Introduction
The incidence of esophagogastric junction (EGJ) adenocarcinoma has increased during
recent decades, especially in Western countries. This is mainly related to the increasing
prevalence of gastroesophageal reflux disease and obesity, and the decreasing prevalence
of Helicobacter pylori infection. The incidence of EGJ adenocarcinoma is also increasing in Eastern countries,
including Japan, and is expected to continue doing so for similar reasons. Endoscopic
submucosal dissection (ESD) is a minimally invasive and curative treatment for superficial
gastrointestinal cancer, as it provides a high rate of en bloc resection, regardless
of lesion size, and facilitates accurate histopathological diagnosis [1]. Several retrospective studies have recently indicated that ESD is safe and effective
for EGJ adenocarcinomas [2]
[3]
[4]
[5]
[6].
EGJ adenocarcinomas can extend laterally and invade the neighboring normal esophageal
squamous epithelium; this is known as subsquamous tumor extension (STE) [7]. Because STE may not be observable at the esophageal epithelium, it is difficult
to accurately define the proximal tumor margin during ESD for EGJ adenocarcinomas;
this can lead to postoperative positive lateral margins [3]
[5]
[6]. In a previous study, we retrospectively showed that an extra 1-cm safety margin
achieved 100 % negative lateral margins during ESD for EGJ adenocarcinomas, even when
the STE was < 7 mm [4].
In response to the lack of prospectively collected data, the present prospective feasibility
study aimed to evaluate whether considering an extra 1-cm safety margin was effective
in preventing positive lateral margins during ESD for EGJ adenocarcinomas.
Patients and methods
Patients
The protocol for this single-center prospective feasibility study was approved by
the ethics committee of the Osaka City University Hospital (approval: 3064) and was
registered in the University Hospital Medical Network Clinical Trial Registry (UMIN000017120).
All patients provided written informed consent before enrollment. The inclusion criteria
comprised: (1) an EGJ tumor diagnosed as Siewert’s type II classification; (2) pathological
diagnosis of definite or suspected adenocarcinoma; (3) cT1a-M (invasion limited to
the muscularis mucosae); (4) no signs of lymph node or distant metastasis; (5) age
> 20 years; and (6) provided written informed consent. The exclusion criteria comprised:
(1) no informed consent; (2) possibility of pregnancy or lactation; (3) severe mental
illness; (4) continuous systemic steroid use; (5) active bacterial or fungal infection;
(6) cardiac infarction or unstable angina pectoris within the last 3 months; (7) uncontrollable
hypertension or diabetes mellitus; (8) pulmonary disease requiring oxygen; or (9)
judged to be ineligible at the discretion of a study investigator. The sample size
was, based on our previous experience, calculated as per the number of patients that
we could feasibly enroll during the study period.
Outcome measures
The primary outcome of interest was the rate of complete resection (R0 resection rate),
i. e., an en bloc resection with histologically cancer-free margins. The secondary
outcomes of interest were the treatment outcomes (curative resection and adverse events
[AEs]) and the incidence, extension length, and preoperative diagnostic rate of the
STE. Preoperative diagnostic rate was defined as accuracy rate. We determined that
STE could be diagnosed if the presence and length of STE were consistent with preoperative
endoscopic diagnosis.
Defining STE and the 1-cm safety margin
Preoperative endoscopy was performed using an Evis Lucera Elite System (Olympus, Tokyo,
Japan) with a magnifying upper gastrointestinal endoscope (GIF-H260Z; Olympus). We
delineated the lesion’s area via magnifying endoscopy with narrow-band imaging (NBI)
([Fig. 1a]). In all cases, the presence of STE was determined on the basis of the following
modalities: 1) white light endoscopy (WLE) for visualizing a slight elevation ([Fig. 1b]) [4]; 2) non-magnifying endoscopy with NBI for visualizing a pale brownish area ([Fig. 1c]) [5]; 3) magnifying endoscopy with NBI for visualizing an irregular microvascular pattern
under the normal squamous epithelium ([Fig. 1d]) [7]; and 4) magnifying endoscopy with acetic acid application for visualizing a small
white sign ([Fig. 1e]) [8].
