ORIGINAL ARTICLE: Clinical Endoscopy

EUS for detection of the hepatocellular carcinoma: results

of a prospective study
Pankaj Singh, MD, Richard A. Erickson, MD, FACP, FACG, Phalguni Mukhopadhyay, MD,
Shanthi Gopal, MD, Alex Kiss, PhD, Ahmed Khan, MD, T. Ulf Westblom, MD
Temple, Texas, USA

Background: Early detection of hepatocellular carcinoma (HCC) and accurate determination of the number of
lesions are critical in determining eligibility for liver transplantation or resection. Current diagnostic modalities
(CT and magnetic resonance imaging [MRI]) often miss small lesions.
Objective: To compare the accuracy of the EUS with CT for the detection of primary tumors of the liver.
Design: Prospective single-center study.
Setting: Academic medical center.
Patients: Subjects at high risk of HCC (hepatitis B, hepatitis C, or alcoholic cirrhosis) were enrolled.
Interventions: US, CT, MRI, and EUS examinations of the liver were performed. Liver lesions identified during
EUS underwent EUS-guided FNA (EUS-FNA).
Results: Seventeen patients were enrolled in the study. Nine of these patients had liver tumors (HCC, 8; chol-
angiocarcinoma, 1). EUS-FNA established a tissue diagnosis in 8 of the 9 cases. The diagnostic accuracy of US, CT,
MRI, and EUS/EUS-FNA were 38%, 69%, 92%, and 94%, respectively. EUS detected a significantly higher number
of nodular lesions than US (P Z.03), CT (P Z.002), and MRI (P Z.04). For HCC lesions, a trend was observed in
favor of EUS for the detection of more lesions than US (8 vs 2; P Z .06) and CT (20 vs 8; P Z .06). No compli-
cations were observed as a result of EUS-FNA.
Limitations: Small sample size.
Conclusions: EUS-FNA is a safe and accurate test for the diagnosis of HCC. EUS increases the accuracy of intra-
hepatic staging of the HCC by delineation of lesions, which are missed by CT and MRI. We recommend EUS for
suspected HCC, particularly in cases that are being considered for liver transplantation. (Gastrointest Endosc
2007;66:265-73.)

The incidence of hepatocellular carcinoma (HCC) is in-
creasing in the United States.1 Most patients with HCC
present when the tumor is at an advanced stage and
when surgical intervention has a low cure rate.2,3 Detect-
ing HCC early, when the lesions are small, is critical for
successful surgical therapy.4 Serial alpha-fetoprotein
(AFP) and transcutaneous US (TUS) and/or CT are cur-
rently recommended for the early detection of HCC.5
The sensitivity of AFP varies from 39% to 64%.6-8 TUS
and CT miss 42% and 32% of tumors, respectively.6-11
Copyright ª 2007 by the American Society for Gastrointestinal Endoscopy
0016-5107/$32.00
doi:10.1016/j.gie.2006.10.053
Once diagnosed, accurate staging (determination of
the size and the number of HCC lesions) decides the
eligibility for liver transplantation or liver resection.12
Though magnetic resonance imaging (MRI) appears to
have a high sensitivity (84%) for the detection of the nod-
ules between 1 and 2 cm, lesions!1 cm are missed in 70%
of cases, which lowers the accuracy of intrahepatic staging
of HCC.13,14 The excellent long-term survival with liver
transplantation in patients with HCC has increased the
demand for, and put an additional strain on, an already
short supply of donor livers.15,16 Liver transplantation in
individuals with advanced HCC may not only be futile
but withholds the liver from other eligible patients
who might have a higher success rate. Therefore, it is
important to identify the patient who is truly eligible for
liver transplantation. EUS is capable of detecting liver

