Matthijs Swannie
Oudkerk, L. Tjiam,
Liver
MD, MD
PhD Paul
#{149}
Bart van Ooijen, MD, PhD #{149}Sybrand P. M. Mali, MD I. M. Schmitz, MD, PhD #{149} Theo Wiggers, MD, PhD
#{149}
Metastases
Detection
with
A diagnostic
from Continuous
approach to assess liver from colorectal carcinoma was prospectively evaluated in 30 patients with and without metastases on the basis of findings at conventional computed tomography (CT). With the technique, termed continuous CT angiography (CCTA), CT data were continuously sampled for 24 seconds at the same section level after initiation of a 3-second injection of 10-20 mL of contrast medium in the common hepatic artery. The procedure was repeated for each contiguous section level of the liver. Findings at preoperative ultrasound (US), conventional CT, and CCTA were compared with those at intraoperative US and surgical exploration as the standard of reference. Forty-four liver metastases were identified in 16 patients, and 14 patients had no metastases. CCTA had a sensitivity of 98% (43 lesions identified) and higher accuracy (81% 154 of 67 diagnoses]) than US and conventional CT. The data indicate that CCTA can supplement information obtained with conventional imaging techniques in patients who must undergo hepatic surgery because of metastases from colorectal carcinoma. metastases
Index
terms:
tomography (CT), tomography (CT), technology #{149} Liver neoplasms, angiography, 761.124 #{149} Liver neoplasms, CT, 761.1211, 761.332 #{149} Liver neoplasms, secondary, 761.332 comparative
Radiology
Computed
studies
1992;
#{149} Computed
185:157-161
I From the Departments of Diagnostic Radiology (MO., S.P.M.M., S.L.T.), Surgical Oncology (B.v.O., T.W.), and Statistics (P.I.M.S.), Dr. Daniel den Hoed Cancer Center, Groene Hilledijk 301, 3075 EA Rotterdam, The Netherlands. Received November 12, 1991; revision requested December 26; final revision received May 11, 1992; accepted May 14. Address reprint requests to MO. C RSNA, 1992
T
Colorectal
Carcinoma:
CT Angiography’
HE presence
or absence of liver metastases is a major determinant of prognosis in patients with colorectal carcinoma (1). Surgical resection of liver tumors has become a widely accepted and utilized therapy in selected patients (2-4). Five- and 10-year survival data have shown a plateau after resection, and long-term survivors seem to have the same risk of death as their age-matched populalion without cancer (2). Over the past decade, the standard modalities for preoperative evaluation of hepatic disease have been computed tomography (CT) and ultrasonography (US) (5,6). Sensitivity of only 38% (14 of 37) to 68% (65 of 95), however, has been reported for diagnosis of hepalic metastatic disease with these modalities
(7-10). Because of the low efficacy of US and CT in predicting liver disease and/or resectability, other preoperalive imaging techniques have been investigated, including angiography, CT angiography, and magnetic resonance (MR) imaging (10-12). CT angiography may be more sensitive for tumor detection than all other methods (10). Intraoperative US is advocated in many institutions as an adjunct to hepalic resection. In one study, intraoperative US resulted in the detection of 25% (40 of 167) and 35% (59 of 150) more lesions compared with preoperalive US and CT, respectively (13). In another study, intraoperative US a!lowed identification of 23% (six of 26) of metastases in cases in which palpalion, preoperative US, or CT did not show any lesions at all (14). Intraoperalive US in combination with surgical palpation is at present considered to be the best method for staging of liver disease (13-15). To our knowledge, CT angiography and intraoperalive US have not been compared in any type of study. In the present study, we prospectively assessed preoperalive US, conventional contrast-me-
dium-enhanced CT, and CT angiography performed with a continuously rotating CT scanner (CCTA) in palients with or without liver metastases from colorectal carcinoma on the basis of findings at conventional CT. Imaging modalities were used independently by different investigators, and all findings were compared with findings at surgical palpation and intraoperalive US as the standard of reference. Correlation with resected specimens was attempted whenever possible. PATIENTS
AND
METHODS
Thirty consecutive patients with a current or previous carcinoma of the colon or rectum (with or without liver involvement) and selected for laparotomy on the
basis of conventional subject of the study.
