The authors have declared that no competing interests exist.
Conceived and designed the experiments: MF EB TM. Performed the experiments: MF EB RF DA SK MS TM. Analyzed the data: MF EB HR TM. Contributed reagents/materials/analysis tools: TM MS PS. Contributed to the writing of the manuscript: MF EB TM. Analyzed the biopsy specimens, revised the manuscript and gave technical support: PS. Reviewed the manuscript for important intellectual content: VE. Supervised the statistical analysis: AH. Senior author and principal investigator, study concept and design, analysed and interpreted the data, wrote the manuscript, performed the statistical analysis, supervised the study and finally approved the version to be published: TM.
The Endocuff is a device mounted on the tip of the colonoscope to help flatten the colonic folds during withdrawal. This study aimed to compare the adenoma detection rates between Endocuff-assisted (EC) colonoscopy and standard colonoscopy (SC).
This randomized prospective multicenter trial was conducted at four academic endoscopy units in Germany. Participants: 500 patients (235 males, median age 64[IQR 54–73]) for colon adenoma detection purposes were included in the study. All patients were either allocated to EC or SC. The primary outcome measure was the determination of the adenoma detection rates (ADR).
The ADR significantly increased with the use of the Endocuff compared to standard colonoscopy (35.4%[95% confidence interval{CI} 29–41%] vs. 20.7%[95%CI 15–26%], p<0.0001). Significantly more sessile polyps were detected by EC. Overall procedure time and withdrawal time did not differ. Caecal and ileum intubation rates were similar. No major adverse events occurred in both groups. In multivariate analysis, age (odds ratio [OR] 1.03; 95%[CI] 1.01–1.05), male sex (OR 1.74; 95%CI 1.10–2.73), withdrawal time (OR 1.16; 95%CI 1.05–1.30), procedure time (OR 1.07; 95%CI 1.04–1.10), colon cleanliness (OR 0.60; 95%CI 0.39–0.94) and use of Endocuff (OR 2.09; 95%CI 1.34–3.27) were independent predictors of adenoma detection rates.
EC increases the adenoma detection rate by 14.7%(95%CI 6.9–22.5%). EC is safe, effective, easy to handle and might reduce colorectal interval carcinomas.
ClinicalTrials.gov
Colorectal cancer derived from precancerous colonic lesions is one of the most frequently occurring tumors in the industrialized world. In a recently published study by Nishihara et al., it could be shown that endoscopic removal of polyps was associated with reduced incidences of colorectal cancer mortality
We performed a prospective randomized controlled multicenter trial, conducted at four academic endoscopy units in Germany (HELIOS Medical Centers Northeim, Siegburg, Helmstedt and the Department of Gastroenterology at University Medical Center Göttingen), comparing EC and SC for colonic adenoma detection. The protocol for this trial and supporting CONSORT checklist are available as supporting information; see
A total of 818 consecutive patients underwent colonoscopy between February 2014 and July 2014, of whom 500 patients fulfilling the inclusion criteria were prospectively enrolled. After giving written informed consent, all patients underwent general physical examination and laboratory testing. Within a time interval of two weeks after enrollment, patients underwent either EC-assisted colonoscopy or standard colonoscopy. Block randomization (block size: 50) of the two groups (EC and SC) was performed by using computerized randomization lists. An independent researcher, who was not part of the endoscopy team, generated the allocation sequence and assigned participants to the groups. Participants were eligible for the study if they were at least 18 years of age and were scheduled for screening, surveillance or diagnostic colonoscopy (anemia or abdominal discomfort). Patients with known colonic strictures, status post partial colonic resection, acute diverticulitis within six weeks before the examination, acute exacerbation of chronic inflammatory bowel disease, pregnancy or inability to give informed consent were excluded from the study. The enrollment flow chart is shown in
Variable | EC | SC | p-Value |
Patients, N | 249 | 243 | |
Median age (years) [IQR] | 64 [54–73] | 63 [53–73] | 0.572 |
Gender (male/female), N | 122/127 | 109/134 | 0.358 |
First time colonoscopy, N (%) | 85 (34) | 92 (38) | 0.390 |
Diabetes, N (%) | 11 (4) | 7 (3) | 0.364 |
ASS medication, N (%) | 28 (11) | 33 (14) | 0.432 |
Prior abdominal surgery, N (%) | 42 (17) | 44 (18) | 0.718 |
ASS = acetylsalicylate; IQR = interquartile range.