www.giejournal.org Volume 66, No. 2 : 2007 GASTROINTESTINAL ENDOSCOPY 265

EUS for detection of the hepatocellular carcinoma
metastases that are missed on the CT.17-19 Case reports
show that EUS can detect HCC lesions that are missed
on a CT.20-22 However, there are no studies that evaluated
the accuracy of EUS for the detection of HCC.
We hypothesized that, given the greater sensitivity of
EUS to detect small liver lesions, it is likely to detect
HCC in high-risk subjects who have normal CTs. The pur-
pose of this study was to conduct a prospective trial to
compare the accuracy of EUS and EUS-guided FNA
(EUS-FNA) with a CT for the detection of primary liver
tumors.
PATIENTS AND METHODS
This study was conducted at the Central Texas Veterans
Health Care System, Temple, Texas. The institutional re-
view board of the Central Texas Veterans Health Care Sys-
tem, Texas, approved the protocol. Informed consent was
obtained from all participants.
Subjects with hepatitis B, hepatitis C, or alcoholic cir-
rhosis who were at high risk for HCC were prospectively
enrolled in the study from February 2005 to January
2006. High risk was defined as the presence of an elevated
AFP (AFP O 8.1 ng/mL) and/or abnormal radiologic find-
ings. Abnormal radiologic findings were defined as focal
lesions in the liver that were suggestive of HCC. Patients
were referred for the study from both the medical inpa-
tient service and outpatient hepatitis C and gastroenterol-
ogy clinics. Eligible participants underwent US, CT, MRI,
and EUS examination of the liver. EUS-FNA was performed
in subjects who had EUS evidence of liver lesions.
The primary objective of the study was to compare the
accuracy of EUS and CT for the detection of primary car-
cinoma of the liver. A secondary objective was to study
the safety of EUS-FNA for liver lesions. The criterion stan-
dard for the diagnosis of the HCC was cytologic confirma-
tion of the presence of malignant cells. Lesions were
considered benign when there was no progression in
the size of the lesion on follow-up imaging for a period
of 6 to 12 months.
EUS procedure and FNA technique
EUS was done by a single endoscopist. The curved
linear-array echoendoscope (GF-UCT/P 1140; Olympus
America Corp, Melville, NY) was used for all EUS examina-
tions and FNA. Images were obtained at 7.5 MHz, and
EUS-FNA was performed with a 22-gauge needle (FNA
needle; Olympus). A pathologist and a cytotechnologist
were present in the room to provide an on-site prelimi-
nary diagnosis.
Abdominal US
A US of the liver was performed with a US machine
(model HDI 5000; Advanced Technology Laboratories,
Bothell, Wash) with 3.5-MHz frequency.
266 GASTROINTESTINAL ENDOSCOPY Volume 66, No. 2 : 2007
Singh et al
Capsule Summary
What is already known on this topic
d CT and MRI may miss small HCC lesions important for
determining patient eligibility for liver transplantation.
What this study adds to our knowledge
d In a prospective single-center study of patients at high
risk of HCC, the diagnostic accuracy of US, CT, MRI, and
EUS/EUS-FNA was 38%, 69%, 92%, and 94%, respectively,
with EUS detecting a higher number of nodular lesions
than the other modalities.
CT
All CT examinations were performed with the use of
a helical CT scanner (model PQ-5000; Picker International,
Cleveland, Ohio). Images were acquired with the use of
10-mm collimation. A total of 100 mL Omnipaque 300
(GE Healthcare AS, Oslo, Norway) was administered intra-
venously to the patients. An injection rate of 3 mL/sec and
a scan delay of 60 to 70 seconds was used.
MRI
MRI imaging was performed on a 1.5 T system. (Magne-
tom; Siemens, Erlangen, Germany). The MRI protocol
included acquisition in T1-weighted in-phase and out-of-
phase and T2-weighted fat-suppressed images in transverse
planes, conventional HASTE (half-Fourier acquired single-
shot turbo spin echo) coronal images, and dynamic post
gadolinium injection transverse images. A paramagnetic con-
trast agent, gadopentetate dimeglumine (Magnevist; Scher-
ing, Berlin, Germany), at a dose of 0.2 mmol per kilogram of
body weight, and a flow rate of 3 mL/sec was used. This was
followed by subtraction imaging in the transverse plane.
Follow-up
Follow-up consisted of a patient interview by a trained
endoscopy nurse (for outpatients) or a personal visit by
a physician (for inpatients), communication with the
patient’s primary care physician, collection of additio-
nal radiologic test results, and a review of cytopathologic
findings.
Statistics
Sensitivity; specificity; positive predictive value; nega-
tive predictive value; and accuracy of CT, MRI, and EUS/
EUS-FNA were calculated and compared. A McNemar
test was performed to assess the differences among ac-
curacies of tests of interest. The accuracy of ‘‘intention
to perform MRI’’ was also calculated. The results were an-
alyzed on a per-lesion and a per-patient basis. A nonpara-
metric test (Wilcoxon signed rank test) was used to
compare the number of lesions among EUS and other
tests. If 1 lesion in a subject was cytologically established
www.giejournal.org
Singh et al
Figure 1. Enrollment of the subjects for the participation in the study.
to be HCC, then all other lesions with similar imaging char-
acteristics were also considered to be malignant.
Lesions were defined as ‘‘false positive’’ when the cytology
was negative for malignancy and when there was no pro-
gression on follow-up imaging. A Fisher exact test was per-
formed to assess the relation between Child class and the
diagnosis of HCC. The interobserver reliability for the num-
ber of lesions between 2 endoscopists was assessed by
means of a kappa statistic. A P value of less than .05 was con-
sidered to indicate statistical significance. All analyses were
performed by using SAS Version 9.1 (SAS Institute, Cary,
NC).
RESULTS
Seventeen subjects were enrolled in the study. The me-
dian (standard deviation [SD]) age of the participants was
56 years (10.9 years; range, 43-85). During the study
period, cytologically confirmed primary liver tumors
were detected in 9 of the 17 cases (HCC, 8; cholangiocar-
cinoma, 1). EUS, EUS-FNA, CT, MRI, and US were per-
formed in 17, 16, 15, 14, and 13 cases, respectively (Fig. 1).
www.giejournal.org
EUS for detection of the hepatocellular carcinoma
A review of data from the tumor registry showed that
only 2 cases of HCC were not enrolled during the enroll-
ment period. One patient was not seen in the gastro-
enterology clinic and, therefore, was missed by the
investigators. This patient underwent a CT-guided biopsy,
which established the diagnosis of HCC. The second patient
was diagnosed as having HCC by EUS-FNA. However, con-
sent could not be obtained for participation in the study;
therefore, this patient was not included for the analysis.
Child-Pugh score
The Child-Pugh score was 5, 6, 7, 8, and 9, in 6, 5, 2, 1,
and 2 cases, respectively; and Child class was A and B in
11 and 5 cases, respectively. Assessment of the relation be-
tween the Child class and the diagnosis of HCC showed sig-
nificant association (P Z.03) in favor of HCC and class B.
AFP levels
Sixteen of the 17 patients had AFP levels done. Of the
8 patients found to have HCC, AFP levels were within
normal or near normal in 5 and were elevated more
than 8-fold in 3 patients.
Volume 66, No. 2 : 2007 GASTROINTESTINAL ENDOSCOPY 267
EUS for detection of the hepatocellular carcinoma Singh et al

