ORIGINAL ARTICLE: Clinical Endoscopy
Usefulness of magnifying endoscopy with narrow band imaging for the detection of specialized intestinal metaplasia in columnar-lined esophagus and Barrett’s adenocarcinoma Ken-ichi Goda, MD, Hisao Tajiri, MD, Masahiro Ikegami, MD, Mitsuyoshi Urashima, MD, Takashi Nakayoshi, MD, Mitsuru Kaise, MD Tokyo, Japan
Background: Barrett’s esophagus with specialized intestinal metaplasia (SIM) from columnar-lined esophagus is difficult to distinguish with routine endoscopy. Objective: To examine the values of fine mucosal patterns and the capillary patterns observed by magnifying endoscopy with narrow band imaging (MENBI) for the detection of SIM in columnar-lined esophagus and superficial Barrett’s adenocarcinoma. We also undertook a histologic investigation regarding whether the capillary pattern observed by MENBI corresponds to the 3-dimensional (3D) structure of the capillary as depicted by using a confocal laser-scanning microscope (CLSM). Design: To compare the findings of MENBI, at 217 sites of columnar-lined esophagus, with histologic findings. Capillaries of the superficial mucosal layer were observed and were analyzed by 3D with a CLSM in 45 biopsied specimens. Patients: Fifty-eight patients, including 4 with superficial Barrett’s adenocarcinoma. Setting: Jikei University Hospital, Tokyo, Japan. Results: Upon observation, all 6 adenocarcinoma sites were classified as irregular patterns in both the fine mucosal patterns and capillary patterns. The most characteristic endoscopic patterns of SIM were revealed to be the cerebriform fine mucosal pattern (sensitivity, 56%; specificity, 79%; odds ratio, 4.78) and ivy- or deoxyribonucleic acid (DNA)-like capillary pattern (sensitivity, 77%; specificity, 94%; odds ratio, 51.6). The addition of capillary patterns to fine mucosal patterns improved the accuracy of diagnosing SIM (P ! .0001). Conclusions: MENBI was able to precisely visualize the structure of capillaries in the superficial mucosal layer. The addition of capillary patterns to fine mucosal patterns appeared to improve the diagnostic value for detecting SIM and superficial Barrett’s adenocarcinoma upon observation by MENBI. (Gastrointest Endosc 2007;65:36-46.)
Among the malignant esophageal tumors, the incidence of Barrett’s adenocarcinoma has been steadily increasing in Western countries. In Japan, squamous-cell carcinoma accounts for at least 90% of malignant esophageal tumors.1 Particularly in white men, the proportion of Barrett’s adenocarcinoma in all cases of esophageal cancer reportedly exceeds 50%.2 It has been believed that the main cause of the high percentage of Barrett’s adenocarcinoma is the increased incidence of Barrett’s esophagus, which shows a close causal relation with reflux esophagiCopyright ª 2007 by the American Society for Gastrointestinal Endoscopy 0016-5107/$32.00 doi:10.1016/j.gie.2006.03.938
36 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
tis, the incidence of which has rapidly increased in the last 20 to 30 years. However, the incidence of reflux esophagitis is increasing dramatically in Japan, because of the aging of society, with an increasing elderly population and decreased frequency of infection with Helicobacter pylori in younger individuals.3,4 Consequently, there has been apprehension about the possibility that the incidence of Barrett’s esophagus and Barrett’s adenocarcinoma will increase in Japan in the future. Columnar-lined esophagus and Barrett’s esophagus are the states in which stratified squamous epithelium is continuously replaced by columnar epithelium from the esophagogastric junction. The columnar-lined esophagus is histologically classified into (1) gastric-fundic type (fundic www.giejournal.org
Goda et al
type), (2) junctional type (cardiac type), and (3) specialized columnar epithelium or specialized intestinal epithelium (SIM).5 SIM is similar to incomplete intestinal metaplasia of the gastric mucosa,6 and specific attention has been paid to SIM as the origin of Barrett’s adenocarcinoma.7 Because SIM and dysplasia cannot be readily differentiated by routine endoscopic observation, a stepwise, 4-quadrant biopsy procedure (biopsies in 4 directions randomly selected at 1- to 2-cm intervals) has been recommended in Western countries.8 However, the use of this procedure, which results in sampling errors and relatively high cost, has been intensely debated for so long because the number of biopsied sites is large. Since 1994, when first reported by Stevens et al,9 attempts have been made to elucidate the endoscopic findings characteristic of SIM by combining a dye method, ie, indigo carmine solution or methylene blue staining, and an acetic acid method with magnifying endoscopy10-14; and favorable results have been obtained. However, magnifying endoscopy with these dyes and acetic acid methods are bound to be influenced by staining and by the degree of color change. There is also the risk of procedural complications with these combined methods. With methylene blue staining, the binding of deoxyribonucleic acid (DNA) that then