Introduction
Gastric cancer is the fifth most commonly diagnosed cancer and the third leading cause
of cancer-related mortality [1]. Despite a worldwide decrease in incidence and mortality, gastric cancer remains
quite prevalent in parts of Asia, Eastern Europe, and South America [1]. Therefore, early detection and diagnosis of gastric cancer remains a key element
for improved prognosis and survival outcomes.
Although the gold standard for diagnosis of gastric cancer is histopathological examination
[2], recent endoscopic technologies have been developed to optimize detection and diagnosis
of early-stage gastric cancer. Among these, magnifying endoscopy with narrow-band
imaging (NBI) and chromoendoscopy are more commonly applied whereas others such as
confocal endomicroscopy and endocytoscopy, although promising, are rarely used.
Endocytoscopy (EC) is an ultra-high magnification endoscopic technique that allows
microscopic visualization of the mucosal surface of gastrointestinal cells by using
intraprocedural stains [3]. First-generation EC was developed in 2003 [4]
[5]. Since then, several modifications and refinements allowed the development of fourth-generation
endocytoscopes that exhibit a magnification power of × 500, thus enabling in vivo
tissue observation, and visualization of individual cell nuclei [4]
[5].
To date, EC has been used for assessment of pharyngeal, esophageal, duodenal, and
colonic lesions. However, there are a limited number of studies examining gastric
lesions because the quality of EC images in the stomach is far from excellent due
to poor staining, resulting from presence of gastric secretions [6]. Bearing that in mind, the idea of performing a diagnostic accuracy study utilizing
fourth-generation EC while at the same time addressing these issues seemed quite appealing.
This study primarily aimed to evaluate the diagnostic performance and accuracy of
EC in the diagnosis of early gastric cancer.
Patients and methods
Study population
This is a retrospective analysis of data collected prospectively at a single tertiary
referral center, Showa University Koto Toyosu Hospital, between December 2015 and
October 2018. All individuals who underwent gastric EC examination and magnifying
endoscopy work-up before or after endoscopic submucosal dissection (ESD) were included.
Endocytoscopy technical considerations
EC examinations were carried out using a fourth-generation endocytoscope with a single
lens reflection, consecutive increase in magnification to × 500, observation range
of 570 × 500 μm, and an outer diameter of 9.7 mm (GIF-H290EC; Olympus Medical Systems
Corp., Tokyo, Japan). A black silicone cap (Distal Hood MAJ-1989; Olympus Medical
Systems Corp., Tokyo, Japan) was placed at the tip of the endocytoscope, secured by
a white tape in an oblique manner in order for the lens to come in contact with the
gastric mucosa.
Endocytoscopic assessment
All gastric EC examinations were performed under intravenous sedation using propofol
(200 mg/20 mL) by board-certified fellows of the Japan Gastroenterological Endoscopy
Society. All subjects received 40 mL of water-based solution also containing the mucolytic
agent pronase (20 000 U; Kaken Pharmaceutical, Tokyo, Japan) and the defoaming substance
dimethicone (Balgin 2 % Antifoaming Oral Solution; Kaigen Pharmaceutical, Tokyo, Japan)
per os prior to endoscopy. The suitable lesions were initially examined with white-light
endoscopy (WLE) and NBI, followed by EC using 0.05 % crystal violet and 0.1 % methylene
blue (CM) double staining. The staining solution is a mixture of 10 cc crystal violet
and 1 cc methylene blue. In two to three separate 10-cc syringes, 1 cc each is aspirated
with 9 cc of air ([Fig. 1]) which is then used for spraying multiple times until a satisfactory staining is
achieved. The average total procedure time was approximately 30 to 40 minutes, with
endocytoscopy taking around 15 to 20 minutes. All images were retrieved from the database
and an all-EC image catalog was created. EC images were assessed separately by two
expert endoscopists, with expertise in third-generation EC (> 50 × examinations),
blinded to patient’s clinical information, WLE and NBI findings, and histopathological
results. EC diagnosis was made by assessing cellular and glandular morphology, lumen
changes, loss of polarity of nuclei, and nuclear structural changes, and was based
on an adaptation of EC classification, originally created for colorectal lesions [7], for gastric mucosa ([Table1]). In brief, EC1 lesions are considered as non-neoplastic, EC2 as adenomas, and EC3
as carcinomas ([Fig. 2]). Hyperchromatic nuclei occupying the vast majority of the cell that are disarranged
and with heterogenous shape (large and elongated, large with rough edges) were detected
in the process and termed as the “enlarged nuclear sign” (ENS). The efficiency of
ENS in distinguishing between malignant and non-malignant lesions was also evaluated
by the two expert endoscopists. Biopsy specimens were available for all cases, however,
for gastric cancer, histopathologic diagnosis was made on ESD specimens. Accuracy
of expert EC-based diagnosis was assessed while using the gold standard histopathology
as a reference.