TABLE 1. Characteristics of liver ‘‘lesions’’ that TABLE 2. Comparison of US, CT, MRI, and EUS-FNA
underwent EUS-FNA for primary liver tumors

Malignant* Benign No. lesions

Location of lesions Test that
established
Left lobe 6 4 Tissue cytologic
Right lobe 8 3 Subjects US CT MRI EUS type diagnosis

Size of the lesion, mm 1 d 0 0 6 HCC EUS-FNA

!5 1 0 2 1 IE 1 1 HCC EUS-FNA

6-10 1 1 3 d 2 CI 3 HCC EUS-FNA

11-20 1 1 4 0 1 1 1 HCC EUS-FNA

21-30 5 1 5 0 1 1 2 HCC EUS-FNA

O30 6 4 6 0 0 3 2 HCC EUS-FNA

Echogenicity 7 3 3 HCC d

Hyperechoic 10 4 8 1 1 1 2 HCC EUS-FNA

Hypoechoic 3 2 9 d d IE 1 Cholangio EUS-FNA

carcinoma
Isoechoic 1 1
10 0 2 0 7 Dysplasia/ EUS-FNA
*Eleven lesions were definitely malignant, and 3 lesions were highly HCC
suspicious for malignant cells. cannot
be ruled
out
Abdominal US/CT/MRI
EUS was performed before a CT and a MRI in 53% and No. 2/5 8/7 7/6 21/8
60% of cases, respectively. The sequence of tests was not malignant
lesions/
preplanned and was not determined by the investigators. no. cases
Timing of the tests was dependent on the scheduling
office at the department of radiology. CI, Contraindicated; IE, incomplete examination.