Colonoscopy using the EC device (Endocuff AEC120 or AEC140, Arc Medical, Leeds, UK) was performed with the device mounted on the tip of the endoscope. For the Fujifilm endoscopes (EC-590 WM4/WL4, Fujifilm Europe, Düsseldorf, Germany), we used the AEC120 device, for the Olympus colonoscopes (CF-H180AI/AL, Olympus, Tokyo, Japan), the AEC140 device was attached at the tip of the scope.
Procedure time was defined as: beginning with the insertion of the colonoscope, including therapeutic interventions, and ending with removal of the endoscope.
Withdrawal time was defined as: beginning with withdrawal of the colonoscope from the caecal pole minus time spent for interventions, and ending with removal of the endoscope.
Colon cleansing was achieved by using Moviprep (Norgine, Amsterdam, Netherlands) according to a standard bowel preparation protocol provided by the manufacturer. For sedation purposes, propofol was applied at an initial dose of 60 to 80 mg and repeated at fractionated doses of 40 to 80 mg, on demand, during the examination. If necessary, fentanyl at 50 to 100 µg for analgesia was intravenously administered. The colonoscopy started with the patient lying in the left lateral position and the colonoscope with or without the EC was pushed straight forward to the caecal pole. Routinely, 3 to 5 attempts for intubating the terminal ileum were made. A microchronometer was used to determine the overall procedure time and withdrawal time. Every detected polyp during withdrawal which was removed by standard forceps or snare, was put into a separate pot for histological analysis and individual assignment to the particular colonic location. The colonoscopies were performed by 10 highly experienced board-certified gastroenterologists, each of which having performed at least 3000 colonoscopies. All endoscopists had experience with at least 5 EC-assisted colonoscopies before participating in the study.
The specimens were evaluated by three institutes of pathology. All lesions were classified as hyperplastic polyps, or serrated, tubular, tubulovillous, villous adenomas or carcinoma. A five-tiered classification and grading system according to the Vienna classification
The primary outcome measure of the present study was direct comparison of EC-assisted colonoscopy with standard colonoscopy with regard to the adenoma detection rate (defined as the number of patients in whom at least one adenoma could be found). Secondary outcome measures included the polyp detection rate, number of polyps detected per colonoscopy, number of adenomas detected per colonoscopy, analysis of polyp distribution and polyp morphology in the different parts of the colon (rectum, sigmoid, descending colon, left flexure, transverse colon, right flexure, ascending colon and caecum). Secondary outcomes also included polyp detection proved by histology (hyperplastic polyp, LGIN, HGIN, invasive carcinoma), rate of caecal intubation, intubation rates of the ileum, overall colon cleansing level (graded as either poor, fair or good, according to the criteria of Leighton and colleagues)
In modification of the statistical methods of Rastogi
We prospectively enrolled a total of 500 patients into the study who fulfilled the inclusion criteria. 8 patients were excluded, either because of study withdrawal, stenosis or active inflammation (Enrollment flow chart see
The caecum could be reached in almost all patients, regardless of whether or not the EC was used (EC: 96%[95%CI 93–98%] vs. SC: 94%[95%CI 91–97%]; p = 0.624). The main procedural outcomes are presented in
Variable | EC | SC | p-Value |
Cecum intubation, N (%) | 238 (96) | 229 (94) | 0.624 |
Ileum intubation, N (%) | 165 (66) | 173 (71) | 0.239 |
Procedure time (min), median [IQR] | 17 |
17 |
0.959 |
Withdrawal time (min), median [IQR] | 6.32 [5.5–8.0] | 6.05 [5.5–8.0] | 0.524 |
Cleanliness score, median [IQR] | 1 |
1 |
0.797 |
1 = good, N (%) | 185 (74) | 176 (72) | |
2 = fair, N (%) | 53 (21) | 54 (22) | |
3 = poor, N (%) | 12 (5) | 12 (5) | |
Propofol dosage (mg), median (IQR) | 150 [100–200] | 170 [130–210] |
IQR: interquartile range.
