Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
65
Sonography with Intraarterial Infusion of Carbon Dioxide Microbubbles (Sonographic Angiography): Value in Differential Diagnosis
of Hepatic
Tumors
: Masatoshi Kudo1 Shusuke Tomita1 Hitoshi Tochio2 Jun Mimura1 Yoshihiro Okabe1 Hiroshi Kashida1 Masahiro Hirasa1 Yasuyosi Ibuki1 Akio Todo1
Differential diagnosis of small liver tumors is important, but is not always possible, even with angiography. To solve this problem, we introduced sonographic angiography, which combines sonography and angiography. The vascular pattern of a variety of hepatic nodules was evaluated with sonographic angiography, and the results were compared with those of conventional angiography. Sonographic angiography (sonography performed during intraarterial infusion of carbon dioxide microbubbles) was performed in 184 patients with a total of 222 hepatic nodules: 139 hepatocellular carcinomas, nine adenomatous hyperplasias, seven regenerative nodules, 21 hemangiomas, 33 metastases, seven lymphomas, one granuloma, and five focal nodular hyperplasias. Sonographic angiography detected a hypervascular pattern with penpheral blood supply in cases of hepatocellular carcinoma (sensitivity, 90%; specificity, 89%). Typical vascular patterns of adenomatous hyperplasia, hemangioma, metastasis, and focal nodular hyperplasia on sonographic angiography were hypovascularity (sensitivity, 100%; specificity, 91%), spotty pooling (sensitivity, 100%; specificity, 100%), peripheral hypervascularity (sensitivity, 64%; specificity, 100%), and a central arterial supply (sensitivity, 100%; specificity, 100%), respectively. The detectability of hyper-
vascularity
was greater
with sonographic
angiography
than with conventional
angiog-
raphy in hepatocellular carcinoma, metastasis, and hemangioma. Our experience indicates that sonographic angiography depicts characteristic vascular features that reflect the vascular anatomy of specific types of hepatic tumors, and thus is useful in the differential diagnosis of these lesions. AJR
158:65-74,
The recent tion
January
advances
of small
1992
in several
asymptomatic
hepatic
imaging
techniques
tumor
nodules.
have led to improved Confirmation
of the
detec-
pathologic
type of such nodules is important but is not always possible when the nodules are extremely small or do not have characteristic findings on conventional angiograms or histopathologic examinations. Several attempts have been made to improve the sensitivity
Received May 16, 1 991 : accepted July31, 1991.
after revision
This work was supported in part by the 1989 Grant-in-Aid for Medical Research of Kobe City General Hospital. 1 Department of Medicine, Division of Gastroenterology, Kobe City General Hospital, 4-6, Minatojima-Nakamachi, Chuo-ku, Kobe 650, Japan. Address reprint requests to M. Kudo. 2 Section of Abdominal Ultrasound Service, Kobe City General
Hospital,
Kobe
650, Japan.
0361 -803X/92/1 581-0065 © American Roentgen Ray Society
of
angiography
in diagnosing
hepatic
tumors.
These
include
digital
subtraction angiography, infusion hepatic angiography [1 ], CT arteriography [2], and CT after intraarterial injection of iodized oil (Lipiodol CT) [3, 4]. Although those techniques improve sensitivity in detecting hypervascular nodules, they are still of limited value in the diagnosis of hypovascular lesions. Moreover, in hypovascular tumors, they do not provide diagnostic information about the specific histology of the lesions. In order to solve this problem, we studied the usefulness of sonographic angiography
(sonography
bubbles) in detecting hepatic tumors.
Materials From patients
with
tumor
intraarterial
vascularity
infusion
of carbon
dioxide
as an aid in the differential
[CO2]
micro-
diagnosis
of
and Methods March with
1989 222
hepatic
to
January tumors:
1991
,
sonographic
1 39 hepatocellular
angiography carcinomas
was (HCC5),
performed nine
in
adenomatous
184
KUDO
66
AJA:158,
(FNH5). Of 1 1 8 patients
selected for the conventional angiographic examination because of the need for a differential diagnosis as well as for intraarterial therapy. The 222 hepatic nodules had been detected with conventional so-
The total period of the procedure was classified into three phases: early, middle, and late, depending on the degree of enhancement of the liver parenchyma with CO2 microbubbles. The period until the surrounding liver parenchyma was filled with CO2 microbubbles was defined as the early phase, which lasted approximately 5-1 0 sec.
nography,
The period the CO2 microbubbles
cirrhosis
and
in this
11 had
chronic
and these
study.
proved
133 nodules).
confirmed
with
sion
CT (SPECT),
peared
hyperechoic,
listed
were
nodules
seven
originally
were
included
hemangiomas
were
of the seven
‘Tc-IabeIed
hemangiomas
RBC
and/or
clinical
was
single-photon
follow-up
emis-
of more
than
tumors,
132 appeared hypoechoic, 28 apthe remaining 62 had a mosaic pattern on The size ranges of the 222 tumors are
and
sonography.
in Table
patients
82 nodules; biopsy under sonographic
angiography,
Of 222 hepatic
visible except
Diagnosis
delayed
1 year.
unenhanced
nodules
(surgery,
guidance,
MA,
All 184
sonographically
All hepatic
histologically
hepatitis.