Of the 9 cases with liver tumors, an MRI could not be

done in 2 cases: 1 because of a cardiac pacemaker, and
the other because of a penile implant. Of the remaining
7 subjects who underwent MRI, imaging could not be ob-
tained in 1, because of the inability of the patient to hold
breaths in a satisfactory manner, and, in 1 case, the patient
had to be referred to another institution for an open MRI
because of obesity. Of 7 cases without HCC, MRI was
contraindicated in 1 case.
EUS/EUS-FNA
EUS-FNA was performed in 16 of the 17 patients with
identifiable liver lesions, and a cytologic diagnosis of pri-
mary liver tumor was established in 8 cases (HCC, 7; chol-
angiocarcinoma, 1) (Tables 1 and 2). The median number
of needle passes to establish the cytologic diagnosis of ma-
lignancy was 2 (SD, 1.1; range, 1-4). The median depth
of liver lesion that underwent FNA was 30 mm (range,
11-60 mm). Among the patients found to have HCC, 6
had multiple lesions and 1 had a single lesion.
Characteristics of lesions that underwent FNA
In total, EUS-FNA was performed on 21 lesions (left
lobe, 10; right lobe, 11). The echogenicity of the malignant
lesions was variable, predominantly being hyperechoic.
The smallest lesion that was cytologically confirmed to be
malignant on EUS-FNA was!4 mm in size. (Table 1, Fig. 2).
Comparisons of EUS with other tests
for accuracy
The diagnostic accuracies of US, CT, MRI, and EUS/EUS-
FNA were 38%, 69%, 92%, and 94%, respectively. The dif-
ference was not statistically significant. The accuracy of
combined EUS and EUS-FNA was significantly superior
to EUS alone (P Z .01) (Table 3).
Comparison of EUS with other tests
for the number of benign
and malignant lesions
Total lesions (benign and malignant). EUS de-
tected a significantly higher number of total lesions than
US (24 vs 10; P Z .03), CT (33 vs 12; P Z .002), and MRI
(31 vs 11; P Z.04). Stratifying the lesions to left and right
lobes of the liver showed that EUS detected a significantly
higher number of lesions than CT in the left lobe of the
liver (16 vs 1; P Z .0078). There was no significant differ-
ence in the number of lesions in the left lobe between EUS

268 GASTROINTESTINAL ENDOSCOPY Volume 66, No. 2 : 2007 www.giejournal.org

Singh et al EUS for detection of the hepatocellular carcinoma

Figure 2. A, Very small dysplasia lesion. EUS image, showing 5-mm dysplastic lesion in the left lobe of the liver that was missed on CT and MRI. EUS
examination showed 3 hypoechoic lesions (%5 mm) in the liver and FNA of 1 of the 3 lesions confirmed the lesion to be dysplasia. B, Very small HCC
lesion. EUS image, showing 1 of the 6 hypoechoic lesions (5-7 mm) that were detected on EUS examination of the left lobe of the liver. MRI showed
multiple small nodules in the left lobe of the liver, consistent with regenerating nodules. CT did not detect any lesion in the left lobe of the liver. The
EUS-FNA confirmed the lesion to be HCC. The patient was referred for radiofrequency ablation (RFA); however, the lesions could not be visualized on
the radiologic imaging. RFA was cancelled, and the patient underwent chemoembolization. C, Deep HCC lesion in the right lobe of the liver. EUS
showed a ‘‘deep subdiaphragmatic’’ hyperechoic lesion in the right lobe of the liver. EUS-FNA was successful in the acquisition of cytology and the
confirmation of the diagnosis of HCC. D, Hemangioma. EUS image, showing hyperechoic lesion (hemangioma).

and US and MRI (P Z.25). There was no significant differ-
ence between EUS and CT for the number of lesions in
the right lobe (17 vs 11; P Z.25). There was a marginally sig-
nificant difference between EUS and US (P Z .05) in the
number of lesions in the right lobe.
HCC lesions. For HCC lesions, a trend was observed
in favor of EUS for the detection of more lesions than
US (8 vs 2; P Z .06) and CT (19 vs 8; P Z .06). EUS de-
tected more lesions than MRI (14 vs 7; P Z.25); however,
the difference was not statistically significant (Fig. 2).
Interobserver variation for the identification
of the lesions
There was a high degree of reliability for the number of
lesions identified by EUS (kappa 0.941, 95% confidence
interval, 0.916-0.966).
Safety of EUS-FNA
Sixteen patients underwent EUS-FNA. Forty-one needle
passes were made in 21 lesions. No minor or major com-
plications were observed as a result of EUS-FNA.
DISCUSSION
This prospective study showed that EUS-FNA is a safe
and accurate test for the detection of HCC in high-risk
patients. Importantly, EUS detected small HCC lesions
that were missed by CT and MRI. EUS-FNA helped in the
determination of the cytologic nature of liver nodular
lesions that were indeterminate on CT and MRI.
The high sensitivity of EUS at detecting small HCC le-
sions was an advantage over CT. The early diagnosis of
HCC is highly dependent on the size of the lesion. Tumors
smaller than 2 cm in size are difficult to image and even
more difficult to biopsy.23,24 The high miss rate of HCC
by CT shown in this study was previously reported.10-14
EUS was successful in both visualizing and establishing
the cytologic diagnosis in cases that were missed by the
CT. The smallest lesion that was visualized by EUS and
cytologically confirmed to be HCC by EUS-FNA was
4 mm in size (Figs. 2 and 3). The high specificity was
another reason for the high accuracy of EUS/EUS-FNA. A
CT detected lesions that were too small to be character-
ized and, therefore, were called indeterminate for HCC.
EUS-FNA cytologically established the nature of the inde-
terminate lesions.
MRI has played a critical role in the preoperative evalu-
ation of liver transplantation, because it detects lesions
that are missed by CT.13 In this study, though the accuracy
of the EUS was statistically comparable with MRI, many
patients could not have MRI for a variety of reasons.

