Acta Radiologica 43 (2002) 71–76 Copyright C Acta Radiologica 2002

Printed in Denmark ¡ All rights reserved

AC TA R A D I O L O G I C A
ISSN 0284-1851

ACCURACY OF CT IN LOCAL STAGING OF

GALLBLADDER CARCINOMA

B. S. K1, H. K. H1, I.-J. L1, J. H. K1, H. W. E1, I. Y. B1, A. Y. K1, T. K. K1,

M. H. K2, S. K. L2 and W. K3
Departments of 1Diagnostic Radiology, 2Internal Medicine and 3Biostatistics and Research, University of Ulsan, Asan Medical
Center, Seoul, Korea.
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10

Abstract
Purpose: To evaluate the accuracy of CT for staging gallbladder cancers, Key words: Gallbladder, neoplasms;
especially the T-factor of the TNM staging system. CT.
Material and Methods: CT investigations of 100 patients with surgically
proven gallbladder cancers were retrospectively analyzed. Dynamic helical CT Correspondence: Hyun Kwon Ha,
For personal use only.

was performed in 16 patients and conventional CT in the remaining 84. On Department of Diagnostic
CT, three radiologists attempted tumor staging for these patients; the majority Radiology, University of Ulsan, Asan
opinion was used for final decision. According to CT protocols (dynamic helical Medical Center, 388–1, Poongnap-
CT vs. conventional CT) and each tumor type (thickened wall/intraluminal dong, Songpa-ku, Seoul, 138–736,
mass/massive), the accuracy of CT staging was compared. The CT staging was Korea.
correlated with the surgico-pathologic results. FAX π82 2 476 4719.
Results: The overall accuracy of CT for staging gallbladder cancers was 71%;
it was 79% for T1 and T2 tumors, 46% for T3 tumors, and 73% for T4 tumors. Accepted for publication 9 October
For all three readers, the poorest accuracy was obtained in T3 tumors. No 2001.
statistically significant difference was noted in the accuracy between the groups
undergoing conventional CT and dynamic helical CT. A statistically significant
difference was noted in the accuracy for staging thickened wall and intraluminal
mass types of tumors (p⬍0.05); the highest accuracy was obtained in the intra-
luminal mass type (89%) and the massive type (83%), while it was 54% in the
thickened wall type.
Conclusion: The accuracy of tumor staging with CT in patients with gallblad-
der cancer depends on the morphological type of tumor. The poorest result is
obtained in the thickened wall type.

Early diagnosis of primary gallbladder cancer is dif-
ficult to make pre-operatively because the patients
are often asymptomatic or present with signs or
symptoms of chronic cholecystitis or cholelithiasis
(15). Therefore, most investigators have reported
that the prognosis for these patients is poor even
with surgical intervention (11). Delay in the diag-
nosis of this disease is the main reason for unsatis-
factory results following surgery (11). Recently, with
the development of the imaging and treatment mod-
alities, it is recognized that the prognosis for gall-
bladder cancer and selection of operative procedure
depend upon the depth of tumor invasion (10). Tu-
mors confined within the mucosa or muscularis
have a relatively favorable prognosis and may be
cured using simple cholecystectomy or extended
cholecystectomy. However, lesions spreading be-
yond the muscularis are associated with a poor out-
come, and in those a more aggressive surgical ap-
proach should be taken (10). For this reason, more
accurate assessment of the depth of gallbladder can-
cer invasion is important. However, there have been
limits in the literature regarding the accuracy of CT
in the staging of gallbladder cancers (12).
The purpose of this study was to evaluate the
accuracy of CT in staging gallbladder cancers.