107 had liver
time
needed
preparing
Two
proper, right, or left hepatic artery) after conventional superselective hepatic angiography, including infusion hepatic angiography [1 J. Superselective infusion hepatic angiography was performed in 138
cases
a total
with
of
a total
155
Berlin)
(Toshiba
angiography
nodules
after
at a rate
of 2-4
of 40-60
injection mI/sec.
Digital
DFP 40A unit) was performed
of 67 nodules
after
injection
of 1 0-20
ml of
in 46 patients
ml of iopamidol
at
were prepared by vigorously normal saline, and 5 ml of the patient’s own blood. The patient’s own blood was added to increase the surface tension of the bubbles so that microbubbles by hand
could
be easily
1 0 ml of CO2, 10 ml of heparinized
obtained. 1 -2
of approximately
The
mI/sec,
microbubbles
were
injected
permitting
the accurate
at a rate
evaluation
of
tional
when
ancy
arose
preters of the
a long
period,
which
permits
a precise
observation
their
of the
breath
and
the nodule
videotape until
In all cases,
recorder
the
bubbles
preparation
of
dynamic
images
from before the injection were
washed
out
CO2 microbubbles
from
the
was
quite
were
recorded
on a
parenchyma.
easy
and
out
knowledge
1: Histologic
Varieties
of Liver
Nodules
Hepatocellular
Adenomatous
carcinoma
hyperpla-
3 cm
Total
38
36
139
2
0
0
7
0
7
0
0
7
0 0 0
6 8 0
15 21 5
0 4 2
21 33 7
0
1
0
0
1
1
3
1
0
5
spotty
(4) central
pooling
arterial
supply
9
sia
Regenerative
Most
interpreters.
angiograms
ing on the tumor
:
Type of Lesion
liver
knowledge
of vascularity;
was isovascular
angiography
peripheral
TABLE
10 than
phases.
sonographic
The
needed
the vascular
without
the case and came to a consensus.
sonographic
late
of CO2 microbubbles liver
other
in the surrounding
including
results.
diagnosis
the
=
Tokyo).
extra
than
Vasculanty on both sonographic and conventional angiography was quantified by comparing tumor vascularity with the vascularity of the surrounding liver parenchyma. Quantification of the vascularity on sonographic and conventional angiography was done blindly with-
sion
vascular
total less
technique
blindly
results,
biopsy
pattern of the small nodule in the early arterial phase. A convex and/ or linear-array electronic scanner with a 3.5-MHz (n = 1 74) or 5.0MHz (n 48) transducer was used for the studies (SSD 650 model, Aloka,
analyzed
imaging
or
reevaluated
raphy
to hold
remained
the videotape
on the
between
graphic
them
other
agreed
neous
to enable
was
required.
interpreters
angiography;
interpreters
The total volume injected was determined by monitoring sonographic images so that strong attenuation of the deeper liver parenchyma was not produced. The total volume injected ranged from 5 to 20 ml in each study. Oxygen was given to the patients during the sonoprocedure
The
usually
No special
was
by replaying
history;
parenchyma:
for
microbubbles
independent
the CO2 flow within the liver and nodule from the early arterial phase to the late phase. The total injection time ranged from 5 to 10 sec.
angiographic
of hemangioma.
sec (early phase) and thereafter
subtraction
a rate of 2-4 mI/sec. CO2 microbubbles mixing
CO2
angiography.
angiography
parenchyma was defined as the middle phase, which lasted approximately 1 0-60 sec after injection of CO2. The late phase was defined as the period when CO2 was almost washed out from the liver parenchyma, which usually was 1-3 mm after injection of CO2. The duration of each phase varied from case to case, depending on the placement of the catheter, total volume of CO2 injected, hepatic blood-flow velocity, and severity of liver disease.
patient’s
iopamidol(Schering,
for conventional
sonographic
in cases
of each nodule
1.
with
for
mm except
Sonographic angiography was performed by injecting microbubbIos of CO2 through a 6.5-French preshaped hepatic catheter (Clinical Supply Co., Gifu, Japan) placed into the hepatic artery (either the
patients
used
1992
other
with a total of 139 HCCs,
solution
January
seven regenerative nodules, 21 hemangiomas, 33 metastases (colorectal cancer 1 9; stomach cancer, five; pancreatic cancer, four; lung cancer, three; and breast cancer, two), seven lymphomas, one granuloma, and five focal nodular hyperplasias hyperplasias,
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
ET AL.
