- Email: [email protected]
Endoscopic trimodal imaging versus standard video endoscopy for detection of early Barrett's neoplasia: a multicenter, randomized, crossover study in general practice
ORIGINAL ARTICLE: Clinical Endoscopy
Endoscopic trimodal imaging versus standard video endoscopy for detection of early Barrett’s neoplasia: a multicenter, randomized, crossover study in general practice Wouter L. Curvers, MD, PhD, Frederike G. van Vilsteren, MD, Lubertus C. Baak, MD, PhD, Clarisse Böhmer, MD, PhD, Rosalie C. Mallant-Hent, MD, PhD, Anton H. Naber, MD, PhD, Arnout van Oijen, MD, PhD, Cyriel Y. Ponsioen, MD, PhD, Pieter Scholten, MD, Ed Schenk, MD, PhD, Erik Schoon, MD, PhD, Cees A. Seldenrijk, MD, PhD, Gerrit A. Meijer, MD, PhD, Fiebo J. ten Kate, MD, PhD, Jacques J. Bergman, MD, PhD Amsterdam, Hoofddorp, Almere, Hilversum, Alkmaar, Zwolle, Eindhoven, Nieuwegein, the Netherlands
Background: Endoscopic trimodal imaging (ETMI) may improve detection of early neoplasia in Barrett’s esophagus (BE). Studies with ETMI so far have been performed in tertiary referral settings only. Objective: To compare ETMI with standard video endoscopy (SVE) for the detection of neoplasia in BE patients with an intermediate-risk profile. Design: Multicenter, randomized, crossover study. Setting: Community practice. Patients and Methods: BE patients with confirmed low-grade intraepithelial neoplasia (LGIN) underwent both ETMI and SVE in random order (interval 6-16 weeks). During ETMI, BE was inspected with high-resolution endoscopy followed by autofluorescence imaging (AFI). All visible lesions were then inspected with narrowband imaging. During ETMI and SVE, visible lesions were sampled followed by 4-quadrant random biopsies every 2 cm. Main Outcome Measurements: Overall histological yield of ETMI and SVE and targeted histological yield of ETMI and SVE. Results: A total of 99 patients (79 men, 63 ⫾ 10 years) underwent both procedures. ETMI had a significantly higher targeted histological yield because of additional detection of 22 lesions with LGIN/high-grade intraepithelial neoplasia (HGIN)/carcinoma (Ca) by AFI. There was no significant difference in the overall histological yield (targeted ⫹ random) between ETMI and SVE. HGIN/Ca was diagnosed only by random biopsies in 6 of 24 patients and 7 of 24 patients, with ETMI and SVE, respectively. Abbreviations: AFI, autofluorescence imaging; AMC, Academic Medical Center; BE, Barrett’s esophagus; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, highresolution endoscopy; ID, indefinite for dysplasia; LGIN, low-grade intraepithelial neoplasia; NBI, narrow-band imaging; ND, no dysplasia; SVE, standard video endoscopy. DISCLOSURE: The following author disclosed a financial relationship relevant to this publication: Dr. Bergman: research grant from Olympus Inc, Tokyo, Japan. The other authors disclosed no financial relationships relevant to this publication. Copyright © 2011 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2010.10.014 Received August 31, 2010. Accepted October 8, 2010. Current affiliations: Department of Gastroenterology and Hepatology (W.L.C., F.G.v.V., C.Y.P., J.J.B.), Academic Medical Center, Amsterdam, Department of Gastroenterology and Hepatology (L.C.B.), Onze Lieve Vrouwe Gasthuis, Amsterdam, Department of Gastroenterology and Hepatology (C.B.), Spaarne Hospital, Hoofddorp, Department of Internal
www.giejournal.org
Medicine (R.C.M.-H.), Flevohospital, Almere, Department of Internal Medicine (A.H.N.), Tergooi Hospitals, Hilversum, Department of Gastroenterology and Hepatology (A.v.O.), Medical Center Alkmaar, Alkmaar, Department of Gastroenterology and Hepatology (P.S.), St. Lucas Andreas Hospital, Amsterdam, Department of Gastroenterology and Hepatology (E.S.), Isala Clinics, Zwolle, Department of Gastroenterology and Hepatology (E.S.), Catharina Hospital, Eindhoven, Department of Pathology (C.A.S.), St. Antonius Hospital, Nieuwegein, Department of Pathology (G.A.M.), Free University Medical Center, Amsterdam, Department of Pathology (F.J.t.K.), Academic Medical Center, Amsterdam, Amsterdam Gastroenterological Association (W.L.C., F.G.v.V., L.C.B., C.B., R.C.M.-H., A.H.N., A.v.O., C.Y.P., P.S., F.J.t.K., J.J.B.), Amsterdam, the Netherlands. Presented at Digestive Disease Week, 2010, New Orleans, Louisiana (Gastrointest Endosc 2010;71:AB142). Reprint requests: Wouter Curvers, MD, Department of Gastroenterology and Hepatology, Academic Medical Center Meibergdreef 9, 1105 AZ, Amsterdam, the Netherlands. If you would like to chat with an author of this article, you may contact Dr Curvers at [email protected].
Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
195
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
Limitations: Inspection, with high-resolution endoscopy and AFI, was performed sequentially. Conclusion: ETMI performed in a community-based setting did not improve the overall detection of dysplasia compared with SVE. The diagnosis of dysplasia is still being made in a significant number of patients by random biopsies. Patients with a confirmed diagnosis of LGIN have a significant risk of HGIN/Ca. (Clinical trial registration number: ISRCTN91816824; NTR867.) (Gastrointest Endosc 2011;73:195-203.)
Barrett’s esophagus (BE) is a precursor lesion of esophageal adenocarcinoma, the cancer with the fastest increasing incidence in the Western world.1,2 Esophageal adenocarcinoma has a poor prognosis when diagnosed at a symptomatic stage, whereas the prognosis is excellent for early yet generally asymptomatic stages.3 Malignant transformation of BE into carcinoma (Ca) is a gradual process that takes place through several histopathological grades classified as no dysplasia (ND), low-grade intraepithelial neoplasia (LGIN), and high-grade intraepithelial neoplasia (HGIN).4,5 Patients with known BE, therefore, undergo regular endoscopic surveillance for the detection of early neoplastic lesions.6,7 These early lesions are, however, poorly visible with standard endoscopy, and therefore random biopsy samples are obtained, which is labor intensive and neoplastic areas may still be missed because of sampling error. Advances in endoscopic imaging may improve the detection of these early neoplastic lesions.8 Recently, a new endoscopic system has become commercially available that incorporates white-light highresolution endoscopy (HRE), autofluorescence imaging (AFI), and narrow-band imaging (NBI)— endoscopic trimodal imaging (ETMI). We performed several uncontrolled studies showing that AFI may improve the detection of early neoplasia in BE, whereas NBI subsequently decreases the high false-positive rate of AFI.9-11 In a recent multicenter, randomized, crossover study that compared ETMI with standard video endoscopy (SVE), AFI significantly improved the targeted detection of early neoplasia in BE.12 This study was performed in a tertiary referral setting by experienced endoscopists in a high-risk Barrett’s neoplasia population. In a BE population with a lower rate of neoplasia and with endoscopists with less experience in the detection early Barrett’s neoplasia, the additional value of AFI and NBI may be different. The aim of the current study was therefore to compare the diagnostic potential of ETMI with that of SVE in a community practice setting including only BE patients with an intermediate-risk profile.