A total of 501 polyps were detected during colonscopy within the study period: In the EC-group, 312 polyps were found while in the SC-group 189 polyps could be retrieved. The polyp detection rate (PDR), that is the number of patients in whom at least one polyp was detected, was significantly higher in the EC-group compared to the SC-group (55.4%[95%CI 49–62%] vs. 38.4%[95%CI 32–44%], p<0.0001) resulting in a PDR increase of 17% (95%CI 6.4–23.8%). Significantly more polyps smaller than 1 cm in size as well as more sessile polyps were detected in the EC group (N = 272 vs. 172, p<0.0001 and N = 250 vs. 145, p<0.0001 for sessile polyps). The detection rates for flat or pedunculated polyps were not different. For details see
EC |
SC |
p-Value | |
Polyp detection rate, N (%) | 138 (55.4) | 93 (38.4) | |
Polyps per patient, median [IQR] |
2 |
1 |
0.250 |
Polyp by size analysis, N (%) | |||
≥1 cm | 40 (12.8) | 17 (9) | |
<1 cm | 272 (87.2) | 172 (91) | |
Polyp by morphology analysis, N (%) | |||
Sessile | 250 (80.1) | 145 (76.7) | |
Flat | 48 (15.4) | 33 (17.4) | 0.072 |
Pedunculated | 14 (4.5) | 11 (5.9) | 0.275 |
Adenoma detection rate (ADR), N (%) | 87 (35.4) | 50 (20.7) | |
Number of adenomas (LGIN), N (%) | 138 (93) | 87 (99) | |
Number of adenomas (HGIN), N (%) | 6 (4) | 1 (1) | 0.061 |
Number of carcinomas, N (%) | 5 (3) | 0 (0) | 0.061 |
Adenomas per patient, median [IQR] |
1 |
1 |
0.851 |
Hyperplastic Polyp detection rate, N (%) | 71 (28.7) | 51 (21.2) | 0.053 |
Hyperplastic polyps per patient, median [IQR] |
1 [0–2] | 1 [0–1] | 0.922 |
*only patients considered in whom at least one adenoma or polyp, respectively were detected.
3 polyps (EC) and two polyps (SC) not retrieved endoscopically.
When using the Endocuff, the sigmoid polyp detection rates (30.1%[95%CI 24–36%] vs. 17.3%[95%CI 12–22%], p = 0.001) and the ascending colon polyp detection rates (12.4%[95%CI 8–17%] vs. 5.8%[95%CI 3–7%], p = 0.010) were significantly higher. In all other colonic segments, no difference in polyp detection rates could be observed. For absolute number of polyps, significantly more polyps <1 cm in size could be detected in the sigmoid (108 vs. 52, p = 0.001) and ascending colon (31 vs. 14, p = 0.036) with the use of EC. When analyzing the polyp morphology, EC-assisted colonoscopy detected significantly more sessile polyps in the sigmoid (99 vs. 45, p = 0.002) and caecal region (29 vs. 7, p = 0.003) as well as significantly more flat polyps in the transverse colon (10 vs. 0, p = 0.015) compared to SC. Details are presented in
Variable, N | EC | SC | P-Value |
Rectum polyp detection rate, N (%) | 40 (16) | 31 (12.8) | 0.297 |
No of Rectum polyps <1 cm, N (%) | 52 (90) | 45 (96) | 0.544 |
No of Rectum polyps >1 cm, N (%) | 6 (10) | 2 (4) | 0.165 |
sessile, N (%) | 47 (81) | 35 (74) | 0.279 |
flat, N (%) | 8 (14) | 7 (15) | 0.549 |
pedunculated, N (%) | 3 (5) | 5 (11) | 0.455 |
Sigmoid polyp detection rate, N (%) | 75 (30.1) | 41 (17.3) | |
No of Sigmoid polyps <1 cm, N (%) | 108 (91) | 52 (95) | |
No of Sigmoid polyps >1 cm, N (%) | 11 (9) | 3 (5) | 0.053 |