www.giejournal.org Volume 66, No. 2 : 2007 GASTROINTESTINAL ENDOSCOPY 269

EUS for detection of the hepatocellular carcinoma
TABLE 3. Comparison of US, CT, MRI, EUS, and EUS-FNA for the detection of the primary liver tumors
Sensitivity Specificity
N % (95% CI) % (95% CI)
US 13 50 (23-77) 29 (9-59)
CT 13 71 (41-91) 67 (36-88)
MRI 13 86 (55-98) 100 (71-100)
EUS 17 100 (77-100) 25 (9-52)
EUS/EUS-FNA 17y 89 (76-100) 100 (77-100)
CI, Confidence interval.
*Accuracy of the ‘‘intention to perform MRI’’ was 81% (53-95).
yEUS-FNA was not performed in 1 case, with EUS finding of cystic lesion.
Overall, in 18% of cases, successful imaging with MRI
could not be obtained. EUS/EUS-FNA was successful in im-
aging and establishing the tissue diagnosis in all of these
cases. These results indicate that, in situations when
a MRI is contraindicated or difficult to perform, EUS-
FNA is useful as an alternative procedure for pre–liver-
transplantation evaluation.
Accurate delineation of the number of HCC lesions is
critical for determining the eligibility for liver transplanta-
tion. In this study, EUS detected 11 HCC lesions that were
missed by CT and 7 HCC lesions that were missed by MRI
(Fig. 2). The ability of EUS to detect occult metastatic le-
sions in the liver that are missed by CT was reported ear-
lier.17-19 The imaging of the liver for HCC differs from that
of metastases from primary tumors of other organs, be-
cause HCC frequently coexists with cirrhotic regenerative
and dysplastic nodules. Identification of liver lesions
against the background of diffuse fibrosis and regenerative
nodules is difficult when using imaging modalities.
Regenerative nodules, dysplastic nodules, and HCC
may be impossible to distinguish without FNA, particularly
when they are small.25 In this study, there were 2 cases
with multiple subcentimeter lesions, which required tissue
acquisition to identify their nature. EUS-FNA was success-
ful in tissue acquisition and helped to confirm a diagnosis
of HCC in 1 case and dysplasia in another (Figs. 2 and 3).
EUS-FNA is the only test that has the capability of sam-
pling such small lesions and, thus, the ability to differenti-
ate a malignant lesion from a nonmalignant one. Though
it is difficult to draw conclusions from just 2 cases, these
2 cases do show that EUS-FNA has the potential to dis-
tinguish dysplasia from HCC. It is important to identify
dysplastic lesions in the liver, because they are precancer-
ous, just like dysplasia in other conditions, eg, esophageal
and cervical adenocarcinoma. Anthony et al26 showed
a strong association between liver-cell dysplasia and
HCC, which was confirmed by other studies.27-34 The abil-
ity to detect a small lesion and to cytologically confirm the
precancerous nature of a lesion is of potential importance
270 GASTROINTESTINAL ENDOSCOPY Volume 66, No. 2 : 2007
Singh et al
Positive predictive Negative predictive Accuracy
value % (95 CI) value % (95 CI) % (95 CI)
37 (14-67) 40 (16-69) 38 (15-68)
71 (41-91) 67 (36-88) 69 (39-90)
100 (71-100) 86 (55-98) 92 (62-100)*
60 (34-81) 100 (77-100) 65 (39-85)
100 (77-100) 89 (63-98) 94 (69-100)
for using EUS as a screening test for HCC in a high-risk
population and for follow-up in patients with nodular dys-
plastic lesions.
It is difficult to differentiate benign from malignant liver
lesions by using US, because HCC lesions can be hypere-
choic, hypoechoic, or isoechoic. Because US does not def-
initely distinguish a liver lesion as benign or malignant, we
recommend that, once a lesion is identified, FNA be per-
formed, regardless of echogenic features. A distinct advan-
tage of EUS is that EUS examination and EUS-FNA can be
performed simultaneously, and, therefore, confirmation of
malignancy can usually be accomplished in a single pro-
cedure, whereas, conventional imaging studies typically
require 2 sessions to accomplish these tasks, 1 for detec-
tion and another for FNA.
The role for establishing a tissue diagnosis has been
questioned in cases where the lesions are large and there
is marked elevation of AFP.23 Tumor seeding from the
needle track during the biopsy evaluation of the HCC is
one of the concerns. Consensus diagnostic criteria can be
used for establishing the noninvasive diagnoses in such
cases.6,22 In this study, the majority of subjects did not
meet the noninvasive diagnostic criteria for HCC. AFP levels
were near normal in 50% of the cases, and a CT missed
the diagnosis in 40% of cases. It has been well established
that various types of benign nodules and pseudolesions
are identified on imaging scans performed for the diag-
nosis of HCC. Regenerating nodules, in particular, can
closely resemble HCC on imaging tests. An accurate dif-
ferentiation between true HCC and regenerative nodules
is critical for correct patient management. Tissue diagno-
sis plays a pivotal role, particularly when therapeutic op-
tions include major surgical procedures.
There is a theoretical risk that, with increases in the
number of passes, needle-track seeding may increase.
Though the final diagnosis is based on H&E stained sec-
tions of cell block, on-site cytopathology assessment of
the tissue sample for the preliminary diagnosis is helpful
in avoiding the excessive needle passes (Fig. 3). Dysplastic
www.giejournal.org
Singh et al EUS for detection of the hepatocellular carcinoma