71
 B. S. KIM ET AL.
Material and Methods
A computerized search was conducted at our insti-
tution from May 1996 to June 2000 to identify
cases of gallbladder cancer. Of the 197 patients
identified, 97 were excluded from the study for one
of the following reasons: radical surgery was not
performed (nΩ51); CT was not available (nΩ40);
or pathologic specimens were not available (nΩ6).
Therefore, a total of 100 patients who underwent
surgery were analyzed. They included 36 men and
64 women with ages ranging from 19 to 85 years
(mean 57 years).
CT was obtained using Somatom Plus-S (nΩ65)
(Siemens) or Somatom Plus-4 (nΩ35) (Siemens)
units with 8- or 10-mm slices at intervals of 8 or 10
mm from the diaphragm to the symphysis pubis.
Contrast material (600–900 ml; E-Z-CAT, 2% iod-
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10
inated, water soluble; E-Z-EM, Westbury, NY,
USA) was given orally to all patients 1 h before
examination. In all patients, 100–120 ml of iopam-
idol (Iopamiro 300; Bracco) or iopromide (Ul-
travist 300; Schering) were given intravenously as
a bolus (rate 2.5–3.0 ml/s). Examination was
started 70 s after the beginning of the injection in
For personal use only.
84 patients. In the remaining 16 patients, a dy-
namic study using helical CT was performed with
the same volume and injection rate of contrast ma-
terial; dual-phase helical CT was performed with
arterial and portal phases. Arterial phase images
were obtained at 30 s with 8- or 10-mm slices at
intervals of 8- or 10-mm from the diaphragm to
the third portion of the duodenum. Portal phase
images were obtained at 70 s with 8- or 10-mm
slices at intervals of 8- or 10-mm from the dia-
phragm to the symphysis pubis.
On CT, the primary features of gallbladder can-
cers were divided into three types: intraluminal
mass (polypoidal mass protruding into the lumen);
thickened gallbladder wall (infiltrating mass that
manifested as wall thickening without an obvious
mass); and massive type (mass almost filling the
gallbladder lumen) (6). According to the surgico-
pathological TNM classification (1), the tumors
were staged as follows: T1 (tumor invading the mu-
cosa or muscle layer); T2 (tumor invading the per-
imuscular connective tissue without extension be-
yond the serosa); T3 (tumor invading beyond the
serosa but less than 2 cm into the liver); and T4
(tumor extending more than 2 cm into the liver).
Because it was impossible to discriminate tumor
invasion within the gallbladder wall, we did no
attempt to separate T1 from T2 tumors. In ad-
dition, both N and M stagings were also not con-
sidered for tumor staging. Pericholecystic infil-
tration was considered to be present if the fat sur-
72
rounding the gallbladder was infiltrated by linear
strandings. Hepatic invasion was considered to be
present if the hepatic parenchyma near the gall-
bladder bed revealed a mass with altered attenu-
ation and contour deformity.
Without knowledge of the final surgico-path-
ological results, three radiologists independently
performed tumor staging on CT for the 84 patients
who underwent conventional CT. In the remaining
16 patients (who underwent dynamic helical CT),
tumor staging as well as contrast enhancement
pattern of the lesions (i.e., hyperattenuated, hy-
poattenuated, or isoattenuated compared with ad-
jacent hepatic parenchyma) were independently
interpreted by the same readers on each arterial
and portal phase in a blinded fashion. In addition,
the incidence of transient enhancement of hepatic
parenchyma adjacent to the gallbladder fossa was
analyzed on the arterial phase. We also compared
the accuracy of CT staging according to the CT