METHODS Study setting This study was performed in 8 teaching hospitals in the Netherlands and the Academic Medical Center (AMC), Amsterdam. All centers participate in an ongoing prospective BE registration. Before initiating the prospective reg196 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
Take-home Message ●
●
Multimodality imaging cannot yet replace random sampling in Barrett’s esophagus. Autofluorescence imaging may be helpful in the workup of patients with early Barrett’s neoplasia or for future risk stratification in Barrett’s surveillance. Patients with a confirmed diagnosis of low-grade intraepithelial neoplasia have an increased risk of neoplastic progression and should undergo intensified endoscopic follow-up or may be eligible for endoscopic therapy.
istration, a retrospective BE registration was performed. In each of the hospitals, all BE patients were identified by the pathology databases in the participating centers.13
Patients All patients with a diagnosis of LGIN made in the preceding 5 years were identified from the retrospective and prospective BE registration database. From these patients, pathology specimens of all procedures with a diagnosis of LGIN were retrieved. Two of 3 panel GI pathologists (C.A.S., G.A.M., F.J.t.K.), each with extensive experience in Barrett’s neoplasia, independently reviewed all retrieved slides.14,15 The presence of intestinal metaplasia and the presence and degree of dysplasia were assessed and graded according to the Vienna classification of GI epithelial neoplasia into D, indefinite for dysplasia (ID), LGIN, HGIN, or Ca.16 Patients were enrolled when they met the following inclusion criteria: (1) age older than 18 years, (2) previous diagnosis of BE defined as the presence of columnar-lined epithelium in the tubular esophagus with intestinal metaplasia on histology, (3) previous diagnosis of LGIN confirmed by at least 1 panel pathologist, (4) written informed consent provided. Patients were excluded for the following reasons: (1) presence of active erosive esophagitis grade B or worse according to the Los Angeles classification of erosive esophagitis; (2) at first endoscopy, the presence of an advanced lesion (eg, type 0-I or 0-III) suspicious for neoplasia that would not allow a delay in intervention for 6 weeks; (3) the presence of conditions precluding histological sampling of the esophagus. www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Endoscopists
were noted. Finally, all suspicious areas were inspected in detail with NBI by using magnification endoscopy. The overall NBI appearance was classified into 3 categories: suspicious for neoplasia, not suspicious for neoplasia, or indeterminate, based on the assessment of the mucosal pattern, the vascular pattern, and the presence of abnormal blood vessels. The inspection times with HRE, AFI, and NBI, as well as the targeted and random sampling time were recorded. SVE. SVE was performed with various standard video endoscopes (Olympus GIF-140, GIF-160 or GIFQ-160; Olympus Inc). The Barrett’s segment was inspected for the presence of areas suspicious for dysplasia. For all detected suspicious areas, the locations were recorded. The inspection time, targeted sampling time, and random sampling time were recorded. Biopsy protocol. From all abnormalities detected during ETMI and SVE, 2 targeted biopsy specimens were obtained. Finally, 4-quadrant random biopsy samples were obtained for every 2 cm of the Barrett’s segment, not taking samples from areas with suspicious lesions.
All procedures were performed by 9 board-certified gastroenterologists from the participating teaching hospitals (L.C.B., C.B., R.C.H.-M., A.H.N., A.v.O., C.Y.P., P.S., E.S., E.S.) who had no specific expertise in BE or advanced imaging technique before this study. To prepare for the study, all participating endoscopists received a training DVD with examples of neoplastic and non-neoplastic lesions in BE with HRE, AFI, and NBI and visited dedicated endoscopy programs at the AMC Amsterdam, where patients referred for early Barrett’s neoplasia were investigated with the ETMI system. All participating endoscopists performed at least 5 dedicated BE surveillance endoscopy lists with the ETMI system at their own center and participated in interobserver studies on NBI.17-19
Technical background of the ETMI endoscopy system The ETMI system consisted of a high-resolution whitelight endoscope with optical zoom (magnification ⫻85 when displayed on a 19-inch monitor, XGIF-Q240/GIFFQ260FZ; Olympus Inc, Tokyo, Japan) equipped with AFI and NBI modes. This endoscope had 2 separate highquality monochromatic charge-coupled devices, 1 for high-resolution white-light imaging and NBI and 1 for AFI.11 All 3 imaging modalities of the ETMI system provided real-time endoscopic images. The endoscopist was able switch from one modality to another in 1 to 2 seconds by pushing control buttons on the handle of the endoscope. The ETMI system is currently marketed in the United Kingdom and Asia.
Study design and randomization All patients underwent 2 consecutive upper endoscopies with ETMI and SVE in random order at an interval of 6 to 16 weeks. Both consecutive procedures were performed by 2 different endoscopists. Before the procedures, both endoscopists had the same clinical information, and the endoscopist assigned to the second procedure was blinded to the results of the first procedure. Randomization of the technique was done before the first endoscopy by means of sealed opaque envelopes. Envelopes were provide by the study coordinator at the AMC and were not stratified per center.
Endoscopic procedures Patients were sedated with intravenous midazolam (2.5-15 mg) and, if necessary, supplemented with fentanyl (0.1-0.2 mg). ETMI endoscopy. First, the esophagus was inspected with HRE without using magnification for the presence of areas suspicious for dysplasia. Subsequently, the Barrett’s segment was inspected with AFI for the detection of additional suspicious areas. For all suspicious areas, the location and technique that primarily led to their detection www.giejournal.org
Histological assessment Biopsy samples were processed and stained by using standard methods in the pathology laboratories of the participating centers and were routinely assessed by the centers’ pathologists. For study purposes, all specimens were subsequently reviewed by a single GI pathologist with extensive experience in Barrett’s neoplasia (F.J.t.K.). All pathologists were blinded to the allocated endoscopic modality. The histological outcome was classified according to the Vienna classification of GI epithelial neoplasia into the following categories: ND, ID, LGIN, HGIN, or Ca.16
Outcomes parameters The primary outcomes were the overall histological yield of ETMI and SVE defined as the highest grade of neoplasia diagnosed in any biopsy specimen (ie, targeted or random) obtained during ETMI and SVE, respectively, and the targeted histological yield of ETMI and SVE defined as the highest grade of neoplasia diagnosed in any targeted biopsy specimen obtained from visible abnormalities identified during ETMI or SVE. The secondary outcomes were the number of patients diagnosed with LGIN/HGIN/Ca by ETMI and SVE, the number of visible abnormalities with HGIN/Ca detected by ETMI and SVE, the number of visible abnormalities with LGIN/HGIN/Ca detected by ETMI and SVE, the accuracy of NBI for detailed inspection of visible abnormalities identified with HRE and/or AFI, and procedure times.
Sample size and statistical analysis Based on cohort studies in patients with histologically confirmed LGIN in BE, 25% of these patients were expected to have HGIN/Ca detected.20,21 From preliminary Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
197
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
Figure 1. Patient flow-diagram. SVE, standard video endoscopy; ETMI, Endoscopic Tri-Modal Imaging; LGIN, low-grade intra-epithelial neoplasia.
results, we expected the detection rate for early neoplastic lesions with ETMI to be 80% and that with SVE to be 40%.9,10 To detect this difference with a power of 80% and a significance level of 5%, the sample size was calculated at 96 patients. All statistical analyses were performed by using a statistical software package (Statistical Package for the Social Sciences version 18.0.1; SPSS Inc, Chicago, Ill). To compare the primary outcome measures as paired evaluations with multiple outcome categories, the generalized McNemar-Bowker test of symmetry was used. For the secondary outcome parameters, the McNemar test, Wilcoxon signed-rank test, independent t test, or paired t test was used to compare both groups. The current study and the retrospective and prospective Barrett registration were approved by the Medical Ethical Reviewing Boards from all participating centers (ISRCTN91816824).
as ND and 83 as ID). In total, 103 patients consented to participate in the study and were included. Finally, 99 patients (circumferential Barrett’s extent of 2.0 cm [interquartile range 1.0-5.0] and median maximum extent of 5.0 cm [interquartile range 3.0-7.0]) were analyzed (Fig. 1). The mean interval between both procedures was 10 weeks (standard deviation [SD] 3.9). The overall histological outcome of both procedures was ND in 14 patients, ID in 18 patients, LGIN in 43 patients, and HGIN/Ca in 24 patients.
RESULTS
Targeted histological yield of ETMI and SVE
Patients Between January 2006 and November 2009, 417 patients with a previous diagnosis of LGIN were reviewed. LGIN was confirmed by at least 1 of the panel pathologists in 129 patients who were approached for participation in the current study (205 patients were classified 198 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
Overall histological yield of ETMI and SVE The overall histological yield (targeted sampling ⫹ random sampling) of ETMI and SVE was the same for both procedures in 60 patients. In 23 patients, ETMI yielded a higher histological grade than SVE, whereas in 16 patients, SVE had a higher histological yield (Table 1, P ⫽ .541).