exacerbates damage to Barrett’s epithelium has been recognized as a potential problem.15 The narrow band imaging system is a video endoscopic imaging technique for enhanced display of mucosal microstructure and capillaries of the superficial mucosal layer obtained when using narrow-band filters (wavelength ranges of the new RGB filters: 485-515 nm for red [R], 430-460 nm for green [G], and 400-430 nm for blue [B]), which are different from conventional filters, to the RGB of a plane sequential endoscope, and by changing the spectral feature of the observation light relative to that of the narrow band filters.16 Combining the narrow band imaging system and a magnifying endoscope allows simple and clear visualization of micropatterns of the superficial mucosal layer, its fine mucosal pattern and its capillary pattern, simply by switching the lighting system, without the need for dye or acetic acid. One study demonstrated that observing fine mucosal patterns by using this combined approach is useful when making the diagnosis of Barrett’s esophagus,17 whereas there have been no reports that discuss capillary patterns. Therefore, we divided the images of capillaries into patterns, in addition to fine mucosal patterns visualized by combining magnifying endoscopy with narrow band imaging (MENBI), in columnar-lined esophagus and superficial Barrett’s adenocarcinoma. The relation between fine mucosal patterns and capillary patterns to histological results, including SIM and adenocarcinoma, was prospectively examined. Simultaneously, by using confocal laser-scanning microscopy (CLSM), we also examined whether MENBI precisely visualized the structures of capillaries of the superficial mucosal layer. www.giejournal.org
Magnifying endoscopy with narrow band imaging
Capsule Summary What is already known on this topic d
Distinguishing Barrett’s esophagus that contains specialized SIM from columnar-lined esophagus is difficult when using routine endoscopy, but combining a narrow-band imaging system and a magnifying endoscope allows simple and clear visualization of micropatterns of the superficial mucosal layer.
What this study adds to our knowledge d
d
When using magnifying endoscopy with narrow-band imaging at sites of columnar-lined esophagus, the most characteristic endoscopic patterns of SIM were cerebriform, fine mucosal pattern (sensitivity, 56%; specificity, 79%), and an ivy- or DNA-like capillary pattern (sensitivity, 77%; specificity, 94%). The combined presence of a capillary pattern and a fine mucosal pattern improved the accuracy.
PATIENTS AND METHODS Patients The present study involved patients with columnar-lined esophagus and in whom MENBI was carried out during the period from December 2003 to September 2004 at the Jikei University Hospital. Patients were eligible for the study if they were formerly diagnosed or were referred to our department for workup of diagnosed columnar-lined esophagus, with or without superficial Barrett’s adenocarcinoma. Patients were also enrolled in regular surveillance for columnar-lined esophagus or had been previously treated by endoscopy for superficial Barrett’s adenocarcinoma. None of the patients had ever previously received a detailed examination by MENBI. We excluded patients (1) who had very short columnar-lined esophagus, less than 5 mm in length; (2) who were receiving antiplatelet anticoagulation therapy; (3) who had severe underlying disease, including liver cirrhosis and renal failure; and (4) in whom tissue sampling by biopsy was contraindicated. Two patients in whom the fine mucosal pattern of the entire columnar-lined esophagus or capillary pattern could not be divided further into patterns because of difficulties with visualization, were also excluded. As a result, the present analysis included 58 of the originally enrolled 60 patients. Written informed consent was obtained from all participants.
Endoscopic procedure Conscious sedation was undertaken with 35 mg pethidine hydrochloride and 0.3 to 0.5 mg flunitrazepam for all subjects. To remove mucus, 20,000 U of pronase (Pronase MS; Kaken Pharmaceutical Products Inc, Tokyo, Japan) was administered to all patients before endoscopy. After insertion of the endoscope, esophageal lavage was conducted with approximately 60 mL of water including Volume 65, No. 1 : 2007 GASTROINTESTINAL ENDOSCOPY 37
Magnifying endoscopy with narrow band imaging
a small amount of Gascon (0.04% solution of dimethicone; Kissei Pharmaceutical Products Inc, Tokyo, Japan). The magnifying electronic endoscope used was a GIF-Q240Z (maximum magnification ratio, 80; Olympus Medical Systems Co, Tokyo, Japan). A disposable transparent hood (Olympus) was attached approximately 2 mm distal to the tip of the endoscope for the purpose of maintaining the focal distance during the procedure. Conventional and magnifying endoscopic observations were conducted by 1 endoscopist (K.G.) with extensive experience in MENBI. We sequentially performed MENBI after conventional observation. On all occasions, endoscopic photographs were taken and recorded on digital video tapes.
