Eur Radiol (2016) 26:407416

DOI 10.1007/s00330-015-3846-5

HEPATOBILIARY-PANCREAS

Intrahepatic mass-forming cholangiocarcinoma: prognostic value

of preoperative gadoxetic acid-enhanced MRI
Jieun Koh 1 & Yong Eun Chung 1 & Ji Hae Nahm 2 & Ha Yan Kim 3 & Kyung-Sik Kim 4 &
Young Nyun Park 2 & Myeong-Jin Kim 1 & Jin-Young Choi 1

Received: 29 September 2014 / Revised: 11 May 2015 / Accepted: 12 May 2015 / Published online: 23 May 2015
# European Society of Radiology 2015

Abstract
Objectives To assess whether gadoxetic acid-enhanced MRI
could be used as a prognostic factor for intrahepatic massforming cholangiocarcinomas (IMCCs).
Methods Forty-one patients with pathologically proven IMCCs
who underwent preoperative gadoxetic acid-enhanced MRI were
included. The signal intensity of the IMCCs on hepatobiliary
phase (HBP) MRI was qualitatively analyzed by two radiologists, and categorized into intermediate or hypointense groups.
Analysis of clinicopathological prognostic factors and correlations of imaging and histology were also performed. Survival
time and time to recurrence (TTR) were analyzed.
Results Of the 41 IMCCs, 23 were in the intermediate group
and 18 were in the hypointense group on HBP MRI. IMCCs
in the intermediate group were associated with shorter survival time (P=0.048) and TTR (P=0.002) than the IMCCs of the
hypointense group. Only the intermediate group on HBP MRI
had a significantly shorter TTR on multivariate analysis
(P=0.012). The IMCCs of the intermediate group showed a
tendency for more abundant tumour fibrous stroma than those
of the hypointense group (P=0.027).

* Jin-Young Choi
gafield2@gmail.com
1

Department of Radiology, Research Institute of Radiological

Science, Yonsei University, College of Medicine, 50 Yonsei-ro,
Seodaemun-gu, 120-752 Seoul, Korea

Department of Pathology, Yonsei University, College of Medicine,

Seoul, Korea

Biostatistics Collaboration Unit, Severance Hospital, Yonsei

University, College of Medicine, Seoul, Korea

Department of General Surgery, Yonsei University, College of

Medicine, Seoul, Korea

Conclusions The enhancement of IMCCs on HBP gadoxetic

acid-enhanced MRI appears to correlate with tumour aggressiveness and outcomes due to the tumour fibrous stromal component. Thus, HBP images could be a useful prognostic factor
for IMCCs after surgery.
Key points
The enhancement of IMCCs on HBP correlates with the
tumour fibrous stroma.
The enhancement of IMCCs on HBP MRI appears to correlate with prognosis.
Gadoxetic acid-enhanced MRI is helpful for predicting prognosis of IMCCs after surgery.

Keywords Intrahepatic cholangiocarcinoma . MRI . Contrast

agent . Prognosis . Gadoxetic acid

Abbreviations
IMCC
Intrahepatic mass-forming cholangiocarcinoma
HBP
Hepatobiliary phase
TTR
Time to recurrence
GRE
Gradient echo
ADC
Apparent diffusion coefficient
SI
Signal intensity
SNR
Signal-to-noise ratio
CNR
Contrast-to-noise ratio
ICG 15
Indocyanine green 15
AFP
Alpha-fetoprotein
CA 19-9
Carbohydrate antigen 19-9
CEA
Carcinoembryonic antigen
PIVKA-II Proteins induced by vitamin K absence or
antagonist-II
AJCC
American Joint Committee on Cancer

408

Introduction
Cholangiocarcinoma arises from the ductular epithelium of
the biliary tree [1], and can be classified into three types according to growth characteristics: mass-forming, periductalinfiltrating, and intraductal-growing [2, 3]. Approximately
510 % of cholangiocarcinomas arise from the intrahepatic
bile duct [4], most of which are the mass-forming type [5].
Intrahepatic mass-forming cholangiocarcinoma (IMCC) often
spreads along the microvessel system, leading to intrahepatic
metastasis and decreased long-term survival [6, 7]. Surgery is
the only curative treatment modality for IMCC [1, 8], and the
5-year survival rate remains low, at less than 5 % [9]. Previous
studies have reported several prognostic factors for IMCC,
including lymph node metastasis, curative resection, tumour
stage, vascular invasion, intrahepatic metastasis, and preoperative tumour markers [7, 10, 11].
There have been a few reports indicating that the imaging
findings of IMCC correlate with prognosis [1215]. IMCCs
with abundant tumour fibrous stroma, defined as scirrhous
type, exhibit a delayed enhancement area on CT, which correlates with poor prognosis [12]. In one study, the enhancement patterns of IMCCs on gadoxetic acid-enhanced MRI
appeared to be related to the extent of stromal fibrosis, and
IMCCs showing a target appearance on hepatobiliary phase
(HBP) MRI were associated with central stromal fibrosis [16].
In another study, the relative enhancement of IMCC on the
HBP correlated with histological differentiation, showing
higher enhancement in moderately differentiated than in poorly differentiated tumours. In addition, tumours with higher
enhancement on HBP were correlated with fewer lymph node
metastases [17]. Therefore, we hypothesized that the enhancement pattern and degree of IMCC on HBP MRI may correlate
with the amount of tumour fibrous stroma, possibly affecting
prognosis. The aim of this study was to assess whether
gadoxetic acid-enhanced MRI could be used as a prognostic
factor for IMCCs.

