Abstracts 1080a Endoscopic Tri-Modal Imaging (ETMI) for the Detection and Classification of Early Colorectal Neoplasia: A Multi-Centre Randomized Controlled Trial T. Kuiper, Frank J. Van Den Broek, Anton H. Naber, Ellert Van Soest, Pieter Scholten, Wouter L. Curvers, Rosalie C. Mallant-Hent, Jan Van Den Brande, Jeroen M. Jansen, Arnoud H. Van Oijen, Willem Marsman, Jacques Bergman, Paul Fockens, Evelien Dekker INTRODUCTION:Endoscopic Tri-modal Imaging (ETMI) is a novel technique which combines high-resolution white light endoscopy (HRE), autofluorescence imaging (AFI) and narrow band imaging (NBI) and may improve the detection and differentiation of colonic lesions. ETMI has only been subjected to research in expert settings. This randomized controlled multi-centre study compared ETMI with standard video endoscopy (SVE) for the detection and differentiation of early colorectal neoplasia in a non-expert setting. METHODS:Patients undergoing colonoscopy in non-academic hospitals for screening or surveillance were randomized to undergo either ETMI or SVE. All segments were inspected twice. In the ETMI group, first inspection was done with HRE followed by inspection with AFI. In the SVE group, inspection was performed twice with SVE. All detected lesions were inspected by AFI for colour and by NBI for pit pattern analysis. The calculated sample size of this study was 234 patients (117 per group). Enrolment is anticipated to be complete in January 2010. RESULTS:Up to date, 210 patients have been randomised (119 males, mean age 57 yrs), 102 to the ETMI group and 108 to the SVE group. In the ETMI group, 175 lesions (80 adenomas) were detected during the first inspection with HRE. During the second inspection with AFI, 68 additional lesions (35 adenomas) were found.In the SVE group, 181 lesions (75 adenomas) were detected during the first inspection and 89 additional lesions (25 adenomas) were found during second inspection. There was no significant difference in adenoma detection rate between the two groups (p⫽ 0.386), despite a significantly longer withdrawal time in the ETMI group (12.10 vs. 14.40 minutes, p⬍0.001).The adenoma miss rates of HRE and SVE were 30.4% and 25.0 %, respectively (p⫽ 0.172). The diagnostic values of NBI, AFI and the combination of AFI and NBI for differentiating adenomatous from non-adenomatous lesions are shown in the table below. CONCLUSION:Our preliminary results show that ETMI does not improve the adenoma detection rate compared to SVE in a non-expert setting. Furthermore, the additional value of AFI on top of HRE proved unsatisfactory. Both NBI and AFI had high sensitivities and low specificities in the differentiation of colonic lesions. Also, the accuracy of the algorithm combining AFI and NBI for differentiating lesions was not or only marginally higher than the accuracy of AFI or NBI alone. Technique AFI NBI AFI & NBI
Sensitivity
Specificity
Accuracy
91% 88% 88%
36% 66% 58%
62% 76% 74%
1080b Narrow Band Imaging (NBI) for Adenoma Detection in High Risk Patients: A Randomised, Controlled Trial James E. East, Noriko Suzuki, Thomas Guenther, Nicola Palmer, Ana Ignjatovic, Brian P. Saunders INTRODUCTION: Randomised data so far suggest that NBI provides limited or no benefit for adenoma detection at colonoscopy in patients at standard risk of adenomas[1]; however a back-to-back study in high risk patients with hereditary non-polyposis colorectal cancer (HNPCC) did show improved adenoma detection[2]. No randomised data exist for other higher risk groups where NBI might be more effective. AIMS & METHODS: We aimed to test whether NBI could improve adenoma detection in a patient cohort at high risk for adenomas, defined as: 3 or more adenomas at last colonoscopy or 1 adenoma ⫽⬎10mm in size; cancer follow up; fecal occult blood test (FOBT) positive. This cohort have an estimated adenoma prevalence of 40-50%. The primary outcome measure was the number of patients with at least one adenoma detected. Patients were randomised in a 1:1 ratio to examination with NBI or white light (WLE), with high definition (HDTV) colonoscopes (Olympus, Lucera), by one of 3 experienced endoscopists. Minimum withdrawal time was set at 8 minutes. The trial design had 80% power to detect an increase in adenoma detection from 40 to 60% (20% absolute increase, 50% relative increase) at 5% significance level. RESULTS: 214 patients were randomised (planned sample size). Groups were well matched for baseline variables with the exception of a greater proportion of male subjects in the WLE group. There was no significant difference in the proportion of patients with at least one adenoma detected between groups, relative risk [benefit] 1.09 (95% CI 0.92-1.30), Table. Nor was there a significant difference in the number of patients with 3 or more or 5 or more adenomas. CONCLUSION: Adenoma detection rates with both NBI and HDTV white light were high. NBI did not improve adenoma detection in a cohort at