II
isovascular
a
hypovascular
pattern
Fig. 1.-vascular
pattern
patterns
seen
on sonographic
angiography
of hepatic
masses after intraarterial infusion of carbon dioxide (C02) microbubbles. Vascularity was evaluated at three phases: early, middle, and late.
AJR:158,
(n
=
and
January
55). The sensitivity,
Side
vascular
effects
symptoms,
of each
and
or complications
by whether
physical
angiography.
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
pattern
specificity,
bles were defined
were
SONOGRAPHIC
1992
Blood
monitored
pressure, the
tumor
predictive
values
caused
by injection
was
were
or
laboratory
pulse
rate,
study
tests respiration,
specified,
determined.
of CO2 microbub-
or not there was a change
findings,
during
hepatic
ANGIOGRAPHY
after and
in subjective sonographic temperature
period.
The chi-square test was used for statistical analysis. less than .01 was considered statistically significant.
A p value
of
Results Table 2 and Figure 1 summarize the vascular patterns of the 222 hepatic tumors in the 1 84 patients. No complications or abnormal changes in laboratory tests, including liver function tests, associated with intraarterial injection of CO2 were observed in any of the 1 84 patients in whom sonographic angiography was performed.
Hepatocellular
HEPATIC
TUMORS
67
middle phases and did not enhance in the late (capillary) phase. In 24 of 35 hypervascular HCCs in which tumor vessels were not detected, the vascular direction was from the periphery to the center. The remaining 1 1 hypervascular HCCs showed simultaneous filling within the small nodule. A peripheral hypoechoic layer (halo), which was observed in 91 of 125 hypervascular HCCs, was not enhanced with CO2 microbubbles (Fig. 3). Pathologic examination revealed that eight of the nine isovascular HCCs and all five hypovascular HCCs were well differentiated (grade I) HCCs on the EdmondsonSteiner scale [6], whereas only eight of 1 25 hypervascular HCCs were grade I. The typical vascular pattern of HCCs was a peripheral or simultaneous arterial supply and a homogeneous or mosaic hypervascular pattern (Figs. 2 and 3). Sensitivity and specificity of this finding were 90% (1 25/1 39) and 89% (74/83), respectively. Positive and negative predictive values of this finding were 93% (1 25/1 34) and 84% (74/88), respectively.
Carcinoma
HCC had a hypervascular pattern (Figs. 2 and 3) in 90% (1 25/1 39), an isovascular pattern in 6% (9/1 39), and a hypovascular pattern in 4% (5/1 39) on sonographic angiography (Table 2). In contrast, 64% (61/96) of HCCs were hypervascular on infusion hepatic angiography and 74% (32/43) were hypervascular on digital subtraction angiography (Table 3). Forty-six nodules showed no abnormality on infusion hepatic angiography or digital subtraction angiography. Sensitivity of sonographic angiography for the detection of hypervascularity was superior to that of conventional angiography (p < .01). Forty-six (37%) of the 125 hypervascular HCCs had a mosaic hypervascular pattern on sonographic angiography, and the remaining 79 had a homogeneous hypervascular pattern. Tumor vessel, a neovascularization within the nodule that enhances in a linear pattern with CO2 injection, was observed within the nodule in 90 (72%) of 1 25 hypervascular HCCs; the vascular direction was from the periphery to the center. An arterial tumor vessel was observed corresponding to the septum [5] within the nodule in the early arterial phase in 28 of 1 25 hypervascular HCCs. Eleven of 1 25 hypervascular HCCs showed hypervascularity only in the early and/or
TABLE
OF
2: Relation
Between
Pattern
Arterial
Vascular
Hepatocellular Carcinoma
Pattern
Adenomatous Hyperplastic
Nodule
on Sonographic Regenerative
Nodule
Adenomatous
Hyperplasia
All nine nodules of adenomatous hyperplasia were hypovascular on sonographic angiography (Table 2, Fig. 4). Sensitivity and specificity of this finding were 1 00% (9/9) and 91 % (193/213), respectively. Positive and negative predictive values of this finding were 31 % (9/29) and 1 00% (193/193), respectively. None of the lesions were identifiable on conventional angiography.
Regenerative
Nodule
Four regenerative nodules were hypovascular on sonographic angiography and three were isovascular. No lesion was hypervascular on conventional angiography. When the hypovascular pattern is defined as a typical vascular pattern for regenerative nodule, the sensitivity and specificity of this finding were 57% (4/7) and 88% (190/215), respectively. Positive and negative predictive values of the hypovascular pattern for the diagnosis of regenerative nodule were 1 6% (4/25) and 98% (1 90/1 93), respectively.
