Introduction
The sole possible curative treatment for extrahepatic bile duct cancer is surgical
resection [1]
[2]
[3]. Unresectability of this cancer is associated with poor prognosis. However, lateral
spreading of this cancer along the longitudinal axis of the bile duct often results
in non-curative resection because of detection difficulties encountered in imaging
examinations [4]
[5]. Diagnosing the precise borderlines of the disease is key to achieve curative resection.
Of the two types of lateral spread, intraepithelial spread is extremely difficult
to diagnose with cross-sectional images, whereas intramural spread is relatively easy
to identify on contrast-enhanced computed tomography (CT). Although intraepithelial
spread can be diagnosed with endoscopic ultrasound (EUS) or intraductal ultrasound
(IDUS) [6]
[7] in most cases, it is undetectable when the neoplasm only causes low epithelial changes.
Cholangioscopy can possibly result in a proper diagnosis for difficult intraepithelial
spread [8]
[9]
[10]. After several brilliant studies confirmed the value of both percutaneous transhepatic
cholangioscopy and peroral duodenoscope-assisted cholangioscopy (POCS) with fiberscopes
[11]
[12]
[13]
[14]
[15], digital scopes with higher imaging quality became available. For instance, a CHF-B260
scope from Olympus Corporation (Tokyo, Japan), which has the highest level of imaging
quality among existing cholangioscopes, has been available and is reported to have
higher capability to accurately diagnose such lateral extent [10]
[16].
Recently, the SpyGlass DS system from Boston Scientific Corporation (Marlborough,
Massachusetts, United States) has become commercially available. This system consists
of a disposable digital scope with a field of view of 120 degrees. Although the spatial
quality of the system is considered to be relatively low in comparison with that of
the Olympus scope, the scope has a tapered tip, a four-way tip deflection system,
and a dedicated channel for water irrigation, which allow unimpeded observation and
biopsy procedures [17]
[18]
[19]
[20]
[21]. There have been no reports comparing these two types of digital cholangioscopes.
Therefore, we conducted such a comparison in a retrospective design and reported the
results herein.
Patients and methods
Patients
All data on patients who underwent surgical resection for extrahepatic bile duct cancer
after preoperative examinations, including POCS, to diagnose the lateral extent of
extrahepatic cholangiocarcinoma at Sendai City Medical Center were extracted from
a prospectively maintained database of endoscopic retrograde cholangiopancreatography.
The following patients were excluded: (1) those in whom the clinical record on findings
of the POCS examination could not be obtained or was insufficient for evaluation;
(2) those who did not undergo surgical resection after examinations; and (3) those
in whom the resected specimen was inappropriate to precisely evaluate the lateral
extent. Patients who underwent examinations between January 2004 and September 2015
were included in the POCS by CHF-B260 (CHF) group (group A) and those who underwent
examinations since the SpyGlass DS system was employed in October 2015 were included
in the SpyGlass DS group (group B).
Procedures
Mapping biopsy under x-ray fluoroscopy (endoscopic retrograde cholangiography [ERC])
without POCS was performed in all patients in group A ([Fig. 1]). In this group, POCS was performed when diagnosis of tumor spread by mapping biopsy
findings was questionable because of possible discrepancy between ERC-guided mapping
biopsy findings and other examinations, such as ERC, IDUS, EUS, CT, and magnetic resonance
imaging (MRI) ([Fig. 2]). This group included patients who did not undergo POCS because of absence of such
discrepancy. Radial jaw biopsy forceps (Boston Scientific) were used for ERC-guided
mapping biopsy.
Fig. 1 Endoscopic cholangiography-guided mapping biopsy. When the extrahepatic cholangiocarcinoma
is obviously and largely involved the right hepatic duct, the border lines in the
left hepatic duct and the lower bile duct should be identified. If the mapping biopsy
revealed that the cancer extent did not reach the B3/B4 bifurcation and the level
of the upper edge of the pancreas, this cholangiocarcinoma would be curatively removed
by resection at the shown lines.
Fig. 2 Diagnostic strategy during each study period. a In group A, fluoroscopy-guided mapping biopsy (ERC-MB) was performed in all patients.
Peroral cholangioscopy (POCS) was performed only when the diagnosis of the tumor spread
by ERC-MB findings was questionable because of possible discrepancy among examinations.
b In group B, POCS and POCS-guided mapping biopsy (POCS-MB) was performed without reference
of ERC-MB in all patients. CT, computed tomography; MRI, magnetic resonance imaging;
US, ultrasonography; EUS, endoscopic ultrasonography; ERC, endoscopic retrograde cholangiography;
IDUS, intraductal ultrasonography
In group B, POCS using SpyGlass DS was performed without reference of ERC-guided mapping
biopsy. The SpyGlass DS system was considered to be important as a biopsy instrument
with visual confirmation of the target sites rather than a diagnostic device for obtaining
endoscopic optical findings. SpyBite biopsy forceps (Boston Scientific) were used
for POCS-guided biopsy in this group.