The targeted histological yield (targeted sampling only) of ETMI and SVE was the same for both procedures in 64 patients. In 25 patients, ETMI yielded a higher histological grade than SVE, whereas in 10 patients, SVE had a higher histological yield (Table 2, P ⫽ .053). When we grouped the histological outcome in 2 clinically relevant outcome groups (no lesions/ND/ID and LGIN/HGIN/Ca), ETMI www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
TABLE 1. Overall histological yield of ETMI and SVE (P ⴝ .541)
TABLE 3. Targeted histological yield of endoscopically visible abnormalities detected with ETMI and SVE
ETMI NDBE
ID
ETMI
LGD
HGD/Ca
No lesion/ND/ID
LGIN/HGIN/Ca
No lesion/ND/ID
58
18
LGIN/ HGIN/Ca
6
17
SVE NDBE
15
6
10
1
ID
5
6
2
0
LGD
3
3
25
4
HGD/Ca
1
0
4
14
When ID cases were grouped with NDBE or LGIN the results were comparable (P ⫽ .572 and P ⫽ .446, respectively). Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intrapeithelial neoplasia; NDBE, non-dysplastic Barrett’s esophagus; SVE, standard video endoscopy.
TABLE 2. Targeted histological yield of endoscopically visible abnormalities detected with ETMI and SVE (P ⴝ .053) ETMI No lesion/ND
ID
LGIN
HGIN/Ca
No lesions/ND
44
7
13
5
ID
3
4
0
0
LGIN
3
0
8
0
HGIN/Ca
3
0
1
8
SVE
Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; ND, no dysplasia; SVE, standard video endoscopy.
yielded a higher histological yield in 18 patients, whereas SVE yielded a higher histological yield in 6 patients (Table 3, P ⫽ .023).
Patients diagnosed with LGIN/HGIN/Ca by ETMI and SVE In total, 67 patients (68%) were diagnosed with LGIN/ HGIN/Ca by ETMI and/or SVE (Table 4). ETMI detected a total of 60 patients with LGIN/HGIN/Ca. HRE detected 21 patients by targeted sampling. Subsequent inspection with AFI yielded an additional 14 patients with targeted sampling, resulting in a total of 35 patients with LGIN/ HGIN/Ca detected by targeted sampling during ETMI. The diagnosis in 25 patients was solely made by random sampling. SVE detected 54 patients with LGIN/HGIN/Ca, of whom the diagnosis was made in 23 by targeted biopsy sampling. The diagnosis was made in 31 patients solely by random sampling. www.giejournal.org
SVE
Histological outcome grouped in 2 clinically relevant categories (ID with no lesions/ID and LGIN with HGIN/Ca) (P ⫽ .023). Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; ND, no dysplasia; SVE, standard video endoscopy.
In 13 patients, LGIN/HGIN/Ca was detected by ETMI, whereas the diagnosis was missed with SVE. HRE detected LGIN in 3 patients, whereas AFI detected Ca in 1 patient and LGIN in 4 patients. In the other 5 patients, LGIN/HGIN was found in random biopsy samples obtained during ETMI. In 7 patients, LGIN/HGIN/Ca was detected by SVE but missed by ETMI. In 3 patients, SVE detected suspicious lesions containing LGIN (n ⫽ 2) and HGIN (n ⫽ 1), whereas in the other 4 patients, LGIN/HGIN was detected in the random biopsy samples of the SVE procedure.
Visible abnormalities detected with ETMI and SVE SVE detected a total of 82 suspicious lesions (Table 5). Fourteen contained HGIN/Ca and 15 LGIN, resulting in a false-positive rate of SVE for HGIN/Ca of 83% (68/82). ETMI detected a total of 156 suspicious lesions. Eighty-five lesions were detected with HRE, and AFI detected an additional 71 suspicious lesions. Of the HRE-detected lesions, 11 contained HGIN/Ca and 16 lesions contained LGIN. Of the 71 additional lesions detected with AFI, 5 lesions contained HGIN/Ca and 17 lesions contained LGIN. The overall false-positive rate of ETMI for HGIN/Ca was 90% (140/156).
NBI evaluation ETMI detected 156 suspicious lesions. Detailed inspection with NBI was successfully performed in 152 lesions (97.4%) (Table 6). Of the 16 lesions containing HGIN/Ca, 15 were classified as having a suspicious appearance on NBI. One lesion was judged to have an indeterminate pattern on NBI. Of the 136 lesions that contained ND/ID/ LGIN, detailed NBI inspection resulted in a nonsuspicious appearance on NBI in 55 lesions. The sensitivity and specificity of NBI for HGIN/Ca were 93.8% and 40.4%, respectively. Detailed inspection with NBI resulted in a decrease of false-positive rate in ETMI from 89.7% to 51.9% (81/156). Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
199
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
TABLE 4. Detection of patients with HGIN/Ca and LGIN/HGIN/Ca by random plus targeted biopsies and by targeted sampling SVE
ETMI
Total
No. of patients detected with HGIN/Ca
19
19
24
No. of patients with HGIN/Ca detected with targeted biopsies
12
HRE: 11 HRE⫹AFI: 13
17
No. of patients detected with LGIN/HGIN/Ca
54
60
67
No. of patients with LGIN/HGIN/Ca detected with targeted biopsies
23
HRE: 21 HRE⫹AFI: 35
41
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; LGIN, low-grade intraepithelial neoplasia; SVE, standard video endoscopy.
TABLE 5. Number of visible abnormalities detected with SVE and ETMI SVE
ETMI
Total no. of suspicious lesions detected
82
HRE: 85 HRE⫹AFI: 156
No. of lesions with HGIN/Ca detected
14
HRE: 11 HRE⫹AFI: 16
No. of lesions with LGIN/HGIN/Ca detected
29
HRE: 27 HRE⫹AFI: 49
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; LGIN, low-grade intraepithelial neoplasia; SVE, standard video endoscopy.
TABLE 6. Results of detailed inspection of the mucosal and vascular pattern with NBI of 152 endoscopically visible abnormalities detected with HRE and AFI during ETMI HGIN/Ca
ND/ID/LGIN
NBI ⫹ (suspicious)
15
81
NBI – (indeterminate ⫹ nonsuspicious)
1
55
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; NBI, narrow-band imaging; ND, no dysplasia.
Procedure times The mean procedure time of SVE was 12:22 minutes (SD 3:24 minutes) compared with 19:41 minutes (SD 7:24 minutes) for ETMI (P ⬍ .001). This difference in procedure times was mainly attributed to increased inspection times during ETMI (ETMI, 12:06 minutes [SD 4:59 minutes] vs SVE, 6:04 minutes [SD 01:45 minutes]; P ⫽ .015). There was no difference in mean inspection times between SVE 200 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
(6:04 minutes [SD 1:45 minutes]) and HRE (6:13 minutes [SD 1:46 minutes]), meaning that the additional inspection time during ETMI was attributed to extra inspection with AFI (mean 3:10 minutes [SD 1:45 minutes]) and NBI (mean 2:36 minutes [SD 2:45 minutes]).
DISCUSSION Almost all endoscopic imaging studies on BE to date have been conducted in tertiary referral settings involving selected patient populations (generally with a high-risk profile) and performed by endoscopists with extensive experience in advanced imaging techniques and the recognition of subtle early neoplastic lesions. Results of these studies, therefore, cannot be translated into general endoscopy practice. To overcome these disadvantages, the current study was performed in a unique setting. All procedures were performed by general endoscopists based in teaching hospitals with no particular expertise in BE or advanced imaging techniques. Furthermore, all patients included in the current study had a confirmed diagnosis of LGIN and thus an intermediate-risk profile because a general Barrett’s surveillance population would necessitate an unfeasibly large sample size. We found no significant difference in the overall detection of dysplasia by ETMI and SVE, but ETMI showed a significant increase in the targeted detection of dysplasia (eg, LGIN/HGIN/Ca) compared with SVE. These results are comparable to the results of a recent international randomized, multicenter, crossover study that compared ETMI with SVE in a tertiary referral setting.12 The increased targeted detection of dysplasia was solely attributable to the additional targeted detection by AFI (Fig. 2). The targeted detection of dysplastic lesions with HRE and SVE was comparable, but AFI detected an additional 22 lesions, resulting in an increase in the number of patients detected by targeted sampling from 21 to 35 patients. The increase in the targeted detection of dysplasia by ETMI resulted from detecting more lesions with LGIN. What is the clinical relevance of detecting focal lesions containing www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Figure 2. In images A, B, and C, a patient is depicted with no lesions seen during HRE, AFI, and SVE. D and E show a clearly visible early carcinoma located at the 12 o’clock position. This lesion was detected with HRE as well as SVE. F and G show an AFI positive lesion (arrow) containing HGIN that was not seen during HRE. H, NBI showed irregular mucosal and vascular patterns and abnormal blood vessels suspicious for dysplasia. AFI, autofluorescence imaging; NBI, narrow band imaging; HGIN, high-grade intra-epihtelial neopalsia; HRE, high-resolution endoscopy.