AB142 GASTROINTESTINAL ENDOSCOPY
Volume 71, No. 5 : 2010
high risk for adenomas, nor was there benefit seen in any sub-group. It seems unlikely that NBI will provide substantial increases in adenoma detection rates. ClinicalTrials.gov identifier: NCT00279357REFERENCE(S): 1. Endoscopy 2009;41:59-632. Gut 2008;57:65-70 Adenoma and polyp detection rates with NBI or HDTV white light examination NBI (nⴝ108)
WLE (nⴝ106)
P Value
66 74(70%) 10.40 71(67%) 201 12 17 282
0.16** 0.005* 0.068** 0.37* 0.47** 1.0** 0.15** 0.45**
Age (median, yrs) 65 Gender (male) 55(51%) Extubation time (min.sec) 11.34 1 or more adneomas 79(73%) Total Adenomas 217 Advanced adenomas 12 Flat Adenomas 46 Total Polyps 321 * Fisher exact test; ** Mann-Whitney
1080c A Multicenter, Prospective, Randomized Controlled Trial Comparing Standard Definition White Light (SDWL), High Definition White Light (HDWL) and Narrow Band Imaging (NBI) Colonoscopy for the Detection of Colon Polyps and Real Time Prediction of Histology Amit Rastogi, Dayna S. Early, Neil Gupta, Ajay Bansal, Vikas Singh, Michael Ansstas, Sreenivasa S. Jonnalagadda, Christine E. Hovis, Srinivas Gaddam, Sachin B. Wani, Steven A. Edmundowicz, Prateek Sharma Background: HDWL colonoscopes and NBI are recent developments in colon imaging. Their ability to detect and characterize colon polyps compared to SDWL is unclear. Aims: To compare SDWL, HDWL, and NBI for 1.The proportion of pts with adenomas (TA)2.The number of TA detected per subject3.Real time polyp histology prediction Methods:Pts referred for screening or surveillance colonoscopy were enrolled and randomized to SDWL, HDWL, or NBI at 2 tertiary referral centers (6 endoscopists). Following intubation of cecum, colonic mucosa was carefully inspected with the respective imaging modality during withdrawal of colonoscope. Polyp location, size, morphology and surface mucosal/vascular pattern (defined a priori) were documented and polyps were then removed. Surface patterns used to predict histology have been described by us previously using NBI and were standardized amongst all endoscopists. Colonic withdrawal times (time spent in inspecting the mucosa only) were measured. Primary outcome was proportion of pts with TA. Additional outcomes included number of TA detected per subject and Sensitivity (Se), Specificity (Sp) and Accuracy (Ac) for real-time polyp histology prediction. Sample size of 210 pts in each arm was calculated to detect a 15% higher proportion of pts with TA with HDWL and NBI compared to SDWL. Fisher’s exact and ANOVA were used for statistical analysis.Results:630 pts were enrolled - 210 in each arm. Mean age61 yrs, males-411(65%), Caucasian-451(72%), screening-415(66%). Proportion of males, number of screening procedures, quality of bowel prep were not different between the 3 gps. Mean age was higher in the HDWL arm (62 yrs) compared to SDWL (60 yrs) and NBI (60 yrs) {p⫽0.001}. Withdrawal time was higher for NBI (7.5 min) compared to SDWL (6.9 min) and HDWL (6.6 min) {p⫽0.0002}. Prevalence of TA, number of TA per subject and Se, Sp and Ac for predicting TA for the three imaging modalities are shown in table.Conclusions: 1.Proportion of pts with TA was not significantly different between the 3 gps; however, both HDWL and NBI detected 7% more pts with TA compared to SDWL. 2.HDWL and NBI detected significantly higher number of TA overall (59%) and TA per patient compared to SDWL. 3.Ac of NBI for predicting TA real time during colonoscopy was significantly superior to SDWL and HDWL.
Pts with TA Total number of TA Number of TA per pt Se for predicting TA Sp for predicting TA Ac for predicting TA
SDWL Nⴝ210
HDWL Nⴝ210
NBI Nⴝ210
p value
81 (39%) 145 0.69 52% 91% 69%
96 (46%) 235 1.1 67% 85% 73%
97 (47%) 237 1.1 90% 68% 83%
0.22 0.019 0.019 ⬍0.001 ⬍0.001 ⬍0.001
1080d Cap-Fitted Colonoscopy Reduces Adenoma Miss Rates: A Randomized Tandem Study David G. Hewett, Douglas K. Rex Polyps are missed during colonoscopy, even with meticulous technique. AIM: To investigate the impact of cap-fitted colonoscopy (CFC) on the miss rate for
www.giejournal.org