protocol (dynamic helical CT vs. conventional CT)
and types of gallbladder cancer. Lastly, CT staging
was correlated with surgico-pathological results.
When CT interpretations for staging and other
findings differed among the three readers, the ma-
jority opinion was applied as the final decision.
To compare the accuracy of staging of CT pro-
tocol and each tumor type, a statistical analysis
was performed using the logistic regression model
for tumor type and Fisher’s exact two-tailed test
for CT protocol. A p-value of less than 0.05 was
considered to indicate a statistically significant dif-
ference.
Results
Of 100 patients, 50 were of the thickened gallblad-
der wall type, 44 the intraluminal mass type, and
6 the massive type (Table 1). The surgico-patholog-
ical tumor staging for these 100 patients was as
follows: T1 (nΩ19) and T2 (nΩ48) tumors in 67;
T3 tumors in 22; and T4 tumors in 11. Of 67 pa-
tients with T1 or T2 tumors, there was the thicken-
Table 1
Surgical-pathologic tumor stages according to types of
gallbladder cancer
Surgical- Tumor types Patients, n
pathol. stage
Thickened Intraluminal Massive
wall mass
T1, T2 29 33 5 67
T3 15 7 0 22
T4 6 4 1 11
 LOCAL STAGING OF GALLBLADDER CARCINOMA
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10
Fig. 1. Gallbladder cancer (T1 tumor on surgico-pathologic
staging) in a 49-year-old woman. Contrast-enhanced CT shows
a soft tissue mass (») in the gallbladder lumen without any
evidence of disruption of the wall and pericholecystic infil-
tration. (Correctly staged by all three readers.)
For personal use only.
Fig. 2. Gallbladder cancer (T2 tumor on surgico-pathologic
staging) in a 49-year-old woman. Contrast-enhanced CT shows
diffuse, uneven thickening (») of the gallbladder wall with
focal disruption and pericholecystic infiltration ( ). Gallstone
in the gallbladder. (Two of three readers overstaged this case as
a T3 tumor.)
ed wall type in 29, the intraluminal mass type in
33, and the massive type in 5. Of 22 patients with
T3 tumor, the thickened wall and intraluminal
mass types were seen in 15 and 7, respectively. Of
11 patients with T4, the thickened wall type was
noted in 6, the intraluminal type in 4, and the
massive type in 1 (Table 1).
Table 2 shows the overall accuracy of tumor sta-
ging by the three readers, 70% for reader I, 67%
for reader II, and 64% for reader III. For all three
readers, the poorest result was obtained for T3 tu-
mors.
Table 3 shows the results of the surgico-patho-
logic and CT tumor staging. The overall accuracy
of CT for staging gallbladder cancers was 71% (71/
100). Sixty-seven patients with T1 or T2 tumors
were correctly staged on CT in 53 cases (79%) (Fig.
1) and overstaged in 14 cases (21%) (Fig. 2).
Twenty-two patients with T3 tumor were correctly
staged in 10 cases (46%) (Fig. 3), understaged in 8
(36%) (Fig. 4), and overstaged in 4 (18%). Eleven
Fig. 3. Gallbladder cancer (T3 tumor on surgico-pathologic
staging) in a 58-year-old woman. Contrast-enhanced CT shows
an irregularly shaped soft tissue mass (P) in the gallbladder
lumen with evidence of an irregular serosal surface and minimal
pericholecystic infiltration (»). (Correctly staged by all three
readers.)
Fig. 4. Gallbladder cancer with focal hepatic invasion (T3 tu-
mor on surgico-pathologic staging) in a 53-year-old woman.
Contrast-enhanced CT shows irregular thickening ( ) of the
gallbladder wall without any definite evidence of hepatic tumor
invasion. At surgical biopsy the focal lesion in the liver (»)
proved to be a hemangioma. (Two of three readers understaged
this case as stage T1 or T2.)
73
 B. S. KIM ET AL.

