MAGNIFYING ENDOSCOPY WITH NARROW BAND IMAGING FOR EARLY DIFFERENTIATED GASTRIC ADENOCARCINOMA

Abstract

We have been using magnifying endoscopy with narrow band imaging (NBI) to study early differentiated gastric adenocarcinomas and to assess the relationship between microvessel pattern, pit pattern and histological pattern. The magnified view of the cancerous area showed three types of pattern: (i) a mesh pattern, consisting of mesh-like connected microvessels; (ii) a loop pattern, consisting of loop-like microvessels that were not connected and had tubule-like or villus-like mucosal structures along them; and (iii) an interrupted pattern, consisting of interrupted thick or thin vessels without mucosal structures. The mesh type of microvascular pattern showed a round pit pattern in 88.9% of cases (32/36) and the loop type of microvascular pattern showed a non-round pit pattern in 100% of cases. Among lesions that showed a mesh pattern or a loop pattern, 94.9% (56/59) were mucosal cancer and 5.1% (3/59) were submucosal cancer. However, 92.3% (12/13) of lesions that showed an interrupted pattern were submucosal differentiated adenocarcinoma and 7.7% (1/13) were mucosal differentiated adenocarcinoma. The present findings provide basic data on the characteristics of mucosal differentiated gastric adenocarcinoma revealed by magnifying endoscopy with NBI, as well as invasive changes such as submucosal invasion.

Full text

Page 1

REVIEW
MAGNIFYING ENDOSCOPY WITH NARROW BAND IMAGING FOR EARLY
DIFFERENTIATED GASTRIC ADENOCARCINOMA
Kazuyoshi Yagi,1Atsuo Nakamura,1Atsuo Sekine1and Hajime Umezu2
1Department of Internal Medicine, Niigata Prefectural Yoshida Hospital and2Division of Pathology, Niigata University
Medical and Dental Hospital, Niigata, Japan
We have been using magnifying endoscopy with narrow band imaging (NBI) to study early differentiated gastric adeno-
carcinomas and to assess the relationship between microvessel pattern, pit pattern and histological pattern.The magnified
view of the cancerous area showed three types of pattern:(i) a mesh pattern,consisting of mesh-like connected microvessels;
(ii) a loop pattern, consisting of loop-like microvessels that were not connected and had tubule-like or villus-like mucosal
structures along them; and (iii) an interrupted pattern, consisting of interrupted thick or thin vessels without mucosal
structures.The mesh type of microvascular pattern showed a round pit pattern in 88.9% of cases (32/36) and the loop type
of microvascular pattern showed a non-round pit pattern in 100% of cases.Among lesions that showed a mesh pattern or
a loop pattern, 94.9% (56/59) were mucosal cancer and 5.1% (3/59) were submucosal cancer. However, 92.3% (12/13) of
lesions that showed an interrupted pattern were submucosal differentiated adenocarcinoma and 7.7% (1/13) were mucosal
differentiated adenocarcinoma. The present findings provide basic data on the characteristics of mucosal differentiated
gastric adenocarcinoma revealed by magnifying endoscopy with NBI, as well as invasive changes such as submucosal
invasion.
Key words: gastric cancer, interrupted pattern, magnifying endoscopy, narrow band imaging.
INTRODUCTION
Recently, there have been a number of reports describing
endoscopic diagnosis of lesions in the gastrointestinal tract
using the narrow band imaging (NBI) system.1–3NBI is a
unique sequential electronic endoscopy technique with an
ordinary lighting system, in which the wavelength ranges of
the new RGB filters are 485–515 nm for red (R),430–460 nm
for green (G), and 400–430 nm for blue (B).4This system, in
combination with a magnifying endoscope, can yield very
clear images of microvessels on the mucosal surface.1Details
of the magnified appearance of microvessels in gastric cancer
have been reported byTajiri et al.5andYao et al.6Magnifying
endoscopy with NBI has been reported to give more detailed
information on microvessels in gastric mucosa including
cancerous tissue.1,7–9Sumiyama et al. reported that magnified
views of differentiated adenocarcinoma showed a fine
network pattern in 66.1% of cases.7
We have been studying the magnified view of differenti-
ated gastric adenocarcinomas using NBI and the relationship
between microvessels pattern, pit pattern and histological
pattern. Here we report the characteristics of mucosal differ-
entiated gastric adenocarcinoma revealed by magnifying
endoscopy with NBI,including invasive changes such as sub-
mucosal invasion.
PATIENTS AND METHODS
Between October 2005 and November 2007, 105 patients
with early-stage gastric cancer were consecutively enrolled.
All the patients had been found to have early-stage gastric
differentiated adenocarcinoma by endoscopic examination
and biopsy.After conventional endoscopy, magnifying endo-
scopy with NBI was carried out to clarify the extent of the
cancer in order to prepare for endoscopic submucosal dissec-
tion (ESD) or surgical operation.The protocol was approved
by the ethics committee of our institution. All enrolled
patients gave their written informed consent to participate
in the study. Thirty-three patients among the 105 were not
chosen for this study because of insufficient quality of the
magnifying endoscopic imaging. The endoscopic pictures of
these patients did not have detailed information of microvas-
cular pattern.Therefore, the magnifying endoscopic and his-
tological findings for 72 patients were studied.
Endoscopic procedure
To dissolve the mucus layer of the stomach, each patient
ingested a solution containing 20 000 U pronase (Pronase
MS; Kaken Pharmaceutical Products Inc.,Tokyo, Japan) and
1 g NaHCO2 in 50 mL water, 10 min before endoscopy. The
procedure was carried out with a magnifying endoscope
(GIF-Q240Z or GIF-H260Z; Olympus Optical Co., Tokyo,
Japan). Gastric carcinomas were initially observed by con-
ventional endoscopy,followed by magnifying endoscopy with
NBI.The microvascular pattern of the cancerous mucosa and
the non-cancerous mucosa was observed by magnifying
endoscopy with NBI.
Correspondence: Kazuyoshi Yagi, Department of Internal Medicine,
Niigata-PrefecturalYoshida Hospital,Tsubame,Niigata 959-0242,Japan.
Email: yagikazu@pop12.odn.ne.jp
Received 13 January 2008; accepted 3 March 2008.
Digestive Endoscopy (2008) 20, 115–122doi:10.1111/j.1443-1661.2008.00788.x
© 2008 The Authors
© 2008 Japan Gastroenterological Endoscopy Society