Fig. 1 CM double staining. a An 11-cc staining solution is prepared consisting of 10 cc of crystal violet and
1 cc of methylene blue. b In a separate 10-cc syringe, 1 cc of the solution is aspirated with 9 cc of air and
is used for spraying.
Table 1
EC classification of gastric mucosa.
|
Classification
|
Structure
|
Nuclei
|
|
EC1
(Non-neoplastic)
|
Smooth surfaces, soft edges, regular pit pattern, wide lumen
|
Regular pattern and arrangement of small uniformly sized, rounded nuclei with poor
staining
|
|
EC2
(Adenoma)
|
Dense slit-like lumen (lumen narrowing)
|
Regular pattern of small, uniformly sized, rounded nuclei with poor staining
|
|
EC3
(Cancer)
|
Well-differentiated adenocarcinoma
|
Irregular, rough, distorted gland formation, absence or fusion of lumen
|
Hyperchromatic, disarranged, swollen nuclei with heterogenous shape (large and elongated,
large with rough edges)
= “enlarged nuclear sign”
|
EC, endocytoscopy
Fig. 2 EC classification of gastric mucosa. Representative pictures differentiating EC1
(normal), EC2 (adenoma), and EC3 (carcinomas).
Statistical analysis
The Shapiro-Wilk test and distributional plots were used to assess normality. Continuous
variables are reported as mean ± standard deviation (SD) when normality was met. Categorical
variables are described as absolute and relative frequencies. Sensitivity, specificity,
and accuracy for diagnosis of early gastric cancer were calculated for each endoscopist.
The Kappa statistic was used to estimate the inter-observer agreement. Kappa values of ≤ 0.40, 0.41 – 0.75, and ≥ 0.75 were considered poor, good, and excellent,
respectively. The Wilson method was used to calculate 95 % confidence intervals. All
analyses were performed using STATA software version 14.1 (StataCorp. Texas, United
States).
Ethical considerations
The study protocol adhered to the principles of the Declaration of Helsinki and was
approved by the Ethics Committee of Showa University Koto Toyosu Hospital (IRB Registration
No: 17T7040). Written informed consent was obtained from all participants.
Results
A total of 43 patients (mean age ± SD, 72.6 ± 12.1 years, 31 (72 %) males) were enrolled.
Population characteristics are presented in [Table 2]. Based on histopathology, 23 were diagnosed with well-differentiated adenocarcinomas,
four with adenomas, and 16 with non-neoplastic lesions. The non-neoplastic lesions
comprised two hyperplastic polyps, two chronic gastritis, one hyperplastic mucosa,
and 11 normal gastric mucosa.
Table 2
Patient clinicopathological characteristics (n = 43).
|
Variable
|
Non-malignancy
(n = 20)
|
Malignancy
(n = 23)
|
|
Age (mean age ± SD)
|
69 ± 14.07
|
75.25 ± 9.44
|
|
Sex
|
|
|
14
|
17
|
|
|
6
|
6
|
|
Macroscopic Type
|
|
|
1
|
2
|
|
|
2
|
8
|
|
|
2
|
1
|
|
|
2
|
12
|
|
|
0
|
0
|
|
Location
|
|
|
6
|
13
|
|
|
2
|
8
|
|
|
1
|
2
|
|
Histopathology
|
|
|
16 (37.2 %)
|
N/A
|
|
|
4 (9.3 %)
|
N/A
|
|
|
N/A
|
23 (53.5 %)
|
SD, standard deviation; N/A, not applicable.