CT scans Thirty-one
were the patients
had given informed consent over a 1-year period in the Dr. Daniel den Hoed Cancer Center, but one patient was withdrawn because peritoneal carcinomatosis made it impossible to perform intraoperative US of the liver. There were 16 women and 14 men, with a mean age of 58 years (range, 40-73 years). Seventeen patients were treated for primary colon cancer or local recurrence of a previously treated colorectal carcinoma, and 13 patients underwent surgery for isolated liver metastases, 12 for elective resection and one for selective hepatic artery infusion. All patients underwent preoperative evaluation with several imaging modalities. The
routine
included
performance
of
3.5- and 5-MHz US (128 XP; Acuson, Mountain View, Calif) and conventional and contrast-medium-enhanced CT (Somatom Plus; Siemens, Erlangen, Germany) with 5-mm contiguous sections, 1-second scanning time, and interscanning delay of 5 seconds. Contrast-enhanced CT, with administration of nomonic jopromide (Ultravist 300 [300 mg of iodine per miffi-
Abbreviation: raphy.
CCTA
=
continuous
CT angiog-
157
Table
1 Distribution Segment
of Meta
by Liver
stases,
No.
Segment* I
Figure 1. (a) Conventional contrast-enhanced old woman. (a) No liver lesions are evident scan at the same level shows a hypovascular rectly interpreted as a metastasis.
CT and (b) CCTA scans obtained in a 64-year on the conventional enhanced CT scan. (b) CCTA lesion (arrow) in segment VI, which was cor-
0
II
1
HI IV
3 8
II-Iv
2
V VI VII VIII
3 5 7 6
v_vmt
9t
Total
segment
onds; aging
phase with
2-50 mL all modalities
independently,
with
knowledge
than
in
100 was
a history
of colorectal
either
solid
all solid
lesions
Im-
anamnestic
cinoma, and findings with were described independently. anechoic
in
seconds. performed
no other
and
car-
range,
4-10
same
moment
mL/sec)
data
were
sampled
level
over
24 seconds
hold. The total did not exceed with
echo-
mL
both
started
data
at the
same
of jopromide
CT breath
dose
medium
and
(range,
ing to the segmentation of the liver as described by Couinaud (16). All lesions were coded and classified as certainly be-
to the liver
total contrast
CCTA
dose was
190-300
245
mL).
Time between the onset of each set of seclions was less than 3 minutes. An average of 13 sections (range, 10-18 sections) was
ating
obtained.
lesions,
and
lesions
tastases
at CT.
fatty
livers
der,
compatible
enhancing
were
and
classified
nonen-
as me-
Hyperattenuating
and
lesions
adjacent with
in
to the gallbladfocal
Section
nonsteatosis,
were not included. On the following day, angiography was performed and a selective catheter (5-F celiac, Simmons 1-3, or special 5-F; Cordis Europa, Roden, The Netherlands) was placed in the common hepatic artery. In two patients with separate left and right hepatic arteries (insertion from the superior mesenteric artery), the right hepatic artery was chosen. For the comparison with the standard of reference, those parts
of the
sessed
with
liver
that
had
not
all diagnostic
been
as-
techniques
were
excluded. patients
were rotating
before
angiography,
examined
with
CT
system
the
tiguous sections. was chosen for nods (corresponding lion
ond),
fact
represent
solid
lesions
artery These
raphy.
flow rate were amination neous)
injector
nati)
suspended
into
the
were established at angiogsame values for dose and
adjusted
to obtain filling of the
system
backflow
for the CCTA
optimal hepatic
(Liebel-Flarsheim,
from
ex-
(ie, homogevessels. A CT
Cincin-
the ceiling
and filled
with 70 mL of jopromide (300 mg of iodine per milliliter) diluted with 180 mL of norma! saline solution to a solution of 84 mg of iodine per milliliter was connected with the selective catheter after the patient was positioned in the gantry. For each CT sec-
tion, contrast seconds
158
(volume
#{149} Radiology
medium range,
injection 10-20
during mL;
flow
2-3
in
16 images
phases.
images
in
angiography
involves images
as opposed
vessels,
was
(hyper-
per
at
circular
the
capillary
over
peripheral
was
of other
classified reported
All
arterial
data
other
rectal carcinoma. laparotomy was
knowl-
results
than
a history
On the following performed without
of CCTA findings. means of a 5-MHz
by
or a sharp
without
areas (610; Aloka, by a radiologist
or anof cob-
day,
was who
no knowledge of previous imaging The liver surface was inspected
sonographically
by
means
of a water-filled
balloon. The intraoperative US findings were correlated with findings at surgical palpalion and visual inspection of surface besions.