Definitions As a rule, the oral end of the fold of the stomach was defined as the esophagogastric junction.18 Sixteen of the 58 patients (28%) had severe atrophic gastritis that spread through the cardiac region from the antrum. In these patients, severe atrophic gastritis made it difficult for the oral end of the gastric folds to be used as a marker of the gastroesophageal junction. Therefore, the lower end of the longitudinal vessels of the lower esophagus was taken as the esophagogastric junction according to the definition of the Japanese Society for Research on Esophageal Diseases.19 We diagnosed the lower end of the longitudinal vessels of the lower esophagus by conventional endoscopy and by MENBI. Cases were defined as having a longsegment columnar-lined esophagus for those in which the full circumference (R3 cm) of the squamous epithelium was replaced by columnar epithelium on the esophageal side from the esophagogastric junction, regardless of the presence or the absence of SIM. Other cases were defined as having a short-segment columnar-lined esophagus. In the present study, only columnar-lined esophagus with histologically confirmed SIM was defined as Barrett’s esophagus.20 The histologic diagnosis of Barrett’s adenocarcinoma conformed to the Guidelines for the Treatment of the Gastric Cancer21 established by the Japan Gastric Cancer Association. The Los Angeles classification22 was used for reflux esophagitis. Histology and the rapid urease test (Helicocheck; Institute of Immunology, Co, Ltd, Tochigi, Japan) were used for diagnosing Helicobacter pylori infection. A case positive for either test was considered to be positive for H pylori. A case negative for both tests was considered to be negative for H pylori. Fine mucosal patterns observed by MENBI at 217 sites of columnar-lined esophagus and superficial Barrett’s adenocarcinoma were divided into the following 5 fine mucosal patterns: 1, spotty round or oval; 2, long, straight; 3, digital to foliaceous villous; 4, cerebriform, showing complicated branching and unions; and 5, irregular in shape and branching (Fig. 1A to E). On the other hand, capillary patterns observed by MENBI of columnar-lined esophagus and superficial Barrett’s adenocarcinoma were also classified into the following 5 capillary patterns: I, honeycomb38 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
Goda et al
like or meshlike; II, vinelike, showing smooth curves and branching; III, coiled or curly haired, capillaries occasionally showing intertwining finer than that of capillary pattern II, similar to that seen in capillary pattern IV, as will be described later, but without ‘‘close connections’’ to adjacent capillaries as in a chain or net; IV, ivy-like or DNAspiral–like, showing spirals and complicated branching (the adjacent capillaries show interconnections resembling a chain or net); and V, irregular microvessels, with distinctly irregular courses and uneven forms (Fig. 2A to E). When a glandular duct with goblet cells was recognized even in only a portion of the biopsied specimens, the epithelium was considered to be SIM. Columnar-lined esophagus was histologically classified into the following 4 types for assessment: fundic type, parietal cells in glandular layer; mixed type, mucous gland with parietal cell; cardiac type, mucous gland with no parietal cell; and SIM. Histologically, cases were defined as having foveolar hyperplasia23 when they consisted of elongated foveolae in which apparent branching was frequent and cystic dilatation almost invariable. The lining cells consist of a single layer of regularly arranged hypertrophied foveolar epithelium containing abundant mucus and indicating no atypia.
Protocol for biopsy Tissue specimens were obtained by using standard cup size Radial Jaw 3 Biopsy Forceps (Microvasive Endoscopy, Boston Scientific Corp, Natick, Mass) under magnifying endoscopic observation. The fine mucosal pattern was observed in a half-zoom mode, and the capillary pattern was observed in a full-zoom mode. One or 2 specimens were biopsied for each individual pattern combination. When taking into consideration the increased hemorrhage risk, a biopsy was avoided from sites of severe inflammation as long as neoplasia was not suspected. After biopsy, 10,000 U thrombin was sprayed onto the site, and the test was completed after confirmation of hemostasis. When the biopsy was designed to make a diagnosis of H pylori infection, 2 sites each in the antrum and the greater curvature of the gastric body (a total of 4 sites) were biopsied for histology and the rapid urease test.
Histopathologic analysis All biopsied specimens were fixed in 4% formalin, embedded in paraffin for the preparation of thinly sliced specimens, and subjected to H&E staining. The biopsied specimens collected for histology were stained with Giemsa for the diagnosis of H pylori infection. One pathologist (M.I.), with extensive experience in GI tract histopathology, evaluated all of the tissue specimens without prior knowledge of the endoscopic findings.
Three-dimensional reconstruction of capillaries Capillaries of the superficial mucosal layer, which were observed by using CLSM, were 3-dimensionally (3D) www.giejournal.org
Goda et al
Magnifying endoscopy with narrow band imaging
Figure 1. Fine mucosal pattern of columnar-lined esophagus and superficial Barrett’s adenocarcinoma was divided into the 5 patterns observed by MENBI. A, Fine mucosal pattern 1, round or oval. B, Fine mucosal pattern 2, long straight. C, Fine mucosal pattern 3, villous. D, Fine mucosal pattern 4, cerebriform. E, Fine mucosal pattern 5, irregular.