nodules exhibit cellular atypia, but the architecture of liver

cell plates is only mildly altered, whereas HCC is associ-
ated with cellular atypia and a trabecular (3-4 liver cell
plates surrounded by layer of a flattened endothelial cells)
architectural pattern (Fig. 3).
A large multicenter survey showed that EUS-FNA is safe
for liver lesions.35 This study also showed the safety of ac-
quiring tissue samples in patients with underlying chronic
liver disease. However, the patients in our study were at
a higher risk of bleeding complications because of their
underlying chronic liver disease. Contrary to popular be-
lief, lesions in the right lobe of the liver could be visualized
and successfully sampled during EUS examination. The
number of lesions detected in the right lobe of the liver
on EUS and a CT were 17 and 11, respectively. Of the 21
lesions that underwent EUS-FNA, 11 were performed
from the bulb and the second portion of the duodenum.
The smallest lesion that was sampled in the right lobe was
smaller than 10 mm. Acoustic shadows from gallstones
posed a problem in the complete examination of the right
lobe of the liver.
A small sample size is the main limitation of this study.
However, we decided to report our findings, because the
results showed the diagnostic superiority of EUS/EUS-FNA
over CT. The study showed the ability of the EUS to detect
small, as well as large occult HCC lesions that were missed
on radiologic imaging and to determine the nature of
lesions by using EUS-FNA (in a single session) when the
findings of the imaging tests are indeterminate. Interest-
ingly, the majority of the patients with HCC had normal
or near-normal AFP and normal US, which suggests that
EUS may be used as an adjunct screening test for HCC.
When considering these findings, we propose an algo-
rithm (Fig. 4).
In conclusion, this study showed that EUS/EUS-FNA is
an excellent test for the diagnosis of HCC in high-risk
patients. Because EUS can detect lesions that are missed
by conventional imaging modalities, it should be consid-
ered in patients who are candidates for liver transplanta-
tion or curative liver resection. The ability to perform
imaging and EUS-FNA in a single session and to sample
subcentimeter lesions is a distinct advantage over US,
CT, and MRI, and, therefore, should also be considered
to differentiate small hepatoma lesions from regenerative
nodules and dysplasia when imaging studies detects
lesions that are too small to be characterized.

<
Figure 3. Cytopathology of samples obtained by using EUS-FNA of liver
lesions. A, Liver, dysplastic nodule: normal and enlarged liver cell plates,
composed of hepatocytes with small and scattered large nuclei (H&E,
orig. mag. 40). B, Hepatocellular carcinoma: pseudoglandular forma-
tion of atypical hepatocytes (on-site Diff-Quik smear, orig. mag. 40).
C, Hepatocellular carcinoma: sheet of markedly pleomorphic cells with
enlarged nuclei and prominent nucleoli (H&E, orig. mag. 40).