LGIN? It has been demonstrated that biopsies of endoscopically visible lesions may underestimate the true histology and that endoscopic resection of the focal lesions results in upstaging to a higher grade of neoplasia.22,23 It is, however, not known whether this histological upgrading also holds for flat lesions that are detected by their AFI appearance. Our study does not allow us to assess the impact of AFI-detected lesions www.giejournal.org
on decisions regarding the therapeutic management on a per-patient level. In this study, NBI was not used as a primary detection tool but solely for detailed inspection of lesions detected with HRE and/or AFI. Both ETMI and SVE exhibited a high false-positives rate (90% and 83%, respectively), which likely reflects the lower prevalence of neoplasia in the Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
201
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
study population and the fact that endoscopists had less experience with early Barrett’s neoplasia. Furthermore, biopsy artifacts may have influenced the results of the second procedure, especially with ETMI, by possibly generating more false-positive findings. However, all patients were on high-dose proton pump inhibitor treatment, and the chosen minimal interval (6 weeks) between consecutive endoscopies is generally believed to be sufficient for healing of biopsy sites. Detailed inspection with NBI only marginally decreased the false-positive rate of ETMI to 52%, and 1 lesion with HGIN/Ca was misclassified as nonsuspicious on NBI. In the aforementioned randomized, crossover study in a tertiary referral setting, detailed NBI evaluation misclassified 17% of HGIN/Ca lesions as “not suspicious.”12 In that study, 66 HGIN/Ca lesions were detected during ETMI, whereas the current study only found 16 HGIN/Ca lesions. Therefore, the difference in false-negative rates between both studies may be because of a chance effect. An alternative explanation is that endoscopists in the current study had a low threshold for classifying lesions as suspicious on NBI given their inexperience with early Barrett’s neoplasia. This may have led to a low rate of misclassifying HGIN/Ca as not suspicious on NBI yet was also associated with a decrease in the false-positive rate that is clinically less relevant. Several interobserver studies have questioned the additional value of NBI over HRE for identifying images with neoplasia or improving interobserver agreement.17-19 Because NBI magnification endoscopy is operator dependent and labor intensive, we believe that just obtaining targeted biopsy samples of suspicious areas is likely to be easier and faster. Another important finding of the current study is the high yield of HGIN/Ca (24%) in our patient population. This confirms the results of previous studies on LGIN suggesting that patients with a confirmed histological diagnosis of LGIN carry a significant risk of progressing to HGIN/Ca.15,20,21 The current study design has some limitations that need to be addressed. First, during ETMI endoscopy, HRE and AFI inspection were performed sequentially by the same endoscopist. AFI assessment may, therefore, have been biased by the HRE findings. Second, the inspection time with HRE and AFI was longer compared with the inspection time with SVE (9:23 minutes vs 6:04 minutes). In the current design, SVE was not controlled for double inspection time, and this may have led to an overestimation of the targeted detection rate of AFI. Third, despite the facts that all participating endoscopists were sufficiently trained before the study, we cannot exclude the possibility of a learning effect given the relatively small numbers of endoscopies per endoscopist. Fourth, the results may have been biased by the fact that the participating endoscopists were aware of the intermediate-risk profile of the included patients. In conclusion, the findings of our study suggest that ETMI currently has no role in detecting early Barrett’s neoplasia in general endoscopy practice. The increase in targeted detection of dysplasia by AFI in recent random-
ized, crossover studies may indicate that AFI is of additional value in the workup of patients with early Barrett’s neoplasia yet the impact of this on patient management has yet to be shown. Detailed inspection of suspicious lesions with NBI with magnification appears to be of limited value as a clinical decision-making tool in BE.
202 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
REFERENCES 1. Spechler SJ. Clinical practice. Barrett’s esophagus. N Engl J Med 2002; 346:836-42. 2. Devesa SS, Blot WJ, Fraumeni JF Jr. Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer 1998; 83:2049-53. 3. van Sandick JW, van Lanschot J, Kuiken BW, et al. Impact of endoscopic biopsy surveillance of Barrett’s oesophagus on pathological stage and clinical outcome off Barrett’s carcinoma. Gut 1998;43:216-22. 4. Hameeteman W, Tytgat GN, Houthoff HJ, et al. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989; 96:1249-56. 5. Haggitt RC. Pathology of Barrett’s esophagus. J Gastrointest Surg 2000; 4:117-8. 6. Sampliner RE. Updated guidelines for the diagnosis, surveillance, and therapy of Barrett’s esophagus. Am J Gastroenterol 2002;97:1888-95. 7. Hirota WK, Zuckerman MJ, Adler DG, et al. ASGE guideline: the role of endoscopy in the surveillance of premalignant conditions of the upper GI tract. Gastrointest Endosc 2006;63:570-80. 8. Curvers WL, Kiesslich R, Bergman JJ. Novel imaging modalities in the detection of esophageal neoplasia. Best Pract Res Clin Gastroenterol 2008;22:687-720. 9. 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. 10. Kara MA, Peters FP, Fockens P, et al. Endoscopic video-autofluorescence imaging followed by narrow band imaging for detecting early neoplasia in Barrett’s esophagus. Gastrointest Endosc 2006;64:176-185. 11. 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. 12. Curvers WL, Herrero LA, Wallace MB, et al. Endoscopic tri-modal imaging is more effective than standard endoscopy in targeting early-stage neoplasia in Barrett’s esophagus. Gastroenterology 2010;139:1106-14. 13. Casparie M, Tiebosch AT, Burger G, et al. Pathology databanking and biobanking in The Netherlands, a central role for PALGA, the nationwide histopathology and cytopathology data network and archive. Cell Oncol 2007;29:19-24. 14. Hulscher JB, Haringsma J, Benraadt J, et al. Comprehensive Cancer Centre Amsterdam Barrett Advisory Committee: first results. Neth J Med 2001;58:3-8. 15. Curvers WL, Rosmolen W, Elzer B, et al. Low-grade intra-epithelial in Barrett’s esophagus: Over-diagnosed but underestimated. Am J Gastroenterol 2010;105:1523-30. 16. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;47:251-5. 17. Curvers W, Baak L, Kiesslich R, et al. Chromoendoscopy and narrowband imaging compared with high-resolution magnification endoscopy in Barrett’s esophagus. Gastroenterology 2008;134:670-9. 18. Curvers WL, Bohmer CJ, Mallant-Hent RC, et al. Mucosal morphology in Barrett’s esophagus: interobserver agreement and role of narrow band imaging. Endoscopy 2008;40:799-805. 19. Herrero LA, Curvers WL, Bansal A, et al. Zooming in on Barrett oesophagus using narrow-band imaging: an international observer agreement study. Eur J Gastroenterol Hepatol 2009;21:1068-75.
www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
20. Skacel M, Petras RE, Gramlich TL, et al. The diagnosis of low-grade dysplasia in Barrett’s esophagus and its implications for disease progression. Am J Gastroenterol 2000;95:3383-7. 21. Montgomery E, Goldblum JR, Greenson JK, et al. Dysplasia as a predictive marker for invasive carcinoma in Barrett esophagus: a follow-up study based on 138 cases from a diagnostic variability study. Hum Pathol 2001;32:379-88.