Hepatic
lesions
were
located
of surgical
palpation
by and
vi-
sual inspection of the liver surface combined with intraoperative US. Without intraoperalive US, 38 lesions were identified. The additional six
accord-
detected
with
intraoper-
In 10 patients (14 !eintraoperative US data were
100% that the lesions were metastatic. Most of the other lesions were so superficially located that the macroscopic appearance and one or two positive
biopsy
garded as Of the six intraoperative proved as three, two same
proved
Intraoperatransducer
Tokyo) (S.L.T.)
as malig-
sions), correlated with resected specimens (5-mm slices), and concordance was
as a perfusion
examination
coded
in 16 patients
section phase,
pattern
means
found
were
circular
in the
lesions
were
US only.
time.
demarcated
nant
alive
or hypoattenuating)
with
abnormality. Results were
for small performed
ma-
lesions
were classified as metastases. Proirregular enhancement without a
knowledge live US
uncertainly
to the delinea-
seen
enhancement
demarcation
had data.
of differ-
These
level,
lesions
amnestic
benign,
Forty-four
1.5-second pe540#{176} reconstruc-
with
veins
con-
reconstruction
contiguous
section
hepatic
uncertainly
lignant, and certainly malignant. All diagnostic data were collected and supervised by a trial management office to guarantee independent reporting of each procedure.
exami37-54
with
CT angiography of only one or two
same
and
edge
without
Image
lion of arterial
mum
rate
10 mm
16-24 [maximum, 110] images obtained with CCTA). Detailed
clear
flow
CCTA (range,
transversal
(conventional obtainment the
was
resulting
angiographic
with the catheter in the hepatic artery. The maximum contrast medium dose that could be washed out from the liver parenchyma in 24 seconds and the maxi-
hepatic
data),
ent
phase longed
sec-
average total 50 minutes
nign,
RESULTS
peripheral
a continu-
(360#{176} per
was
thickness
and/or
Immediately ously
The
nation time minutes).
in
section
in a single
angiography
located
IV.
at the
collection.
radiation 280 mGy.
The average
modality At US, all
was
as CT
genic but not sharply demarcated lesions were classified as malignant. All solid lesions, hyperor hypoattenuhancing
44
* Couinaud classification. t Some lesions were also partially
liter]; Schering, Berlin) was performed two phases: phase 1-50 mL in 25 sec-
of
Lesions
tients
specimens
conclusive lesions
re-
detected
only
with
US, three were biopsy metastases; of the other in one patient were of the
appearance
as other
metastases. with
were
for metastalis.
no
There
metastases
biopsy-
were
14 pa-
detected,
nine patients with only one detectable mass, two with two metastases, two with three metastases, and three with four metastases. The mean diameter of solitary liver lesions
(in
In case
of multiple
nine
patients)
liver
was
6.2 cm.
metastases
(seven patients), the mean diameter was 2.3 cm. Overall, 19 lesions were less than 1 cm in diameter and 25
October
1992
S
4.r#{149}’
:
a.
b.
Figure 3. (a) Conventional contrast-enhanced CT and (b) CCTA scans obtained in a 55-yearold woman. A cyst (curved arrow) and a metastasis (open straight arrow) are evident on both scans. CCTA at the same section also revealed a hypovascubar solid lesion (solid arrow in b)
a.
with a hypervascular center. This was correctly interpreted flow of contrast medium in the splenic artery on the CCTA end of the series.
a 5’? b.
in diameter With
CCTA, with
were
detected,
F
20s e.
Figure
2.
(b-e) man.
(a) Contrast-enhanced
shows
no
lesions
at this
level.
CT
Lesions
were also not apparent at adjacent levels. No lesion is visible at (b) 16 seconds of CCTA, but after (c) 18 seconds, a small hypervascubar ring (arrow) of less than 1 cm in diameter became sharply visible. (d, e) The ring (arrow in d) gradually
The lesion
disappears
was correctly
after
2 seconds.
interpreted
as a me-
tastasis.
more
Thirty
than
lesions
1 cm in diameter. were
ated in the right V-VIII), whereas cated
in the
II-IV). caudate bution
left
is given
exclusively
liver lobe 14 lesions liver
lobe
situ-
(segments were !o(segments
in Table
enabled identification of 98% of all lesions and was the sensitive diagnostic modality
of 44)
most (P
the
< .001
two
vs CT
patients
right
artery
the
unperfused
was
or US)
(Figs
1-3).