analyzed in 45 biopsied specimens from 12 patients (10 with columnar-lined esophagus and 2 with Barrett’s adenocarcinoma). The biopsy samples were not sliced and used for CLSM observation; the biopsied tissue was used without change. The outline of the method for preparing the specimens is presented below. After fixation in 4% paraformaldehyde and 0.1 M phosphate buffered saline solution (pH 7.4), pretreatment was conducted with Triton X-100 (Nacali Tesque Inc, Kyoto, Japan) as a nonionic detergent. For immunostaining, the specimens were imwww.giejournal.org
mersed in 1% normal goat serum (0.1 M, diluted with phosphate buffered saline solution) at room temperature, then allowed to react with murine antihuman endothelial cell monoclonal antibody (CD34; Novocastra Laboratories, Ltd, Newcastle, UK) as the primary antibody and with fluorescein isothiocyanate-labeled goat antimouse immunoglobulin G antibody (Molecular Probes, Inc, Eugene, Ore) as the secondary antibody. After being reacted with the fluorescein isothiocyanate-labeled antibodies, the specimens were embedded in glycerin and were observed Volume 65, No. 1 : 2007 GASTROINTESTINAL ENDOSCOPY 39
Magnifying endoscopy with narrow band imaging
Goda et al
Figure 2. Capillary pattern of columnar-lined esophagus and superficial Barrett’s adenocarcinoma was classified into the 5 patterns observed by MENBI. A, Capillary pattern I, honeycomb-like or meshlike. B, Capillary pattern II, vinelike. C, Capillary pattern III, coiled or curly haired. D, Capillary pattern IV, ivy-like or DNA-spiral-like. E, Capillary pattern V, irregular.
by using a CLSM (LSM 510; Carl Zeiss Co, Jena, Germany). Excitation of fluorescein isothiocyanate was achieved with an argon laser at a wavelength of 488 nm. An image analysis processing software LSM 5 Image Browser (Carl Zeiss) was used for 3D reconstruction of the immunostained microvessels.
Statistical analysis The associations between the endoscopic findings of fine mucosal patterns 1 to 5, the capillary patterns I to 40 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
V, and the presence of SIM were evaluated in terms of sensitivity, specificity, odds ratio with 95% confidence intervals (CI), and P values. When the P value was less than .05, the difference was considered statistically significant. Diagnostic figures to model the probability of the presence of SIM from the findings of MENBI were compared by analyzing the area under the curve of the receiver operator characteristic curves. All tests were performed by using STATA 8.0 software (STATA Co, College Station, Tex). www.giejournal.org
Goda et al
Magnifying endoscopy with narrow band imaging
TABLE 1. Patient characteristics SSCLE (5-29 mm) (N Z 51) Mean age, y (range) Male/female (male:female ratio)
LSCLE (R30 mm) (N Z 7)
59 (40-81)
64 (43-80)
All (N Z 58) 60 (40-81)
45/6 (7.5:1)
6/1 (6.0:1)
51/7 (7.3:1)
O
31 (61)
4 (57)
35 (60)
A or B
15 (29)
2 (29)
17 (30)
C or D
5 (10)
1(14)
6 (10)
Presence of hiatal hernia, no. (%)
25 (49)
7 (100)
32 (55)
Infection with Helicobacter pylori, no. (%)
16 (31)
1 (14)
17 (29)
Presence of SIM, no. (%)
29 (57)
7 (100)
36 (62)
Presence of superficial adenocarcinoma, no. (%)
3 (8)
Reflux esophagitis, no. (%)*
1 (5)
4 (7)
LSCLE, Long-segment columnar-lined esophagus; SSCLE, short-segment columnar-lined esophagus. *Los Angeles classification of reflux esophagitis.
