- Email: [email protected]
Trimodal imaging endoscopy may improve diagnostic accuracy of early gastric neoplasia: a feasibility study
ORIGINAL ARTICLE: Clinical Endoscopy
Trimodal imaging endoscopy may improve diagnostic accuracy of early gastric neoplasia: a feasibility study Masayuki Kato, MD, PhD, Mitsuru Kaise, MD, PhD, Jin Yonezawa, MD, Kenichi Goda, MD, PhD, Hirobumi Toyoizumi, MD, Noboru Yoshimura, MD, Yukinaga Yoshida, MD, Muneo Kawamura, MD, PhD, Hisao Tajiri, MD, PhD Tokyo, Japan
Background: A considerable number of superficial gastric neoplasias are overlooked with conventional white light imaging (WLI) endoscopy. Objective: The aim was to investigate the diagnostic potential of trimodal imaging endoscopy (TME), which combines WLI, autofluorescence imaging (AFI), and narrow-band imaging (NBI), for superficial gastric neoplasia. Design: Feasibility study. Setting: Single academic center. Patients: Sixty-two patients with or without gastric neoplasia. Intervention: Each patient serially assessed with WLI, AFI, and magnifying endoscopy with NBI (ME-NBI) by an endoscopist blinded for clinical information. ME-NBI over WLI and AFI was designated as TME. Histopathology of biopsy and ESD specimens was evaluated and used as the gold standard. Main Outcome Measurements: Sensitivity and specificity of endoscopic diagnosis of pathology-proven neoplasia by per-patient and per-lesion analyses. Results: The study included 47 pathology-proven neoplasias and 44 pathology-proven nonneoplasias that were detected as neoplasias with any of the modalities. By a per-lesion analysis, the sensitivity of TME (89.4%) was higher than that of WLI (76.6%) and AFI (68.1%). The specificity of TME (98.0%) was higher than that of WLI (84.3%) and AFI (23.5%). By a per-patient analysis, the sensitivity of TME (90.9%) was higher than that of WLI (75%) and AFI (68.2%). The specificity of TME (100%) was higher than that of WLI (72.2%) and AFI (44.4%). Limitations: Case-enriched population at a single center. Conclusions: Higher diagnostic accuracy of TME over conventional WLI indicates the feasibility of TME for the efficacious diagnosis of early gastric neoplasia. (Gastrointest Endosc 2009;70:899-906.)
Efficacious detection of early neoplasia with endoscopy increases the curability of GI neoplasias and decreases
Copyright ª 2009 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2009.03.1171
associated morbidities. Autofluorescence endoscopy (AFE) for the detection of early neoplasias has attracted considerable attention since the development of a fluorescence detection method for cancerous lesions.1 AFE differs from white light imaging (WLI) in that AFE has been shown to be more accurate for the diagnosis of neoplasia, as described in studies of early neoplasia of the esophagus,2-6 stomach,7-10 and colon.11-14 We previously performed a prospective feasibility study to compare WLI with AFE by using autofluorescence imaging (AFI) for the detection of early gastric neoplasia.15 We found that (1) WLI missed or misdiagnosed 25% of early gastric neoplasias and AFI detected 10% more neoplasias that were not identified by WLI and (2) the sensitivity of
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 899
Abbreviations: AFE, autofluorescence endoscopy; AFI, autofluorescence imaging; BE, Barrett’s esophagus; CI, confidence interval; ESD, endoscopic submucosal dissection; FMS, fine mucosal structure; MENBI, magnifying endoscopy with narrow-band imaging; NBI, narrowband imaging; path-neoplasia, pathology-proven neoplasia; TME, trimodal imaging endoscopy; WLI, white light imaging. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
TGE may improve diagnostic accuracy of early gastric neoplasia
WLI and that of AFI were comparable, whereas the specificity of AFI was considerably lower than that of WLI because of a high number of false positives. Therefore, we concluded that AFI is of limited clinical value. It is noteworthy that AFI detected 10% more lesions compared with WLI, which missed or misdiagnosed 25% of early gastric neoplasias. A combination of AFI and a different modality that would decrease the high rate of false positives might be a promising endoscopic technique for the early detection of gastric neoplasia. A candidate modality for improving specificity is magnifying endoscopy with narrow-band imaging (ME-NBI).16 Therefore, we conducted a prospective feasibility study to investigate the diagnostic potential of trimodal imaging endoscopy (TME), which combines WLI, AFI, and NBI, for early gastric neoplasia.
PATIENT AND METHODS Patient population Patients with early gastric neoplasia who were referred for endoscopic submucosal dissection (ESD) by affiliated hospitals or clinics from April to December 2007 were eligible for this study. Patients with obvious advanced gastric carcinomas and cancerous lesions with deep invasion to the gastric submucosa were excluded. Eligible patients who gave written informed consent were serially enrolled in the study and scheduled for a preoperative advanced workup including TME. To decrease the bias of only including patients with known gastric neoplasia, patients who were undergoing follow-up endoscopy after ESD were also recruited as controls because these subjects were suspected of being free of neoplasia.
Kato et al
Capsule Summary What is already known on this topic d
Autofluorescence endoscopy (AFI) has been shown to be more accurate than white light imaging (WLI), which misses or misdiagnoses 25% of early gastric neoplasias.
What this study adds to our knowledge d
In 62 patients serially assessed with WLI, AFI, and magnifying endoscopy with narrow band imaging (ME-NBI), the addition of AFI and ME-NBI to WLI increased the detection rate of early gastric neoplasias by 12.8%.
the superficial mucosa. For NBI observations, endoscopy at a magnification of as high as 80 was performed. In the AFI mode, light emitted from the xenon lamp is directed to the rotary filter, which splits it into excitation wavelengths of 390 to 470 nm and green light of 540 to 560 nm wavelengths. The AFI endoscope (XGIF-Q240FZ; Olympus Medical Systems) incorporates a monochrome charged couple device with a barrier filter to exclude the excitation light and capture only the weak autofluorescence. A pseudocolored image is reconstructed based on the autofluorescence input signals, such that high-intensity autofluorescence is green and low-intensity autofluorescence is magenta. In general, the AFI system displays normal pyloric mucosa or atrophic fundic mucosa in green, blood vessels in dark green, and hypertrophic fundic mucosa or elevated neoplasia in magenta.