Materials and methods

The institutional review board approved this retrospective
study, with a waiver of informed consent.
Study population
We searched the electronic medical record database from
January 2008 to March 2013 using the following search parameters: 1) a diagnosis of intrahepatic cholangiocarcinoma, 2) patients who were scheduled for hepatectomy, and
3) patients who underwent preoperative gadoxetic acidenhanced MRI. Initially, 185 patients were included. Patients
without mass-forming intrahepatic cholangiocarcinoma (n=

Eur Radiol (2016) 26:407416

144), such as Klatskin tumour or periductal infiltrating type,

were excluded, and a total of 41 patients were included in this
study (mean age, 62.2 years, range 3679 years; 25 men, 16
women).
MRI protocol
A magnetic strength of 3 T was used for 35 patients
(Magnetom Trio a Tim, Siemens Medical Solutions,
Erlangen, Germany, n = 15; Achieva; Philips Medical
Systems, Best, the Netherlands, n = 18; Discovery, GE
Medical Systems, Milwaukee, Wisconsin, USA, n=2), while
1.5 T was used for six patients (Achieva 1.5 T, Philips Medical
Systems, Best, the Netherlands, n=4; Magnetom Vision,
Siemens Medical Solutions, Erlangen, Germany, n=2). A single dedicated body phased-array coil was applied anteriorly
and spine array coils were applied posteriorly.
After localizing, in-phase (repetition time [TR]/effective
echo time [TE]/flip angle [FA], 150192 ms/2.30 and
2.46 ms/6580 for 3T, 167 ms/2.30 ms/80 for 1.5T) and
opposed-phase (TR/TE/FA 150192 ms/1.14 and 1.23 ms/
6580 for 3T, 167 ms/4.60 ms/80 for 1.5T) images were
obtained using two-dimensional dual-echo breath-hold T1weighted spoiled gradient echo (GRE) with a matrix of
256192 or 256256, 7-mm slice thickness, and a 0.71mm gap. A navigator-triggered T2-weighted turbo spin-echo
sequence (29235371 ms/7388 ms; flip angle, 140150;
matrix, 256192 or 302202; slice thickness, 5 or 4 mm;
section gap, 1 mm) and a breath-hold, heavily T2-weighted
turbo spin-echo image (450470 ms/96190 ms; flip angle, 90
or 150; matrix, 256192 or 320179; slice thickness, 4 mm;
section gap, 1 mm) were sequentially obtained. Dynamic images were acquired using three-dimensional T1-weighted
spoiled GRE sequence with chemically selective fat suppression (3.34.48 ms/1.162.19 ms; flip angle, 1315; matrix,
320224 or 256256; slice thickness, 2.5 mm; zero intersection gap) at 2535 s (arterial phase), 5565 s (portal phase),
8595 s (transitional phase), and 1520 min (hepatobiliary
phase) after a bolus injection of gadoxetic acid (Primovist,
Bayer Schering Pharma, Berlin, Germany) at 0.025 mmol/
Kg body weight, followed by a saline flush of 1520 mL at
an injection rate of 2 mL/s, and acquisition duration was 15 s
for each sequence. Diffusion-weighted images (1500
6100 ms/4470 ms; echo train length, 57 or 108; slice thickness and gap, 5/1 mm; matrix, 192108 or 124124, b-values
of 50 s/mm2, 400 s/mm2, and 800 s/mm2) were also obtained,
and apparent diffusion coefficient (ADC) maps were created
from the diffusion-weighted images.
Image interpretation
Two abdominal imaging radiologists (J.Y.C. and Y.E.C., with
13 and 7 years of experience, respectively) retrospectively and