Both types of scopes were managed in the mother-baby style, that is, handled through
a duodenoscope (TJF260V, Olympus). Endoscopic sphincterotomy was performed in the
same or previous session in all patients. When the POCS procedure, which required
injection of a large amount of water into the bile duct, was considered to be inappropriate
in the first session because of accompanying acute cholangitis, it was performed in
the next session after drainage with a plastic stent. After insertion of the cholangioscope
with or without guidewire assistance, turbid bile and contrast were replaced by saline
injected through the working channel of the CHF-B260 scope or through the irrigation
channel of the SpyGlass DS scope. A guidewire or contrast solution was used to confirm
the anatomical branch, if necessary. When it was difficult to insert the cholangioscope
through the stricture, a bougie dilator or a 10-Fr plastic stent was used for dilation.
Endoscopic optical findings obtained from POCS images, including the location of the
cancer borderlines, were reported in detail by the examiner.
Outcome measurements and definitions
The main outcome measurement was defined as accuracy of the overall preoperative diagnosis
of tumor lateral extent by the biopsy results and POCS findings in each group. Because
there would be selection bias due to the different diagnostic strategies in each group,
diagnostic accuracy of pure POCS findings was included as a secondary outcome. The
overall preoperative diagnosis was determined in the institutional preoperative conference
composed of more than 10 surgeons wherein pancreatobiliary expert endoscopists reviewed
all diagnostic information.
Secondary outcome measures were the accuracy of ERC-guided mapping biopsy in group
A, the ability of POCS to diagnose the borderline(s) of the cancer in the two groups,
and the reasons for wrong diagnosis. Borderlines at the liver or ampullary side were
not evaluated if the evaluation was deemed unnecessary due to predetermined resection
by the required surgical procedure. Additionally, borderlines at the liver and ampullary
sides may vary in difficulty. Therefore, results for the liver side and those for
the ampullary side were evaluated separately in this study.
Accuracy was defined as precise diagnosis of benignancy or malignancy at the points
of interest. For example, when absence of neoplasm at the left/right bifurcation (first
bifurcation) was diagnosed in cases of distal bile duct cancer, biopsy results or
POCS findings were accurate if the examination confirmed absence of a tumor at that
site. When a tumor spread beyond the left/right bifurcation, tumor involvement at
the second bifurcation should be estimated for determining the resection line. In
such cases, accuracy of the examinations was judged at the second bifurcation, rather
than the first. In cases of perihilar cancer that necessitate the decision for pancreatoduodenectomy,
the point of interest was determined to be the distal bile duct at the level of the
upper margin of the pancreas. Accuracy was not defined as precise diagnosis of the
length of the lateral spread or location of the borderline between the neoplasm and
non-neoplastic mucosa.
Extent of intraepithelial spread of the neoplasm was judged by referring to the following
mucosal findings continuous from the main tumor: (1) fine irregular papillary or granular
changes ([Fig. 3]); (2) fine protrusions with so-called fish egg-like appearance (i. e., like salmon
caviar) ([Fig. 4]); (3) vessels with irregularity in diameter ([Fig. 5]); and (4) a line demarcating the height of the mucosa ([Fig. 6]).
Fig. 3 Endoscopic images indicating fine irregular or granular changes. a Image obtained by CHF-B260. b Image obtained by SpyGlass DS. Arrow head, main tumor.
Fig. 4 Endoscopic images indicating fine protrusions with so-called fish egg-like appearance.
Both images are obtained by CHF-B260.
Fig. 5 Endoscopic images indicating vessels with an irregular diameter. a Image obtained by CHF-B260. b Image obtained by SpyGlass DS (arrow shows an irregular vessel).
Fig. 6 Endoscopic images indicating a demarcation line (arrow heads). Both images are obtained
by CHF-B260.
Resected specimens were evaluated by total segmentation and the extent of the cancer
was mapped on a macroscopic picture with segmenting lines. Then, preoperative diagnosis
by mapping biopsy plus POCS observation was evaluated as “accurately diagnosed,” “underdiagnosed”
(false negative), or “overdiagnosed” (false positive).
Sensitivity, specificity, and accuracy were compared between group A and group B.
Analytic methods
Categorical data were compared by Fisher’s exact test. Continual data were compared
by t tests. A P value < 0.05 was considered statistically significant. SPSS software ver. 24 (IBM
Japan, Tokyo, Japan) was used for all analyses.