Overall study design
The endoscopic system consists of a high-resolution white light endoscope with optical zoom (80 magnification, XGIF-Q240FZ; Olympus Medical Systems Co, Tokyo, Japan), which can be switched easily to the AFI or NBI modes. Although the basic configuration is identical to that of the standard video-endoscopy system (LUCERA CV-260/CLV-260; Olympus Medical System), this trimodal system allows either red, green, blue illumination (for WLI and NBI) or an excitation/reflected light illumination combination (for AFI). The light source incorporates a rotary filter, which is designed in a double-wheel configuration with 2 concentric wheels: a red, green, blue filter wheel for WLI and NBI and an AFI filter wheel. In the NBI mode, the light source (CLV-U40D; Olympus Medical Systems) for this endoscope is equipped with narrowband filters corresponding to red (485-515 nm), green (430-460 nm), and blue (400-430 nm) light. Short wavelength light in the blue range is absorbed by hemoglobin in vivo, thereby enhancing the appearance of capillaries in
Although the endoscopists were informed beforehand that the study population was enriched and included patients with and without gastric neoplasia, they had no access to any clinical information before endoscopy. Patients underwent TME under conscious sedation with pethidine hydrochloride and midazolam. An examiner serially performed WLI, AFI, and ME-NBI under the supervision of the study coordinator, who organized the patient data. The examiner recorded the presence or absence of lesions suspected to be neoplasias with WLI, AFI, and TME, in accordance with the aforementioned criteria. All lesions recorded as neoplasias with any of the 3 imaging modalities (endoscopy neoplasias) and neoplasias referred for ESD but not detected by any of the imaging modalities underwent biopsies under the supervision of the study coordinator, who attended all endoscopic examinations and was aware of the endoscopist’s diagnoses. The pathologic results obtained from the biopsy samples and ESD specimens were used as the criterion standard for diagnostic accuracy. The sensitivity and specificity for endoscopic diagnosis of pathology-proven neoplasias (path-neoplasias) were evaluated by per-patient and
900 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
The TME system
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
Figure 1. Diagnostic criteria of AFE. We defined a neoplasia-suspected lesion by AFE (AFE-positive) as an area or lesion that was clearly different from the surrounding mucosa in color and that showed a clearly defined margin. A, WLE image of a superficial gastric carcinoma. B, AFE image displaying superficial carcinoma as a magenta lesion with defined margins within green-colored mucosa, thus diagnosed as AFE-positive.
per-lesion analyses. If a patient had 2 or more path-neoplasias, with at least 1 diagnosed as endoscopy-nonneoplasia, the diagnosis was defined as missed on a per-patient analysis. The study was approved by the ethics committee of the Jikei University Hospital and was conducted in accordance with the revised Declaration of Helsinki (1989). Sensitivity and specificity were calculated by standard formulas for a binominal proportion, and 95% confidence intervals (CIs) were calculated by the Wald interval method.17 Because of the preliminary focus of this study and the synthetic mixture of the composition of the study population, we considered that the statistical analysis results should be characterized as being descriptive only. Therefore, we only presented sensitivities and specificities and their 95% CIs and did not analyze the statistical differences among sensitivities and specificities of these tested endoscopic modalities.
Diagnostic criteria for gastric neoplasia A lesion was recorded as WLI endoscopy neoplasia when it was recognized or suspected to be early gastric neoplasia based on color and appearance when using WLI. We defined an AFI endoscopy neoplasia as an area or lesion that was different from the surrounding mucosa in the color of AFI and had a defined circumferential margin (Fig. 1). If a lesion diagnosed as WLI endoscopy neoplasia did not fulfill the definition of AFI endoscopy neoplasia, it was diagnosed as AFI endoscopy-nonneoplasia. We assessed ME-NBI endoscopy neoplasia by using our diagnostic criteria for early gastric neoplasia.18 A superficial lesion demarcated from the surrounding mucosa with differences in the fine mucosal structure (FMS) and/or microvasclature was diagnosed as endoscopy neoplasia if FMS disappearance, microvascular dilation, and heterogeneity (the triad of microstructure) were seen (Fig. 2). Superficial elevated lesions were diagnosed as endoscopy neoplasias if the triad was observed or an irregular FMS with micrification and heterogeneity www.giejournal.org
was demonstrated. If a lesion diagnosed as WLI endoscopy neoplasia and/or AFI endoscopy neoplasia did not fulfill the definition of ME-NBI endoscopy neoplasia, it was diagnosed as TME endoscopy nonneoplasia. Although ME-NBI generally targeted lesions that were diagnosed as endoscopy neoplasia by WLI and/or AFI, ME-NBI observation was also done for lesions diagnosed as WLI and AFI endoscopy nonneoplasia. If a lesion diagnosed as WLI and AFI endoscopy nonneoplasia fulfilled the definition of ME-NBI endoscopy neoplasia, it was diagnosed as TME endoscopy neoplasia. Because AFI and ME-NBI differ considerably from WLI, endoscopists who had extensive experience in diagnosing early gastric neoplasia by AFI and ME-NBI in more than 100 cases were selected.
Histopathology All biopsy specimens were fixed in 10% formalin and embedded in paraffin, serially sectioned, and stained with hematoxylin and eosin. Lesions diagnosed as category 3 low-grade neoplasias, according to the revised Vienna classification,19,20 were classified as gastric adenomas. Lesions diagnosed as category 4 neoplasias were defined as gastric carcinomas. Patients with category 3 or 4 neoplasias were enrolled in this study as having gastric neoplasias. ESD specimens were pinned flat onto a corkboard with adequate tension and immersed in formalin. After fixation, specimens were cut into 2-mmwide longitudinal strips and examined pathologically. All histopathologic findings were reviewed by expert GI pathologists.
RESULTS Sixty-two patients, composed of 44 patients with gastric neoplasia referred for ESD and 18 patients who were undergoing follow-up EGD after ESD, were enrolled in this Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 901
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
Figure 2. We defined diagnostic criteria for superficial gastric neoplasia by ME-NBI. If there were disappearance of fine mucosa structure and microvascular irregularities showing dilation, abrupt caliber alteration, heterogeneity in shape, and tortousness, the lesion was diagnosed as neoplastic.
Figure 3. Flow diagram of patients enrolled for analysis.
study. Eighty-eight gastric lesions in 55 patients were diagnosed as endoscopy neoplasias; 44 and 44 of these 88 lesions were pathologically defined as neoplasias and nonneoplasias, respectively. Three early gastric neoplasias that were diagnosed and referred for ESD by affiliated hospitals were not diagnosed as endoscopy neoplasias by any of the imaging modalities in the study. These 3 neoplasias
were included in 7 patients who were diagnosed as not having gastric neoplasia (Fig. 3). Five of 44 patients with gastric neoplasia enrolled in the study were diagnosed after previous ESD. These neoplasias were considered to be metachronous gastric neoplasias after ESD. Forty-three of 44 patients with gastric neoplasia had gastric atrophy and 17 of 44 patients had
902 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
Figure 4. Schematic flow diagram of 47 neoplastic lesions.
concomitant benign gastric ulcer scars other than ESD scars. Three of 44 patients with gastric neoplasia had double gastric neoplasias. Of the 47 gastric neoplasias found in 44 patients with gastric neoplasia, 35 were gastric cancers (category 4) and 12 were gastric adenomas (category 3).
Tree diagram analysis WLI identified 36 (76.6%) of 47 enrolled path-neoplasias, and AFI identified 3 (6.4%) additional neoplasias not detected by WLI. TME identified 42 (89.4%) patients with 6 (12.8%) additional neoplasias that were not detected by WLI (Fig. 4). WLI and AFI diagnosed 8 and 39 pathology-proven nonneoplasias as endoscopy-neoplasias, respectively. Forty-three of the 44 false-positive lesions by WLI and/or AFI were finally diagnosed as nonneoplasia by ME-NBI, and only one pathology-proven nonneoplasia was diagnosed as endoscopy neoplasia by TME (Fig. 5).