sessile, N (%) | 99 (83) | 45 (82) | |
flat, N (%) | 12 (10) | 7 (13) | 0.118 |
pedunculated, N (%) | 8 (7) | 3 (5) | 0.214 |
Descending colon polyp detection rate, N (%) | 14 (6) | 18 (7) | 0.422 |
No of descending colon polyps <1 cm, N (%) | 13 (87) | 15 (75) | 0.782 |
No of descending colon polyps >1 cm, N (%) | 2 (13) | 5 (25) | 0.241 |
sessile, N (%) | 11 (73) | 15 (75) | 0.371 |
flat, N (%) | 3 (20) | 4 (20) | 0.972 |
pedunculated, N (%) | 1 (7) | 1 (5) | 0.986 |
Left flexure polyp detection rate, N (%) | 4 (1.6) | 5 (2) | 0.709 |
No of left flexure polyps <1 cm, N (%) | 3 (75) | 6 (100) | 0.456 |
No of left flexure polyps >1 cm, N (%) | 1 (25) | 0 (0) | 0.324 |
sessile, N (%) | 3 (75) | 5 (83) | 0.676 |
flat, N (%) | 1 (25) | 1 (17) | 0.986 |
pedunculated, N (%) | 0 (0) | 0 (0) | 1.000 |
Transverse colon detection rate, N (%) | 23 (9.2) | 15 (6.2) | 0.203 |
No of transversum polyps <1 cm, N (%) | 24 (80) | 16 (94) | 0.405 |
No of transversum polyps >1 cm, N (%) | 6 (20) | 1 (6) | 0.107 |
sessile, N (%) | 19 (63) | 17 (100) | 0.900 |
flat, N (%) | 10 (33) | 0 (0) | |
pedunculated, N (%) | 1 (4) | 0 (0) | 0.323 |
Right flexure polyp detection rate, N (%) | 13 (5) | 10 (4) | 0.561 |
No of right flexure polyps <1 cm, N (%) | 15 (94) | 11 (92) | 0.159 |
No of right flexure polyps >1 cm, N (%) | 1 (6) | 1 (8) | 0.986 |
sessile, N (%) | 13 (81) | 11 (92) | 0.301 |
flat, N (%) | 3 (19) | 1 (8) | 0.328 |
pedunculated, N (%) | 0 (0) | 0 (0) | 1.000 |
Ascending colon polyp detection rate, N (%) | 31 (12.4) | 14 (5.8) | |
No of ascendens polyps <1 cm, N (%) | 31 (84) | 14 (88) | |
No of ascendens polyps >1 cm, N (%) | 6 (16) | 2 (12) | 0.166 |
sessile, N (%) | 29 (78) | 11 (69) | 0.015 |
flat, N (%) | 7 (19) | 3 (19) | 0.329 |
pedunculated, N (%) | 1 (3) | 2 (12) | 0.984 |
Caecum polyp detection rate, N (%) | 24 (9.6) | 14 (5.8) | 0.107 |
No of caecum polyps <1 cm, N (%) | 26 (79) | 13 (81) | 0.110 |
No of caecum polyps >1 cm, N (%) | 7 (21) | 3 (19) | 0.221 |
sessile, N (%) | 29 (88) | 7 (44) | |
flat, N (%) | 4 (12) | 9 (56) | 0.335 |
pedunculated, N (%) | 0 (0) | 0 (0) | 1.000 |
No = number; bold = significant differences.
The adenoma detection rate (meaning all patients in whom at least one adenoma was detected during colonoscopy) increased by 14.7%[95%CI 6.9–22.5%] when using the Endocuff. In the EC group, the ADR was 35.4%[95%CI 29–41%] compared to 20.7%[95%CI 15–26%] in the SC group (p<0.0001) (
In the majority of colonoscopies, no adverse events occurred in either group. Significantly more minor mucosal lacerations without any clinical impact were observed in EC-assisted procedures in comparison to SC (EC: 18 vs. SC: 2, p<0.0001). No major bleeding events, perforations or cuff losses occurred (
Adverse Event | EC | SC | p-Value |
Minor mucosal lacerations, N (%) | 18 (7.3) | 2 (0.8) | |
Major bleeding | 0 | 0 | ––––– |
Perforation | 0 | 0 | –––– |
Loss of Cuff | 0 | 0 | –––– |
SpO2 decline (<90%) | 0 | 0 | –––– |
SpO2 = Saturation of peripheral Oxygen; bold = significant differences.