Eur Radiol (2016) 26:407416

qualitatively analysed the preoperative MR images and

achieved consensus. A tumour was categorized in the intermediate group when there were more than 50 % hyper- or
isointense areas as compared to the spleen, and it was categorized in the hypointense group when there was less than 50 %
hyper- or isointensity as compared to the spleen on visual
inspection on hepatobiliary phase; equivocal cases were
assessed using region-of-interest signal intensity measurements [18, 19]. On dynamic study images, tumours were categorized as hypovascular when the tumour had less than twothirds arterial enhancement or peripheral rim enhancement,
and as hypervascular when there was more than two-thirds
arterial enhancement. The temporal enhancement pattern
was also analyzed [17]. The two radiologists were blinded to
all clinical history as well as laboratory and pathology reports.
For quantitative image analysis, regions of interest were
drawn by one radiology resident (J.E.K.) measuring the signal
intensity (SI) of the whole tumour area at the level of the
largest tumour area on hepatobiliary phase. The signal-tonoise ratios (SNRs) of the liver and the tumour were calculated as follows: liver SNR=liver SI/SD of background noise,
and tumour SNR=tumour SI/SD of background noise. The
contrast enhancement ratio and contrast-to-noise ratio (CNR)
of the tumour were calculated as follows: tumour contrast
enhancement ratio=[(enhanced tumour SNRunenhanced tumou r SNR)/une nhan ced tu mour SNR] 10 0, a nd
CNR=(tumour SIliver SI)/SD of background noise [20].
On ADC maps, regions of interest were drawn by the same
radiology resident onto the largest tumour area.
Patient characteristics
The level of serum CEA was below 2.5 ng/mL in 16 patients
and was equal to or above 2.5 ng/mL in 19 patients. Other
patient characteristics, including laboratory data, are presented
in Table 1. Four patients had underlying liver cirrhosis.
Preoperative concurrent chemoradiotherapy or chemotherapy
was performed for seven patients. There was no further postoperative treatment for 25 patients, and the remaining patients
received adjuvant chemotherapy (n=12), adjuvant chemotherapy and radiotherapy (n=2), or adjuvant radiotherapy (n=2).
Clinical data analysis
Clinical data were assessed by reviewing electronic medical
records, including the date of surgery and surgical procedure,
the date of recurrence, date of death, recurrence pattern, preoperative and postoperative treatment, and treatment after recurrence. Laboratory data were also recorded, including the
levels of indocyanine green 15 (ICG 15), alpha-fetoprotein
(AFP), carbohydrate antigen 19-9 (CA 19-9),
carcinoembryonic antigen (CEA), proteins induced by

409
Table 1

Preoperative laboratory and clinical data of 41 patients

Category

Patient number

ICG 15 (%)
<10

17

10
Not available

6
18

<7.0

25

7.0

AFP (IU/mL)

Not available

13

<2.5
2.5

16
19

CEA (ng/mL)

Not available

CA19-9 (U/mL)
<37

13

37
Not available

23
5

<40
40
Not available

21
3
17

<3
3

2
39

<10
10

41
0

PIVKA-II (mAU/mL)

Albumin (g/dL)

Total bilirubin (mg/dL)

Receiving adjuvant chemotherapy

Yes
No

7
34

ICG 15 indocyanine green 15, AFP alpha-fetoprotein, CEA

carcinoembryonic antigen, CA 19-9 carbohydrate antigen 19-9, PIVKAII proteins induced by vitamin K absence or antagonist-II

vitamin K absence or antagonist-II (PIVKA-II), albumin,

and bilirubin.
Pathologic evaluation
Pathology reports were retrospectively reviewed for gross
morphology, including tumour diameter, presence of tumour
necrosis, haemorrhage or peliosis, and the presence of portal
vein or bile duct invasion. Microscopic features were also
recorded, including tumour differentiation (i.e. well-differentiated, moderately differentiated, poorly differentiated, and
undifferentiated) as well as resection margin, serosal invasion,
portal vein invasion, bile duct invasion, hepatic vein invasion,
microvessel invasion, multi-centric occurrence, lymph node
metastasis, and non-tumour liver pathology. Pathologic

410

TNM classification was evaluated, and the final staging was

determined according to the American Joint Committee on
Cancer (AJCC) staging criteria.
For evaluation of imagingpathology correlations, a pathologist (J.H.N.) who was blinded to the clinical information
reviewed the IMCC slides. The amount of tumour fibrosis was
evaluated using the tissue sections of the largest cut surface of
the tumour, and the necrotic area was excluded for evaluating
fibrosis. The amount of fibrosis was graded into four scales;
grade I represented<25 % of stromal fibrosis, grade II25 and
<50 %, grade III50 % and <75 %, and grade IV75 %.
Data and statistical analysis
Continuous variables were analyzed using the Student t test,
and categorical variables were analyzed by the Fisher exact
test and Pearson chi-square test. Survival rates and time to
recurrence were evaluated with the Kaplan-Meier method
and log-rank test. The Cox proportional hazards model was
used in the multivariate survival analysis. Patients with
follow-up loss or death from unrelated causes were regarded
as censored data. For the evaluation of the imaginghistological correlations linear by linear association was used.
A value of P<0.05 was regarded as significant. All statistical
analyses were performed using SAS version 9.2 software
(SAS Institute Inc., Cary, NC, USA).