patients with T4 tumor were correctly staged in 8

cases (73%) and understaged in 3 (27%). Overall,
overstaging (nΩ18/100) was more common than
understaging (nΩ11/100).
In 16 patients who underwent dynamic helical
CT, the gallbladder cancers appeared to be isoat-
tenuated in 9, hyperattenuated in 6, and hypoat-
tenuated in 1 on the arterial CT phase. Transient
enhancement in the hepatic parenchyma near the
gallbladder was identified in 10 patients on the ar-
terial phase. On portal phase CT, the gallbladder
tumors appeared as hypoattenuated lesions in 12
patients, hyperattenuated in 2, and isoattenuated
in 2. Therefore, the lesion-to-liver contrast was in-

Table 2
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10

Tumor staging on CT, comparison of accuracy of the

three readers
Surgical- Reader
pathol.
stage I II III

n % n % n %

T1, T2 53/67 79 49/67 73 47/67 70

T3 9/22 41 10/22 45 9/22 41 Fig. 5. Gallbladder cancer (T2 tumor on surgico-pathologic
For personal use only.

T4 8/11 73 8/11 73 8/11 73 staging) in a 62-year-old woman. a) Arterial phase CT shows

Overall 70/100 70 67/100 67 64/100 64 isoattenuated, poorly marginated soft tissue mass ( ) in the
gallbladder lumen. Faintly enhanced hepatic parenchyma (»)
adjacent to the gallbladder fossa. Mild dilatation of the com-
mon bile duct ( ) caused by benign stricture of the distal com-
mon bile duct was proven with ampullary biopsy. (The arterial
Table 3 phase alone overstaged by all three readers as T3.) b) Portal
CT and surgical-pathologic stages, comparison in 100 phase CT showing the tumor mass ( ) in the gallbladder hy-
patients with gallbladder cancer poattenuated and more clearly demonstrated on this image
than in the arterial phase. Hyperattenuated hepatic paren-
Surgical- CT stage Pat., chyma seen on (a) becomes isoattenuated on (b). (The portal
pathol. n phase correctly staged by all three readers.)
stage T1, T2 T3 T4

n % n % n %

T1, T2 53 79 10 15 4 6 67
T3 8 36 10 46 4 18 22
creased on the portal phase. The tumors were cor-
T4 1 9 2 18 8 73 11 rectly staged in 9 patients on the arterial phase and
in 12 patients on the portal phase (Fig. 5). Com-
bined use of both arterial and portal phase CT
resulted in a correct staging in 12 patients; the ar-
Table 4 terial phase helical CT did not improve the results
CT accuracy for staging gallbladder cancer, comparison with CT of tumor staging. The results of the comparison of
protocol CT accuracy for staging gallbladder cancers ac-
Surgical- CT protocol cording to the CT protocol are shown in Table 4.
pathol. There was no statiscally significant difference in
stage Conventional CT, nΩ84 Dynamic CT, nΩ16 the overall accuracy between the groups who
U C O U C O underwent conventional CT and dynamic CT in-
T1 or T2 0 42 13 0 11 1
vestigations (p⬎0.05).
T3 3 10 6 2 0 1 When considering the relationship between the
T4 3 7 0 0 1 0 morphological tumor types and CT tumor staging,
Total, n and (%) 6 (7) 59 (70) 19 (23) 2 (13) 12 (75) 2 (13) there were some noticeable differences (Table 5).
Of the 50 patients with the thickened wall type, 27
UΩunderstaging, CΩcorrect, OΩoverstaging.
(54%) were correctly staged on CT. In contrast, 39