Page 2

All of the magnifying endoscopic examinations and
recordings of all of the endoscopic findings were carried out
by a single endoscopist (KY). On all occasions, endoscopic
photographs were taken and recorded on digital videotape.
Examination of pit pattern of gastric differentiated
adenocarcinomas by magnifying endoscopy
after acetic acid sprinkling
All specimens resected by ESD were extended on boards
with pins and observed by magnifying endoscopy after acetic
acid sprinkling. Close attention was paid to the cancerous
mucosa, non-cancerous mucosa and the borderline between
them and several pictures were taken. To compare the
microvascular pattern of the cancerous area recorded using
NBI with the pit pattern of cancerous tissue on the resected
specimen,it was important to identify the same area that had
been observed by magnifying endoscopy using NBI. For
identification of the same area, the shape of the cancerous
margin was recorded initially in detail and then two endos-
copists (KY and AN) confirmed the area when the resected
specimen was later observed by magnifying endoscopy.
Histopathology
After pictures had been obtained by magnifying endoscopy,
the resected specimens were fixed in 20% formalin. The
lesions, together with the surrounding non-carcinomatous
mucosa, were cut into 2 mm-wide serial-step sections on the
specimens resected by ESD and 5 mm-wide serial-step sec-
tions on the specimens resected surgically; these dimensions
were determined by the endoscopist who performed the
study (KY) to correspond to the correct portion of the mag-
nified endoscopic images. The histopathological diagnosis
was done by a pathologist (HU) who was blinded to the
endoscopic findings.
Statistical analysis
The correspondence of each pit pattern with each microvas-
cular pattern was studied using a contingency table.When the
P-value was less than 0.05, the difference was considered
statistically significant.
RESULTS
Macroscopic features of the lesions
Ten of the 72 lesions were elevated, 27 were flat and 35 were
depressed.All of the lesions were diagnosed as differentiated
adenocarcinomas from biopsy specimens and resected
specimens.
Appearance of microvessels by magnifying
endoscopy using NBI
The magnified view of the cancerous area was divided into
three patterns: (1) a mesh pattern, consisting of mesh-like
connected microvessels;(2) a loop pattern,consisting of loop-
like microvessels that were not connected and had tubule-
like or villus-like mucosal structures along them; and (3) an
interrupted pattern, consisting of interrupted thick or thin
vessels without mucosal structures. Furthermore, the mesh
pattern was subdivided into three types: (1a) a complete
mesh pattern; (1b) an incomplete mesh pattern and (1c) an
irregular mesh pattern.
The complete mesh pattern consisted entirely of connected
microvessels with a mesh-like appearance (Fig. 1a) and was
similar to the fine network pattern reported by Nakayoshi
et al.1The incomplete mesh pattern consisted of microvessels
with an incomplete mesh-like appearance (Fig. 1b) and low
density resembling the microvascular pattern of adenoma.
The irregular mesh pattern basically showed a mesh-forming
microvessel pattern, but the microvessels were often broken
and irregular (Fig. 1c).All of the lesions with an incomplete
mesh pattern and a complete mesh pattern were mucosal-
depth cancers, although three lesions with irregular mesh
pattern were submucosal invasive cancers (Table 1).
The loop pattern showed loop-forming microvessels at the
tipsoftubule-likeorvillus-likemucosalstructureswhichwere
associated with branched pits. This pattern was subdivided
intothreetypesonthebasisofmucosalstructures,andnotthe
appearanceofthemicrovessels:(2a)thevilliformpattern;(2b)
the gyriform pattern;and (2c) the fusion pattern.
The villiform pattern consisted of small granule-like or