Fig. 1 Endoscopic sign of EGJ adenocarcinomas and placement of the 1-cm safety margin. a Demarcation line with an irregular microvascular or microsurface pattern during magnifying
endoscopy with NBI. b Slight elevation (red arrow) during white light endoscopy. c Pale brownish area during non-magnifying endoscopy with NBI. d Irregular microvascular pattern under the normal squamous epithelium during magnifying
endoscopy with NBI. e Small white sign (orange arrow) during magnifying endoscopy with acetic acid application.
f One-cm marking on the tip of the device’s sheath. g Measurement of the 1-cm safety margin using the device. h Marking dots around the lesion. EGJ, esophagogastric junction; NBI, narrow-band imaging
The marking dots were generally positioned at least 2 mm outside the tumor margin.
However, these dots were placed 1cm away from the oral side of any suspected STE or
1 cm away from the squamous-columnar junction in cases without endoscopic signs of
STE. The 1-cm safety margin was measured using a 1-cm marking on the tip of the device’s
sheath ([Fig. 1f], [Fig. 1g], [Fig. 1h]). ESD was performed as previously reported [4]. A flush knife (Flush knife, DK2620JN; Fujifilm Medical, Tokyo, Japan) was used
as the main electrosurgical knife.
Pathological examination
The fresh and post-formalin fixation specimens were photographed with a scale. The
resected specimens were cut into 2-mm slices from the oral side to the anal side,
and each slice was evaluated to determine the histological type, size, depth of invasion,
lateral and vertical margins, and lymphovascular invasion status of the EGJ tumor.
In cases with STE, the length of the extension was determined with a scale using microscopy,
and the lesion was then mapped on the photographs of the fresh specimen. The length
of the STE was based on the photograph of the fresh specimen. In the same case as
described in [Fig. 1], an EGJ adenocarcinoma arising from the Barrett’s esophagus was defined as a Barrett’s
adenocarcinoma and was identified from the presence of esophageal glands at the anal
side of the carcinoma ([Fig. 2]).
Fig. 2 Calculation of STE length. a Length of the STE as calculated using microscopy with a scale. The red arrow depicts
the range of the lesion, and the blue arrow depicts the range of STE. b,c STE. d Length calculated by mapping the lesion on the photographs of the fresh specimen
using the aforementioned reference length. STE, subsquamous tumor extension
Gastroesophageal reflux disease or proton pump inhibitor (PPI) use sometimes makes
it difficult to assume that double muscularis mucosa and columnar epithelial island.
Thus, in this study, esophageal glands on the anal side of the lesion were considered
as an accurate definition of Barrett’s adenocarcinoma. All adenocarcinomas that were
not considered as Barrett’s adenocarcinomas were classified as non-Barrett’s adenocarcinomas.
Definitions
PPI intake was defined as taking PPI before endoscopy. It was defined as Helicobacter pylori infection, if any following test was positive, the urea breath test, serum anti-Helicobacter pylori body titer, and stool antigen test. The presence of circular Barrett mucosa extending
longitudinally for 3 cm or more was defined as long-segment Barrett’s esophagus (LSBE),
and the others were defined as short-segment Barrett’s esophagus (SSBE) [9]. Gastric atrophy was definition based on Kimura-Takemoto Classification. Mild atrophy
was defined as C-1 and C-2, moderate was as C-3 and O-1, severe was O-2 and O-3, and
no was as C0.
An en bloc resection was defined as the resection of a single specimen that included
all the marking dots. Complete resection was defined as an en bloc resection with
histologically cancer-free margins. Curative resection of a Barrett’s adenocarcinoma
was defined as the complete resection of a differentiated adenocarcinoma with an invasion
depth of T1a-DMM (deep muscularis mucosa) and no lymphovascular invasion, as these
patients have a very low risk of metastasis [10]. Curative resection of a non-Barrett’s adenocarcinoma was defined as the complete
resection of an adenocarcinoma with an invasion depth of < 500 µm from the muscularis
mucosa, no lymphovascular invasion, and a poorly differentiated component [11].