(100%; 95% CI, 79.3%-100%) was higher than that of WLI (72.2%; 95% CI, 51.5%-92.9%) and AFI (44.4%; 95% CI, 21.5%-67.4%) (Table 1). We also analyzed the diagnostic accuracy for 35 gastric cancers when 12 gastric adenomas were excluded. By a per-lesion analysis for the diagnosis of gastric cancers, the sensitivity of TME (91.4%; 95% CI, 82.2%-100%) was higher than that of WLI (77.1%; 95% CI, 63.2%-91.1%) and AFI (62.9%; 95% CI, 46.8%-78.9%). The specificity of TME (98.0%; 95% CI, 94.2%-100%) was higher than that of WLI (84.3%; 95% CI, 74.3%-94.3%) and AFI (23.5%; 95% CI, 11.9%-35.2%). By a per-patient analysis, the sensitivity of TME (90.6%; 95% CI, 80.5%-100%) was higher than that of WLI (75%; 95% CI, 60%-90%) and AFI (62.5%; 95% CI, 45.7%-79.3%). The specificity of TME (100%; 95% CI, 79.3%-100%) was higher than that of WLI (72.2%; 95% CI, 51.5%-92.9%) and AFI (38.9%; 95% CI, 16.4%-61.4%) (Table 2).
DISCUSSION Comparison of the diagnostic accuracy of WLI and TME for gastric neoplasias By a per-lesion analysis, the sensitivity of TME (89.4%; 95% CI, 80.5%-98.2%) was higher than that of WLI (76.6%; 95% CI, 64.5%-88.7%) and AFI (68.1%; 95% CI, 54.8%-81.4%). The specificity of TME (98.0%; 95% CI, 94.2%-100%) was higher than that of WLI (84.3%; 95% CI, 74.3%-94.3%) and AFI (23.5%; 95% CI, 11.9%-35.2%). By a per-patient analysis, the sensitivity of TME (90.9%; 95% CI, 82.4%-99.4%) was higher than that of WLI (75%; 95% CI, 62.2%-87.8%) and AFI (68.2%; 95% CI, 54.4%-81.4%). The specificity of TME
Early gastric neoplasias are good indications for endoscopic resection21 and are targets for endoscopic surveillance to decrease the morbidity associated with gastric neoplasia. In this study, the conventional endoscopic modality of WLI identified 76.6% of early gastric neoplasias, of which approximately one fourth were not correctly diagnosed, even by using high-resolution endoscopy for WLI. The detection rate is comparable to that found in our previous study,15 confirming the limited value of WLI for diagnosing early gastric neoplasias. This study shows that TME
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 903
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
Figure 5. Schematic flow diagram of 44 non-neoplastic lesions.
TABLE 1. Per-lesion and per-patient analyses for category 3 and 4 gastric cancers Sensitivity (95% CI)
TABLE 2. Per-lesion and per-patient analyses for category 4 gastric cancers
Specificity (95% CI)
Per-lesion analysis for gastric cancers (categories 3 and 4)
Sensitivity (95% CI)
Specificity (95% CI)
Per-lesion analysis for gastric cancers (category 4)
WLE general diagnosis
76.6% (64.5%-88.7%)
84.3% (74.3%-94.3%)
AFI diagnosis
68.1% (54.8%-81.4%)
23.5% (11.9%-35.2%)
TME diagnosis
89.4% (80.5%-98.2%)
98% (94.2%-100%)
Per-patient analysis for gastric cancers (categories 3 and 4)
WLE general diagnosis
77.1% 84.3% (63.2%-91.1%) (74.3%-94.3%)
AFI diagnosis
62.9% 23.5% (46.8%-78.9%) (11.9%-35.2%)
TME diagnosis
91.4% 98% (82.2%-100%) (94.2%-100%)
Per-patient analysis for gastric cancers (category 4) WLE general diagnosis
75% (60%-90%)
72.2% (51.5%-92.9%)
WLE general diagnosis
75% (62.2%-87.8%)
72.2% (51.5%-92.9%)
AFI diagnosis
62.5% 38.9% (45.7%-79.3%) (16.4%-61.4%)
AFI diagnosis
68.2% (54.4%-81.4%)
44.4% (21.5%-67.4%)
TME diagnosis
90.6% 100% (80.5%-100%) (79.3%-100%)
TME diagnosis
90.9% (82.4%-99.4%)
100% (79.3%-100%)
CI, Confidence interval; WLE, white light endoscopy; AFI, autofluorescence imaging; TME, trimodal imaging endoscopy.
CI, Confidence interval; WLE, white light endoscopy; AFI, autofluorescence imaging; TME, trimodal imaging endoscopy.
increases the rate of detection by 13% of early gastric neoplasias (89.4%) over WLI. ME-NBI is considered the mainstay of TME in enhancing diagnostic accuracy for early gastric neoplasia because it re-
duced the high rate of false positives associated with WLI and AFI to a negligible level (2.0% of the enrolled pathneoplasias by a per-lesion analysis). Diagnoses reached with WLI and AFI rely on alterations in color and shape of target lesions, which are relatively nonspecific for
904 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
neoplasias. Thus, a significant number of false positives is inevitable when using these imaging modalities. In contrast, ME-NBI has the advantage of yielding clear images of the FMS and microvessels, which represent neoplasia-specific abnormalities, and can achieve endoscopic pathology.16 Therefore, in TME, ME-NBI shows high specificity for the diagnosis of early gastric neoplasia and correctly diagnoses the false-positive lesions obtained with WLI and AFI. To realize the advantages of ME-NBI, endoscopists need the advanced techniques of magnifying endoscopy to yield clear images of the FMS and microvessels of target lesions. In this study, we selected 2 endoscopists who were well trained and experienced in using the modality. Moreover, the Japanese population used in this study was highly enriched for cases of early gastric neoplasia. All enrolled patients except one had atrophic gastritis, which is very different from a Western population. Therefore, the current results derived for TME cannot be applied readily to standard clinical settings with general endoscopists in countries where the risk of gastric neoplasia is not as high. Furthermore, endoscopists must have the ability to diagnose gastric neoplasia by recognizing abnormalities of the microstructures obtained by ME-NBI. Thus, an additional multicenter trial in which endoscopists from various hospitals with different experience participate is necessary to confirm the clinical relevance of TME, which includes the advanced modality of ME-NBI. Furthermore, we need clear and definite diagnostic criteria for ME-NBI of early gastric neoplasia to overcome potential interobserver differences in diagnosing with MENBI. We applied the criteria for ME-NBI diagnosis in this study, ie, FMS disappearance and irregular microvessels with dilation and heterogeneity. Because irregular microvessels and FMS disappearance are often undetectable in elevated gastric neoplasias, irregular and micrified FMS was used as an additional criterion for elevated lesions. Although we found a high level of diagnostic accuracy of ME-NBI when using these criteria,18 we need to establish definitive and universal diagnostic criteria for ME-NBI of early gastric neoplasia. The contribution of AFI to TME was found to be limited in this study. The rate of additional detection of early gastric neoplasias with AFI over WLI was low at 6%. In a recent first report on TME for the detection of early neoplasia in Barrett’s esophagus (BE), AFI exhibited considerable additive value (62%) over WLI in detecting early BE neoplasia; WLI detected 63 early BE neoplasias, whereas AFI detected 102 early BE neoplasias that were not detected by WLI. It remains unclear why the contributions of AFI to TME are different in these studies.22 Possible explanations are differences in target organs and neoplasias and/or differences in study design (multicenter study vs single-center study). A multicenter study of TME for the detection of early gastric neoplasia may disclose the contribution of AFI and the clinical relevance of TME. In this study, we enrolled some patients without neoplasias but who were undergoing follow-up endoscopy
after ESD. Detecting such an ESD scar may have caused the participating endoscopists to know that the patient had been treated already and therefore conclude that the likelihood of finding lesions was small. Conversely, when such a scar was not found, the endoscopist might have performed a better/longer inspection to find lesions. This possible bias is one limitation of this study. However, we know that metachronous gastric neoplasias after ESD are not infrequent,23 and gastric ulcer scars are frequently found in patients with gastric cancers. In this study, 5 of 44 patients with gastric neoplasia had metachronous neoplasias after ESD, meaning that they had ESD scars. Moreover, 17 of 44 patients with gastric neoplasia had benign gastric ulcer scars other than from ESD. Therefore, the existence of ulcer scars might not have caused a bias that affected the results of this study. In conclusion, the addition of AFI and ME-NBI to WLI increases the detection rate of gastric neoplasia by 12.8%. Most of the false-positive lesions obtained by AFI and WLI are correctly diagnosed by ME-NBI. Higher diagnostic accuracy of TME over conventional WLI indicates the feasibility of TME for the efficacious diagnosis of early gastric neoplasia.