22. Peters FP, Brakenhoff KP, Curvers WL, et al. Histologic evaluation of resection specimens obtained at 293 endoscopic resections in Barrett’s esophagus. Gastrointest Endosc 2008;67:604-9. 23. Mino-Kenudson M, Brugge WR, Puricelli WP, et al. Management of superficial Barrett’s epithelium-related neoplasms by endoscopic mucosal resection: clinicopathologic analysis of 27 cases. Am J Surg Pathol 2005;29:680-6.
www.giejournal.org
Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
203
Endoscopic trimodal imaging versus standard video endoscopy for detection of early Barrett’s neoplasia: a multicenter, randomized, crossover study in general practice Wouter L. Curvers, MD, PhD, Frederike G. van Vilsteren, MD, Lubertus C. Baak, MD, PhD, Clarisse Böhmer, MD, PhD, Rosalie C. Mallant-Hent, MD, PhD, Anton H. Naber, MD, PhD, Arnout van Oijen, MD, PhD, Cyriel Y. Ponsioen, MD, PhD, Pieter Scholten, MD, Ed Schenk, MD, PhD, Erik Schoon, MD, PhD, Cees A. Seldenrijk, MD, PhD, Gerrit A. Meijer, MD, PhD, Fiebo J. ten Kate, MD, PhD, Jacques J. Bergman, MD, PhD Amsterdam, Hoofddorp, Almere, Hilversum, Alkmaar, Zwolle, Eindhoven, Nieuwegein, the Netherlands
Background: Endoscopic trimodal imaging (ETMI) may improve detection of early neoplasia in Barrett’s esophagus (BE). Studies with ETMI so far have been performed in tertiary referral settings only. Objective: To compare ETMI with standard video endoscopy (SVE) for the detection of neoplasia in BE patients with an intermediate-risk profile. Design: Multicenter, randomized, crossover study. Setting: Community practice. Patients and Methods: BE patients with confirmed low-grade intraepithelial neoplasia (LGIN) underwent both ETMI and SVE in random order (interval 6-16 weeks). During ETMI, BE was inspected with high-resolution endoscopy followed by autofluorescence imaging (AFI). All visible lesions were then inspected with narrowband imaging. During ETMI and SVE, visible lesions were sampled followed by 4-quadrant random biopsies every 2 cm. Main Outcome Measurements: Overall histological yield of ETMI and SVE and targeted histological yield of ETMI and SVE. Results: A total of 99 patients (79 men, 63 ⫾ 10 years) underwent both procedures. ETMI had a significantly higher targeted histological yield because of additional detection of 22 lesions with LGIN/high-grade intraepithelial neoplasia (HGIN)/carcinoma (Ca) by AFI. There was no significant difference in the overall histological yield (targeted ⫹ random) between ETMI and SVE. HGIN/Ca was diagnosed only by random biopsies in 6 of 24 patients and 7 of 24 patients, with ETMI and SVE, respectively. Abbreviations: AFI, autofluorescence imaging; AMC, Academic Medical Center; BE, Barrett’s esophagus; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, highresolution endoscopy; ID, indefinite for dysplasia; LGIN, low-grade intraepithelial neoplasia; NBI, narrow-band imaging; ND, no dysplasia; SVE, standard video endoscopy. DISCLOSURE: The following author disclosed a financial relationship relevant to this publication: Dr. Bergman: research grant from Olympus Inc, Tokyo, Japan. The other authors disclosed no financial relationships relevant to this publication. Copyright © 2011 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2010.10.014 Received August 31, 2010. Accepted October 8, 2010. Current affiliations: Department of Gastroenterology and Hepatology (W.L.C., F.G.v.V., C.Y.P., J.J.B.), Academic Medical Center, Amsterdam, Department of Gastroenterology and Hepatology (L.C.B.), Onze Lieve Vrouwe Gasthuis, Amsterdam, Department of Gastroenterology and Hepatology (C.B.), Spaarne Hospital, Hoofddorp, Department of Internal
www.giejournal.org
Medicine (R.C.M.-H.), Flevohospital, Almere, Department of Internal Medicine (A.H.N.), Tergooi Hospitals, Hilversum, Department of Gastroenterology and Hepatology (A.v.O.), Medical Center Alkmaar, Alkmaar, Department of Gastroenterology and Hepatology (P.S.), St. Lucas Andreas Hospital, Amsterdam, Department of Gastroenterology and Hepatology (E.S.), Isala Clinics, Zwolle, Department of Gastroenterology and Hepatology (E.S.), Catharina Hospital, Eindhoven, Department of Pathology (C.A.S.), St. Antonius Hospital, Nieuwegein, Department of Pathology (G.A.M.), Free University Medical Center, Amsterdam, Department of Pathology (F.J.t.K.), Academic Medical Center, Amsterdam, Amsterdam Gastroenterological Association (W.L.C., F.G.v.V., L.C.B., C.B., R.C.M.-H., A.H.N., A.v.O., C.Y.P., P.S., F.J.t.K., J.J.B.), Amsterdam, the Netherlands. Presented at Digestive Disease Week, 2010, New Orleans, Louisiana (Gastrointest Endosc 2010;71:AB142). Reprint requests: Wouter Curvers, MD, Department of Gastroenterology and Hepatology, Academic Medical Center Meibergdreef 9, 1105 AZ, Amsterdam, the Netherlands. If you would like to chat with an author of this article, you may contact Dr Curvers at [email protected].
Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
195
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
Limitations: Inspection, with high-resolution endoscopy and AFI, was performed sequentially. Conclusion: ETMI performed in a community-based setting did not improve the overall detection of dysplasia compared with SVE. The diagnosis of dysplasia is still being made in a significant number of patients by random biopsies. Patients with a confirmed diagnosis of LGIN have a significant risk of HGIN/Ca. (Clinical trial registration number: ISRCTN91816824; NTR867.) (Gastrointest Endosc 2011;73:195-203.)
Barrett’s esophagus (BE) is a precursor lesion of esophageal adenocarcinoma, the cancer with the fastest increasing incidence in the Western world.1,2 Esophageal adenocarcinoma has a poor prognosis when diagnosed at a symptomatic stage, whereas the prognosis is excellent for early yet generally asymptomatic stages.3 Malignant transformation of BE into carcinoma (Ca) is a gradual process that takes place through several histopathological grades classified as no dysplasia (ND), low-grade intraepithelial neoplasia (LGIN), and high-grade intraepithelial neoplasia (HGIN).4,5 Patients with known BE, therefore, undergo regular endoscopic surveillance for the detection of early neoplastic lesions.6,7 These early lesions are, however, poorly visible with standard endoscopy, and therefore random biopsy samples are obtained, which is labor intensive and neoplastic areas may still be missed because of sampling error. Advances in endoscopic imaging may improve the detection of these early neoplastic lesions.8 Recently, a new endoscopic system has become commercially available that incorporates white-light highresolution endoscopy (HRE), autofluorescence imaging (AFI), and narrow-band imaging (NBI)— endoscopic trimodal imaging (ETMI). We performed several uncontrolled studies showing that AFI may improve the detection of early neoplasia in BE, whereas NBI subsequently decreases the high false-positive rate of AFI.9-11 In a recent multicenter, randomized, crossover study that compared ETMI with standard video endoscopy (SVE), AFI significantly improved the targeted detection of early neoplasia in BE.12 This study was performed in a tertiary referral setting by experienced endoscopists in a high-risk Barrett’s neoplasia population. In a BE population with a lower rate of neoplasia and with endoscopists with less experience in the detection early Barrett’s neoplasia, the additional value of AFI and NBI may be different. The aim of the current study was therefore to compare the diagnostic potential of ETMI with that of SVE in a community practice setting including only BE patients with an intermediate-risk profile.
METHODS Study setting This study was performed in 8 teaching hospitals in the Netherlands and the Academic Medical Center (AMC), Amsterdam. All centers participate in an ongoing prospective BE registration. Before initiating the prospective reg196 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
Take-home Message ●
●
Multimodality imaging cannot yet replace random sampling in Barrett’s esophagus. Autofluorescence imaging may be helpful in the workup of patients with early Barrett’s neoplasia or for future risk stratification in Barrett’s surveillance. Patients with a confirmed diagnosis of low-grade intraepithelial neoplasia have an increased risk of neoplastic progression and should undergo intensified endoscopic follow-up or may be eligible for endoscopic therapy.