RESULTS The characteristics of the patients by group according to the length of columnar-lined esophagus are shown in Table 1. Short-segment columnar-lined esophagus (range, 5-28 mm; mean, 15 mm) was observed in 51 patients, and long-segment columnar-lined esophagus (range, 30-130 mm; mean, 69 mm) was observed in 7 patients. With regard to age, sex, and reflux esophagitis, a similar trend was recognized in both short- and long-segment columnar-lined esophagus. The presence of hiatal hernia and SIM was more frequent in long- than in short-segment columnarlined esophagus. The frequency of H pylori infection in short-segment columnar-lined esophagus was higher than in long-segment columnar-lined esophagus. Three of the 4 patients with Barrett’s adenocarcinoma showed short-segment columnar-lined esophagus, and 1 had longsegment columnar-lined esophagus. Each of the 4 cancer patients had a single superficial Barrett’s adenocarcinoma. One of these 4 adenocarcinomas was a completely flat lesion, 2 were depressed lesions, and the remaining 1 was a depressed lesion with elevated and flat components. By using conventional endoscopy, we suspected that 2 of the 4 adenocarcinomas were neoplastic lesions, except for a tiny erosive lesion and a completely flat lesion. As a result of sequential observation by MENBI, all adenocarcinomatous extent, including 2 subtle lesions, were clearly recognized. A total of 217 biopsied specimens were obtained from the 58 patients: 33 from the 7 patients with long-segment columnar-lined esophagus and 184 from the 51 patients with short-segment columnar-lined esophagus. Of the www.giejournal.org
217 biopsied specimens, 211 showed benign mucosa, and 6 had adenocarcinoma. All of these 6 adenocarcinoma sites showed fine mucosal pattern 5 and capillary pattern V upon observation by MENBI, whereas none of other 211 benign columnar-lined esophagus sites showed either the fine mucosal pattern 5 or the capillary pattern V. The sensitivity and the specificity of irregular patterns (fine mucosal pattern 5 and capillary pattern for detecting adenocarcinoma were both 100%). Adenocarcinoma sites were clearly distinguished from benign columnar-lined esophagus sites under MENBI observation. Six adenocarcinoma sites were excluded from the following analyses. The association between endoscopic patterns and the presence of SIM at the 211 benign sites are shown in Table 2. According to the fine mucosal patterns, the ratios of SIM were highest in fine mucosal pattern 4 (61%), followed by fine mucosal pattern 3 (43%); the sensitivity for SIM was highest in fine mucosal pattern 4 (71%); and the specificity was highest in fine mucosal pattern 2 (93%). According to capillary patterns, the ratios of SIM were highest in capillary pattern IV (88%), and both sensitivity and specificity for SIM were highest in capillary pattern IV (77% and 94%). As a single variate, the finding of fine mucosal pattern 1 (odds ratio, 0.28), capillary pattern I (odds ratio, 0.00), capillary pattern II (odds ratio, 0.14), or capillary pattern III (odds ratio, 0.41) significantly lowered the possibility of SIM; in contrast, that of fine mucosal pattern 4 (odds ratio, 4.78) or capillary pattern IV (odds ratio, 51.6) significantly raised the possibility of SIM. The diagnostic value of adding capillary patterns to fine mucosal patterns was evaluated by comparing the data Volume 65, No. 1 : 2007 GASTROINTESTINAL ENDOSCOPY 41
Magnifying endoscopy with narrow band imaging
Goda et al
TABLE 2. Associations of MENBI patterns and presence of SIM MENBI patterns*
Frequency
SIM/non-SIM (%)
Sensitivity (95% CI)
Specificity (95% CI)
OR (95% CI)
P value
FMP-1
57 (26%)
10/57 (18%)
13% (6-23)
65% (56-73)
0.28 (0.12–0.61)
FMP-2
12 (6%)
2/12 (17%)
3% (0-9)
93% (87-96)
0.33 (0.03–1.62)
Not significant
FMP-3
71 (34%)
22/71 (43%)
29% (19-40)
63% (55-72)
0.69 (0.36–1.32)
Not significant
FMP-4
71 (34%)
43/71 (61%)
56% (44-67)
79% (71-86)
4.78 (2.48–9.26)
!.0001
CP-I
41 (19%)
0/41 (0%)
0% (0-5)
69% (61-77)
0.00 (0.00–0.11)
!.0001
CP-II
42 (20%)
3/42 (7%)
5% (1-13)
72% (63-79)
0.14 (0.03–0.41)
!.0001
CP-III
61 (29%)
15/61 (25%)
18% (10-29)
65% (56-73)
0.41 (0.19–0.84)
!.0001
CP-IV
67 (32%)
59/67 (88%)
77% (66-86)
94% (89-97)
51.6 (19.9–141.9)
!.0001
.0005
CP, Capillary pattern; FMP, fine mucosal pattern; OR, odds ratio. *FMP and CP observed by MENBI: FMP-1, round or oval; FMP-2, long straight; FMP-3, villous; FMP-4, cerebriform; CP-I, honeycomb-like or meshlike; CP-II, vinelike; CP-III, coiled or curly haired; CP-IV, ivy-like or DNA-spiral-like.
with a fine mucosal pattern alone (Fig. 3). The area under the curve of the receiver operator characteristic curves in detecting SIM for fine mucosal pattern alone and a fine mucosal pattern plus a capillary pattern was 0.71, 95% CI 0.64–0.78, and 0.92, 95% CI 0.88–0.97, respectively, the difference being significant, P ! .0001). The histologic type of columnar-lined esophagus at 195 of the 211 sites at which noncancerous biopsied tissues could be histologically evaluated is shown in Figure 4. Many specimens of the fundic type showed endoscopic patterns of fine mucosal pattern 1 and capillary pattern I. The frequencies of the cardiac type in fine mucosal pattern 2 and capillary pattern II were high, and those of SIM in fine mucosal pattern 4 and capillary pattern IV were also high. Each pattern of MENBI and the presence or the absence of foveolar hyperplasia in 192 of the 211 sites at which noncancerous biopsied tissues could be histologically evaluated was evaluated (Fig. 5). The frequency of foveolar hyperplasia was relatively high in fine mucosal pattern 3 and fine mucosal pattern 2, and in the capillary pattern II. Finally, we examined the biopsied specimens obtained from the sites showing capillary pattern I to V by MENBI. We observed these specimens by using CLSM and attempted to produce 3D reconstructions of capillaries of the superficial mucosal layer by using 3D reconstruction software. All 3D reconstruction patterns were similar to the capillary patterns I to V observed by MENBI (Fig. 6A to E). The biopsied specimens that showed capillary pattern I yielded a pattern of reconstruction of capillaries, which formed a relatively regular honeycomb network. The biopsied specimens that showed capillary pattern II yielded a vinelike pattern of reconstruction, which showed a relatively slow but quite remarkable meandering of capillaries. The biopsied specimens that showed capillary pattern III yielded a coiled or curly-haired pattern of 42 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
small meandering loops. In the biopsied specimens that showed capillary pattern IV, an ivy-like or DNA-spiral–like structure of capillaries, which showed branching during fine complicated turns and meandering, was observed with CLSM as well. These capillaries were interconnected and formed chain- or netlike structures. The findings of Barrett’s adenocarcinoma when using CLSM included dense capillaries of different diameters, with remarkably irregular courses.