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 905
REFERENCES 1. Rall DP, Loo TL, Lane M, et al. Appearance and persistence of fluorescent material in tumor tissue after tetracycline administration. J Natl Cancer Inst 1957;19:79-85. 2. Panjehpour M, Overholt BF, Vo-Dinh T, et al. Endoscopic fluorescence detection of high-grade dysplasia in Barrett’s esophagus. Gastroenterology 1996;111:93-101. 3. Vo-Dinh T, Panjehpour M, Overholt BF. Laser-induced fluorescence for esophageal cancer and dysplasia diagnosis. Ann N Y Acad Sci 1998;838:116-22. 4. Mayinger B, Neidhardt S, Reh H, et al. Fluorescence induced with 5-aminolevulinic acid for the endoscopic detection and follow-up of esophageal lesions. Gastrointest Endosc 2001;54:572-8. 5. Mayinger B, Horner P, Jordan M, et al. Light-induced autofluorescence spectroscopy for the endoscopic detection of esophageal cancer. Gastrointest Endosc 2001;54:195-201. 6. Kara MA, Peters FP, Ten Kate FJ, et al. Endoscopic video autofluorescence imaging may improve the detection of early neoplasia in patients with Barrett’s esophagus. Gastrointest Endosc 2005;61: 679-85. 7. Kobayashi M, Tajiri H, Seike E, et al. Detection of early gastric cancer by a real-time autofluorescence imaging system. Cancer Lett 2001;165: 155-9. 8. Chwirot BW, Chwirot S, Jedrzejczyk W, et al. Ultraviolet laser-induced fluorescence of human stomach tissues: detection of cancer tissues by imaging techniques. Lasers Surg Med 1997;21:149-58. 9. Berk JE, Burlant WJ, Irland MJ, et al. Description of fluorometer for use in fluorescence test for gastric cancer. Bull Gastrointest Endosc 1964;10:16-7. 10. Bhunchet E, Hatakawa H, Sakai Y, et al. Fluorescein electronic endoscopy: a novel method for detection of early stage gastric cancer not evident to routine endoscopy. Gastrointest Endosc 2002;55:562-71. 11. Izuishi K, Tajiri H, Fujii T, et al. The histological basis of detection of adenoma and cancer in the colon by autofluorescence endoscopic imaging. Endoscopy 1999;31:511-6.
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
12. Wang TD, Crawford JM, Feld MS, et al. In vivo identification of colonic dysplasia using fluorescence endoscopic imaging. Gastrointest Endosc 1999;49:447-55. 13. Schomacker KT, Frisoli JK, Compton CC, et al. Ultraviolet laser-induced fluorescence of colonic polyps. Gastroenterology 1992;102:1155-60. 14. Kapadia CR, Cutruzzola FW, O’Brien KM, et al. Laser-induced fluorescence spectroscopy of human colonic mucosa. Detection of adenomatous transformation. Gastroenterology 1990;99:150-7. 15. Kato M, Kaise M, Yonezawa J, et al. Autofluorescence endoscopy versus conventional white light endoscopy for the detection of superficial gastric neoplasia: a prospective comparative study. Endoscopy 2007;39:937-41. 16. 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. 17. Agresti A, Coull BA. Approximate is better than ‘‘exact’’ for interval estimation of binomial proportions. Am Stat 1998;52:119-26. 18. Kaise M, Kato M, Urashima M, et al. Magnifying endoscopy combined with narrow-band imaging for differential diagnosis of superficial depressed gastric lesions. Endoscopy 2009;41:310-5. 19. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;47:251-5. 20. Dixon MF. Gastrointestinal epithelial neoplasia: Vienna revisited. Gut 2002;51:130-1.
21. Gotoda T, Yamamoto H, Soetikno RM. Endoscopic submucosal dissection of early gastric cancer. J Gastroenterol 2006;41:929-42. 22. Curvers WL, Singh R, Song LM, et al. Endoscopic tri-modal imaging for detection of early neoplasia in Barrett’s oesophagus: a multi-centre feasibility study using high-resolution endoscopy, autofluorescence imaging and narrow band imaging incorporated in one endoscopy system. Gut 2008;57:167-72. 23. Nakajima T, Oda I, Gotoda T, et al. Metachronous gastric cancers after endoscopic resection: how effective is annual endoscopic surveillance? Gastric Cancer 2006;9:93-8.
906 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Received September 30, 2008. Accepted March 27, 2009. Current affiliations: Department of Endoscopy (M. Kato, M. Kaise, J.Y., K.G., H. Toyoizumi, N.Y., Y.Y., M.K., H. Tajiri), Department of Gastroenterology and Hepatology (H. Tajiri), Jikei University School of Medicine, Tokyo, Japan. Reprint requests: Mitsuru Kaise, MD, Department of Endoscopy, Jikei University School of Medicine, 3-25-8 Nishishinbashi, Minato-ku, Tokyo 105-8461, Japan. If you would like to chat with an author of this article, you may contact him at [email protected].