Results
Clinicopathologic analysis
The mean tumour diameter was 4.82.4 cm (range 1.5
13 cm). Tumours were well differentiated in seven patients,
moderately differentiated in 22 patients, poorly differentiated
in seven patients, and undifferentiated in one patient; differentiation was not reported in the pathology report in four patients. The resection margin was clear in 30 patients.
Microvessel invasion was present in 30 patients and perineural
invasion in 13 patients. Intrahepatic metastasis was present in
three patients. The majority of patients were in the advanced
stage (stage 1, n=6; stage 2, n=11; stage 3, n=11; and stage 4,
n=13). Most patients were above pT2, and 13 patients showed
lymph node metastasis (Table 2).
Tumour recurrence was observed in 17 patients; among
these, eight patients died from the recurrence. One patient died
outside the hospital, and further follow-up was not possible.
The organs where tumours recurred were the liver (n=13),
lymph node (n=3), lung (n=5), peritoneum (n=3), and bone
(n=1). Treatment after recurrence was systemic chemotherapy
(n=8), systemic chemotherapy and radiotherapy (n=2), hepatic resection (n=1), radiofrequency ablation (n=1), and supportive care (n=5).

Eur Radiol (2016) 26:407416

Qualitative and quantitative image analysis

The signal intensity of the tumour as compared to the spleen
on HBP MRI was qualitatively analyzed. The intermediate
group included 23 patients (56.1 %; Fig. 1), and one tumour
exhibited hyperintense signal intensity compared to liver parenchymal enhancement. The remaining 18 patients (43.9 %)
were regarded as belonging to the hypointense group (Fig. 2).
Baseline characteristics and pathology data were analyzed between the two groups, and no significant differences were
found (Table 2). The intermediate group showed variable patterns of enhancement on HBP MRI, including heterogeneous
(n=10), central (n=9), and homogeneous enhancement (n=
4). Two lesions showed T2 hypointense signal intensity and
the remaining 21 tumours showed T2 hyperintense signal intensity compared to surrounding liver parenchyma.
On dynamic study, 13 (31.7 %) tumours were
hypervascular and 28 (68.3 %) were hypovascular on the arterial phase. The most common temporal enhancement pattern
was gradual enhancement, which demonstrated contrast enhancement that increased over time (n=25). This was followed by persistent enhancement, which showed contrast enhancement that remained constant throughout the dynamic
studies (n=8), arterial enhancement with delayed washout
(n=5), and studies with no enhancement or minimal enhancement (n=3).
The mean contrast enhancement ratio among all IMCCs
was 60.6 %; the mean contrast enhancement ratio was
73.1 % for the intermediate group and 44.5 % for the
hypointense group (P = 0.175). The mean CNR of the
IMCCs was 11281.2.
The mean ADC value among all IMCCs was 1.210
103 mm2/s. The mean ADC value in the intermediate group
was 1.209103 mm2/s, while in the hypointense group it was
1.210103 mm2/s (P=0.283).
There was a statistically significant correlation between the
grade of tumour fibrous stroma and MR signal intensity on
HBP MRI of the IMCCs (P=0.027; Fig. 3). The hypointense
group had a tendency toward a low grade of tumour fibrous
stroma, whereas a high grade of tumour fibrous stroma was
more frequently observed in the intermediate group.
Correlation of imaging and outcomes
The 5-year survival rate was 87 % in the hypointense group
and 53 % in the intermediate group. On the log-rank test, the
survival rate was greater in the hypointense group than the
intermediate group (P=0.048). The Cox proportional hazard
model revealed a hazard ratio of 4.35, although this was not
statistically significant (0.8921.22 with 95 % confidence interval, P=0.069). In univariate analysis of factors including
tumour marker, pathologic tumour characteristics, staging,
pre- and postoperative treatment, and grade of tumour fibrous

Eur Radiol (2016) 26:407416

Table 2

411

Demographic, laboratory, and pathologic findings according to enhancement pattern on hepatobiliary phase imaging

Age (years)
ICG 15 (%)
AFP (IU/mL)
CEA (ng/mL)
CA19-9 (U/mL)

pT

pN
Tumour differentiation
Positive resection margin (R1) (n=11)
Gross pathology
Tumour diameter (cm)

Intermediate
(n=21)

P value

6013

6310

0.385

7.93.8

8.73.6

0.599

4.65.0
32.091.8

8.824.2
7.713.7

0.582
0.282

4202.26598.4

0.451

66.3122.4
4.20.7

27.011.2
4.10.5

0.313
0.518

0.80.3
5

0.60.3
1

0.094
0.857

2562.75993.7

PIVKA-II (mAU/mL)
Albumin (g/dL)
Total bilirubin (mg/dL)
AJCC staging

Hypointense
(n=18)

1 (n=6)
2 (n=11)

10

3 (n=11)
4 (n=13)

6
6

5
7

1 (n=7)
2 (n=20)

5
6

2
14

3 (n=14)
0 (n=28)
1 (n=13)

7
12
6

7
16
7

Well to moderate (n=29)

Poor or undifferentiated (n=8)
4

14
2
7

15
6
0.726

0.18

4.62.1

5.02.5

0.624

7
3
0

12
2
0

0.397
0.638

6
2
7

11
4
11

0.350
0.679
0.752

0
11
5
0

0
19
9
3

0.164
0.52
0.243

Tumour necrosis (n=19)