villus-likestructures containing
(Fig. 2a).The gyriform pattern (Fig. 2b) consisted of tubular
or gyriform structures surrounded by branched pits. Loop-
like microvessels appeared along these branched pits. The
fusion pattern showed fusion of the villiform pattern and
loop-like microvessels
a
b
c
Fig. 1.
arrows).
(a) Complete mesh pattern (white arrows). (b) Incomplete mesh pattern (white arrows). (c) Irregular mesh pattern (white
116K YAGI ET AL.
© 2008 The Authors
© 2008 Japan Gastroenterological Endoscopy Society

Page 3

gyriform pattern due to the absence of branched pits
(Fig. 2c).Loop-like microvessels appeared inside these fused
structures. All of the lesions with the loop pattern were
mucosal-depth cancers (Table 1).
The distribution of the lesions comprising each mesh
pattern and loop pattern subgroup is demonstrated in
Table 2.
The interrupted pattern showed broken and short irregular
microvessels, which differed from the mesh pattern or the
loop pattern, and the mucosal structure could not be imaged
from this microvessel pattern (Fig. 3a,b; see also Correlation
between microvascular pattern and pit pattern below).
Among the lesions that showed the interrupted pattern,
92.3% (12/13) were submucosal differentiated gastric cancers
(Table 1).
The illustrations of a mesh pattern, a loop pattern and an
interrupted pattern are shown (Fig. 3c–e).
Correlation between microvascular pattern and pit pattern
The pit patterns revealed by acetic acid sprinkling in resected
specimensweredivisibleintotwotypes;roundandnon-round
(branch-like pit). The round pit pattern was subdivided into
three patterns; small round pits (Fig. 4a), oval round pits
(Fig. 4b) and long round pits (Fig. 4c). The non-round pit
pattern was subdivided into granular and tubular types
accordingtothemucosalstructuresurroundedbythebranch-
like pits. The granular pattern showed granular elevations
surrounding branched pits (Fig. 4d), whereas the tubular
pattern showed tubular elevations surrounding branched pits
(Fig. 4e).Among the lesions with a mesh-type microvascular
pattern,88.9% (32/36) showed a round pit pattern,and 100%
of the lesions with a loop-type microvascular pattern showed
a non-round pit pattern (Table 3). The two cancerous pit
patterns differed in their microvascular patterns (P < 0.001).
This result suggests that mesh-pattern microvascular vessels
surround cancerous ducts with round pits.
Correlation between microvascular pattern and
histological findings including of depth of cancer
Most cases with the complete mesh pattern (Fig. 5a) showed
a round pit pattern (Fig. 5b) and,histologically,showed short
and straight cancerous ducts (Fig. 5c). Furthermore, the his-
tological type in all cases showing the complete mesh pattern
was well-differentiated adenocarcinoma. Cases with the
a
bc
Fig. 2.
(a) Villiform pattern (yellow arrows). (b) Gyriform-like pattern (yellow arrows). (c) Fusion pattern (yellow arrows).
Table 1.
Depth of cancerous lesion for each microvascular pattern
DepthMesh patternLoop pattern Interrupted patternTotal
Mucosal cancer
Sm1 (submucosal scanty invasion)
Sm2–3 (submucosal massive invasion)
33
1†
2†
23
0
0
1
1
57
2
1311
Total 36 2313 72
†Three cases with submucosal invasion showed an irregular mesh pattern.
Table 2.
Distribution of the lesions showing the mesh pattern and loop pattern
Mesh patternLoop pattern
Complete
mesh pattern
Incomplete
mesh pattern
Irregular
mesh pattern
Villiform
pattern
Gyriform
pattern
Fused
pattern
Total
19 (0)6 (0)
36
11 (3)†
12 (0)8 (0)
26
3 (0)
( ) number of submucosal cancers.
†Two lesions with an irregular mesh pattern were massive submucosal cancers and one lesion was a scanty submucosal cancer. All of the lesions
except those with an irregular mesh were mucosal cancers.
MAGNIFYING ENDOSCOPY WITH NBI 117
© 2008 The Authors
© 2008 Japan Gastroenterological Endoscopy Society