www.giejournal.org Volume 66, No. 2 : 2007 GASTROINTESTINAL ENDOSCOPY 271

EUS for detection of the hepatocellular carcinoma
Figure 4. Algorithmic approach to the detection and staging of the HCC.
ACKNOWLEDGMENTS
We thank Tushar Patel, MD, for an insightful critique
and for editing the manuscript.
DISCLOSURE
None of the authors have any disclosures to make.
REFERENCES
1. El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma
in the United States. N Engl J Med 1999;340:745-50.
2. Pawarode A, Voravud N, Sriruranpong K, et al. Natural history of un-
treated primary hepatocellular carcinoma: a retrospective study of
157 patients. Am J Clin Oncol 1998;21:386-91.
272 GASTROINTESTINAL ENDOSCOPY Volume 66, No. 2 : 2007
Singh et al
3. Trevisani F, D’Intino PE, Grazi GL, et al. Clinical and pathologic features
of hepatocellular carcinoma in young and older Italian patients.
Cancer 1996;77:2223-32.
4. Mazzafero V, Regalia E, Doci R, et al. Liver transplantation for the treat-
ment of small hepatocellular carcinomas in patients with cirrhosis.
N Engl J Med 1996;334:693-9.
5. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepato-
cellular carcinoma. Conclusions of the Barcelona-2000 EASL confer-
ence. J Hepatol 2001;35:421-30.
6. Gambarin-Gelwan M, Wolf DC, Shapiro R, et al. Sensitivity of com-
monly available screening tests in detecting hepatocellular carcinoma
in cirrhotic patients undergoing liver transplantation. Am J Gastroen-
terol 2000;95:1535-8.
7. Nguyen MH, Garcia RT, Simpson PW, et al. Racial differences in effec-
tiveness of alpha-fetoprotein for diagnosis of hepatocellular carci-
noma in hepatitis C virus cirrhosis. Hepatology 2002;36:410-7.
8. Dodd GD 3rd, Miller WJ, Baron RL, et al. Detection of malignant tumors
in end-stage cirrhotic livers: efficacy of sonography as a screening
technique. AJR Am J Roentgenol 1992;159:727-33.
9. Trevisani F, D’Intino PE, Morselli-Labate AM, et al. Serum alpha-feto-
protein for diagnosis of hepatocellular carcinoma in patients with
www.giejournal.org
Singh et al
chronic liver disease: influence of HBsAg and anti-HCV status. J Hepa-
tol 2001;34:570-5.
10. Miller WJ, Baron RJ, Dodd GD 3rd, et al. Malignancies in patients with
cirrhosis: CT sensitivity and specificity in 200 consecutive transplant
patients. Radiology 1994;193:645-50.
11. Peterson MS, Baron RL, Marsh JW Jr, et al. Pretransplantation surveil-
lance for possible hepatocellular carcinoma in patients with cirrhosis:
epidemiology and CT-based tumor detection rate in 430 cases with
surgical pathologic correlation. Radiology 2000;217:743-9.
12. Schwartz M. Liver transplantation for hepatocellular carcinoma. Gas-
troenterology 2004;127(Suppl 1):S268-76.
13. Burrel M, Llovet JM, Ayuso C, et al. MRI angiography is superior to he-
lical CT for detection of HCC prior to liver transplantation: an explant
correlation. Hepatology 2003;38:1034-42.
14. Stark DD, Wittenberg J, Butch RJ, et al. Hepatic metastases: random-
ized, controlled comparison of detection with MR imaging and CT.
Radiology 1987;165:399-406.
15. Mazaferro V, Regalia E, Docci R, et al. Liver transplantation for the
treatment of small hepatocellular carcinomas in patients with cirrho-
sis. N Engl J Med 1996;334:693-9.
16. Wiesner RH, Freeman RB, Mulligan DC. Liver transplantation for hepa-
tocellular cancer: the impact of the MELD allocation policy. Gastroen-
terology 2004;127(Suppl 1):S261-7.
17. Nguyen P, Feng JC, Chang KJ. Endoscopic ultrasound (EUS) and EUS-
guided fine-needle aspiration (FNA) of liver lesions. Gastrointest
Endosc 1999;50:357-61.
18. Prasad P, Schmulewitz N, Patel A, et al. Detection of occult liver metas-
tases during EUS for staging of malignancies. Gastrointest Endosc 2004;
59:49-53.
19. DeWitt J, LeBlanc J, McHenry L, et al. Endoscopic ultrasound-guided
fine needle aspiration cytology of solid liver lesions: a large single-
center experience. Am J Gastroenterol 2003;98:1976-81.
20. Hollerbach S, Willert J, Topalidis T, et al. Endoscopic ultrasound-