Angiography
Hemangioma
and Histologic
Metastasis
Diagnosis
Lymphoma
Granuloma
Hyperplasia Focal Nodular
Hypervascular Peripheral arterial supply (mosaic or homoge-
neous) Spotty pooling Peripheral hypervascularity Central arterial supply
0
0
0
0
0
0
21
0
0
0
0
21
1 25
7
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
Isovascular
9
0
3
0
0
0
0
0
Hypovascular
5
9
4
0
5
5
1
0
1 39
9
7
21
33
7
1
5
Total
KUDO
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
68
A
ET AL.
AJR:158,
1992
D
C
B
January
Fig. 2.-Hepatocellular carcinoma. A, Conventional transverse sonogram
hypoechoic (arrow). B-D,
lesion at left lateral segment
Sonographic
shows
of liver
angiograms clearly show which runs adjacent at 2 (B) and 4 (C) sec. Tumor
feeding
artery (arrowheads),
to portal
vein branch
vessels arising from periphery and infiltrating to center of tumor (arrows) are also seen at 2 and 4 sec (early phase). Mosaic hypervascular pattern of tumor (arrow) is seen in middle phase (10 sec, 0). E, Arterial phase of common hepatic arteriogram shows hypervascular tumor in left lateral segment. F, Venous phase of angiography shows inho-
E
(mosaic)
stain.
was
characteristic
F
hypervascular
Hemangioma
specificity
Hemangioma had an extremely tern: central hypovascularity with in the
mogeneous
early
arterial
phase,
gradually
characteristic vascular patperipheral echogenic spots infiltrating
to the
center
over time, and spotty pooling in the very late phase, lasting up to 30-60 mm after injection (Fig. 5). The earliest time that this spotty pooling was observed was 2-3 mm after injection. Sensitivity and specificity of this finding were both 100%, whereas the detection of hypervascularity with infusion hepatic angiography and digital subtraction angiography were 88% (1 4/1 6) and 1 00% (5/5), respectively (Table 3). A cotton wool appearance was observed on conventional angiography in 90% (1 9/21); the specificity was 1 00%.
Metastasis Sonographic angiography of hepatic metastases showed a hypovascular pattern in five cases (1 S%), a homogeneous hypervascular pattern in five (1 5%), a mosaic or inhomogeneous hypervascular pattern in two (6%), and a peripheral hypervascular pattern in 21 (64%) (Fig. 6). The peripheral
pattern was
1 00%
(1 89/1 89) (Table
for metastasis; 2). Positive
and
the nega-
tive predictive values of this finding were 1 00% and 94%, respectively. In contrast, detection of hypervascularity in metastases was 63% (1 7/27) with infusion hepatic angiography and 67% (4/6) with digital subtraction angiography (Table 3). Sixteen nodules were hypervascular with a lucent center and five
nodules
pattern.
The
showed
a faint
detection
rate
graphic angiography was angiography (p < .01).
homogeneous
hypervascular
of
hypervascularity
superior
to that
with
with
sono-
conventional
Lymphoma Five lymphomas were hypovascular giography and two were homogeneously
on sonographic hypervascular.
an-
Granuloma The one granuloma in this series tern on sonographic angiography.
had a hypovascular
pat-
AJR:158,
January
SONOGRAPHIC
1992
ANGIOGRAPHY
OF
;-‘.m
HEPATIC
:
.#{149}..:‘c-
.
69
TUMORS
--: ,‘. .
L.
.-.
-
a
-
‘H’-’
,:
.‘
“fl-
.
k,‘-
L
;-
-‘
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
h%”: 5.
---4:-‘?
-*
I
;,
#{149}-_:‘
,r
t.._(.s.
.
-.
‘g5
-
D
C
B
..
Fig. 3.-Hepatocellular carcinoma. A, Conventional intercostal sonogram shows hypoechoic tumor (arrow) with peripheral hypoechoic layer (halo). B-D, Sonographic angiograms at 2 (B), 4 (C), and 30 (0) sec show mosaic hypervascular pattern (arrows). Halo is not enhanced with carbon dioxide microbubbles (arrowheads). E, Resected specimen confirms nodule is hepatocellular carcinoma. Mosaic structure is evident. Hemorrhagic necrotic area within nodule corresponds to unenhanced area in C.