Page 4

incomplete mesh pattern also showed short and straight can-
cerous ducts, which were diagnosed as well-differentiated
adenocarcinoma. However, 50% (3/6) of the cases with an
incomplete mesh pattern had been diagnosed as adenoma by
biopsy a few years ago.
Among cases with an irregular mesh pattern, 90.9% (10/
11) were moderately differentiated adenocarcinomas. Fur-
thermore, two cases with an irregular mesh pattern showed
massive submucosal invasion and one case showed scanty
submucosal invasion (Table 1).
The histological hallmark of the villiform pattern was can-
cerous tissue with club-shaped stroma (Fig. 5d),whereas that
of the gyriform pattern was branched and had long cancerous
glands (Fig. 5e).
Among the lesions showing an interrupted pattern, 92.3%
(12/13)weresubmucosaldifferentiatedadenocarcinomasand
7.7% (1/13) were mucosal differentiated adenocarcinomas.
All of the cases showing an interrupted pattern were inva-
sive differentiated adenocarcinomas, including intramucosal
cancer. In cases of submucosal invasive cancer, submucosal
a
b
cd
e
Fig. 3.
Interrupted pattern. (c) Illustration of a
mesh pattern. (d) Illustration of a loop
pattern.(e) Illustration of an interrupted
pattern.
(a) IIc lesion on the antrum.(b)
118K YAGI ET AL.
© 2008 The Authors
© 2008 Japan Gastroenterological Endoscopy Society

Page 5

invasion was found approximately in the sites of interrupted
pattern. Fig. 6a shows the endoscopic appearance of cardiac
cancer, and it was difficult to diagnose submucosal invasive
cancer from conventional endoscopic findings.The magnify-
ingendoscopicviewofFig. 6ashowedtheinterruptedpattern
(Fig. 6b). Histological examination of the resected specimen
showedsubmucosalinvasivecancer(Fig.6c).Fig. 6dshowsthe
endoscopic appearance of a IIc lesion in the antrum. The
magnifying endoscopic view in Fig. 6d shows an interrupted
pattern (Fig. 6e). Although histological examination of the
resected specimen showed mucosal cancer, the cancerous
tissue showed intramucosal invasive cancer (Fig. 6f).
a
b
c
d
e
Fig. 4.
Oval round pit pattern. (c) Long round
pit pattern. (d) Granular pattern. (e)
Tubular pattern.
(a) Small round pit pattern. (b)
Table 3.
Relationship between microvascular pattern and pit pattern
Pit patternMesh pattern Loop patternTotal
Round pit pattern
Non-round pit pattern
(Branch-like pit pattern)
Impossible to classify
320 32
25223
202
Total 3623 59
Contingency table: P < 0.001.
MAGNIFYING ENDOSCOPY WITH NBI 119
© 2008 The Authors
© 2008 Japan Gastroenterological Endoscopy Society