Trimodal imaging endoscopy may improve diagnostic accuracy of early gastric neoplasia: a feasibility study Masayuki Kato, MD, PhD, Mitsuru Kaise, MD, PhD, Jin Yonezawa, MD, Kenichi Goda, MD, PhD, Hirobumi Toyoizumi, MD, Noboru Yoshimura, MD, Yukinaga Yoshida, MD, Muneo Kawamura, MD, PhD, Hisao Tajiri, MD, PhD Tokyo, Japan
Background: A considerable number of superficial gastric neoplasias are overlooked with conventional white light imaging (WLI) endoscopy. Objective: The aim was to investigate the diagnostic potential of trimodal imaging endoscopy (TME), which combines WLI, autofluorescence imaging (AFI), and narrow-band imaging (NBI), for superficial gastric neoplasia. Design: Feasibility study. Setting: Single academic center. Patients: Sixty-two patients with or without gastric neoplasia. Intervention: Each patient serially assessed with WLI, AFI, and magnifying endoscopy with NBI (ME-NBI) by an endoscopist blinded for clinical information. ME-NBI over WLI and AFI was designated as TME. Histopathology of biopsy and ESD specimens was evaluated and used as the gold standard. Main Outcome Measurements: Sensitivity and specificity of endoscopic diagnosis of pathology-proven neoplasia by per-patient and per-lesion analyses. Results: The study included 47 pathology-proven neoplasias and 44 pathology-proven nonneoplasias that were detected as neoplasias with any of the modalities. By a per-lesion analysis, the sensitivity of TME (89.4%) was higher than that of WLI (76.6%) and AFI (68.1%). The specificity of TME (98.0%) was higher than that of WLI (84.3%) and AFI (23.5%). By a per-patient analysis, the sensitivity of TME (90.9%) was higher than that of WLI (75%) and AFI (68.2%). The specificity of TME (100%) was higher than that of WLI (72.2%) and AFI (44.4%). Limitations: Case-enriched population at a single center. Conclusions: Higher diagnostic accuracy of TME over conventional WLI indicates the feasibility of TME for the efficacious diagnosis of early gastric neoplasia. (Gastrointest Endosc 2009;70:899-906.)
Efficacious detection of early neoplasia with endoscopy increases the curability of GI neoplasias and decreases
Copyright ª 2009 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2009.03.1171
associated morbidities. Autofluorescence endoscopy (AFE) for the detection of early neoplasias has attracted considerable attention since the development of a fluorescence detection method for cancerous lesions.1 AFE differs from white light imaging (WLI) in that AFE has been shown to be more accurate for the diagnosis of neoplasia, as described in studies of early neoplasia of the esophagus,2-6 stomach,7-10 and colon.11-14 We previously performed a prospective feasibility study to compare WLI with AFE by using autofluorescence imaging (AFI) for the detection of early gastric neoplasia.15 We found that (1) WLI missed or misdiagnosed 25% of early gastric neoplasias and AFI detected 10% more neoplasias that were not identified by WLI and (2) the sensitivity of
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 899
Abbreviations: AFE, autofluorescence endoscopy; AFI, autofluorescence imaging; BE, Barrett’s esophagus; CI, confidence interval; ESD, endoscopic submucosal dissection; FMS, fine mucosal structure; MENBI, magnifying endoscopy with narrow-band imaging; NBI, narrowband imaging; path-neoplasia, pathology-proven neoplasia; TME, trimodal imaging endoscopy; WLI, white light imaging. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication.
TGE may improve diagnostic accuracy of early gastric neoplasia
WLI and that of AFI were comparable, whereas the specificity of AFI was considerably lower than that of WLI because of a high number of false positives. Therefore, we concluded that AFI is of limited clinical value. It is noteworthy that AFI detected 10% more lesions compared with WLI, which missed or misdiagnosed 25% of early gastric neoplasias. A combination of AFI and a different modality that would decrease the high rate of false positives might be a promising endoscopic technique for the early detection of gastric neoplasia. A candidate modality for improving specificity is magnifying endoscopy with narrow-band imaging (ME-NBI).16 Therefore, we conducted a prospective feasibility study to investigate the diagnostic potential of trimodal imaging endoscopy (TME), which combines WLI, AFI, and NBI, for early gastric neoplasia.
PATIENT AND METHODS Patient population Patients with early gastric neoplasia who were referred for endoscopic submucosal dissection (ESD) by affiliated hospitals or clinics from April to December 2007 were eligible for this study. Patients with obvious advanced gastric carcinomas and cancerous lesions with deep invasion to the gastric submucosa were excluded. Eligible patients who gave written informed consent were serially enrolled in the study and scheduled for a preoperative advanced workup including TME. To decrease the bias of only including patients with known gastric neoplasia, patients who were undergoing follow-up endoscopy after ESD were also recruited as controls because these subjects were suspected of being free of neoplasia.
Kato et al
Capsule Summary What is already known on this topic d
Autofluorescence endoscopy (AFI) has been shown to be more accurate than white light imaging (WLI), which misses or misdiagnoses 25% of early gastric neoplasias.
What this study adds to our knowledge d
In 62 patients serially assessed with WLI, AFI, and magnifying endoscopy with narrow band imaging (ME-NBI), the addition of AFI and ME-NBI to WLI increased the detection rate of early gastric neoplasias by 12.8%.
the superficial mucosa. For NBI observations, endoscopy at a magnification of as high as 80 was performed. In the AFI mode, light emitted from the xenon lamp is directed to the rotary filter, which splits it into excitation wavelengths of 390 to 470 nm and green light of 540 to 560 nm wavelengths. The AFI endoscope (XGIF-Q240FZ; Olympus Medical Systems) incorporates a monochrome charged couple device with a barrier filter to exclude the excitation light and capture only the weak autofluorescence. A pseudocolored image is reconstructed based on the autofluorescence input signals, such that high-intensity autofluorescence is green and low-intensity autofluorescence is magenta. In general, the AFI system displays normal pyloric mucosa or atrophic fundic mucosa in green, blood vessels in dark green, and hypertrophic fundic mucosa or elevated neoplasia in magenta.