Based upon expert endoscopists’ ratings, high-quality images could be obtained in
81.4 % of cases. Comparison of EC diagnosis made by the two experts is presented in
[Table 3]. Based on these data, the sensitivity, specificity, and accuracy of EC for diagnosis
of early gastric cancer were calculated, as shown in [Table 4]. EC findings of 11 normal gastric mucosa showed regular glands, smooth surfaces,
and soft edges, well-preserved lumen, and small uniformly sized rounded nuclei with
poor staining ([Fig. 3]). No infiltrating cells, necrotic tissue, or debris were identified. Chronic gastritis
revealed more stained crypts and presence of infiltrating cells and debris, with regular
glands and well-preserved lumen. Hyperplastic polyps demonstrated wider, oval lumen
with small uniformly sized rounded nuclei.
Table 3
Comparison of expert EC-based diagnosis and histopathologic diagnosis.
|
Histopathologic diagnosis
|
Endocytoscopic diagnosis
|
|
Endoscopist A
(n = 43)
|
Endoscopist B
(n = 43)
|
|
EC1
|
EC2
|
EC3
|
EC1
|
EC2
|
EC3
|
|
Non-neoplastic
|
13
|
2
|
1
|
12
|
2
|
2
|
|
Adenoma
|
0
|
1
|
3
|
0
|
1
|
3
|
|
Cancer
|
0
|
3
|
20
|
0
|
2
|
21
|
EC, endocytoscopy.
Table 4
Diagnostic yields of endocytoscopy for early gastric cancer.
|
Test
|
Endoscopist A
|
Endoscopist B
|
|
Results
|
95 % CI
|
Results
|
95 % CI
|
|
Sensitivity
|
87.0 %
|
67.9 – 95.5
|
91.3 %
|
73.2 – 97.6
|
|
Specificity
|
80.0 %
|
58.4 – 91.9
|
75.0 %
|
53.1 – 88.8
|
|
Accuracy
|
83.7 %
|
70.0 – 91.9
|
83.7 %
|
70.0 – 91.9
|
CI, confidence interval.
Fig. 3 Normal gastric mucosa. a Endocytoscopic findings of normal gastric mucosa showing regular glands, smooth surfaces
and soft edges, well-preserved lumen, and small uniformly sized rounded nuclei with
poor staining. b Histologic appearance of normal gastric mucosa
Average sensitivity, specificity, and accuracy of EC for both endoscopists were 89.0 %,
77.5 %, and 83.7 %, respectively. Interobserver agreement, Kappa statistic = 0.71 (95 % CI: 0.50 – 0.93), was good.
ENS was observed in malignant lesions and its sensitivity, specificity, and accuracy
for diagnosis of early gastric cancer were calculated and presented in [Table 5]. Average sensitivity, specificity, and accuracy of ENS for both endoscopists were
84.8 %, 90.0 %, and 87.2 %, respectively. Interobserver agreement, Kappa statistic = 0.68 (95 % CI: 0.51 – 0.89), was good.
Table 5
Diagnostic yields of enlarged nuclear sign for early gastric cancer.
|
Test
|
Endoscopist A
|
Endoscopist B
|
|
Results
|
95 % CI
|
Results
|
95 % CI
|
|
Sensitivity
|
87.0 %
|
67.9 – 95.5
|
82.6 %
|
62.9 – 93.0
|
|
Specificity
|
95.0 %
|
76.4 – 99.1
|
85.0 %
|
64.0 – 94.8
|
|
Accuracy
|
90.7 %
|
78.4 – 96.3
|
83.7 %
|
70.0 – 91.9
|
CI, confidence interval.
Discussion
Diagnostic performance and accuracy of fourth-generation EC to discern early gastric
cancer among endoscopically identified lesions was evaluated in this study. When compared
to the gold-standard histopathology, fourth-generation EC showed satisfactory accuracy
in diagnosing early gastric cancer. Identification of ENS seems to further aid the
differentiation of gastric cancer from non-malignancy.