74
 LOCAL STAGING OF GALLBLADDER CARCINOMA
CT accuracy for staging gallbladder cancer, comparison to tumor types
Surgical-
pathol.
stage Thickened wall, nΩ50
U C O U
T1, T2 0 17 12
T3 6 7 2 2
T4 3 3 0 0
Total, n and (%) 9 (18) 27 (54) 14 (28)
UΩunderstaging, CΩcorrect, OΩoverstaging
(89%) of the 44 patients with the intraluminal mass
type were correctly staged. Of the 6 patients with
the massive type of tumor, 5 cases were correctely
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10
staged by CT (83%). There was a statistically sig-
nificant difference of the accuracy in CT staging
between the groups of the intraluminal mass and
the thickened wall type (p⬍0.05). However, there
was no statistically significant difference in the ac-
curacy between the groups of intraluminal mass
and massive type (p⬎0.05) and between the groups
For personal use only.
with the thickened wall and the massive type
(p⬎0.05).
Discussion
The need for accurate pre-operative staging of
gallbladder cancer cannot be overemphasized, not
only for the patient’s prognosis but also for se-
lecting the optimal surgical strategy. The median
survival for T1 and T2 tumors has been reported
to be significantly better than for T3 and T4
lesions (3). P et al. (14) have also shown
that the average survival time of patients with gall-
bladder cancer dropped from 21.8 months for
stage III to 3.5 months for stage IV. Therefore, it
is generally agreed that the single most important
factor for survival is the extent of tumor at the
time of diagnosis.
Despite of the merits of CT for characterizing
and defining tumor extent, there have been rela-
tively few reports evaluating the accuracy of CT in
the staging of gallbladder cancers (especially of the
T-factor of the TNM staging system) (9). In our
study, the overall accuracy of CT for staging gall-
bladder cancers was 71%; 79% for T1 and T2 tu-
mors, 46% for T3 tumors, and 73% for T4 tumors.
As CT overstaged in 18 of our 100 patients and
understaged in 11, overstaging appears to be a
more common problem. All three interpreters
showed similar overall accuracy rates for tumor
staging with the poorest results for T3 tumors.
Table 5
Tumor types
Intraluminal mass, nΩ44 Massive, nΩ6
C O U C O
0 32 1 0 4 1
3 2 0 0 0
4 0 0 1 0
2 (4) 39 (89) 3 (7) 0 5 (83) 1 (17)
This indicates that CT has a significant limitation
for determining the presence or absence of mini-
mal pericholecystic tumor infiltration. It may be
attributed to the fact that the CT interpreters com-
monly confused pericholecystic inflammatory in-
filtration or partial volume averaging-related mar-
ginal blurring of the gallbladder wall with tumor
infiltration. Furthermore, microinvasion of a
lesion toward the liver is commonly missed on CT.
In fact, there was a selection bias in our study as
we excluded patients who were inoperable due to
advanced tumor stages. However, although the in-
clusion of these patients might have improved our
results, use of conventional CT may not overcome
its inherent limitation.
In order to improve the accuracy of pre-operat-
ive staging of gallbladder cancer, various other in-
vestigative modalities have been utilized. One re-
port (4) showed that the overall accuracy of sono-
graphy was 38%; the accuracy for T3 tumors was
69% but was only 31% for T4 tumors. According
to that report, only 1 patient was overstaged while
the majority of patients were understaged. Thus,
H et al. (4) concluded that sonography
underestimated tumor status. O et al.
(13), who compared sonography and CT in the sta-
ging of gallbladder cancers, reported the sensitivity
of sonography for determining liver invasion to be
68%. They did not find CT to be superior to sono-
graphy, and sonography also appeared to have
considerable limitations for staging gallbladder
cancers.
To improve the accuracy of CT for tumor sta-
ging, other techniques can be used. Recently, the
dynamic CT study has been widely used for char-
acterizing abdominal lesions. However, our study
showed that application of dynamic CT did not
improve the diagnostic accuracy. In 9 of our 16
patients, the gallbladder cancers appeared as isoat-
tenuated lesions on the arterial phase. In contrast,
they were hypoattenuated on the portal phase in
75
 B. S. KIM ET AL.
12 patients. Therefore, the lesion-to-liver contrast
was increased in the portal phase, thereby improv-
ing the CT determination of tumor invasion into
the pericholecystic space and the liver. Moreover,
the common occurrence of transient hyperattenu-
ation in the hepatic parenchyma adjacent to the
gallbladder, which might result from coexisting
cholecystitis, caused some confusion in staging the
