Professional Documents
Culture Documents
EUS has assumed an important role in the diag- staging of malignant GI or pulmonary tumors). Fourteen
nosis and staging of malignant GI tumors over the (10 men, 4 women, age range 38 to 88 years) of the 574
past few decades. The addition of EUS-guided fine- patients had focal solid liver lesions (5 right lobe, 9 left
needle aspiration (FNA) has increased the diagnostic lobe) and underwent EUS-guided FNA. Among these 14
patients, the indications for EUS before FNA included a
and staging accuracy over EUS alone.1 The clinical
pancreatic lesion seen at CT without a tissue diagnosis
utility of EUS and EUS-guided FNA appears to be (6), established pancreatic cancer with referral for EUS
greatest in the diagnosis and staging of pancreatic staging (3), liver mass with referral to assess primary
cancer and in the staging of malignant tumors of the lesion and establish tissue diagnosis (2), staging of
esophagus.2-7 EUS, however, is not traditionally esophageal cancer (1), diagnosis of extrinsic esophageal
thought to be applicable in the detection of liver compression (1), and duodenal mass (1). All procedures
metastases. The role of EUS-guided FNA in diag- were performed by one of two endosonographers (K.J.C.,
nosing liver lesions has not been well described. We P. N.). Informed consent was obtained under protocols
report on a series of patients with liver lesions approved by our institutional review board (UCI HS 94-
02, 94-03). Platelet count and prothrombin time were
detected by means of EUS with subsequent EUS-
obtained for all patients before the procedure. Exclusion
guided FNA. criteria included cystic or vascular lesions, prothrombin
PATIENTS AND METHODS time greater than 15 seconds, and platelet count less than
50,000. Antibiotics were not given prophylactically.
The liver was evaluated for a series of 574 consecutive All 574 patients were first examined with a mechanical
patients (April 1, 1995, through February 12, 1998) under- radial scanning echoendoscope at 7.5 MHz (GF-UM3 or
going upper EUS for various indications (diagnosis and UM20; Olympus America, Inc. Melville, N.Y.). EUS was
performed in a sequential manner similar to transabdom-
Received June 10, 1998. For revision September 17, 1998. Accepted inal US in which the order of structures examined was the
January 14, 1999.
same for each patient. The left lobe and hilum of the liver
From the Division of Gastroenterology, University of California, were examined from the gastric body and fundus after
Irvine Medical Center, Orange, California.
other surrounding structures were seen. The tip of the
Presented in abstract form at the American Society of
Gastrointestinal Endoscopy EUS Topic Forum, May 11-14, 1997,
echoendoscope was placed in the gastric antrum, the bal-
Washington, DC, (Gastrointest Endosc 1997;45:AB176). loon inflated with 5 to 10 mL of water, and suction applied
Reprint requests: Phuong Nguyen, MD, UCI Clinical Cancer to remove intragastric air. The echoendoscope was then
Center, 101 The City Drive, Bldg 23, Rt 81, Orange, CA 92868. slowly withdrawn with a right turn and the scope tip
Copyright © 1999 by the American Society for Gastrointestinal deflected up (similar to the shortening technique for
Endoscopy 0016-5107/99/$8.00 + 0 37/1/97208 ERCP). When the liver came into view, the instrument
was rotated to evaluate all portions of the liver. The right guided FNA cytologic diagnosis of benign hepato-
lobe of the liver was imaged from the duodenum and cytes. At autopsy, no liver metastasis was detected.
antrum. The best location for viewing the right lobe is There were no immediate or late complications of
from the duodenal bulb. When liver parenchyma is identi- EUS-guided FNA with a mean follow-up period of
fied, the transducer is rotated to visualize all portions. 5.5 months (range 1 to 23 months).