Although the concept of EC as a tool for gastric cancer diagnosis is not entirely
novel, the number of available studies on the subject is rather limited. The scarcity
of relevant data can be attributed to the challenges related to obtaining good-quality
staining as well as limited availability of the necessary dedicated endocytoscopes,
which are only available in a small number of centers worldwide. As the gastric mucosa
is exhibiting enhanced mucus secretory function, the abundance of mucus poses a challenge
for high-quality staining. In the current study, this issue was addressed by applying
the mucolytic pronase prior to the procedure as well as careful water-jet-assisted
mucosal rinsing using water plus dimethicone solution before CM double staining multiple
times through the scope channel. The combination of the cutting-edge fourth-generation
endocytoscope, which provides higher magnification, with CM to produce a staining
pattern that resembles the traditional hematoxylin-eosin (H&E) stain [8]
[9]
[10]
[11], allowed acquisition of high-quality images in over 80 % of cases in this study.
By obtaining images of higher quality, identification of tissue and cellular characteristics
was optimized, thus leading to more accurate detection of morphological changes associated
with cancer (distortion in gastric tissue architecture, cellular atypia, etc.) ([Fig. 4]). Based on presence or absence of these EC-identified changes, three distinct diagnostic
entities could be identified corresponding to traditional classification of lesions
as being non-neoplastic, neoplastic yet benign, or malignant. Through adaptation of
a previous EC classification that was originally used in colorectal lesions [7], a novel EC classification, this time for the gastric mucosa, was created.
Fig. 4 Well-differentiated adenocarcinoma. a Endocytoscopic findings of well-differentiated adenocarcinoma showing irregular and
distorted glands and the strongly stained “enlarged nuclear sign” (yellow arrows).
b Histologic appearance of well-differentiated adenocarcinoma.
The combination of favorable characteristics (images of high quality and magnification)
with EC classification of gastric mucosa led to good diagnostic accuracy and interobserver
agreement for distinguishing between gastric cancer and non-malignant cases in our
study. Interestingly, the results were comparable to those acquired through standard
histopathology with the exception of adenomas, for which a trend towards over-grading
was recorded. Safe conclusions regarding this potential over-grading of adenomas cannot
be drawn, mainly due to the very limited number of adenoma cases in this study.
Another major finding in our study was detection of disarranged, heterogenous-shaped,
hyperchromatic nuclei large enough to give the impression of “taking over” the entire
cell surface ([Fig. 5]). Although the nucleus:cytoplasm (N:C) ratio has been traditionally used during
histopathologic as well as endocytoscopic evaluation [2]
[6]
[12]
[13], the striking characteristic of nuclear “taking over” observed in the EC images
of gastric adenocarcinoma ([Fig. 6]) in this study encouraged its use as a distinct feature, and the term “enlarged
nuclear sign” was coined to describe it. Using ENS provided good diagnostic accuracy
and interobserver agreement for differentiation of gastric cancer from non-malignancy,
thus supporting its further use for this purpose.
Fig. 5 “Enlarged nuclear sign”. Representative pictures of the “enlarged nuclear sign” (yellow
arrows) by endocytoscopy a and by histopathology b depicting the striking characteristic of nuclear “taking over” the cell surface.
Fig. 6 Early gastric cancer case diagnosed using fourth-generation endocytoscopy. a, b A lesion in the antrum observed by white-light endoscopy. c Endocytoscopic findings revealed non-neoplastic glands of the surrounding mucosa,
d while the highly dysplastic area showed distortion of glands and presence of the
“enlarged nuclear sign” (yellow arrows).
At this point, however, certain study limitations must be acknowledged. To avoid bias
due to preceding WLE and NBI assessment, only still EC images were retrieved from
the database and assessed by the two expert endoscopists. Moreover, the sample size
is relatively small, suggesting that future studies on the same topic may necessitate
a larger number of patients to substantiate the results of this study. Non-neoplastic
lesions that mimic well-differentiated adenocarcinomas in histopathology were not
encountered in this study, therefore, EC assessment and differentiation between these
lesions and adenocarcinomas could not be determined. Finally, all gastric malignancy
cases in this study pertained to well-differentiated adenocarcinomas, therefore, no
conclusions can be drawn about EC assessment of poorly-differentiated adenocarcinomas.
Conclusions
In all, due to its good diagnostic accuracy and identification of the newly-recognized
“enlarged nuclear sign” in this study, fourth-generation EC seems to aid in diagnosis
of early gastric cancer, particularly well-differentiated adenocarcinomas, and deserves
further evaluation in future studies.