Tumour haemorrhage/peliosis (n=5)
Portal vein invasion (n=0)
Bile duct invasion (n=11)
Microscopic pathology
Serosal invasion (n=17)
Portal vein invasion (n=6)
Bile duct invasion (n=18)
Hepatic artery or vein invasion (n=0)
Microvessel invasion (n=30)
Perineural invasion (n=14)
Intrahepatic metastasis (n=3)

0.14

0.843

ICG 15 indocyanine green 15, AFP alpha-fetoprotein, CEA carcinoembryonic antigen, CA 19-9 carbohydrate antigen 19-9, PIVKA-II proteins induced
by vitamin K absence or antagonist-II

stroma, none of these factors showed a statistically significant

influence on survival time (Table 3).
There was a statistically significant difference in TTR between the hypointense and intermediate groups, with the
hypointense group having a longer TTR than the intermediate
group (P=0.002). The 5-year recurrence rate was 24 % in the
hypointense group and 65 % in the intermediate group. In
univariate analysis of TTR, microvessel invasion was another
statistically significant factor (P=0.043). Multivariate Cox regression analysis was performed with HBP images and

microvessel invasion, and only the HBP image remained statistically significant (P=0.012) (Table 4).
Quantitative analysis indicated no correlation of tumour
contrast enhancement ratio and CNR with survival (P =
0.859 and 0.714) or recurrence (P=0.417 and 0.152). Based
on arterial enhancement, there was no statistically significant
difference in survival rate or TTR between the hypervascular
and hypovascular groups (P=0.670 and 0.124), and no statistically significant difference in survival rate or TTR was found
among the four temporal enhancement patterns (P=0.121 and

412

Eur Radiol (2016) 26:407416

Fig. 1 A 67-year-old woman

with IMCC. (a) Gadoxetic acidenhanced T1-weighted 3D
gradient-echo image (TR/TE/
FA=3.3 ms/1.16 ms/13) in the
late arterial phase shows
peripheral rim enhancement of
the tumour. (b) The transitional
phase at 3 min demonstrates
centripetal filling of the contrast
agent. (c) On hepatobiliary phase
MRI, the tumour consisted of
more than 50 % hyperintensity
compared to the spleen (arrow),
and was categorized in the
intermediate group. (d) On visual
assessment of the microscopic
image (H and E, 40), there was
75 % stromal fibrosis, which
corresponds to grade IV fibrosis
(arrow). Recurrence was noted
271 days after surgery, and the
patient died 304 days after
surgery due to recurrence.
IMCC=intrahepatic massforming cholangiocarcinoma

Fig. 2 A 67-year-old man with IMCC. (a) Gadoxetic acid-enhanced T1weighted 3D gradient-echo image (TR/TE/FA=3.3 ms/1.16 ms/13) in
the late arterial phase shows a mass with peripheral irregular
enhancement. (b) The transitional phase at 3 min demonstrates the
subsequent filling of the contrast agent. (c) On the hepatobiliary phase
image, most of the tumour was hypointense compared to the spleen

(arrow), and it was thus categorized in the hypointense group. (d) On

microscopic image (H and E, 40), stromal fibrosis was noted in 35 % of
the whole tumour, and the grade was II (arrows). Although the patient
initially had a large tumour, he survived for 1213 days without tumour
recurrence. IMCC=intrahepatic mass-forming cholangiocarcinoma

Eur Radiol (2016) 26:407416

413

Fig. 3 Correlation of the signal

intensity and grade of tumour
fibrous stroma of IMCCs on
hepatobiliary phase imaging.
Percentage of stromal fibrosis on
visual assessment of the
pathologic slide: Grade I<25 %,
Grade II25 and <50 %, Grade
III50 % and <75 %, Grade IV
75 %. IMCC=intrahepatic massforming cholangiocarcinoma

0.257, respectively). There was no statistically significant difference in survival rate or TTR based on ADC value (P=
0.102 and P=0.868, respectively).

Discussion
Our study demonstrated that IMCCs with greater than 50 %
intermediate signal intensity on HBP imaging had worse outcomes than hypointense tumours. The intermediate signal intensity in IMCCs on HBP imaging was related to the amount
of tumour fibrous stroma in the tumour based on the image
histology correlation, and was also correlated with prognosis.
Therefore, gadoxetic acid-enhanced MRI including HBP imaging is not only useful for preoperative assessment of tumour
staging, but also has incremental value in predicting
prognosis.
Typical CT findings of IMCC show homogeneous attenuation and irregular peripheral enhancement, followed by gradual centripetal enhancement [2123]. The enhancement pattern of IMCC on extracellular contrast agent-enhanced MRI is
similar to that of CT, showing peripheral and then centripetal
enhancement [21]. IMCC may show varying degrees of delayed enhancement on both CT and MR due to tumour fibrous
stroma [21]. A relationship between delayed enhancement
pattern on CT and patient survival has been reported [12]. In
one study, IMCCs with more than two-thirds of the tumour
showing delayed enhancement on CT were found to correlate
with abundant tumour fibrous stroma and frequent perineural
invasion and with a lower survival rate than IMCCs with less
than two-thirds of the tumour showing delayed enhancement