Ethics
Written informed consent for endoscopic procedures was obtained before the procedure
from each patient. This retrospective study was approved by the institutional review
board of Sendai City Medical Center. The registration ID issued by UMIN was UMIN000030583.
All authors had access to the study data and reviewed and approved the final manuscript.
Results
Among 109 patients who underwent surgical resection for extrahepatic bile duct cancer
between January 2004 and December 2017, 56 patients for group A and 20 patients for
group B were eligible to be included in this study. No statistical differences between
the study groups were identified for any baseline characteristics ([Table 1]). Thirty-two patients in group A underwent POCS because of possible discrepancy
between ERC-guided mapping biopsy and other examinations ([Fig. 2]). The cholangioscope could be advanced beyond the tumor in all patients who required
the procedure.
Table 1
Baseline characteristics and lesion description of the patient groups.
|
Group A
n = 56
|
Group B
n = 20
|
P
|
|
Gender, Male:female
|
37:19
|
17:3
|
0.15
|
|
Age, yrs, mean ± SD
|
69 ± 8
|
72 ± 7
|
0.14
|
|
Macroscopic form
|
n.s.
|
|
|
16
|
8
|
|
|
|
35
|
11
|
|
|
|
5
|
1
|
|
|
Location of the main tumor
|
1.00
|
|
|
11
|
4
|
|
|
|
45
|
16
|
|
|
Intraepithelial spread (> 10 mm)
|
|
|
25
|
13
|
0.19
|
|
|
15
|
10
|
0.09
|
n.s., not significant
Final accuracy by overall preoperative examinations was 93 % for the liver side and
100 % for the ampullary side in group A, and 84 % for the liver side and 100 % for
the ampullary side in group B ([Table 2]). Of the four wrongly diagnosed patients in group A, one was overdiagnosed by POCS
because of fine granular changes derived from inflammatory hyperplasia after the biopsy
showed false-positive results due to contamination; one was overdiagnosed by POCS
because of inflammatory changes despite the true negative biopsy results; one was
underdiagnosed by POCS despite the true-positive biopsy results, which was wrongly
considered to be contamination; and one in whom POCS was not performed was underdiagnosed
by the biopsy findings because of sampling errors. The three wrongly diagnosed patients
in group B were underdiagnosed by false-negative results on both POCS and biopsy.
Although these three patients had long lateral extent toward the liver side, POCS
could not identify mucosal irregularity because the laterally spreading neoplasm resulted
in extremely low changes in height. All biopsy specimens obtained under POCS guidance
(6, 7, and 8 specimens for the three patients, respectively) were inappropriate because
they were too small or did not contain the epithelium. Overall, sensitivity and specificity
for liver-side estimation were 92 % and 93 %, respectively, in group A, and 70 % and
100 %, respectively, in group B. Sensitivity and specificity for ampullary-side estimation
were 100 % in both groups. No statistical differences in sensitivity or specificity
were detected between groups.
Table 2
Diagnostic accuracy of overall preoperative examinations.
|
Group A
n = 56
|
Group B
n = 20
|
P
|
|
Liver side
|
|
|
92 % (23/25)
|
70 % (7/10)
|
0.13
|
|
|
93 % (26/28)
|
100 % (12/12)
|
1.00
|
|
|
93 % (49/53)
|
84 % (16/19)
|
0.37
|
|
Ampullary side
|
|
|
100 % (6/6)
|
100 % (2/2)
|
N/A
|
|
|
100 % (8/8)
|
100 % (6/6)
|
N/A
|
|
|
100 % (14/14)
|
100 % (4/4)
|
N/A
|
Accuracy of ERC-guided mapping biopsy in group A was 80 % (44/51) for the liver side
and 92 % (12/13) for the ampullary side. Tumor spread toward the liver side was misdiagnosed
in seven patients due to contamination in five patients (false-positive), inappropriate
biopsy site (i. e., unintended upstream bifurcation of which the tumor spread did
not reach) in one patient (false-negative), and impossibility of advancing the biopsy
forceps beyond the obstruction in one patient (specimen was not obtained). The biopsy
results were false-positive in one patient in whom tumor spread toward the ampullary
side was misdiagnosed.
[Table 3] shows the accuracy of simple optical evaluation by POCS for diagnosing borderline(s)
of cancer in the two groups. When comparing groups, accuracy rates seemed lower in
group B, although statistical significance was not detected.