DISCUSSION A new endoscopic imaging technique, narrow band imaging and magnifying endoscopy were combined to make diagnoses of SIM in columnar-lined esophagus and superficial Barrett’s adenocarcinoma, focusing particularly on morphologic changes in capillaries of the superficial mucosal layer and a fine mucosal pattern. Our results suggest that MENBI may be very useful for assessing SIM and Barrett’s adenocarcinoma. In this study, fine mucosal patterns observed by MENBI were classified into 5 patterns. As reported by Endo et al,12 fine mucosal pattern 2 was present though minimal. The fundic type was frequently recognized as fine mucosal pattern 1, and the cardiac or mixed type or SIM was frequently recognized as fine mucosal pattern 3 or fine mucosal pattern 4. This tendency was similar to that reported by Toyoda et al.14 Accordingly, there appears to be no significant difference between the mucosal pattern observed by magnifying endoscopy with the acetic acid spraying method and the fine mucosal pattern observed by MENBI. Fine mucosal pattern 4 in the present study appeared to correspond to ‘‘pattern IV, ridged’’ described by Guelrud et al11 and ‘‘pit-4, tubular’’ described by Endo et al.12 Fine mucosal pattern 3 in the present study appeared to correspond to ‘‘pattern III, villous’’ described www.giejournal.org
Goda et al
Magnifying endoscopy with narrow band imaging
Figure 3. Receiver operator characteristic curves in detecting SIM for a fine mucosal pattern alone and a fine mucosal pattern plus a capillary pattern.
Figure 4. Correlation between histopathologic classification of columnar-lined esophagus and endoscopic patterns observed by MENBI.
by Guelrud et al,11 and ‘‘pit-5, villous’’ described by Endo et al.12 In the present study, the results of the relation between fine mucosal patterns and SIM differed markedly from previously reported results; the ratios of SIM in fine mucosal pattern 4 and fine mucosal pattern 3 were 61% and 43%, respectively, in the present study, whereas previous studies reported this ratio in fine mucosal pattern 4 and fine mucosal pattern 3 as 100% and 87% to 100%, respectively.11,12 For the low rates of SIM in these 2 types of fine mucosal patterns, our attention was focused on the relatively high frequency with which hyperwww.giejournal.org
plastic changes in the foveolar epithelium were observed histologically at columnar-lined esophagus. Hyperplastic changes in the foveolar epithelium were observed in approximately half (51%) of all biopsied specimens; the frequency of these changes was highest in fine mucosal pattern 3, at 75%. We cannot rule out the possibility that the foveolar epithelium, which histologically showed elongated foveolae or apparent branching, was especially recognized as fine mucosal pattern 3. Our assumption was that this might account for the marked difference in the fine mucosal pattern between past reports and the Volume 65, No. 1 : 2007 GASTROINTESTINAL ENDOSCOPY 43
Magnifying endoscopy with narrow band imaging
Goda et al
Figure 5. Frequencies of histopathologic finding of foveolar hyperplasia according to endoscopic patterns observed by MENBI.