[12]. However, in another study, a higher percentage of enhancement in IMCC gadoxetic acid-enhanced HBP images
correlated with better differentiation of tumours and fewer
lymph node metastases [17]. In the present work, we evaluated whether survival or recurrence of IMCC were related to
gadoxetic acid-enhanced MRI findings, and found that the
intermediate group had a worse prognosis than hypointense
tumours, consistent with a previous study using CT [12]. It
remains to be determined whether delayed enhancement on
CT and gadoxetic acid-enhanced MRI are comparable for
predicting prognosis.
IMCCs usually demonstrate hypointensity on HBP MRI
compared to adjacent liver parenchyma because these tumours
do not take up hepatobiliary agents due to a lack of organic
anionic transporter peptide (OATP) expression [17, 24].
Paradoxically, however, some IMCCs may have intermediate
or mixed signal intensity on HBP MRI because of contrast
agent pooling in the extracellular space or tumour fibrous
stroma. In our study, almost all tumours (97.6 %) appeared
hypointense compared to the adjacent liver parenchyma on
HBP MRI, and only one tumour appeared hyperintense.
With the signal intensity of the spleen set as the reference,
56.1 % of tumours fell within the intermediate group, showing
a hyper or iso-intense area greater than 50 % of the area of the
whole tumour. Since intermediate signal intensity on HBP
MRI reflects the contrast material retained in the tumour fibrous stroma of IMCC, abundant tumour fibrous stroma corresponds to a larger intermediate signal intensity area on HBP
MRI [16]. This finding coincides with the results of our study:
tumours with intermediate signal intensity on HBP MRI exhibited more abundant stroma on histopathologic examination, which correlated with poor prognosis. This evidence

414
Table 3

Eur Radiol (2016) 26:407416

Prognostic factors according to univariate analysis

Factors

Survival rate (%)

1 year

3 years

P value
5 years

Disease-free rate (%)

1 year

3 years

P value
5 years

Hepatobiliary phase image

Hypointense (n=18)

94

87

87

88

76

76

Intermediate (n=21)

74

53

53

41

35

35

85

85

85

65

51

51

75
100

60
50

60
50

63
60

63
60

63
60

67

33

33

33

33

33

Temporal enhancement pattern

Gradual enhancement (n=25)
Persistent enhancement (n=8)
Arterial enhancement with delayed washout (n=5)
No or minimal enhancement (n=3)

0.048

0.121

ADC value
CEA (ng/mL)
<2.5 (n=16)
2.5 (n=19)
Albumin (g/dL)
<3 (n=2)
3 (n=39)
Tumour diameter (cm)
Tumour differentiation
Well (n=7)

0.002

0.257

0.102
0.122
95
75

86
7

86
7

0.868
0.372
71
55

62
55

62
55

100

100

100

100

100

100

85

77

77

62

56

56

0.069

0.100

0.665
100

100

100

Moderate (n=22)
Poor (n=7)
Undifferentiated (n=1)
Resection margin
Clear (R0) (n=30)

86
86
100

82
57
100

82
57
100

87

77

Not clear (R1) (n=11)

Microvessel invasion
Not present (n=11)
Present (n=30)
Perineural invasion

82

82

100
80

100
70

100
70

Not present (n=27)

Present (n=14)
Intrahepatic metastasis
Not present (n=38)

89
79

82
71

82
71

84

76

76

Present (n=3)
pT
1 (n=7)
2 (n=20)
3 (n=14)
pN
0 (n=28)
1 (n=13)
AJCC Staging
1 (n=6)
2 (n=11)
3 (n=11)
4 (n=13)
Preoperative treatment
No (n=34)

100

100

100

0.616
88

75

75

64
43
100

59
43
100

59
43
100

77

67

63

63

82

55

46

46

91
53

91
47

91
47

67
57

67
43

67
43

63

61

61

67

33

33

100
37
79

100
30
71

100
30
71

68
54

61
54

61
54

0.973

0.043

0.662

0.265

0.107

0.800

82
79

100
45
79

100
45
79

89
77

82
69

82
69

0.257

0.118

0.884
100
91
82
77

100
73
82
69

100
73
82
69

82

74

74

0.457
0.192

0.284

0.424

0.356
100
36
82
54

100
27
73
54

100
27
73
54

62

56

56

0.668

Eur Radiol (2016) 26:407416

415

Table 3 (continued)
Factors

Survival rate (%)

P value

1 year

3 years

5 years

100

100

100

91
71

69
71

69
71

1 (n=2)
2 (n=15)

60
91

60
91

3 (n=17)
4 (n=7)

88
67

74
33

Yes (n=7)
Postoperative treatment

Disease-free rate (%)