Perforation was defined as an endoscopically visible hole in the esophageal or gastric
wall that exposed the mediastinal or peritoneal cavity. Delayed bleeding was defined
as bleeding with hematemesis or melena that required endoscopic reintervention or
transfusion after the ESD procedure. Esophageal stricture was defined as diameter
reduction that prevented the passage of a standard 9.2-mm endoscope at the follow-up
endoscopic examination (1 to 2 months after ESD).
Statistical analyses
Continuous variables were reported as median (range) and analyzed using the Mann-Whitney
U test, categorical variables were reported as number (percentage) and analyzed using
Fisher’s exact test. Differences were considered statistically significant at P < 0.05. All statistical analyses were performed using EZR software (version 1.35;
Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical
user interface for R software (The R Foundation for Statistical Computing, Vienna,
Austria).
Results
Clinical characteristics
Fourteen patients presented with a suspected EGJ adenocarcinoma between September
2015 and December 2018. One patient with histologically no cancer and another in whom
an evaluation for STE was impossible, because the specimen was not cut from the oral
side to the anal side, were excluded. Thus, 12 patients ([Table 1]), five of whom were histologically diagnosed with a Barrett’s adenocarcinoma, were
enrolled in this study.
Table 1
Patient characteristics
|
Total
|
Barrett's adenocarcinoma
|
Non-Barrett’s adenocarcinoma
|
|
n = 12
|
n = 5
|
n = 7
|
|
Age, years, median (range)
|
64 (50 – 80)
|
63 (50 – 76)
|
64 (60 – 80)
|
|
Sex, male/ female, n
|
9/3
|
5/0
|
4/3
|
|
ASA PS, 1/2/3, n
|
2/8/2
|
5/0/0
|
2/3/2
|
|
Medication, PPI, n
|
5
|
1
|
4
|
|
Helicobacter pylori infection[1], n
|
3
|
0
|
3
|
|
SSBE/LSBE, n
|
8/0
|
5/0
|
3/0
|
|
Gastric atrophy, no/mild/moderate, n
|
6/2/4
|
5/0/0
|
1/2/4
|
|
Tumor size, mm, median (range)
|
28 (5.0 – 53)
|
28 (13.0 – 37)
|
28(5.0 – 53)
|
|
Circumference, %, median (range)
|
29.2 (8.3 – 91.7)
|
16.7 (8.3 – 58.3)
|
33.3 (8.3 – 91.7)
|
|
Macroscopic appearance, elevated/flat/depressed, n
|
4/3/5
|
1/1/3
|
3/2/2
|
|
Depth of cancer, mucosa/submucosa, n
|
8/4
|
4/1
|
4/3
|
ASA PS, American Society of Anesthesiologists Physical Status; PPI, proton pump inhibitor;
SSBE, short-segment Barrett’s epithelium; LSBE, long-segment Barrett’s epithelium.
1 Status of Helicobacter pylori in two cases was not investigated.
Primary endpoints
The complete resection rate was 91.7 % (95 % confidence interval [CI]: 62.5–100.6)
(11/12) ([Table 2]). The remaining patient with a positive vertical margin had an EGJ adenocarcinoma
with massive submucosal invasion. The rate of negative lateral margins was 100 % (95 %
CI: 71.8–103.9), and all lesions were laterally included within the 1-cm safety margin.
Table 2
Primary and secondary outcomes.
|
Total
|
Barrett's adenocarcinoma
|
Non-Barrett’s adenocarcinoma
|
|
n = 12
|
n = 5
|
n = 7
|
|
Result of resection, n (%)
|
|
|
12 (100)
|
5 (100)
|
7 (100)
|
|
|
11 (91.7)
|
5 (100)
|
6 (85.7)
|
|
|
12 (100)
|
5 (100)
|
7 (100)
|
|
|
11 (91.7)
|
5 (100)
|
6 (85.7)
|
|
|
8 (66.7)
|
4 (80.0)
|
4 (57.2)
|
|
Adverse events of ESD procedure, n (%)
|
|
|
2 (16.7)
|
0 (0)
|
2 (28.6)
|
|
|
0 (0)
|
0 (0)
|
0 (0)
|
|
|
0 (0)
|
0 (0)
|
0 (0)
|
|
Subsquamous tumor extension (STE)
|
|
|
9 (75.0)
|
4 (80.0)
|
5 (71.4)
|
|
|
8 (66.7)
|
5 (100)
|
3 (42.9)
|
|
|
5.7 (1.0–24.8)
|
6.2 (5.7–21)
|
3.6 (1.0–24.8)
|
ESD, endoscopic submucosal dissection; STE, subsquamous tumor extension.