tumors. CT features of homogenous hyperattenu-
ation on the arterial phase and isoattenuation
compared with surrounded hepatic parenchyma on
the portal or late phases would help differentiating
transient hyperattenuation from tumor invasion
(7). Recent advent of multidetector-row CT per-
mits a more rapid acquisition of thinner collimated
images than is possible with conventional CT. In
addition, high quality three-dimensional images
can be obtained with volume data (5). In this re-
Acta Radiol Downloaded from informahealthcare.com by HINARI on 10/08/10
gard, use of narrow collimation of section thick-
ness and three-dimensional images are promising
for improving the diagnostic accuracy of tumor
staging, as this technique may not only improve
the detection of minimal pericholecystic tumor in-
filtration but also minimize the partial volume av-
eraging effect.
For personal use only.
Gallbladder cancers are usually classified into
three types: ‘‘massive’’; ‘‘thickened wall’’; and ‘‘in-
traluminal mass’’ (6). In our study, the type of gall-
bladder cancer affected the accuracy of CT for tu-
mor staging. The accuracy was much higher (89%
vs. 54%) in the intraluminal mass type than in the
thickened wall type. This is attributable to the fact
that in the intraluminal mass type of tumor the inci-
dence of pericholecystic tumor infiltration is low
and the extent of areas which should be observed is
more localized than in the thickened wall type. Ac-
cording to a series of A et al. (2), the intralumi-
nal mass type of tumor is less invasive than the other
types of tumor and rarely invades the serosa if not
grown to a considerable size. Therefore, intralumin-
al types of tumor seem to have a better prognosis.
Our results, demonstrating that 33 of the 44 intralu-
minal mass types belonged to T1 or T2 tumors, may
support this assumption (2).
The size of a polypoid mass is also reported to
be closely correlated with tumor spread and conse-
quently with the prognosis (9). A polypoid tumor
of more than 1 cm in diameter is more likely to be
malignant, whereas tumors less than 1 cm are more
often benign and are commonly cholesterol polyps
(8).
In conclusion, the accuracy of tumor staging in
76
patients with gallbladder cancer by using CT de-
pends on the morphological types of tumor with
the poorest accuracy in the thickened wall type.
ACKNOWLEDGEMENT
We thank Bonnie Hami (Department of Radiology, The Uni-
versity Hospitals Health System, Cleveland, OH, USA) for edi-
torial assistance in preparing this manuscript.
REFERENCES
1. A J C  C: Manual for sta-
ging of cancer, 3rd edn., p. 93. Edited by O. H. Beahrs et
al. Lippincott, Philadelphia 1988.
2. A T., H T., K M. et al.: Intraluminal pap-
illary carcinoma of the gallbladder. Prognostic value of
computed tomography and sonography. Gastrointest.
Radiol. 13 (1988), 261.
3. C P., G A. & C G.: Surgical
treatment of 724 carcinomas of the gallbladder. Results of
a French surgical association survey. Ann. Surg. 219
(1994), 275.
4. H S. P., K V. K., G R. B. et al.: Sta-
ging of carcinoma of the gallbladder. An ultrasonographic
evaluation. Hepatogastroenterology 44 (1997), 1240.
5. H H., H H. D., F W. D. et al.: Four multidetector-
row helical CT. Image quality and volume coverage speed.
Radiology 215 (2000), 55.
6. I Y., A T., Y K. et al.: Computed tomo-
graphy of gallbladder carcinoma. Radiology 137 (1980),
713.
7. I K., A H., M D. G. et al.: Gallbladder dis-
ease. Appearance of associated transient increased attenu-
ation in the liver at biphasic, contrast enhanced dynamic
CT. Radiology 204 (1997), 723.
8. K A., W K., F T. et al.: Diagnosis
and operative indications for polypoid lesions of the gall-
bladder. Arch. Surg. 123 (1988), 26.
9. K A. & A S.: Carcinoma of the gallbladder.
CT findings in 50 cases. Abdom. Imaging 19 (1994), 304.
10. L M. D., H T., N K. et al.: Improved delin-
eation of the gallbladder wall with ultrasonography. Its
value in assessment of the depth of carcinoma invasion. J.
Clin. Ultrasound 19 (1991), 471.
11. O T., S Y., T K. et al.: Carcinoma of the
gallbladder. CT evaluation of lymphatic spread. Radiology
189 (1993), 875.
12. O T., S Y., T K. et al.: Spread of gall-
bladder carcinoma. CT evaluation with pathologic corre-
lation. Abdom. Imaging 21 (1996), 195.
13. O H., P M., L S. et al.: Radiologi-
cal findings in cases of gallbladder carcinoma. Eur. Radiol.
17 (1993), 179.
14. P R., K S. P., S S. S. et al.: Predictors
of survival in patients with carcinoma of the gallbladder.
Cancer 76 (1995), 1145.
15. W S. N., K M., M H. et al.:
Sonography and computed tomography in the diagnosis of
carcinoma of the gallbladder. AJR 142 (1984), 735.