1 year

P value

3 years

5 years

71

71

71

60
60

60
41

60
41

60
91

60
56

60
40

60
40

74
33

76
38

69
38

69
38

0.549

No (n=25)
Yes (n=16)
Histological tumour fibrous stroma grade

0.495

0.510

0.223

CEA carcinoembryonic antigen, ADC apparent diffusion coefficient

suggests a close link between tumour cells and tumourassociated stroma. The stroma actively provides continuous
support to the tumour cells and modulates tumour progression. Histologically, IMCC can be classified into two types
according to the amount of fibrosis in the lesion: tumours with
abundant stroma, the so-called scirrhous type, are known to
have worse prognosis than non-scirrhous-type IMCCs [25].
Scirrhous-type IMCCs frequently overproduce extracellular
matrix, degrade basement membrane, and modulate macromolecules that play important roles in cellular growth, differentiation, cellmatrix adhesion, invasion, and eventually metastasis of cancer cells [26]. Scirrhous-type IMCCs also show
frequent lymphatic permeation, perineural invasion, and
higher proliferative activity [25]. As such, information on
the degree of stroma based on MRI may be useful for improving surgical strategies in treating this neoplasm. Theoretically,
extracellular contrast agent-enhanced MRI may also show delayed enhancement in patients with IMCC due to contrast
pooling in the tumour fibrous stroma. However, given that
gadoxetic acid-enhanced MRI including HBP images provides increased lesion conspicuity as well as better delineation
of daughter nodules and intrahepatic metastasis, this technique
could be advantageous in preoperative staging as well as in the
assessment of the tumour fibrous stroma [17].
Although IMCC is typically hypovascular, IMCC arising
from chronic viral hepatitis and the cirrhotic or pre-cirrhotic
liver is more likely to be hypervascular after the malignant
transformation of the cholangiocytes [27]. It has been reported

Table 4 Multivariate Cox

regression for recurrence

that hypervascular versus hypovascular IMCC on arterial

phase CT correlates with reduced malignant behaviour and
better survival [1315, 28]. In a previous study, welldifferentiated tumours were found to have greater enhancement on gadoxetic acid-enhanced MRI than moderately or
poorly differentiated IMCCs [16]. However, in our study,
hypervascular IMCCs comprised 31.7 % of tumours, and
there was no statistical difference between hypervascular and
hypovascular IMCCs in either survival rate or TTR. Further
evaluation of arterial enhancement and prognosis is needed in
a larger study.
We acknowledge that this study has several limitations.
First, due to the rare incidence of IMCC, only a small number
of patients were included, and the follow-up period was relatively short. Since most studies with IMCC include a small
patient sample, statistical power may be somewhat weak, and
so a larger study is needed to validate these results. Second,
there may have been selection bias, as we included only surgically resected cholangiocarcinoma. Further study is needed
in patients who receive non-surgical treatment. Third, quantitative analysis was performed on the whole tumour region,
without quantitative comparison of the degree of enhancement
among different areas of the tumour. In addition, parallel imaging may have influenced the SNR, which would not account
for the noise throughout the image. However k-space-based
parallel acquisition spreads the noise throughout images, thus
reducing the influence on SNR calculation [20], which may
result in a discrepancy between qualitative and quantitative

Factors
Hepatobiliary phase image
Microvessel invasion

Hazard ratio
Hypointense group
Intermediate group
No
Yes

1
5.088
1
6.947

95 % Confidence Interval

P value
0.012

1.62920.352
0.062
0.91053.058

416

Eur Radiol (2016) 26:407416

analysis. Fourth, over the 5-year period of this study, various

MRI units and sequences were adopted, which may have
weakened reproducibility.
In conclusion, IMCCs with intermediate signal intensity on
HBP MRI are associated with abundant tumour fibrous stroma
and poor prognosis. Therefore, gadoxetic acid-enhanced MRI
could be a useful prognostic tool for IMCCs after surgery.
Acknowledgements The scientific guarantor of this publication is
Jin-Young Choi. The authors of this manuscript declare no relationships
with any companies whose products or services may be related to the
subject matter of the article. This study was supported by a faculty research grant of Yonsei University College of Medicine for 6-2014-0126.
Ha Yan Kim kindly provided statistical advice for this manuscript. One of
the authors has significant statistical expertise. Institutional review board
approval was obtained. Written informed consent was waived by the
institutional review board. Methodology: retrospective, diagnostic or
prognostic study, performed at one institution.

12.

13.

14.

15.

16.

17.

18.

References
1.
2.

3.

4.

5.

6.

7.

8.
9.
10.

11.

Khan SA, Thomas HC, Davidson BR, Taylor-Robinson SD (2005)

