugliese V, Machado MA, et al. Hepatic adenoma and focal
nodular hyperplasia: differential diagnosis and treatment. World J Surg.
2000;24(3):372–376.
43. Shortell CK, Schwartz SI. Hepatic adenoma and focal nodular hyperplasia.
Surg Gynecol Obstet. 1991;173(5):426–431.
44. Klatskin G. Hepatic tumors: possible relationship to use of oral contra-
ceptives. Gastroenterology. 1977;73(2):386–394.
45. Gyorffy EJ, Bredfeldt JE, Black WC. Transformation of hepatic cell adenoma
to hepatocellular carcinoma due to oral contraceptive use. Ann Intern Med.
1989;110(6):489–490.
46. Foster JH, Berman MM. The malignant transformation of liver cell adeno-
mas. Arch Surg. 1994;129(7):712–717.
47. Foster JH. Primary benign solid tumors of the liver. Am J Surg. 1977;
133(4):536–541.