Overall study design
The endoscopic system consists of a high-resolution white light endoscope with optical zoom (80 magnification, XGIF-Q240FZ; Olympus Medical Systems Co, Tokyo, Japan), which can be switched easily to the AFI or NBI modes. Although the basic configuration is identical to that of the standard video-endoscopy system (LUCERA CV-260/CLV-260; Olympus Medical System), this trimodal system allows either red, green, blue illumination (for WLI and NBI) or an excitation/reflected light illumination combination (for AFI). The light source incorporates a rotary filter, which is designed in a double-wheel configuration with 2 concentric wheels: a red, green, blue filter wheel for WLI and NBI and an AFI filter wheel. In the NBI mode, the light source (CLV-U40D; Olympus Medical Systems) for this endoscope is equipped with narrowband filters corresponding to red (485-515 nm), green (430-460 nm), and blue (400-430 nm) light. Short wavelength light in the blue range is absorbed by hemoglobin in vivo, thereby enhancing the appearance of capillaries in
Although the endoscopists were informed beforehand that the study population was enriched and included patients with and without gastric neoplasia, they had no access to any clinical information before endoscopy. Patients underwent TME under conscious sedation with pethidine hydrochloride and midazolam. An examiner serially performed WLI, AFI, and ME-NBI under the supervision of the study coordinator, who organized the patient data. The examiner recorded the presence or absence of lesions suspected to be neoplasias with WLI, AFI, and TME, in accordance with the aforementioned criteria. All lesions recorded as neoplasias with any of the 3 imaging modalities (endoscopy neoplasias) and neoplasias referred for ESD but not detected by any of the imaging modalities underwent biopsies under the supervision of the study coordinator, who attended all endoscopic examinations and was aware of the endoscopist’s diagnoses. The pathologic results obtained from the biopsy samples and ESD specimens were used as the criterion standard for diagnostic accuracy. The sensitivity and specificity for endoscopic diagnosis of pathology-proven neoplasias (path-neoplasias) were evaluated by per-patient and
900 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
The TME system
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
Figure 1. Diagnostic criteria of AFE. We defined a neoplasia-suspected lesion by AFE (AFE-positive) as an area or lesion that was clearly different from the surrounding mucosa in color and that showed a clearly defined margin. A, WLE image of a superficial gastric carcinoma. B, AFE image displaying superficial carcinoma as a magenta lesion with defined margins within green-colored mucosa, thus diagnosed as AFE-positive.
per-lesion analyses. If a patient had 2 or more path-neoplasias, with at least 1 diagnosed as endoscopy-nonneoplasia, the diagnosis was defined as missed on a per-patient analysis. The study was approved by the ethics committee of the Jikei University Hospital and was conducted in accordance with the revised Declaration of Helsinki (1989). Sensitivity and specificity were calculated by standard formulas for a binominal proportion, and 95% confidence intervals (CIs) were calculated by the Wald interval method.17 Because of the preliminary focus of this study and the synthetic mixture of the composition of the study population, we considered that the statistical analysis results should be characterized as being descriptive only. Therefore, we only presented sensitivities and specificities and their 95% CIs and did not analyze the statistical differences among sensitivities and specificities of these tested endoscopic modalities.
Diagnostic criteria for gastric neoplasia A lesion was recorded as WLI endoscopy neoplasia when it was recognized or suspected to be early gastric neoplasia based on color and appearance when using WLI. We defined an AFI endoscopy neoplasia as an area or lesion that was different from the surrounding mucosa in the color of AFI and had a defined circumferential margin (Fig. 1). If a lesion diagnosed as WLI endoscopy neoplasia did not fulfill the definition of AFI endoscopy neoplasia, it was diagnosed as AFI endoscopy-nonneoplasia. We assessed ME-NBI endoscopy neoplasia by using our diagnostic criteria for early gastric neoplasia.18 A superficial lesion demarcated from the surrounding mucosa with differences in the fine mucosal structure (FMS) and/or microvasclature was diagnosed as endoscopy neoplasia if FMS disappearance, microvascular dilation, and heterogeneity (the triad of microstructure) were seen (Fig. 2). Superficial elevated lesions were diagnosed as endoscopy neoplasias if the triad was observed or an irregular FMS with micrification and heterogeneity www.giejournal.org
was demonstrated. If a lesion diagnosed as WLI endoscopy neoplasia and/or AFI endoscopy neoplasia did not fulfill the definition of ME-NBI endoscopy neoplasia, it was diagnosed as TME endoscopy nonneoplasia. Although ME-NBI generally targeted lesions that were diagnosed as endoscopy neoplasia by WLI and/or AFI, ME-NBI observation was also done for lesions diagnosed as WLI and AFI endoscopy nonneoplasia. If a lesion diagnosed as WLI and AFI endoscopy nonneoplasia fulfilled the definition of ME-NBI endoscopy neoplasia, it was diagnosed as TME endoscopy neoplasia. Because AFI and ME-NBI differ considerably from WLI, endoscopists who had extensive experience in diagnosing early gastric neoplasia by AFI and ME-NBI in more than 100 cases were selected.
Histopathology All biopsy specimens were fixed in 10% formalin and embedded in paraffin, serially sectioned, and stained with hematoxylin and eosin. Lesions diagnosed as category 3 low-grade neoplasias, according to the revised Vienna classification,19,20 were classified as gastric adenomas. Lesions diagnosed as category 4 neoplasias were defined as gastric carcinomas. Patients with category 3 or 4 neoplasias were enrolled in this study as having gastric neoplasias. ESD specimens were pinned flat onto a corkboard with adequate tension and immersed in formalin. After fixation, specimens were cut into 2-mmwide longitudinal strips and examined pathologically. All histopathologic findings were reviewed by expert GI pathologists.
RESULTS Sixty-two patients, composed of 44 patients with gastric neoplasia referred for ESD and 18 patients who were undergoing follow-up EGD after ESD, were enrolled in this Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 901
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
Figure 2. We defined diagnostic criteria for superficial gastric neoplasia by ME-NBI. If there were disappearance of fine mucosa structure and microvascular irregularities showing dilation, abrupt caliber alteration, heterogeneity in shape, and tortousness, the lesion was diagnosed as neoplastic.
Figure 3. Flow diagram of patients enrolled for analysis.
study. Eighty-eight gastric lesions in 55 patients were diagnosed as endoscopy neoplasias; 44 and 44 of these 88 lesions were pathologically defined as neoplasias and nonneoplasias, respectively. Three early gastric neoplasias that were diagnosed and referred for ESD by affiliated hospitals were not diagnosed as endoscopy neoplasias by any of the imaging modalities in the study. These 3 neoplasias
were included in 7 patients who were diagnosed as not having gastric neoplasia (Fig. 3). Five of 44 patients with gastric neoplasia enrolled in the study were diagnosed after previous ESD. These neoplasias were considered to be metachronous gastric neoplasias after ESD. Forty-three of 44 patients with gastric neoplasia had gastric atrophy and 17 of 44 patients had
902 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
Figure 4. Schematic flow diagram of 47 neoplastic lesions.
concomitant benign gastric ulcer scars other than ESD scars. Three of 44 patients with gastric neoplasia had double gastric neoplasias. Of the 47 gastric neoplasias found in 44 patients with gastric neoplasia, 35 were gastric cancers (category 4) and 12 were gastric adenomas (category 3).
Tree diagram analysis WLI identified 36 (76.6%) of 47 enrolled path-neoplasias, and AFI identified 3 (6.4%) additional neoplasias not detected by WLI. TME identified 42 (89.4%) patients with 6 (12.8%) additional neoplasias that were not detected by WLI (Fig. 4). WLI and AFI diagnosed 8 and 39 pathology-proven nonneoplasias as endoscopy-neoplasias, respectively. Forty-three of the 44 false-positive lesions by WLI and/or AFI were finally diagnosed as nonneoplasia by ME-NBI, and only one pathology-proven nonneoplasia was diagnosed as endoscopy neoplasia by TME (Fig. 5).