TABLE
3: Detection
of Hypervascularity
with Sonographic
and Conventional
No. of Nod ules Detected Imaging
Method
Hepatocellular Carcinoma
Sonographic Conventional
angiography angiography
1 25/1 39 (90)
Hepatic infusion
61/96
(64)
Digital
32/43
(74)
subtraction
Total a
Focal
Nodular
Statistically
93/1 39 (67) significant (p
65
Sonography with Intraarterial Infusion of Carbon Dioxide Microbubbles (Sonographic Angiography): Value in Differential Diagnosis
of Hepatic
Tumors
: Masatoshi Kudo1 Shusuke Tomita1 Hitoshi Tochio2 Jun Mimura1 Yoshihiro Okabe1 Hiroshi Kashida1 Masahiro Hirasa1 Yasuyosi Ibuki1 Akio Todo1
Differential diagnosis of small liver tumors is important, but is not always possible, even with angiography. To solve this problem, we introduced sonographic angiography, which combines sonography and angiography. The vascular pattern of a variety of hepatic nodules was evaluated with sonographic angiography, and the results were compared with those of conventional angiography. Sonographic angiography (sonography performed during intraarterial infusion of carbon dioxide microbubbles) was performed in 184 patients with a total of 222 hepatic nodules: 139 hepatocellular carcinomas, nine adenomatous hyperplasias, seven regenerative nodules, 21 hemangiomas, 33 metastases, seven lymphomas, one granuloma, and five focal nodular hyperplasias. Sonographic angiography detected a hypervascular pattern with penpheral blood supply in cases of hepatocellular carcinoma (sensitivity, 90%; specificity, 89%). Typical vascular patterns of adenomatous hyperplasia, hemangioma, metastasis, and focal nodular hyperplasia on sonographic angiography were hypovascularity (sensitivity, 100%; specificity, 91%), spotty pooling (sensitivity, 100%; specificity, 100%), peripheral hypervascularity (sensitivity, 64%; specificity, 100%), and a central arterial supply (sensitivity, 100%; specificity, 100%), respectively. The detectability of hyper-
vascularity
was greater
with sonographic
angiography
than with conventional
angiog-
raphy in hepatocellular carcinoma, metastasis, and hemangioma. Our experience indicates that sonographic angiography depicts characteristic vascular features that reflect the vascular anatomy of specific types of hepatic tumors, and thus is useful in the differential diagnosis of these lesions. AJR
158:65-74,
The recent tion
January
advances
of small
1992
in several
asymptomatic
hepatic
imaging
techniques
tumor
nodules.
have led to improved Confirmation
of the
detec-
pathologic
type of such nodules is important but is not always possible when the nodules are extremely small or do not have characteristic findings on conventional angiograms or histopathologic examinations. Several attempts have been made to improve the sensitivity
Received May 16, 1 991 : accepted July31, 1991.
after revision
This work was supported in part by the 1989 Grant-in-Aid for Medical Research of Kobe City General Hospital. 1 Department of Medicine, Division of Gastroenterology, Kobe City General Hospital, 4-6, Minatojima-Nakamachi, Chuo-ku, Kobe 650, Japan. Address reprint requests to M. Kudo. 2 Section of Abdominal Ultrasound Service, Kobe City General
Hospital,
Kobe
650, Japan.
0361 -803X/92/1 581-0065 © American Roentgen Ray Society
of
angiography
in diagnosing
hepatic
tumors.
These
include
digital
subtraction angiography, infusion hepatic angiography [1 ], CT arteriography [2], and CT after intraarterial injection of iodized oil (Lipiodol CT) [3, 4]. Although those techniques improve sensitivity in detecting hypervascular nodules, they are still of limited value in the diagnosis of hypovascular lesions. Moreover, in hypovascular tumors, they do not provide diagnostic information about the specific histology of the lesions. In order to solve this problem, we studied the usefulness of sonographic angiography
(sonography
bubbles) in detecting hepatic tumors.
Materials From patients
with
tumor
intraarterial
vascularity
infusion
of carbon
dioxide
as an aid in the differential
[CO2]
micro-
diagnosis
of
and Methods March with
1989 222
hepatic
to
January tumors:
1991
,
sonographic
1 39 hepatocellular
angiography carcinomas
was (HCC5),
performed nine
in
adenomatous
184
KUDO
66
AJA:158,
(FNH5). Of 1 1 8 patients
selected for the conventional angiographic examination because of the need for a differential diagnosis as well as for intraarterial therapy. The 222 hepatic nodules had been detected with conventional so-
The total period of the procedure was classified into three phases: early, middle, and late, depending on the degree of enhancement of the liver parenchyma with CO2 microbubbles. The period until the surrounding liver parenchyma was filled with CO2 microbubbles was defined as the early phase, which lasted approximately 5-1 0 sec.
nography,
The period the CO2 microbubbles
cirrhosis
and
in this
11 had
chronic
and these
study.
proved
133 nodules).
confirmed
with
sion
CT (SPECT),
peared
hyperechoic,
listed
were
nodules
seven
originally
were
included
hemangiomas
were
of the seven
‘Tc-IabeIed
hemangiomas
RBC
and/or
clinical
was
single-photon
follow-up
emis-
of more
than
tumors,
132 appeared hypoechoic, 28 apthe remaining 62 had a mosaic pattern on The size ranges of the 222 tumors are
and
sonography.
in Table
patients
82 nodules; biopsy under sonographic
angiography,
Of 222 hepatic
visible except
Diagnosis
delayed
1 year.
unenhanced
nodules
(surgery,
guidance,
MA,
All 184
sonographically
All hepatic
histologically
hepatitis.