Presence of STE
STE was observed in 75 % of patients (9/12; [Table 2]). The median length of the STE was 5.7 mm (range: 1–24.8 mm; [Table 2]). The preoperative diagnostic rates for the four modalities are shown in [Fig. 3]. Magnifying endoscopy with acetic acid application identified STE in 100 % (5/5)
of patients with a Barrett’s adenocarcinoma but in only 28.6 % (2/7) of patients with
a non-Barrett’s adenocarcinoma. The preoperative diagnostic rates for STE of EGJ adenocarcinomas
were 66.7 % using all four modalities, 41.7 % using WLE, 58.3 % using magnifying endoscopy
with NBI, and 58.3 % using magnifying endoscopy with acetic acid application. Four
patients were misdiagnosed during preoperative endoscopy ([Fig. 4]), all with non-Barret’s adenocarcinoma. Sensitivity and specificity rates for STE
of EGJ adenocarcinomas were both 66.7 % using all four modalities ([Table 3]). Histological type was well-differentiated adenocarcinoma in three patients and
moderately in one patient. In all cases, invasion depth of STE was the lamina propria
mucosa.
Fig. 3 Preoperative diagnosis rates for STE using various modalities. Preoperative diagnosis
was defined as accuracy. EGJ, esophagogastric junction; NBI, narrow-band imaging;
STE, subsquamous tumor extension
Fig. 4 Four misdiagnosed cases of STE. The white bar shows preoperative diagnosis, while
the gray bar shows pathological result. The black dots show oral side marking. No
findings were detected in two cases during preoperative endoscopy. The final case
was expected to have 16-mm submucosal tumor extension; however, the pathological extension
was 24.8 mm. SCJ, squamous-columnar junction; STE, subsquamous tumor extension
Table 3
Diagnostic performance (accuracy [preoperative diagnosis], sensitivity, specificity,
PPV and NPV) of STE.
|
Accuracy (%) (95 % CI)
|
Sensitivity (%) (95 % CI)
|
Specificity (%) (95 % CI)
|
PPV (%) (95 % CI)
|
NPV (%) (95 % CI)
|
|
Diagnostic performance using all four modalities
|
|
|
66.7 (45.1–80.0)
|
66.7 (52.3–75.6)
|
66.7 (23.6–93.4)
|
85.7 (67.2–97.2)
|
40.0 (14.1–56.0)
|
|
|
100 (70.2–100)
|
100 (81.4–100)
|
100 (25.5–100)
|
100 (81.4–100)
|
100 (25.5–100)
|
|
|
42.9 (20.5–65.4)
|
40.0 (24.3–55.8)
|
50.0 (10.8–89.5)
|
66.7 (40.6–93.0)
|
25.0 (5.4–44.8)
|
|
Diagnostic performance using each modality of total
|
|
|
41.7 (20.9–55.1)
|
33.3 (19.5–42.3)
|
66.7 (23.6–93.4)
|
85.7 (67.2–97.2)
|
40.0 (14.1–56.0)
|
|
|
58.3 (36.8–71.7)
|
55.6 (41.2–64.5)
|
66.7 (23.6–93.5)
|
83.3 (61.8–96.7)
|
33.3 (11.8–46.7)
|
|
|
58.3 (36.8–71.7)
|
55.6 (41.2–64.5)
|
66.7 (23.6–93.5)
|
83.3 (61.8–96.7)
|
33.3 (11.8–46.7)
|
|
|
58.3 (36.8–71.7)
|
55.6 (41.2–64.5)
|
66.7 (23.6–93.5)
|
83.3 (61.8–96.7)
|
33.3 (11.8–46.7)
|
CI, confidence interval; EGJ, esophagogastric junction; NBI, narrow-band imaging;
NPV, negative predictive value; PPV, positive predictive value; STE, subsquamous tumor
extension
Clinical outcomes
The curative resection rate was 66.7 % (8/12). Two patients had an esophageal stricture.