istration, a retrospective BE registration was performed. In each of the hospitals, all BE patients were identified by the pathology databases in the participating centers.13
Patients All patients with a diagnosis of LGIN made in the preceding 5 years were identified from the retrospective and prospective BE registration database. From these patients, pathology specimens of all procedures with a diagnosis of LGIN were retrieved. Two of 3 panel GI pathologists (C.A.S., G.A.M., F.J.t.K.), each with extensive experience in Barrett’s neoplasia, independently reviewed all retrieved slides.14,15 The presence of intestinal metaplasia and the presence and degree of dysplasia were assessed and graded according to the Vienna classification of GI epithelial neoplasia into D, indefinite for dysplasia (ID), LGIN, HGIN, or Ca.16 Patients were enrolled when they met the following inclusion criteria: (1) age older than 18 years, (2) previous diagnosis of BE defined as the presence of columnar-lined epithelium in the tubular esophagus with intestinal metaplasia on histology, (3) previous diagnosis of LGIN confirmed by at least 1 panel pathologist, (4) written informed consent provided. Patients were excluded for the following reasons: (1) presence of active erosive esophagitis grade B or worse according to the Los Angeles classification of erosive esophagitis; (2) at first endoscopy, the presence of an advanced lesion (eg, type 0-I or 0-III) suspicious for neoplasia that would not allow a delay in intervention for 6 weeks; (3) the presence of conditions precluding histological sampling of the esophagus. www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Endoscopists
were noted. Finally, all suspicious areas were inspected in detail with NBI by using magnification endoscopy. The overall NBI appearance was classified into 3 categories: suspicious for neoplasia, not suspicious for neoplasia, or indeterminate, based on the assessment of the mucosal pattern, the vascular pattern, and the presence of abnormal blood vessels. The inspection times with HRE, AFI, and NBI, as well as the targeted and random sampling time were recorded. SVE. SVE was performed with various standard video endoscopes (Olympus GIF-140, GIF-160 or GIFQ-160; Olympus Inc). The Barrett’s segment was inspected for the presence of areas suspicious for dysplasia. For all detected suspicious areas, the locations were recorded. The inspection time, targeted sampling time, and random sampling time were recorded. Biopsy protocol. From all abnormalities detected during ETMI and SVE, 2 targeted biopsy specimens were obtained. Finally, 4-quadrant random biopsy samples were obtained for every 2 cm of the Barrett’s segment, not taking samples from areas with suspicious lesions.
All procedures were performed by 9 board-certified gastroenterologists from the participating teaching hospitals (L.C.B., C.B., R.C.H.-M., A.H.N., A.v.O., C.Y.P., P.S., E.S., E.S.) who had no specific expertise in BE or advanced imaging technique before this study. To prepare for the study, all participating endoscopists received a training DVD with examples of neoplastic and non-neoplastic lesions in BE with HRE, AFI, and NBI and visited dedicated endoscopy programs at the AMC Amsterdam, where patients referred for early Barrett’s neoplasia were investigated with the ETMI system. All participating endoscopists performed at least 5 dedicated BE surveillance endoscopy lists with the ETMI system at their own center and participated in interobserver studies on NBI.17-19
Technical background of the ETMI endoscopy system The ETMI system consisted of a high-resolution whitelight endoscope with optical zoom (magnification ⫻85 when displayed on a 19-inch monitor, XGIF-Q240/GIFFQ260FZ; Olympus Inc, Tokyo, Japan) equipped with AFI and NBI modes. This endoscope had 2 separate highquality monochromatic charge-coupled devices, 1 for high-resolution white-light imaging and NBI and 1 for AFI.11 All 3 imaging modalities of the ETMI system provided real-time endoscopic images. The endoscopist was able switch from one modality to another in 1 to 2 seconds by pushing control buttons on the handle of the endoscope. The ETMI system is currently marketed in the United Kingdom and Asia.
Study design and randomization All patients underwent 2 consecutive upper endoscopies with ETMI and SVE in random order at an interval of 6 to 16 weeks. Both consecutive procedures were performed by 2 different endoscopists. Before the procedures, both endoscopists had the same clinical information, and the endoscopist assigned to the second procedure was blinded to the results of the first procedure. Randomization of the technique was done before the first endoscopy by means of sealed opaque envelopes. Envelopes were provide by the study coordinator at the AMC and were not stratified per center.
Endoscopic procedures Patients were sedated with intravenous midazolam (2.5-15 mg) and, if necessary, supplemented with fentanyl (0.1-0.2 mg). ETMI endoscopy. First, the esophagus was inspected with HRE without using magnification for the presence of areas suspicious for dysplasia. Subsequently, the Barrett’s segment was inspected with AFI for the detection of additional suspicious areas. For all suspicious areas, the location and technique that primarily led to their detection www.giejournal.org
Histological assessment Biopsy samples were processed and stained by using standard methods in the pathology laboratories of the participating centers and were routinely assessed by the centers’ pathologists. For study purposes, all specimens were subsequently reviewed by a single GI pathologist with extensive experience in Barrett’s neoplasia (F.J.t.K.). All pathologists were blinded to the allocated endoscopic modality. The histological outcome was classified according to the Vienna classification of GI epithelial neoplasia into the following categories: ND, ID, LGIN, HGIN, or Ca.16
Outcomes parameters The primary outcomes were the overall histological yield of ETMI and SVE defined as the highest grade of neoplasia diagnosed in any biopsy specimen (ie, targeted or random) obtained during ETMI and SVE, respectively, and the targeted histological yield of ETMI and SVE defined as the highest grade of neoplasia diagnosed in any targeted biopsy specimen obtained from visible abnormalities identified during ETMI or SVE. The secondary outcomes were the number of patients diagnosed with LGIN/HGIN/Ca by ETMI and SVE, the number of visible abnormalities with HGIN/Ca detected by ETMI and SVE, the number of visible abnormalities with LGIN/HGIN/Ca detected by ETMI and SVE, the accuracy of NBI for detailed inspection of visible abnormalities identified with HRE and/or AFI, and procedure times.
Sample size and statistical analysis Based on cohort studies in patients with histologically confirmed LGIN in BE, 25% of these patients were expected to have HGIN/Ca detected.20,21 From preliminary Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
197
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
Figure 1. Patient flow-diagram. SVE, standard video endoscopy; ETMI, Endoscopic Tri-Modal Imaging; LGIN, low-grade intra-epithelial neoplasia.
results, we expected the detection rate for early neoplastic lesions with ETMI to be 80% and that with SVE to be 40%.9,10 To detect this difference with a power of 80% and a significance level of 5%, the sample size was calculated at 96 patients. All statistical analyses were performed by using a statistical software package (Statistical Package for the Social Sciences version 18.0.1; SPSS Inc, Chicago, Ill). To compare the primary outcome measures as paired evaluations with multiple outcome categories, the generalized McNemar-Bowker test of symmetry was used. For the secondary outcome parameters, the McNemar test, Wilcoxon signed-rank test, independent t test, or paired t test was used to compare both groups. The current study and the retrospective and prospective Barrett registration were approved by the Medical Ethical Reviewing Boards from all participating centers (ISRCTN91816824).
as ND and 83 as ID). In total, 103 patients consented to participate in the study and were included. Finally, 99 patients (circumferential Barrett’s extent of 2.0 cm [interquartile range 1.0-5.0] and median maximum extent of 5.0 cm [interquartile range 3.0-7.0]) were analyzed (Fig. 1). The mean interval between both procedures was 10 weeks (standard deviation [SD] 3.9). The overall histological outcome of both procedures was ND in 14 patients, ID in 18 patients, LGIN in 43 patients, and HGIN/Ca in 24 patients.
RESULTS
Targeted histological yield of ETMI and SVE
Patients Between January 2006 and November 2009, 417 patients with a previous diagnosis of LGIN were reviewed. LGIN was confirmed by at least 1 of the panel pathologists in 129 patients who were approached for participation in the current study (205 patients were classified 198 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
Overall histological yield of ETMI and SVE The overall histological yield (targeted sampling ⫹ random sampling) of ETMI and SVE was the same for both procedures in 60 patients. In 23 patients, ETMI yielded a higher histological grade than SVE, whereas in 16 patients, SVE had a higher histological yield (Table 1, P ⫽ .541).
The targeted histological yield (targeted sampling only) of ETMI and SVE was the same for both procedures in 64 patients. In 25 patients, ETMI yielded a higher histological grade than SVE, whereas in 10 patients, SVE had a higher histological yield (Table 2, P ⫽ .053). When we grouped the histological outcome in 2 clinically relevant outcome groups (no lesions/ND/ID and LGIN/HGIN/Ca), ETMI www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
TABLE 1. Overall histological yield of ETMI and SVE (P ⴝ .541)
TABLE 3. Targeted histological yield of endoscopically visible abnormalities detected with ETMI and SVE
ETMI NDBE
ID
ETMI
LGD
HGD/Ca
No lesion/ND/ID
LGIN/HGIN/Ca
No lesion/ND/ID
58
18
LGIN/ HGIN/Ca
6
17
SVE NDBE
15
6
10
1
ID
5
6
2
0
LGD
3
3
25
4
HGD/Ca
1
0
4
14
When ID cases were grouped with NDBE or LGIN the results were comparable (P ⫽ .572 and P ⫽ .446, respectively). Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intrapeithelial neoplasia; NDBE, non-dysplastic Barrett’s esophagus; SVE, standard video endoscopy.