When a liver lesion was found, a linear array echoen-
Conventional abdominal CT performed before
doscope (FG-32UA or FG-36UX; Pentax, Orangeburg,
N.Y.) was used for FNA. The linear array echoendoscope (5
EUS depicted focal lesions in only 3 of the 14
and 7.5 MHz) also was used to examine the livers of all patients. The average time interval between CT and
patients who had no detectable focal liver lesion during EUS was 35 days (range 3 to 93 days). In most
examination with the radial scanning echoendoscope. All patients CT was performed within 2 months of the
15 lesions were imaged by means of both radial scanning EUS procedure. The mean largest size of the lesions
and linear array echoendoscopy. The vascularity of the detected at CT in these 3 patients was 3.5 cm (range
region was assessed with the linear array echoendoscope 1.0 to 5.2 cm). One patient underwent CT-guided
with color flow mapping and Doppler US of both the lesion FNA of the liver lesion whereas two patients did not.
and surrounding structures. EUS-guided FNA was per- CT-guided FNA yielded reactive hepatocytes and no
formed as previously described.1 A 22-gauge 10 cm needle malignant cells in this patient, and EUS-guided
was used for FNA (GIP; Mediglobe, Tempe, Arizona). After
FNA gave the tissue diagnosis of adenocarcinoma.
each FNA pass, preliminary cytologic assessment of the
specimen was performed before another pass was made.
Seven of 14 patients had a known cancer diagnosis
Clinical outcome was categorized as no change (results before EUS. For the other 7, the initial diagnosis of
of EUS FNA made no difference in clinical management), cancer was made by means of EUS-guided FNA of
avoided surgery (results prevented unnecessary surgical the liver. EUS-guided FNA confirmed the diagnosis
intervention), upstaged tumor (results led to upstaging of of liver metastasis (stage M1) and avoided surgery
primary malignant tumor), or made diagnosis (results led for 7 of the 14 patients (Table 2).
to a diagnosis of malignant tumor).
DISCUSSION
RESULTS
The diagnosis of liver metastasis is important stag-
EUS disclosed focal liver lesions in 14 of 574 ing information for deciding on therapy. Trans-
(2.4%) patients. Multiple lesions were found in 7 of abdominal US and contrast-enhanced CT are the
14 patients (range 1 to 30 lesions). All liver lesions most commonly used imaging methods for evaluating
imaged with EUS in our series were round with reg- liver metastases. EUS and EUS-guided FNA have
ular borders. With color flow mapping and Doppler been shown to be efficacious in the diagnosis and stag-
US, the lesions had no internal vascular structures. ing of pancreatic cancer and in the nodal staging of
The median largest diameter of the liver lesions was malignant GI and pulmonary tumors.1-7 However,
1.1 cm and the mean 1.8 cm (range 0.8 to 5.2 cm). EUS has not had a clinical role in imaging the liver.
Twelve of the 15 lesions detected by EUS and that CT has been found to have an overall sensitivity of
had FNA were less than 2 cm in diameter. The most 68% for the detection of focal liver lesions in patients
difficult portions of the liver to visualize were near with malignant disease.8 However, for lesions less than
the dome of the diaphragm and the inferior portions 1 cm in diameter, the detection rate drops to 49%.8 In
of the right lobe of the liver. our study, EUS disclosed liver lesions in 13 patients
EUS-guided FNA was performed on a total of 15 with a median size of 1.1 cm and a mean size of 1.8 cm;
liver lesions in the 14 patients (Table 1). One patient 12 of the 15 lesions seen with EUS were smaller than
underwent FNA of two separate liver lesions 2 cm. CT detected lesions in only 3 patients (21%).