present study. On the assessment of hyperplasia of the foveolar epithelium, we suggested that it is important to distinguish capillary pattern II and capillary pattern IV of fine mucosal pattern 3 to improve the diagnostic accuracy of SIM. The usefulness of adding capillary patterns to fine mucosal patterns by MENBI for endoscopic detection of SIM was demonstrated by comparing the area under the curve of the receiver operator characteristic curves. However, higher magnification than that for fine mucosal patterns is needed to observe the details of capillary patterns, and the visual field is restricted to an extremely narrow range. Therefore, extensive observation of capillary patterns in a short time period, particularly for a long-segment columnar-lined esophagus, appears to be extremely difficult. Extensive observation of capillary patterns in a short time period appears to be extremely difficult. Under these circumstances, 2 observation methods were considered for the efficient detection of SIM by MENBI. In 1, fine mucosal patterns were observed in a half-zoom mode, which was increased mainly for fine mucosal pattern 4, for the purpose of confirming the presence or the absence of capillary pattern IV in a full-zoom mode. For the other approach, endoscopic diagnoses were made by using the density of capillaries as an indicator. The capillary pattern IV site tends to show a higher density than other types of capillary patterns, not only on endoscopic images of MENBI but also on reconstruction images of CLSM. Therefore, the area of capillary pattern IV is frequently visualized as a slight, deep-brown area in a half-zoom mode, though this is an empirical estimation. Thus, the second method is designed for observing 44 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
capillary patterns by increasing the zoom mode mainly in the area of interest. However, attention should be paid to the possibility of the site being associated with inflammatory changes, which are visualized as a deep-brown area because of hyperplasia of the capillaries. The depth of a superficial Barrett’s adenocarcinoma, as investigated in this study, was intramucosal carcinoma in all of our subjects. This may be considered to correspond to high-grade dysplasia in Western countries when using a histologic classification different from that used in Japan. The MENBI showed fine mucosal pattern 5, which was similar to the results of magnification chromoendoscopy when using an indigo carmine solution provided by Sharma et al.13 Furthermore, capillary pattern V was observed at sites consistent with areas of fine mucosal pattern 5. We also recognized the existence and the extent of all 4 adenocarcinomas, including 2 subtle lesions because of a precise observation of fine mucosal pattern 5 and capillary pattern V by MENBI. Therefore, although dysplastic lesions could not be investigated in the present study, these images were characteristic of superficial Barrett’s adenocarcinoma, indicating that MENBI is useful for early detection and for determining the extent of the adenocarcinoma. The capillary pattern of the mucosal superficial layer observed by MENBI was extremely similar to the reconstruction of capillaries observed when using a CLSM. This strongly supports the notion that MENBI precisely visualized the structure of capillaries of the superficial mucosal layer, which included Barrett’s adenocarcinoma. The classification of capillary patterns, which mostly reflected the histologic types of columnar-lined esophagus, www.giejournal.org
Goda et al
Magnifying endoscopy with narrow band imaging
Figure 6. All 3D reconstruction of the capillaries with a CLSM, using biopsy specimens showing capillary pattern I to V, were similar to the capillary patterns observed by MENBI. A, Capillary pattern I yielded a pattern of reconstruction of capillaries, which formed a relatively regular honeycomb network (orig. mag. 10). B, Capillary pattern II yielded a vinelike pattern of reconstruction, which showed relatively slow but quite remarkable meandering of vessels (orig. mag. 20). C, Capillary pattern III yielded a coiled or curly-haired pattern of small meandering loops (orig. mag. 20). D, Capillary pattern IV, an ivy-like or DNA-spiral–like structure of capillaries, showed branching during fine complicated turns and meandering. These capillaries were interconnected, forming chain- or netlike structures (orig. mag. 10). E, Barrett’s adenocarcinoma included dense capillaries of different diameters and with remarkably irregular courses (orig. mag. 20) (continued next page).
was considered to form a basis for the estimation of histologic types of columnar-lined esophagus in a manner similar to that of the classification of fine mucosal patterns. Recent reports showed MENBI to be useful for diagnosing squamous-cell carcinomas of the head and neck and the esophagus, and early gastric cancer.24-26 It is highly probable that a workup of columnar-lined esophagus and Barrett’s adenocarcinoma by MENBI would be procedurally simple and safe. Thus, we anticipate that this www.giejournal.org
method will be recognized as an extremely useful modality when detailed examinations for squamous-cell carcinomas of the head and the neck and the esophagus, and early gastric cancer are conducted simultaneously with thorough examinations of columnar-lined esophagus and Barrett’s adenocarcinoma. In conclusion, we suggest that MENBI is useful for the detection of SIM and superficial Barrett’s adenocarcinoma. Furthermore, we consider it to be extremely Volume 65, No. 1 : 2007 GASTROINTESTINAL ENDOSCOPY 45
Magnifying endoscopy with narrow band imaging
Figure 6 (continued )
important that the 3D histologic investigation with a CLSM indicated the reliability of the capillary pattern depicted by MENBI. DISCLOSURE The narrow band imaging system was provided by Olympus Medical Systems Corp, Tokyo, Japan. No additional grant or financial incentive was provided from manufacturers mentioned in this study.