107 had liver
time
needed
preparing
Two
proper, right, or left hepatic artery) after conventional superselective hepatic angiography, including infusion hepatic angiography [1 J. Superselective infusion hepatic angiography was performed in 138
cases
a total
with
of
a total
155
Berlin)
(Toshiba
angiography
nodules
after
at a rate
of 2-4
of 40-60
injection mI/sec.
Digital
DFP 40A unit) was performed
of 67 nodules
after
injection
of 1 0-20
ml of
in 46 patients
ml of iopamidol
at
were prepared by vigorously normal saline, and 5 ml of the patient’s own blood. The patient’s own blood was added to increase the surface tension of the bubbles so that microbubbles by hand
could
be easily
1 0 ml of CO2, 10 ml of heparinized
obtained. 1 -2
of approximately
The
mI/sec,
microbubbles
were
injected
permitting
the accurate
at a rate
evaluation
of
tional
when
ancy
arose
preters of the
a long
period,
which
permits
a precise
observation
their
of the
breath
and
the nodule
videotape until
In all cases,
recorder
the
bubbles
preparation
of
dynamic
images
from before the injection were
washed
out
CO2 microbubbles
from
the
was
quite
were
recorded
on a
parenchyma.
easy
and
out
knowledge
1: Histologic
Varieties
of Liver
Nodules
Hepatocellular
Adenomatous
carcinoma
hyperpla-
3 cm
Total
38
36
139
2
0
0
7
0
7
0
0
7
0 0 0
6 8 0
15 21 5
0 4 2
21 33 7
0
1
0
0
1
1
3
1
0
5
spotty
(4) central
pooling
arterial
supply
9
sia
Regenerative
Most
interpreters.
angiograms
ing on the tumor
:
Type of Lesion
liver
knowledge
of vascularity;
was isovascular
angiography
peripheral
TABLE
10 than
phases.
sonographic
The
needed
the vascular
without
the case and came to a consensus.
sonographic
late
of CO2 microbubbles liver
other
in the surrounding
including
results.
diagnosis
the
=
Tokyo).
extra
than
Vasculanty on both sonographic and conventional angiography was quantified by comparing tumor vascularity with the vascularity of the surrounding liver parenchyma. Quantification of the vascularity on sonographic and conventional angiography was done blindly with-
sion
vascular
total less
technique
blindly
results,
biopsy
pattern of the small nodule in the early arterial phase. A convex and/ or linear-array electronic scanner with a 3.5-MHz (n = 1 74) or 5.0MHz (n 48) transducer was used for the studies (SSD 650 model, Aloka,
analyzed
imaging
or
reevaluated
raphy
to hold
remained
the videotape
on the
between
graphic
them
other
agreed
neous
to enable
was
required.
interpreters
angiography;
interpreters
The total volume injected was determined by monitoring sonographic images so that strong attenuation of the deeper liver parenchyma was not produced. The total volume injected ranged from 5 to 20 ml in each study. Oxygen was given to the patients during the sonoprocedure
The
usually
No special
was
by replaying
history;
parenchyma:
for
microbubbles
independent
the CO2 flow within the liver and nodule from the early arterial phase to the late phase. The total injection time ranged from 5 to 10 sec.
angiographic
of hemangioma.
sec (early phase) and thereafter
subtraction
a rate of 2-4 mI/sec. CO2 microbubbles mixing
CO2
angiography.
angiography
parenchyma was defined as the middle phase, which lasted approximately 1 0-60 sec after injection of CO2. The late phase was defined as the period when CO2 was almost washed out from the liver parenchyma, which usually was 1-3 mm after injection of CO2. The duration of each phase varied from case to case, depending on the placement of the catheter, total volume of CO2 injected, hepatic blood-flow velocity, and severity of liver disease.
patient’s
iopamidol(Schering,
for conventional
sonographic
in cases
of each nodule
1.
with
for
mm except
Sonographic angiography was performed by injecting microbubbIos of CO2 through a 6.5-French preshaped hepatic catheter (Clinical Supply Co., Gifu, Japan) placed into the hepatic artery (either the
patients
used
1992
other
with a total of 139 HCCs,
solution
January
seven regenerative nodules, 21 hemangiomas, 33 metastases (colorectal cancer 1 9; stomach cancer, five; pancreatic cancer, four; lung cancer, three; and breast cancer, two), seven lymphomas, one granuloma, and five focal nodular hyperplasias hyperplasias,
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
ET AL.