(100%; 95% CI, 79.3%-100%) was higher than that of WLI (72.2%; 95% CI, 51.5%-92.9%) and AFI (44.4%; 95% CI, 21.5%-67.4%) (Table 1). We also analyzed the diagnostic accuracy for 35 gastric cancers when 12 gastric adenomas were excluded. By a per-lesion analysis for the diagnosis of gastric cancers, the sensitivity of TME (91.4%; 95% CI, 82.2%-100%) was higher than that of WLI (77.1%; 95% CI, 63.2%-91.1%) and AFI (62.9%; 95% CI, 46.8%-78.9%). The specificity of TME (98.0%; 95% CI, 94.2%-100%) was higher than that of WLI (84.3%; 95% CI, 74.3%-94.3%) and AFI (23.5%; 95% CI, 11.9%-35.2%). By a per-patient analysis, the sensitivity of TME (90.6%; 95% CI, 80.5%-100%) was higher than that of WLI (75%; 95% CI, 60%-90%) and AFI (62.5%; 95% CI, 45.7%-79.3%). The specificity of TME (100%; 95% CI, 79.3%-100%) was higher than that of WLI (72.2%; 95% CI, 51.5%-92.9%) and AFI (38.9%; 95% CI, 16.4%-61.4%) (Table 2).
DISCUSSION Comparison of the diagnostic accuracy of WLI and TME for gastric neoplasias By a per-lesion analysis, the sensitivity of TME (89.4%; 95% CI, 80.5%-98.2%) was higher than that of WLI (76.6%; 95% CI, 64.5%-88.7%) and AFI (68.1%; 95% CI, 54.8%-81.4%). The specificity of TME (98.0%; 95% CI, 94.2%-100%) was higher than that of WLI (84.3%; 95% CI, 74.3%-94.3%) and AFI (23.5%; 95% CI, 11.9%-35.2%). By a per-patient analysis, the sensitivity of TME (90.9%; 95% CI, 82.4%-99.4%) was higher than that of WLI (75%; 95% CI, 62.2%-87.8%) and AFI (68.2%; 95% CI, 54.4%-81.4%). The specificity of TME
Early gastric neoplasias are good indications for endoscopic resection21 and are targets for endoscopic surveillance to decrease the morbidity associated with gastric neoplasia. In this study, the conventional endoscopic modality of WLI identified 76.6% of early gastric neoplasias, of which approximately one fourth were not correctly diagnosed, even by using high-resolution endoscopy for WLI. The detection rate is comparable to that found in our previous study,15 confirming the limited value of WLI for diagnosing early gastric neoplasias. This study shows that TME
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 903
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
Figure 5. Schematic flow diagram of 44 non-neoplastic lesions.
TABLE 1. Per-lesion and per-patient analyses for category 3 and 4 gastric cancers Sensitivity (95% CI)
TABLE 2. Per-lesion and per-patient analyses for category 4 gastric cancers
Specificity (95% CI)
Per-lesion analysis for gastric cancers (categories 3 and 4)
Sensitivity (95% CI)
Specificity (95% CI)
Per-lesion analysis for gastric cancers (category 4)
WLE general diagnosis
76.6% (64.5%-88.7%)
84.3% (74.3%-94.3%)
AFI diagnosis
68.1% (54.8%-81.4%)
23.5% (11.9%-35.2%)
TME diagnosis
89.4% (80.5%-98.2%)
98% (94.2%-100%)
Per-patient analysis for gastric cancers (categories 3 and 4)
WLE general diagnosis
77.1% 84.3% (63.2%-91.1%) (74.3%-94.3%)
AFI diagnosis
62.9% 23.5% (46.8%-78.9%) (11.9%-35.2%)
TME diagnosis
91.4% 98% (82.2%-100%) (94.2%-100%)
Per-patient analysis for gastric cancers (category 4) WLE general diagnosis
75% (60%-90%)
72.2% (51.5%-92.9%)
WLE general diagnosis
75% (62.2%-87.8%)
72.2% (51.5%-92.9%)
AFI diagnosis
62.5% 38.9% (45.7%-79.3%) (16.4%-61.4%)
AFI diagnosis
68.2% (54.4%-81.4%)
44.4% (21.5%-67.4%)
TME diagnosis
90.6% 100% (80.5%-100%) (79.3%-100%)
TME diagnosis
90.9% (82.4%-99.4%)
100% (79.3%-100%)
CI, Confidence interval; WLE, white light endoscopy; AFI, autofluorescence imaging; TME, trimodal imaging endoscopy.
CI, Confidence interval; WLE, white light endoscopy; AFI, autofluorescence imaging; TME, trimodal imaging endoscopy.
increases the rate of detection by 13% of early gastric neoplasias (89.4%) over WLI. ME-NBI is considered the mainstay of TME in enhancing diagnostic accuracy for early gastric neoplasia because it re-
duced the high rate of false positives associated with WLI and AFI to a negligible level (2.0% of the enrolled pathneoplasias by a per-lesion analysis). Diagnoses reached with WLI and AFI rely on alterations in color and shape of target lesions, which are relatively nonspecific for
904 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Kato et al
TGE may improve diagnostic accuracy of early gastric neoplasia
neoplasias. Thus, a significant number of false positives is inevitable when using these imaging modalities. In contrast, ME-NBI has the advantage of yielding clear images of the FMS and microvessels, which represent neoplasia-specific abnormalities, and can achieve endoscopic pathology.16 Therefore, in TME, ME-NBI shows high specificity for the diagnosis of early gastric neoplasia and correctly diagnoses the false-positive lesions obtained with WLI and AFI. To realize the advantages of ME-NBI, endoscopists need the advanced techniques of magnifying endoscopy to yield clear images of the FMS and microvessels of target lesions. In this study, we selected 2 endoscopists who were well trained and experienced in using the modality. Moreover, the Japanese population used in this study was highly enriched for cases of early gastric neoplasia. All enrolled patients except one had atrophic gastritis, which is very different from a Western population. Therefore, the current results derived for TME cannot be applied readily to standard clinical settings with general endoscopists in countries where the risk of gastric neoplasia is not as high. Furthermore, endoscopists must have the ability to diagnose gastric neoplasia by recognizing abnormalities of the microstructures obtained by ME-NBI. Thus, an additional multicenter trial in which endoscopists from various hospitals with different experience participate is necessary to confirm the clinical relevance of TME, which includes the advanced modality of ME-NBI. Furthermore, we need clear and definite diagnostic criteria for ME-NBI of early gastric neoplasia to overcome potential interobserver differences in diagnosing with MENBI. We applied the criteria for ME-NBI diagnosis in this study, ie, FMS disappearance and irregular microvessels with dilation and heterogeneity. Because irregular microvessels and FMS disappearance are often undetectable in elevated gastric neoplasias, irregular and micrified FMS was used as an additional criterion for elevated lesions. Although we found a high level of diagnostic accuracy of ME-NBI when using these criteria,18 we need to establish definitive and universal diagnostic criteria for ME-NBI of early gastric neoplasia. The contribution of AFI to TME was found to be limited in this study. The rate of additional detection of early gastric neoplasias with AFI over WLI was low at 6%. In a recent first report on TME for the detection of early neoplasia in Barrett’s esophagus (BE), AFI exhibited considerable additive value (62%) over WLI in detecting early BE neoplasia; WLI detected 63 early BE neoplasias, whereas AFI detected 102 early BE neoplasias that were not detected by WLI. It remains unclear why the contributions of AFI to TME are different in these studies.22 Possible explanations are differences in target organs and neoplasias and/or differences in study design (multicenter study vs single-center study). A multicenter study of TME for the detection of early gastric neoplasia may disclose the contribution of AFI and the clinical relevance of TME. In this study, we enrolled some patients without neoplasias but who were undergoing follow-up endoscopy
after ESD. Detecting such an ESD scar may have caused the participating endoscopists to know that the patient had been treated already and therefore conclude that the likelihood of finding lesions was small. Conversely, when such a scar was not found, the endoscopist might have performed a better/longer inspection to find lesions. This possible bias is one limitation of this study. However, we know that metachronous gastric neoplasias after ESD are not infrequent,23 and gastric ulcer scars are frequently found in patients with gastric cancers. In this study, 5 of 44 patients with gastric neoplasia had metachronous neoplasias after ESD, meaning that they had ESD scars. Moreover, 17 of 44 patients with gastric neoplasia had benign gastric ulcer scars other than from ESD. Therefore, the existence of ulcer scars might not have caused a bias that affected the results of this study. In conclusion, the addition of AFI and ME-NBI to WLI increases the detection rate of gastric neoplasia by 12.8%. Most of the false-positive lesions obtained by AFI and WLI are correctly diagnosed by ME-NBI. Higher diagnostic accuracy of TME over conventional WLI indicates the feasibility of TME for the efficacious diagnosis of early gastric neoplasia.