REFERENCES 1. Rahamim J, Cham CW. Oesophagogastrectomy for carcinoma of the esophagus and cardia. Br J Surg 1993;80:1305-9. 2. Devesa SS, Blot WJ, Frameni JF. Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer 1998;83:2049-53. 3. Fujishiro H, Adachi K, Kawamura A, et al. Influence of Helicobacter pylori infection on the prevalence of reflux esophagitis in Japanese patients. J Gastroenterol Hepatol 2001;16:1217-21. 4. Hongo M, Shoji T. Epidemiology of reflux disease and CLE in East Asia. J Gastroenterol 2003;38(Suppl):25-30. 5. Paull A, Trier JS, Dalton MD, et al. The histologic spectrum of Barrett’s esophagus. N Engl J Med 1976;295:476-80. 6. Spechler SJ, Goral RK. Barrett’s esophagus. N Engl J Med 1986;315: 362-71. 7. Hammeetman W, Tytgat GNJ, Houthouf HJ, et al. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989;96:1249-56. 8. Levine DS. Management of dysplasia in the columnar-lined esophagus. Gastroenterol Clin North Am 1997;26:613-34. 9. Stevens PD, Lightdale CJ, Green PH, et al. Combined magnification endoscopy with chromoendoscopy for the evaluation of Barrett’s esophagus. Gastrointest Endosc 1994;40:747-9.
46 GASTROINTESTINAL ENDOSCOPY Volume 65, No. 1 : 2007
Goda et al 10. Canto MI, Setrakian S, Willis J, et al. Methylene blue-directed biopsies improve detection of intestinal metaplasia and dysplasia in Barrett’s esophagus. Gastrointest Endosc 2000;51:560-8. 11. Guelrud M, Herrera I, Essenfeld H, et al. Enhanced magnification endoscopy: a new technique to identify specialized intestinal metaplasia in Barrett’s esophagus. Gastrointest Endosc 2001;53:559-65. 12. Endo T, Awakawa T, Takahashi H, et al. Classification of Barrett’s epithelium by magnifying endoscopy. Gastrointest Endosc 2002;55: 641-7. 13. Sharma P, Weston AP, Topalovski M, et al. Magnification chromoendoscopy for the detection of intestinal metaplasia and dysplasia in Barrett’s esophagus. Gut 2003;52:24-7. 14. Toyoda H, Rubio C, Befrits R, et al. The detection of intestinal metaplasia in distal esophagus and esophagogastric junction by enhanced magnification endoscopy. Gastrointest Endosc 2004;59: 15-21. 15. Olliver JR, Wild CP, Sahay P, et al. Chromoendoscopy with methylene blue and associated DNA damage in Barrett’s esophagus. Lancet 2003;362:373-4. 16. Gono K, Yamazaki K, Doguchi N, et al. Endoscopic observation of tissue by narrow band illumination. Opt Rev 2003;10:211-5. 17. Hamamoto Y, Endo T, Nosho K, et al. Usefulness of narrow band imaging endoscopy for diagnosis of Barrett’s esophagus. J Gastroenterol 2004;39:14-20. 18. McClave SA, Boyce HW Jr, Gottfried MR. Early diagnosis of columnarlined esophagus: a new endoscopic diagnostic criterion. Gastrointest Endosc 1987;33:413-6. 19. Aoki T, Kawaura Y, Kozu T, et al. Report of the Research Committee on the Definition of Barrett’s Esophagus (epithelium) [Japanese]. In: Sugimachi K, editor. Reports of the research committees of the Japanese Society for Esophageal Diseases. Chiba: Japanese Society of Esophageal Diseases; 2000:20-3. 20. Sampliner RE. Practice guidelines on the diagnosis, surveillance, and therapy of the Barrett’s esophagus. The Practice Parameter Committee of the American College of Gastroenterology. Am J Gastroenterol 1998;93:1028-32. 21. The Japanese Gastric Cancer Association. Guidelines for the treatment of the gastric cancer [English], 1st ed. Tokyo, Japan: Kanehara Publishers; 1995. 22. Lundell LR, Dent J, Bennett JR, et al. Endoscopic assessment of oesophagitis: clinical and functional correlates and further validation of the Los Angeles classification. Gut 1999;45:172-80. 23. Day DW, Jass JR, Price AB, et al. Morson and Dawson’s gastrointestinal pathology, 4th ed. In: Day DW, et al. editors. Tumor-like lesion of the stomach. Boston: Blackwell Publishing. 24. Muto M, Nakane M, Katada C, et al. Squamous cell carcinoma in situ at oropharyngeal and hypopharyngeal mucosal sites. Cancer 2004;101: 1375-81. 25. Yoshida T, Inoue H, Usui S, et al. Narrow band imaging system with magnifying endoscopy for superficial esophageal lesion. Gastrointest Endosc 2004;59:288-95. 26. Nakayoshi T, Tajiri H, Matsuda K, et al. Magnifying endoscopy combined with narrow band imaging system for early gastric cancer: correlation of vascular pattern with histopathology (including video). Endoscopy 2004;36:1080-4.
Received June 11, 2005. Accepted March 31, 2006. Current affiliations: Department of Endoscopy (K.G., H.T., T.N., M.K.), Department of Pathology (M.I.), Division of Clinical Research and Development (M.U.), The Jikei University School of Medicine, Tokyo, Japan. Reprint requests: Ken-ichi Goda, MD, Department of Endoscopy, The Jikei University School of Medicine, 3-25-8 Nshi-shimbashi, Minato-ku, Tokyo 105-8461, Japan.
www.giejournal.org