TABLE 2. Targeted histological yield of endoscopically visible abnormalities detected with ETMI and SVE (P ⴝ .053) ETMI No lesion/ND
ID
LGIN
HGIN/Ca
No lesions/ND
44
7
13
5
ID
3
4
0
0
LGIN
3
0
8
0
HGIN/Ca
3
0
1
8
SVE
Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; ND, no dysplasia; SVE, standard video endoscopy.
yielded a higher histological yield in 18 patients, whereas SVE yielded a higher histological yield in 6 patients (Table 3, P ⫽ .023).
Patients diagnosed with LGIN/HGIN/Ca by ETMI and SVE In total, 67 patients (68%) were diagnosed with LGIN/ HGIN/Ca by ETMI and/or SVE (Table 4). ETMI detected a total of 60 patients with LGIN/HGIN/Ca. HRE detected 21 patients by targeted sampling. Subsequent inspection with AFI yielded an additional 14 patients with targeted sampling, resulting in a total of 35 patients with LGIN/ HGIN/Ca detected by targeted sampling during ETMI. The diagnosis in 25 patients was solely made by random sampling. SVE detected 54 patients with LGIN/HGIN/Ca, of whom the diagnosis was made in 23 by targeted biopsy sampling. The diagnosis was made in 31 patients solely by random sampling. www.giejournal.org
SVE
Histological outcome grouped in 2 clinically relevant categories (ID with no lesions/ID and LGIN with HGIN/Ca) (P ⫽ .023). Ca, Carcinoma; ETMI, endoscopic trimodal imaging; HGIN, highgrade intraepithelial neoplasia; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; ND, no dysplasia; SVE, standard video endoscopy.
In 13 patients, LGIN/HGIN/Ca was detected by ETMI, whereas the diagnosis was missed with SVE. HRE detected LGIN in 3 patients, whereas AFI detected Ca in 1 patient and LGIN in 4 patients. In the other 5 patients, LGIN/HGIN was found in random biopsy samples obtained during ETMI. In 7 patients, LGIN/HGIN/Ca was detected by SVE but missed by ETMI. In 3 patients, SVE detected suspicious lesions containing LGIN (n ⫽ 2) and HGIN (n ⫽ 1), whereas in the other 4 patients, LGIN/HGIN was detected in the random biopsy samples of the SVE procedure.
Visible abnormalities detected with ETMI and SVE SVE detected a total of 82 suspicious lesions (Table 5). Fourteen contained HGIN/Ca and 15 LGIN, resulting in a false-positive rate of SVE for HGIN/Ca of 83% (68/82). ETMI detected a total of 156 suspicious lesions. Eighty-five lesions were detected with HRE, and AFI detected an additional 71 suspicious lesions. Of the HRE-detected lesions, 11 contained HGIN/Ca and 16 lesions contained LGIN. Of the 71 additional lesions detected with AFI, 5 lesions contained HGIN/Ca and 17 lesions contained LGIN. The overall false-positive rate of ETMI for HGIN/Ca was 90% (140/156).
NBI evaluation ETMI detected 156 suspicious lesions. Detailed inspection with NBI was successfully performed in 152 lesions (97.4%) (Table 6). Of the 16 lesions containing HGIN/Ca, 15 were classified as having a suspicious appearance on NBI. One lesion was judged to have an indeterminate pattern on NBI. Of the 136 lesions that contained ND/ID/ LGIN, detailed NBI inspection resulted in a nonsuspicious appearance on NBI in 55 lesions. The sensitivity and specificity of NBI for HGIN/Ca were 93.8% and 40.4%, respectively. Detailed inspection with NBI resulted in a decrease of false-positive rate in ETMI from 89.7% to 51.9% (81/156). Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
199
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
TABLE 4. Detection of patients with HGIN/Ca and LGIN/HGIN/Ca by random plus targeted biopsies and by targeted sampling SVE
ETMI
Total
No. of patients detected with HGIN/Ca
19
19
24
No. of patients with HGIN/Ca detected with targeted biopsies
12
HRE: 11 HRE⫹AFI: 13
17
No. of patients detected with LGIN/HGIN/Ca
54
60
67
No. of patients with LGIN/HGIN/Ca detected with targeted biopsies
23
HRE: 21 HRE⫹AFI: 35
41
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; LGIN, low-grade intraepithelial neoplasia; SVE, standard video endoscopy.
TABLE 5. Number of visible abnormalities detected with SVE and ETMI SVE
ETMI
Total no. of suspicious lesions detected
82
HRE: 85 HRE⫹AFI: 156
No. of lesions with HGIN/Ca detected
14
HRE: 11 HRE⫹AFI: 16
No. of lesions with LGIN/HGIN/Ca detected
29
HRE: 27 HRE⫹AFI: 49
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; LGIN, low-grade intraepithelial neoplasia; SVE, standard video endoscopy.
TABLE 6. Results of detailed inspection of the mucosal and vascular pattern with NBI of 152 endoscopically visible abnormalities detected with HRE and AFI during ETMI HGIN/Ca
ND/ID/LGIN
NBI ⫹ (suspicious)
15
81
NBI – (indeterminate ⫹ nonsuspicious)
1
55
AFI, Autofluorescence imaging; Ca, carcinoma; ETMI, endoscopic trimodal imaging; HGIN, high-grade intraepithelial neoplasia; HRE, high-resolution endoscopy; ID, indefinite for dysplasia; LGIN, lowgrade intraepithelial neoplasia; NBI, narrow-band imaging; ND, no dysplasia.
Procedure times The mean procedure time of SVE was 12:22 minutes (SD 3:24 minutes) compared with 19:41 minutes (SD 7:24 minutes) for ETMI (P ⬍ .001). This difference in procedure times was mainly attributed to increased inspection times during ETMI (ETMI, 12:06 minutes [SD 4:59 minutes] vs SVE, 6:04 minutes [SD 01:45 minutes]; P ⫽ .015). There was no difference in mean inspection times between SVE 200 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
(6:04 minutes [SD 1:45 minutes]) and HRE (6:13 minutes [SD 1:46 minutes]), meaning that the additional inspection time during ETMI was attributed to extra inspection with AFI (mean 3:10 minutes [SD 1:45 minutes]) and NBI (mean 2:36 minutes [SD 2:45 minutes]).
DISCUSSION Almost all endoscopic imaging studies on BE to date have been conducted in tertiary referral settings involving selected patient populations (generally with a high-risk profile) and performed by endoscopists with extensive experience in advanced imaging techniques and the recognition of subtle early neoplastic lesions. Results of these studies, therefore, cannot be translated into general endoscopy practice. To overcome these disadvantages, the current study was performed in a unique setting. All procedures were performed by general endoscopists based in teaching hospitals with no particular expertise in BE or advanced imaging techniques. Furthermore, all patients included in the current study had a confirmed diagnosis of LGIN and thus an intermediate-risk profile because a general Barrett’s surveillance population would necessitate an unfeasibly large sample size. We found no significant difference in the overall detection of dysplasia by ETMI and SVE, but ETMI showed a significant increase in the targeted detection of dysplasia (eg, LGIN/HGIN/Ca) compared with SVE. These results are comparable to the results of a recent international randomized, multicenter, crossover study that compared ETMI with SVE in a tertiary referral setting.12 The increased targeted detection of dysplasia was solely attributable to the additional targeted detection by AFI (Fig. 2). The targeted detection of dysplastic lesions with HRE and SVE was comparable, but AFI detected an additional 22 lesions, resulting in an increase in the number of patients detected by targeted sampling from 21 to 35 patients. The increase in the targeted detection of dysplasia by ETMI resulted from detecting more lesions with LGIN. What is the clinical relevance of detecting focal lesions containing www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Figure 2. In images A, B, and C, a patient is depicted with no lesions seen during HRE, AFI, and SVE. D and E show a clearly visible early carcinoma located at the 12 o’clock position. This lesion was detected with HRE as well as SVE. F and G show an AFI positive lesion (arrow) containing HGIN that was not seen during HRE. H, NBI showed irregular mucosal and vascular patterns and abnormal blood vessels suspicious for dysplasia. AFI, autofluorescence imaging; NBI, narrow band imaging; HGIN, high-grade intra-epihtelial neopalsia; HRE, high-resolution endoscopy.