because FNA of the first lesion in this patient yielded Spiral CT with arterial portography is a newer tech-
normal preliminary cytologic results. The final cyto- nique that has mini-mal motion artifact and section
logic diagnosis of both lesions was adenocarcinoma. misregistration, thus it provides enhanced visualiza-
In these patients, the average needle distance from tion of the liver.9-11 However, this modality frequently
the gastric or duodenal wall to the liver lesion was shows nontumorous perfusion defects and therefore
3.3 cm (range 2.2 to 5.0 cm). The mean number of has a high false-positive rate of diagnosis.9-12
passes per lesion was 2.0 (range 1 to 5 passes). EUS- Transabdominal US has an overall detection rate
guided FNA yielded an adequate specimen in all 14 for liver metastases ranging from 53% to 71%.8,13-15
patients. Fourteen of the 15 lesions sampled by For liver lesions smaller than 1 cm in diameter, the
means of EUS-guided FNA were found to be malig- detection rate falls to 20%.8 This modality appears
nant (11 adenocarcinoma, 1 squamous cell carcinoma, to be inferior to CT and magnetic resonance imaging
2 poorly differentiated carcinoma). One patient with (MRI) for detecting liver metastases.8,16-18 MRI has
pancreatic cancer and one liver lesion had an EUS- been shown to be superior to CT for detecting hepat-
*Not measured.
ic lesions in some comparative studies and inferior Table 2. Clinical outcome with EUS-guided FNA
in others.8,19,20 The conflicting results may relate to of the liver
variations in contrast enhancement techniques and Outcome No. of patients Primary malignant lesion
the many different MRI pulse sequences.21
No change 1/14 (7%) Pancreatic (1)
EUS has been used to detect pancreatic cancer with Avoided surgery 7/14 (50%) Pancreatic (5), lung (1),
high sensitivities, ranging from 88% to 99%.4,22-24 For esophageal (1)
lymph node staging of gastric and esophageal cancer, Upstaged tumor 4/7 (57%) Pancreatic (2), breast (1),
EUS sensitivities are similarly high, reported ranges lung (1)
Made diagnosis 7/7 (100%) Pancreatic (5), hepatoma
being 74% to 80% and 80% to 90%, respectively.25
(1), cholangiocarcinoma (1)
However, to our knowledge there have been no series
reported on the utility of EUS or EUS-guided FNA for
detecting liver lesions. In our series, EUS revealed However, percutaneous FNA is limited by inability
that 14 of a group of 574 patients undergoing EUS for to depict small focal liver lesions. In our series, CT
various indications had focal liver lesions; convention- showed liver lesions in only 3 of 14 patients (21%)
al CT missed focal liver lesions in 11 of these patients. with liver lesions imaged with EUS. Although 2 of
Because patients with obvious liver metastasis at CT the 3 patients did not undergo percutaneous FNA,
were not referred for EUS local staging, EUS may 1 patient did undergo CT-guided FNA of the liver
have greater utility in the detection of liver metasta- with cytologic findings that revealed no malignant
sis before CT. cells. A diagnosis of adenocarcinoma was made for
Various sonographic patterns and echo character- this patient on the basis of EUS-guided FNA find-
istics of liver lesions have been described with trans- ings (Fig. 3).
abdominal US. Tumors with high vascularity such as In our experience, EUS-guided FNA of liver
renal cell and islet cell carcinoma tend to appear lesions located far away from the tip of the echoen-
hyperechoic.26,27 Hypovascular lesions, such as lym- doscope was often more difficult than FNA of lesions
phoma, appear as hypoechoic lesions.28,29 We identi- closer to the probe. The needle may bend or curve as
fied 12 patients with hypoechoic and 2 with hypere- it traverses normal liver to reach the target liver
choic liver lesions (Figs. 1 and 2; Table 1). One of the lesion. The needle tip may then miss the target if the
hyperechoic lesions yielded benign hepatocytes at lesion is small. When this occurred, minor adjust-
FNA, and the other was a cholangiocarcinoma. ments were needed in the needle trajectory to comp-
For the 14 liver lesions detected with EUS, ensate for the bend in the needle. Liver lesions near
EUS-guided FNA had a sensitivity, specificity, and the second or third portion of the duodenum were
diagnostic accuracy of 100%. In comparison, CT difficult to visualize, likely because of positioning of
and US-guided FNA have been reported to have the echoendoscope. In our series, the mean number
sensitivities ranging from 83% to 93.2%. 30-33 of passes per lesion was two, which is in the range of