II
isovascular
a
hypovascular
pattern
Fig. 1.-vascular
pattern
patterns
seen
on sonographic
angiography
of hepatic
masses after intraarterial infusion of carbon dioxide (C02) microbubbles. Vascularity was evaluated at three phases: early, middle, and late.
AJR:158,
(n
=
and
January
55). The sensitivity,
Side
vascular
effects
symptoms,
of each
and
or complications
by whether
physical
angiography.
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
pattern
specificity,
bles were defined
were
SONOGRAPHIC
1992
Blood
monitored
pressure, the
tumor
predictive
values
caused
by injection
was
were
or
laboratory
pulse
rate,
study
tests respiration,
specified,
determined.
of CO2 microbub-
or not there was a change
findings,
during
hepatic
ANGIOGRAPHY
after and
in subjective sonographic temperature
period.
The chi-square test was used for statistical analysis. less than .01 was considered statistically significant.
A p value
of
Results Table 2 and Figure 1 summarize the vascular patterns of the 222 hepatic tumors in the 1 84 patients. No complications or abnormal changes in laboratory tests, including liver function tests, associated with intraarterial injection of CO2 were observed in any of the 1 84 patients in whom sonographic angiography was performed.
Hepatocellular
HEPATIC
TUMORS
67
middle phases and did not enhance in the late (capillary) phase. In 24 of 35 hypervascular HCCs in which tumor vessels were not detected, the vascular direction was from the periphery to the center. The remaining 1 1 hypervascular HCCs showed simultaneous filling within the small nodule. A peripheral hypoechoic layer (halo), which was observed in 91 of 125 hypervascular HCCs, was not enhanced with CO2 microbubbles (Fig. 3). Pathologic examination revealed that eight of the nine isovascular HCCs and all five hypovascular HCCs were well differentiated (grade I) HCCs on the EdmondsonSteiner scale [6], whereas only eight of 1 25 hypervascular HCCs were grade I. The typical vascular pattern of HCCs was a peripheral or simultaneous arterial supply and a homogeneous or mosaic hypervascular pattern (Figs. 2 and 3). Sensitivity and specificity of this finding were 90% (1 25/1 39) and 89% (74/83), respectively. Positive and negative predictive values of this finding were 93% (1 25/1 34) and 84% (74/88), respectively.
Carcinoma
HCC had a hypervascular pattern (Figs. 2 and 3) in 90% (1 25/1 39), an isovascular pattern in 6% (9/1 39), and a hypovascular pattern in 4% (5/1 39) on sonographic angiography (Table 2). In contrast, 64% (61/96) of HCCs were hypervascular on infusion hepatic angiography and 74% (32/43) were hypervascular on digital subtraction angiography (Table 3). Forty-six nodules showed no abnormality on infusion hepatic angiography or digital subtraction angiography. Sensitivity of sonographic angiography for the detection of hypervascularity was superior to that of conventional angiography (p < .01). Forty-six (37%) of the 125 hypervascular HCCs had a mosaic hypervascular pattern on sonographic angiography, and the remaining 79 had a homogeneous hypervascular pattern. Tumor vessel, a neovascularization within the nodule that enhances in a linear pattern with CO2 injection, was observed within the nodule in 90 (72%) of 1 25 hypervascular HCCs; the vascular direction was from the periphery to the center. An arterial tumor vessel was observed corresponding to the septum [5] within the nodule in the early arterial phase in 28 of 1 25 hypervascular HCCs. Eleven of 1 25 hypervascular HCCs showed hypervascularity only in the early and/or
TABLE
OF
2: Relation
Between
Pattern
Arterial
Vascular
Hepatocellular Carcinoma
Pattern
Adenomatous Hyperplastic
Nodule
on Sonographic Regenerative
Nodule
Adenomatous
Hyperplasia
All nine nodules of adenomatous hyperplasia were hypovascular on sonographic angiography (Table 2, Fig. 4). Sensitivity and specificity of this finding were 1 00% (9/9) and 91 % (193/213), respectively. Positive and negative predictive values of this finding were 31 % (9/29) and 1 00% (193/193), respectively. None of the lesions were identifiable on conventional angiography.
Regenerative
Nodule
Four regenerative nodules were hypovascular on sonographic angiography and three were isovascular. No lesion was hypervascular on conventional angiography. When the hypovascular pattern is defined as a typical vascular pattern for regenerative nodule, the sensitivity and specificity of this finding were 57% (4/7) and 88% (190/215), respectively. Positive and negative predictive values of the hypovascular pattern for the diagnosis of regenerative nodule were 1 6% (4/25) and 98% (1 90/1 93), respectively.