There was no bleeding or perforation in any of the patients ([Table 2]).
Discussion
This prospective feasibility study revealed that an extra 1-cm safety margin is safe
and helpful for achieving negative lateral margins during an ESD for EGJ adenocarcinomas.
Despite the difficulty in accurately diagnosing STE preoperatively, all lesions were
included within the 1-cm margin. To the best of our knowledge, this is the first prospective
study in this setting.
Our findings agree with the previously reported R0 resection rates of 79 % to 89.8 %
and AE rates of 3.9 % to 27.3 % after an ESD for EGJ adenocarcinomas [2]
[3]
[5]
[6]. However, these studies included cases with positive lateral margins due to STE
(7.5%–18.0 % of cases) [2]
[3]
[5]
[6]. A multicenter retrospective study also revealed a positive lateral margin rate
of 7.5 % (24/321 patients) [6]. These patients may require additional surgery and may have an increased risk of
local recurrence [5]
[8]. Because positive lateral margin has often been reported only on the oral side,
we considered it important to address such an incomplete resection. Therefore, we
evaluated the STE of the oral side, and we did not identify any patients with positive
lateral margins after we included an extra 1-cm safety margin during ESD for EGJ adenocarcinomas,
and all lesions were included within this 1-cm margin, regardless of the STE status.
Magnifying endoscopy with NBI and acetic acid application is reportedly useful for
diagnosing STE [7]
[8]
[10]. A questionnaire survey revealed that the rates of STE diagnosis were 55 % (97/175
patients) with WLE and 60 % (42/70 cases) with magnifying endoscopy with NBI [8]. In another study, review of endoscopic images after histological examination of
ESD specimens revealed STE diagnostic rates of 50 % (5/10 patients) with WLE, 43 %
(3/7 patients) with magnifying endoscopy and NBI, and 100 % (6/6 patients) with magnifying
endoscopy and acetic acid application [8].
The diagnostic rates in the present study were similar to those reported for WLE and
magnifying endoscopy with NBI previously; however, the diagnostic rate for acetic
acid application was lower than that in a previous report. This might be caused by
the difference in the method of diagnosis. In a previous report, the endoscopic images
of STE were reviewed retrospectively only in cases that yielded pathological results
after ESD [8]. This study revealed that magnifying endoscopy with acetic acid application had
higher accuracy in diagnosing STE of Barrett’s adenocarcinomas than that of non-Barrett’s
adenocarcinomas, with the shorter length of STE of non-Barrett’s adenocarcinoma being
a possible explanation for the reduced accuracy. The median lengths of STE of Barrett’s
adenocarcinoma and non-Barrett’s adenocarcinoma were 6.2 (5.7–21) mm and 3.6 (1.0–24.8)
mm, and in two patients with non-Barrett’s adenocarcinoma, the lengths of STEs were
not detected endoscopically. In addition, because PPI use and preceding biopsy are
known to occasionally result in non-cancerous squamous epithelium covering the existing
tumor [12], they might contribute to a difference in diagnosis.
Nonetheless, our study is unique in that we prospectively applied the four imaging
modalities in all patients, which revealed that magnifying endoscopy with acetic acid
application was more accurate for diagnosing STE of Barrett’s adenocarcinomas than
for diagnosing the STE of non-Barrett’s adenocarcinomas.
This study has several strengths, including its prospective design and the use of
all four imaging modalities for all patients, which allowed us to calculate each modality’s
diagnostic ability for STE. Another strength includes the accurate measurement of
the 1-cm margin using a mark that was placed 1cm from the tip of the device’s sheath.
Finally, it is notable that the length of the STE was effectively evaluated using
a photograph of the fresh specimen, based on the positional relationship between the
STE and the 1-cm safety margin.
The present study also has several limitations, the first being its small single-center
design; hence, there is a need for larger multicenter studies that will validate our
findings. Secondary, we did not evaluate long-term outcomes. Further well-designed
studies are needed to address these issues.
Conclusions
Our findings suggest that an extra 1-cm safety margin is useful for achieving complete
resection of EGJ adenocarcinomas during ESD, warranting validation in a large cohort
study.