www.giejournal.org
Volume 70, No. 5 : 2009 GASTROINTESTINAL ENDOSCOPY 905
REFERENCES 1. Rall DP, Loo TL, Lane M, et al. Appearance and persistence of fluorescent material in tumor tissue after tetracycline administration. J Natl Cancer Inst 1957;19:79-85. 2. Panjehpour M, Overholt BF, Vo-Dinh T, et al. Endoscopic fluorescence detection of high-grade dysplasia in Barrett’s esophagus. Gastroenterology 1996;111:93-101. 3. Vo-Dinh T, Panjehpour M, Overholt BF. Laser-induced fluorescence for esophageal cancer and dysplasia diagnosis. Ann N Y Acad Sci 1998;838:116-22. 4. Mayinger B, Neidhardt S, Reh H, et al. Fluorescence induced with 5-aminolevulinic acid for the endoscopic detection and follow-up of esophageal lesions. Gastrointest Endosc 2001;54:572-8. 5. Mayinger B, Horner P, Jordan M, et al. Light-induced autofluorescence spectroscopy for the endoscopic detection of esophageal cancer. Gastrointest Endosc 2001;54:195-201. 6. Kara MA, Peters FP, Ten Kate FJ, et al. Endoscopic video autofluorescence imaging may improve the detection of early neoplasia in patients with Barrett’s esophagus. Gastrointest Endosc 2005;61: 679-85. 7. Kobayashi M, Tajiri H, Seike E, et al. Detection of early gastric cancer by a real-time autofluorescence imaging system. Cancer Lett 2001;165: 155-9. 8. Chwirot BW, Chwirot S, Jedrzejczyk W, et al. Ultraviolet laser-induced fluorescence of human stomach tissues: detection of cancer tissues by imaging techniques. Lasers Surg Med 1997;21:149-58. 9. Berk JE, Burlant WJ, Irland MJ, et al. Description of fluorometer for use in fluorescence test for gastric cancer. Bull Gastrointest Endosc 1964;10:16-7. 10. Bhunchet E, Hatakawa H, Sakai Y, et al. Fluorescein electronic endoscopy: a novel method for detection of early stage gastric cancer not evident to routine endoscopy. Gastrointest Endosc 2002;55:562-71. 11. Izuishi K, Tajiri H, Fujii T, et al. The histological basis of detection of adenoma and cancer in the colon by autofluorescence endoscopic imaging. Endoscopy 1999;31:511-6.
TGE may improve diagnostic accuracy of early gastric neoplasia
Kato et al
12. Wang TD, Crawford JM, Feld MS, et al. In vivo identification of colonic dysplasia using fluorescence endoscopic imaging. Gastrointest Endosc 1999;49:447-55. 13. Schomacker KT, Frisoli JK, Compton CC, et al. Ultraviolet laser-induced fluorescence of colonic polyps. Gastroenterology 1992;102:1155-60. 14. Kapadia CR, Cutruzzola FW, O’Brien KM, et al. Laser-induced fluorescence spectroscopy of human colonic mucosa. Detection of adenomatous transformation. Gastroenterology 1990;99:150-7. 15. Kato M, Kaise M, Yonezawa J, et al. Autofluorescence endoscopy versus conventional white light endoscopy for the detection of superficial gastric neoplasia: a prospective comparative study. Endoscopy 2007;39:937-41. 16. 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. 17. Agresti A, Coull BA. Approximate is better than ‘‘exact’’ for interval estimation of binomial proportions. Am Stat 1998;52:119-26. 18. Kaise M, Kato M, Urashima M, et al. Magnifying endoscopy combined with narrow-band imaging for differential diagnosis of superficial depressed gastric lesions. Endoscopy 2009;41:310-5. 19. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;47:251-5. 20. Dixon MF. Gastrointestinal epithelial neoplasia: Vienna revisited. Gut 2002;51:130-1.
21. Gotoda T, Yamamoto H, Soetikno RM. Endoscopic submucosal dissection of early gastric cancer. J Gastroenterol 2006;41:929-42. 22. Curvers WL, Singh R, Song LM, et al. Endoscopic tri-modal imaging for detection of early neoplasia in Barrett’s oesophagus: a multi-centre feasibility study using high-resolution endoscopy, autofluorescence imaging and narrow band imaging incorporated in one endoscopy system. Gut 2008;57:167-72. 23. Nakajima T, Oda I, Gotoda T, et al. Metachronous gastric cancers after endoscopic resection: how effective is annual endoscopic surveillance? Gastric Cancer 2006;9:93-8.
906 GASTROINTESTINAL ENDOSCOPY Volume 70, No. 5 : 2009
www.giejournal.org
Received September 30, 2008. Accepted March 27, 2009. Current affiliations: Department of Endoscopy (M. Kato, M. Kaise, J.Y., K.G., H. Toyoizumi, N.Y., Y.Y., M.K., H. Tajiri), Department of Gastroenterology and Hepatology (H. Tajiri), Jikei University School of Medicine, Tokyo, Japan. Reprint requests: Mitsuru Kaise, MD, Department of Endoscopy, Jikei University School of Medicine, 3-25-8 Nishishinbashi, Minato-ku, Tokyo 105-8461, Japan. If you would like to chat with an author of this article, you may contact him at [email protected].