Angiography
Hemangioma
and Histologic
Metastasis
Diagnosis
Lymphoma
Granuloma
Hyperplasia Focal Nodular
Hypervascular Peripheral arterial supply (mosaic or homoge-
neous) Spotty pooling Peripheral hypervascularity Central arterial supply
0
0
0
0
0
0
21
0
0
0
0
21
1 25
7
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
Isovascular
9
0
3
0
0
0
0
0
Hypovascular
5
9
4
0
5
5
1
0
1 39
9
7
21
33
7
1
5
Total
KUDO
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
68
A
ET AL.
AJR:158,
1992
D
C
B
January
Fig. 2.-Hepatocellular carcinoma. A, Conventional transverse sonogram
hypoechoic (arrow). B-D,
lesion at left lateral segment
Sonographic
shows
of liver
angiograms clearly show which runs adjacent at 2 (B) and 4 (C) sec. Tumor
feeding
artery (arrowheads),
to portal
vein branch
vessels arising from periphery and infiltrating to center of tumor (arrows) are also seen at 2 and 4 sec (early phase). Mosaic hypervascular pattern of tumor (arrow) is seen in middle phase (10 sec, 0). E, Arterial phase of common hepatic arteriogram shows hypervascular tumor in left lateral segment. F, Venous phase of angiography shows inho-
E
(mosaic)
stain.
was
characteristic
F
hypervascular
Hemangioma
specificity
Hemangioma had an extremely tern: central hypovascularity with in the
mogeneous
early
arterial
phase,
gradually
characteristic vascular patperipheral echogenic spots infiltrating
to the
center
over time, and spotty pooling in the very late phase, lasting up to 30-60 mm after injection (Fig. 5). The earliest time that this spotty pooling was observed was 2-3 mm after injection. Sensitivity and specificity of this finding were both 100%, whereas the detection of hypervascularity with infusion hepatic angiography and digital subtraction angiography were 88% (1 4/1 6) and 1 00% (5/5), respectively (Table 3). A cotton wool appearance was observed on conventional angiography in 90% (1 9/21); the specificity was 1 00%.
Metastasis Sonographic angiography of hepatic metastases showed a hypovascular pattern in five cases (1 S%), a homogeneous hypervascular pattern in five (1 5%), a mosaic or inhomogeneous hypervascular pattern in two (6%), and a peripheral hypervascular pattern in 21 (64%) (Fig. 6). The peripheral
pattern was
1 00%
(1 89/1 89) (Table
for metastasis; 2). Positive
and
the nega-
tive predictive values of this finding were 1 00% and 94%, respectively. In contrast, detection of hypervascularity in metastases was 63% (1 7/27) with infusion hepatic angiography and 67% (4/6) with digital subtraction angiography (Table 3). Sixteen nodules were hypervascular with a lucent center and five
nodules
pattern.
The
showed
a faint
detection
rate
graphic angiography was angiography (p < .01).
homogeneous
hypervascular
of
hypervascularity
superior
to that
with
with
sono-
conventional
Lymphoma Five lymphomas were hypovascular giography and two were homogeneously
on sonographic hypervascular.
an-
Granuloma The one granuloma in this series tern on sonographic angiography.
had a hypovascular
pat-
AJR:158,
January
SONOGRAPHIC
1992
ANGIOGRAPHY
OF
;-‘.m
HEPATIC
:
.#{149}..:‘c-
.
69
TUMORS
--: ,‘. .
L.
.-.
-
a
-
‘H’-’
,:
.‘
“fl-
.
k,‘-
L
;-
-‘
Downloaded from www.ajronline.org by 94.139.167.167 on 10/06/15 from IP address 94.139.167.167. Copyright ARRS. For personal use only; all rights reserved
h%”: 5.
---4:-‘?
-*
I
;,
#{149}-_:‘
,r
t.._(.s.
.
-.
‘g5
-
D
C
B
..
Fig. 3.-Hepatocellular carcinoma. A, Conventional intercostal sonogram shows hypoechoic tumor (arrow) with peripheral hypoechoic layer (halo). B-D, Sonographic angiograms at 2 (B), 4 (C), and 30 (0) sec show mosaic hypervascular pattern (arrows). Halo is not enhanced with carbon dioxide microbubbles (arrowheads). E, Resected specimen confirms nodule is hepatocellular carcinoma. Mosaic structure is evident. Hemorrhagic necrotic area within nodule corresponds to unenhanced area in C.
TABLE
3: Detection
of Hypervascularity
with Sonographic
and Conventional
No. of Nod ules Detected Imaging
Method
Hepatocellular Carcinoma
Sonographic Conventional
angiography angiography
1 25/1 39 (90)
Hepatic infusion
61/96
(64)
Digital
32/43
(74)
subtraction
Total a
Focal
Nodular
Statistically
93/1 39 (67) significant (p