In
in whom
only
perfused
at CCTA,
lateral
segments
the of
the left lobe were found to be free of metastases (Fig 4). The one false-negative lesion at CCTA was less than 1 cm
Volume
185
Number
#{149}
(18 of 19) of the of 1 cm
as were
100%
patients (n
with
or less
of the three
or
1
in one
(Fig
4), and
a focal
nod-
ties in very inhomogeneously attenuating liver parenchyma; this patient also had seven lesions that were correctly identified. The other three false-positive findings, in one patient,
probably
caused liver
(Fig
of these was interpreted as a satellite lesion but actually belonged the main process, and another with no correlation at surgery was retro-
to
spectively abnormality.
classified
Sensitivity, positive
values
are
as a perfusion
specificity, and negative given
CCTA had the methods (81%;
in Table
accuracy, predictive 4. Over
highest accuracy 54 of 67 diagnoses)
all,
12
HU) over the series of sections obtamed during repeated intraarterial injections. The mean enhancement of liver attenuation in the arterial phase immediately
after
contrast
medium
Therefore,
the
in liver
attenua-
Resection of hepatic lesions secondary to colorectal cancers has proved to be of value (2-4). So far, this is the only
treatment
reported
to be associ-
ated with cure. The most important prognostic factors determining surviva! among patients undergoing hepatic resection are the number of metastatic surgical
deposits margins
in the liver and of the resected
the speci-
men (3,4). Precise evaluation of the number and site of lesions (either unior bibobar) before surgery is therefore essential
to accurate
gery and operations.
avoidance of unnecessary US, CT, hepalic angiogra-
and
trast does
MR
imaging
planning
have
purpose (5-12,17), CT with a high
medium, not show
of sur-
been
used
but even dose of con-
as the best method, high sensitivity in de-
tection of liver metastases from cobrectal cancers, in comparison with that achieved with intraoperative US and surgical resection (13). The discussion on injection vebocity, scanning delay, and quantity contrast medium administered
of all
(Table 4). The overall liver attenuation increased (mean, 5 HU; maximum,
35 HU. increase
DISCUSSION
for this dynamic
by partially
parenchyma
was overall
lion did not influence the detection and conspicuity of the enhancing liver lesions from the beginning of the series to the end.
phy,
4); one
and
1.
CCTA (43
95%
nonperfused
No lesions were found in the lobe (segment I). The distriof lesions according to liver
segment
injection
ret-
ular hyperplasia in another (Fig 4). The other seven false-positive lesions, found in three patients, could not be verified with transection because the liver lobe was not resected. In one patient, four false-positive findings were probably perfusion abnormali-
were were
seen
slight
a diameter
in five
chyma
CT and
CCTA scans obtained in a 70-year-old (a) Conventional contrast-enhanced
scan
barely
5).
more metastases = 18). These percentages were far better than with US and CT (Table 2). CCTA, however, resulted in more false-positive findings (Table 3). Pathologic examination of liver resection specimens revealed a malignant protrusion belonging to the main process in three false-posilive cases, normal hepalic paren-
1
d.
was
(Fig
lesions lesions
,;..
and
rospecively
as a metastasis. Notice the backscan, which was obtained at the
of is not
yet closed (18-20), and differences between patient populations have been reported (7). In the average Dutch patient, in our experience, the injection of 30 g of iodine is adequate for enhancement of the liver parenchyma
and
vasculature.
In most studies, preoperative imaging findings are not correlated with findings at laparotomy or intraoperaRadiology
#{149} 159
Figure 4. False-positive CCTA findings. (a) Liver segments II and III are not perfused. misinterpreted as a metastasis. (b) Segments II and III are not perfused. A malignant sion abnormality (arrow) ventral from the main malignant process was misinterpreted fused. These enhanced after partial withdrawal of the selective catheter in the hepatic
The nonperfused portion in the right lobe (arrow) was lesion is in right dorsal segments VI and VII. The perfuas a satellite lesion. (c) Segments II and III are not perartery (an additional series of CCTA scans of the lateral
segments sion was
was misinterpreted
was obtained). a focal nodular
In segment hyperplasia.
IV, a hypovascular
solid
nodular
lesion
(arrow)
as a metastasis.
Table 2 Sensitivity Techniques
Histologically,
the be-
of Preoperative Diagnostic According to Size and of Lesions
Number
Lesions