Table 3
Diagnostic accuracy of optical evaluation by POCS.
|
Group A
n = 32
|
Group B
n = 20
|
P
|
|
Liver side
|
|
|
88 % (15/17)[1]
|
58 % (7/12)
|
0.09
|
|
|
83 % (10/12)[1]
|
86 % (6/7)
|
1.00
|
|
|
83 % (25/30)[1]
|
68 % (13/19)
|
0.29
|
|
Ampullary side
|
|
|
100 % (5/5)
|
100 % (2/2)[1]
|
N/A
|
|
|
100 % (7/7)
|
100 % (5/5)[1]
|
N/A
|
|
|
100 % (12/12)
|
88 % (7/8)[1]
|
0.40
|
POCS, peroral cholangioscopy system
1 In one case for liver-side evaluation in group A and one case for ampullary-side
evaluation in group B, POCS observation could not be performed for technical reasons.
These cases were included into the accuracy calculation, but excluded from the sensitivity
and specificity calculation.
Discussion
It is difficult to accurately diagnose lateral intraepithelial spread of neoplasms
of the bile duct [1]
[2]. Cross-sectional examinations, including CT and MRI, are insufficient to confirm
such diagnosis due to the lower resolution of the imaging modalities. Although ultrasonography-based
evaluation with EUS and IDUS has been reported to be relatively useful [6]
[7], mucosal changes must be more precisely estimated by endoscopic optical observation,
similar to the estimation of gastrointestinal diseases [21].
The SpyGlass DS system became commercially available on February 2015 in the United
States (Boston Scientific Corporation) and on October 2015 in Japan (Boston Scientific
Japan K.K., Tokyo, Japan). This system provides digital image processing with a 120-degree
field of view, and spectacularly improved image quality in comparison with the previous
system from this corporation [22]. Moreover, this system has a dedicated channel for water irrigation, which enables
unimpeded procedures using a device such as biopsy forceps or an electronic hydraulic
lithotripsy probe without interruption for cleaning. However, this system was considered
to be possibly inappropriate for image diagnosis by endoscopic view because of its
relatively low image quality in comparison with CHF-B260. Operability was the advantage
in the Boston Scientific system, while the image quality was more notable in the Olympus
scope.
The results of this study indicate acceptable capability of the SpyGlass DS system
for endoscopic diagnosis, including POCS-guided biopsy, of the lateral spread. There
is similar diagnostic accuracy despite the different image quality due to the high
manipulability of the SpyGlass DS scope, which enables greater operator control in
visualization, with a four-way tip deflection (the CHF-B260 scope is equipped with
a two-way tip deflection). Moreover, the SpyGlass DS scope has a tapered tip that
enables easy insertion into the liver side through not only the papilla but also the
stricture, resulting in favorable diagnostic outcomes.
In contrast, the capability of optical diagnosis by POCS observation was lower with
SpyGlass DS than with CHF-B260 (accuracy rate: 69 % vs. 81 % for the liver side, 80 %
vs. 100 % for the ampullary side), which might be due to the lower image quality of
the SpyGlass system. Low-growing neoplastic change would be extremely difficult to
identify with SpyGlass DS image resolution.
In this feasibility study with a small sample size, overall diagnostic accuracy did
not significantly differ between groups A and B. Although the SpyGlass DS system would
not be sufficient when only observation is performed, an acceptable strategy involves
the addition of POCS-guided biopsy. The SpyGlass DS system could be considered a perfect
sheath, which enables exact biopsies with small concern for contamination.
However, the SpyBite biopsy forceps would not be sufficient to obtain adequate specimens.
These forceps appear well-developed with better capability to advance through an extremely
thin working channel and superior ability to cut bile duct tissues in comparison with
previous ultrathin forceps from other companies for through-the-POCS procedures. However,
there were two cases in which all obtained specimens were too small to evaluate. Further
improvement in these forceps is desirable.
The small sample size in the retrospective single-center setting is a limitation to
this study. The findings may be subject to Type II errors, which risk under-detection
of differences in small sample sizes. In studies with a larger number of samples,
differences in diagnostic outcomes might be detected. Although the current study of
20 samples per group is able to detect a 38 % difference between two groups with 80 %
power with a two-sided significance level of 0.05 by using the Fisher’s exact test,
274 samples are required for each group to detect a 10 % difference on equal terms.
Moreover, definitive POCS findings for lateral tumor spread have not been fully established,
which means that results might differ depending on the endoscopist or observer. However,
the impact of such inter-observer disparity might be diminished because the final
diagnosis based upon POCS and all other examinations was arrived at after sufficient
discussion during the institutional preoperative conference, which was composed of
pancreatobiliary experts in this study.
Conclusion
In conclusion, the SpyGlass DS system was found to be acceptable for diagnosis of
the lateral extent of extrahepatic cholangiocarcinoma when cholangioscopy-guided biopsy
was applied. Although further improvement in image quality and development of reliable
biopsy forceps are desirable, this system could be a standard approach to determine
resection lines in an era in which preoperative histological confirmation is increasingly
demanded.