Cholangiocarcinoma. Lancet 366:13031314
Liver Cancer Study Group of Japan (2000) The general rules for the
clinical and pathological study of primary liver cancer, 4th edn.
Kanehara, Tokyo
Liu Y, Zhong X, Yan L, Zheng J, Liu Z, Liang C (2015) Diagnostic
performance of CT and MRI in distinguishing intraductal papillary
neoplasm of the bile duct from cholangiocarcinoma with intraductal
papillary growth. Eur Radiol. doi:10.1007/s00330-015-3618-2
Nakeeb A, Pitt HA, Sohn TA et al (1996) Cholangiocarcinoma. A
spectrum of intrahepatic, perihilar, and distal tumors. Ann Surg
224:463473
Murakami Y, Uemura K, Sudo T et al (2011) Prognostic factors
after surgical resection for intrahepatic, hilar, and distal cholangiocarcinoma. Ann Surg Oncol 18:651658
Sasaki A, Aramaki M, Kawano K et al (1998) Intrahepatic peripheral cholangiocarcinoma: mode of spread and choice of surgical
treatment. Br J Surg 85:12061209
Ohtsuka M, Ito H, Kimura F et al (2002) Results of surgical treatment for intrahepatic cholangiocarcinoma and clinicopathological
factors influencing survival. Br J Surg 89:15251531
Lazaridis KN, Gores GJ (2005) Cholangiocarcinoma.
Gastroenterology 128:16551667
Shaib Y, El-Serag HB (2004) The epidemiology of cholangiocarcinoma. Semin Liver Dis 24:115125
Yamamoto M, Takasaki K, Yoshikawa T, Ueno K, Nakano M
(1998) Does gross appearance indicate prognosis in intrahepatic
cholangiocarcinoma? J Surg Oncol 69:162167
Isaji S, Kawarada Y, Taoka H, Tabata M, Suzuki H, Yokoi H (1999)
Clinicopathological features and outcome of hepatic resection for
intrahepatic cholangiocarcinoma in Japan. J Hepatobiliary Pancreat
Surg 6:108116

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

Asayama Y, Yoshimitsu K, Irie H et al (2006) Delayed-phase dynamic CT enhancement as a prognostic factor for mass-forming
intrahepatic cholangiocarcinoma. Radiology 238:150155
Ariizumi S, Kotera Y, Takahashi Y et al (2011) Mass-forming
intrahepatic cholangiocarcinoma with marked enhancement on
arterial-phase computed tomography reflects favorable surgical
outcomes. J Surg Oncol 104:130139
Nanashima A, Sumida Y, Abo T et al (2008) Relationship between
pattern of tumor enhancement and clinicopathologic characteristics
in intrahepatic cholangiocarcinoma. J Surg Oncol 98:535539
Nanashima A, Abo T, Murakami G et al (2013) Intrahepatic cholangiocarcinoma: relationship between tumor imaging enhancement
by measuring attenuation and clinicopathologic characteristics.
Abdom Imaging 38:785792
Jeong HT, Kim M-J, Chung YE, Choi JY, Park YN, Kim KW
(2013) Gadoxetate disodiumenhanced mri of mass-forming
intrahepatic cholangiocarcinomas: imaging-histologic correlation.
Am J Roentgenol 201:W603W611
Kang Y, Lee JM, Kim SH, Han JK, Choi BI (2012) Intrahepatic
mass-forming cholangiocarcinoma: enhancement patterns on
gadoxetic acid-enhanced MR images. Radiology 264:751760
Rummeny EJ, Reimer P, Heindel W (2009) MR imaging of the
body. Thieme, New York
Kim SH, Lee JM, Han JK et al (2006) Hepatic macrosteatosis:
predicting appropriateness of liver donation by using MR imagingcorrelation with histopathologic findings. Radiology 240:
116129
Choi JY, Kim MJ, Park YN et al (2011) Gadoxetate disodiumenhanced hepatobiliary phase MRI of hepatocellular carcinoma:
correlation with histological characteristics. AJR Am J
Roentgenol 197:399405
Chung YE, Kim MJ, Park YN et al (2009) Varying appearances of
cholangiocarcinoma: radiologic-pathologic correlation.
Radiographics 29:683700
Lim JH (2003) Cholangiocarcinoma: morphologic classification
according to growth pattern and imaging findings. AJR Am J
Roentgenol 181:819827
Han JK, Choi BI, Kim AY et al (2002) Cholangiocarcinoma: pictorial essay of CT and cholangiographic findings. Radiographics 22:
173187
Peporte AR, Sommer WH, Nikolaou K, Reiser MF, Zech CJ (2013)
Imaging features of intrahepatic cholangiocarcinoma in Gd-EOBDTPA-enhanced MRI. Eur J Radiol 82:e101106
Kajiyama K, Maeda T, Takenaka K, Sugimachi K, Tsuneyoshi M
(1999) The significance of stromal desmoplasia in intrahepatic
cholangiocarcinoma: a special reference of scirrhous-type and
nonscirrhous-type growth. Am J Surg Pathol 23:892902
Nakanuma Y, Harada K, Ishikawa A, Zen Y, Sasaki M (2003)
Anatomic and molecular pathology of intrahepatic cholangiocarcinoma. J Hepatobiliary Pancreat Surg 10:265281
Xu J, Igarashi S, Sasaki M et al (2012) Intrahepatic
cholangiocarcinomas in cirrhosis are hypervascular in comparison
with those in normal livers. Liver Int Off J Int Assoc Study Liver
32:11561164
Kim SA, Lee JM, Lee KB et al (2011) Intrahepatic mass-forming
cholangiocarcinomas: enhancement patterns at multiphasic CT,
with special emphasis on arterial enhancement patterncorrelation
with clinicopathologic findings. Radiology 260:148157