guided fine-needle aspiration biopsy of liver lesions: histological and
cytological assessment. Endoscopy 2003;35:743-9.
21. Awad SS, Fargan S, Abudayyeh S, et al. Preoperative evaluation of he-
patic lesions for the staging of hepatocellular and metastatic liver car-
cinoma using endoscopic ultrasonography. Am J Surg 2002;184:601-4.
22. Bogstad J, Vilmann P, Burcharth F. Early detection of recurrent hepa-
tocellular carcinoma by endosonographically guided fine-needle aspi-
ration biopsy. Endoscopy 1997;29:322-4.
23. Talwalkar JA, Gores GJ. Diagnosis and staging of hepatocellular carci-
noma. Gastroenterology 2004;127(Suppl 1):S126-32.
24. Durand F, Regimbeau JM, Belghiti J, et al. Assessment of the benefits
and risks of percutaneous biopsy before surgical resection of hepato-
cellular carcinoma. J Hepatol 2001;35:254-8.
25. Krinsky GA, Israel G. Nondysplastic nodules that are hyperintense on
T1-weighted gradient-echo MR imaging: frequency in cirrhotic pa-
tients undergoing transplantation. AJR Am J Roentgenol 2003;180:
1023-7.
www.giejournal.org
EUS for detection of the hepatocellular carcinoma
26. Anthony PP, Vogel CL, Barker LF. Liver cell dysplasia: a premalignant
condition. J Clin Pathol 1973;26:217-23.
27. Ho JCI, Wu PC, Mark TK. Liver cell dysplasia in association with hepa-
tocellular carcinoma, cirrhosis and hepatitis B surface antigen in Hong
Kong. Int J Cancer 1981;28:571-4.
28. Watanabe S, Okita K, Harada T, et al. Morphologic studies of the liver
cell dysplasia. Cancer 1983;51:2197-205.
29. Akagi G, Furuya K, Kanamura A, et al. Liver cell dysplasia and hepatitis
B surface antigen in liver cirrhosis and hepatocellular carcinoma. Can-
cer 1984;54:315-8.
30. Ganne-Carrie N, Chastang C, Chapel F, et al. Predictive score for the
development of hepatocellular carcinoma and additional value of liver
large cel dysplasia in western patients with cirrhosis. Hepatology 1996;
23:1112-8.
31. Nakanuma Y, Kanel GC, Doishita K, et al. Risk lesions in cirrhosis and
development of hepatocellular carcinoma: an autopsy study. Biomed
Pharmacother 1989;43:127-33.
32. Roncalli M, Kobayashi M, Ikeda K, et al. Dysplastic nodules frequently
develop into hepatocellular carcinoma in patients with chronic hepa-
titis C and cirrhosis. Cancer 2006;106:636-47.
33. Podda M, Roncalli M, Battezzati PM, et al. Liver cell dysplasia and he-
patocellular carcinoma. Ital J Gastroenterol 1992;24:39-42.
34. Borzio M, Bruno S, Roncalli M, et al. Liver cell dysplasia is a major risk
factor for hepatocellular carcinoma in cirrhosis: a prospective study.
Gastroenterology 1995;108:812-7.
35. ten Berge J, Hoffman BJ, Hawes RH, et al. EUS-guided fine needle as-
piration of the liver: indications, yield, and safety based on an interna-
tional survey of 167 cases. Gastrointest Endosc 2002;55:859-62.
Received June 26, 2006. Accepted October 23, 2006.
Current affiliations: Section of Gastroenterology and Hepatology (P.S.,
A. Khan), Division of Oncology (P.M.), Department of Pathology (S.G.),
Section of Infectious Diseases (T.U.W.), Central Texas Veterans Health Care
System, Division of Gastroenterology and Hepatology, Scott and White
Memorial Hospital, Temple, Texas (R.A.E.), Department of Biostatistics and
Epidemiology, School of Rural and Public Health, Texas A&M University,
College Station, Texas (P.S.), Department of Internal Medicine, Texas A&M
University System Health Science Center, Temple, Texas (P.S., R.A.E., P.M.,
A. Khan, T.U.W.), USA, Division of Clinical Epidemiology, Department of
Research Design and Biostatistics, Sunnybrook and Women’s College
Health Sciences Center, Toronto, Ontario (A. Kiss), Canada.
This trial is registered at clinicaltrials.gov (NCT00290316).
Presented at Digestive Disease Week 2006, Los Angeles, Calif, USA
(Gastrointest Endosc 2006;63:AB257) and EUS 2006, Amsterdam, The
Netherlands (Endoscopy 2006;38:P43).
Reprint requests: Pankaj Singh, MD, 1901 South 1st St, Central Texas
Veterans Health Care System, Temple, Texas 76504.
Volume 66, No. 2 : 2007 GASTROINTESTINAL ENDOSCOPY 273