LGIN? It has been demonstrated that biopsies of endoscopically visible lesions may underestimate the true histology and that endoscopic resection of the focal lesions results in upstaging to a higher grade of neoplasia.22,23 It is, however, not known whether this histological upgrading also holds for flat lesions that are detected by their AFI appearance. Our study does not allow us to assess the impact of AFI-detected lesions www.giejournal.org
on decisions regarding the therapeutic management on a per-patient level. In this study, NBI was not used as a primary detection tool but solely for detailed inspection of lesions detected with HRE and/or AFI. Both ETMI and SVE exhibited a high false-positives rate (90% and 83%, respectively), which likely reflects the lower prevalence of neoplasia in the Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
201
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
Curvers et al
study population and the fact that endoscopists had less experience with early Barrett’s neoplasia. Furthermore, biopsy artifacts may have influenced the results of the second procedure, especially with ETMI, by possibly generating more false-positive findings. However, all patients were on high-dose proton pump inhibitor treatment, and the chosen minimal interval (6 weeks) between consecutive endoscopies is generally believed to be sufficient for healing of biopsy sites. Detailed inspection with NBI only marginally decreased the false-positive rate of ETMI to 52%, and 1 lesion with HGIN/Ca was misclassified as nonsuspicious on NBI. In the aforementioned randomized, crossover study in a tertiary referral setting, detailed NBI evaluation misclassified 17% of HGIN/Ca lesions as “not suspicious.”12 In that study, 66 HGIN/Ca lesions were detected during ETMI, whereas the current study only found 16 HGIN/Ca lesions. Therefore, the difference in false-negative rates between both studies may be because of a chance effect. An alternative explanation is that endoscopists in the current study had a low threshold for classifying lesions as suspicious on NBI given their inexperience with early Barrett’s neoplasia. This may have led to a low rate of misclassifying HGIN/Ca as not suspicious on NBI yet was also associated with a decrease in the false-positive rate that is clinically less relevant. Several interobserver studies have questioned the additional value of NBI over HRE for identifying images with neoplasia or improving interobserver agreement.17-19 Because NBI magnification endoscopy is operator dependent and labor intensive, we believe that just obtaining targeted biopsy samples of suspicious areas is likely to be easier and faster. Another important finding of the current study is the high yield of HGIN/Ca (24%) in our patient population. This confirms the results of previous studies on LGIN suggesting that patients with a confirmed histological diagnosis of LGIN carry a significant risk of progressing to HGIN/Ca.15,20,21 The current study design has some limitations that need to be addressed. First, during ETMI endoscopy, HRE and AFI inspection were performed sequentially by the same endoscopist. AFI assessment may, therefore, have been biased by the HRE findings. Second, the inspection time with HRE and AFI was longer compared with the inspection time with SVE (9:23 minutes vs 6:04 minutes). In the current design, SVE was not controlled for double inspection time, and this may have led to an overestimation of the targeted detection rate of AFI. Third, despite the facts that all participating endoscopists were sufficiently trained before the study, we cannot exclude the possibility of a learning effect given the relatively small numbers of endoscopies per endoscopist. Fourth, the results may have been biased by the fact that the participating endoscopists were aware of the intermediate-risk profile of the included patients. In conclusion, the findings of our study suggest that ETMI currently has no role in detecting early Barrett’s neoplasia in general endoscopy practice. The increase in targeted detection of dysplasia by AFI in recent random-
ized, crossover studies may indicate that AFI is of additional value in the workup of patients with early Barrett’s neoplasia yet the impact of this on patient management has yet to be shown. Detailed inspection of suspicious lesions with NBI with magnification appears to be of limited value as a clinical decision-making tool in BE.
202 GASTROINTESTINAL ENDOSCOPY
Volume 73, No. 2 : 2011
REFERENCES 1. Spechler SJ. Clinical practice. Barrett’s esophagus. N Engl J Med 2002; 346:836-42. 2. Devesa SS, Blot WJ, Fraumeni JF Jr. Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer 1998; 83:2049-53. 3. van Sandick JW, van Lanschot J, Kuiken BW, et al. Impact of endoscopic biopsy surveillance of Barrett’s oesophagus on pathological stage and clinical outcome off Barrett’s carcinoma. Gut 1998;43:216-22. 4. Hameeteman W, Tytgat GN, Houthoff HJ, et al. Barrett’s esophagus: development of dysplasia and adenocarcinoma. Gastroenterology 1989; 96:1249-56. 5. Haggitt RC. Pathology of Barrett’s esophagus. J Gastrointest Surg 2000; 4:117-8. 6. Sampliner RE. Updated guidelines for the diagnosis, surveillance, and therapy of Barrett’s esophagus. Am J Gastroenterol 2002;97:1888-95. 7. Hirota WK, Zuckerman MJ, Adler DG, et al. ASGE guideline: the role of endoscopy in the surveillance of premalignant conditions of the upper GI tract. Gastrointest Endosc 2006;63:570-80. 8. Curvers WL, Kiesslich R, Bergman JJ. Novel imaging modalities in the detection of esophageal neoplasia. Best Pract Res Clin Gastroenterol 2008;22:687-720. 9. 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. 10. Kara MA, Peters FP, Fockens P, et al. Endoscopic video-autofluorescence imaging followed by narrow band imaging for detecting early neoplasia in Barrett’s esophagus. Gastrointest Endosc 2006;64:176-185. 11. 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. 12. Curvers WL, Herrero LA, Wallace MB, et al. Endoscopic tri-modal imaging is more effective than standard endoscopy in targeting early-stage neoplasia in Barrett’s esophagus. Gastroenterology 2010;139:1106-14. 13. Casparie M, Tiebosch AT, Burger G, et al. Pathology databanking and biobanking in The Netherlands, a central role for PALGA, the nationwide histopathology and cytopathology data network and archive. Cell Oncol 2007;29:19-24. 14. Hulscher JB, Haringsma J, Benraadt J, et al. Comprehensive Cancer Centre Amsterdam Barrett Advisory Committee: first results. Neth J Med 2001;58:3-8. 15. Curvers WL, Rosmolen W, Elzer B, et al. Low-grade intra-epithelial in Barrett’s esophagus: Over-diagnosed but underestimated. Am J Gastroenterol 2010;105:1523-30. 16. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;47:251-5. 17. Curvers W, Baak L, Kiesslich R, et al. Chromoendoscopy and narrowband imaging compared with high-resolution magnification endoscopy in Barrett’s esophagus. Gastroenterology 2008;134:670-9. 18. Curvers WL, Bohmer CJ, Mallant-Hent RC, et al. Mucosal morphology in Barrett’s esophagus: interobserver agreement and role of narrow band imaging. Endoscopy 2008;40:799-805. 19. Herrero LA, Curvers WL, Bansal A, et al. Zooming in on Barrett oesophagus using narrow-band imaging: an international observer agreement study. Eur J Gastroenterol Hepatol 2009;21:1068-75.
www.giejournal.org
Curvers et al
Endoscopic trimodal imaging in Barrett’s esophagus in general practice
20. Skacel M, Petras RE, Gramlich TL, et al. The diagnosis of low-grade dysplasia in Barrett’s esophagus and its implications for disease progression. Am J Gastroenterol 2000;95:3383-7. 21. Montgomery E, Goldblum JR, Greenson JK, et al. Dysplasia as a predictive marker for invasive carcinoma in Barrett esophagus: a follow-up study based on 138 cases from a diagnostic variability study. Hum Pathol 2001;32:379-88.
22. Peters FP, Brakenhoff KP, Curvers WL, et al. Histologic evaluation of resection specimens obtained at 293 endoscopic resections in Barrett’s esophagus. Gastrointest Endosc 2008;67:604-9. 23. Mino-Kenudson M, Brugge WR, Puricelli WP, et al. Management of superficial Barrett’s epithelium-related neoplasms by endoscopic mucosal resection: clinicopathologic analysis of 27 cases. Am J Surg Pathol 2005;29:680-6.
www.giejournal.org
Volume 73, No. 2 : 2011 GASTROINTESTINAL ENDOSCOPY
203