Patients, in whom at least one adenoma was detected, were significantly older compared to those without adenoma detection (median age 69 [61–75] vs. 63 [52–73], p = <0.0001). In the adenoma detection cohort, the proportion of male patients was significantly higher (58%[95%CI 50–67%] vs. 42%[95%CI 37–47%], p = 0.001). Moreover, the withdrawal times and overall procedure times were significantly longer in those patients where adenomas were detected (withdrawal time 7 [6.30–8.25] vs. 6 [5.09–7.00] min, p<0.0001; procedure time: 20
The x-axis represents the Odds ratio on a log scale with the reference line (dashed), Odds ratios (diamond) and 95% CI (whiskers).
Colonoscopy performed within colon cancer prevention programs remains the gold standard for colorectal cancer screening. The adenoma detection rate (ADR) seems to be most crucial in preventing interval colon cancer, as a suboptimal ADR is significantly associated with a higher incidence of colon cancer
Another aspect is the importance of the withdrawal time for the ADR. Earlier studies have shown, that insufficient withdrawal times correlate with a lower ADR and thus with a higher incidence of colorectal cancer
Other technical innovations like cap-assisted colonoscopy, back-to-back colonoscopy or narrow band imaging (NBI) or a combination of several methods were evaluated for higher polyp and adenoma detection rates
It holds true that back-to-back colonoscopy improves polyp and adenoma detection rates
The value of NBI or FICE techniques in the colon is critically discussed, since not all studies have shown improvement of the ADR. In a recently published randomized trial comparing NBI with high definition colonoscopy for the detection of adenomas, NBI was not able to improve the ADR
Water-immersion or water-exchange colonoscopy is a new emerging technique which is still under evaluation. However, in a recently published meta-analysis comparing water immersion and water exchange with air insufflation colonoscopy, a reduction in procedural pain could be observed, but no difference in adenoma detection
The effort and the aspect of losing water through the anus makes its use doubtful for broad colon cancer prevention programs.
Cap-assisted colonoscopy (CAC) showed conflicting results for improving ADR. While in the study by Rastogi et al., ADR improvement was observed
In another recently published randomized controlled trial, conducted by Gralnek et al., the adenoma miss rate was significantly lower with the use of a novel full-spectrum endoscopy platform as compared to standard colonoscopy
A great strength of our study is its multicenter randomized character and its statistical power. So far, this is the study with the largest number of patients included. We further wanted to determine, which independent factors contribute significantly to the variability of the ADR. Therefore we performed multivariate binary logistic regression analysis. Variations of multivariate regression include forward selection and backward elimination methods. We decided to choose backward elimination because all independent factors begin in the model and non-influential variables are eliminated. In our multivariate analysis all included variables remained in the model. We proved the Endocuff, for the first time, to be an independent factor for adenoma detection (OR 2.09). We also found a significantly higher proportion of male patients in whom at least one adenoma was found (OR 1.74). This observation is consistent with previously published data
The overall procedure time with EC did not differ statistically compared to standard colonoscopy.
Statistical differences were seen in the polyp detection rates for polyps less than 1 cm in diameter and to a lesser extent for those larger than 1 cm. Obviously, even larger polyps can better be detected during EC-assisted colonoscopy in critical anatomical positions. The polyp detection rate in the sigmoid region was notably higher with use of EC. Thus, the benefit of the Endocuff is particularily favourable as the majority of polyps occure in the left hemicolon29.
According to the inverse correlation of ADR and interval colon cancer risk (one percent increase of ADR results in three percent reduction of colon cancer risk)13, our observed increase of 14.7% in the ADR does consequently have the potential to reduce the interval colon cancer risk by more than 40%. In our study, for every seven patients screened by EC assisted colonoscopy an additional adenoma could be detected (number needed to screen (NNS = 1/(ADREC−ADRSC)). Assuming a 2 to 5% risk for adenoma to carcinoma progression within the following 10 years, 140 to 350 EC-assisted colonoscopies would be necessary to prevent one colon cancer case. Hence, the use of EC is economically reasonable.
In our study, no significant adverse effects were detected, besides minor mucosal lacerations without any clinical impact. We have not noticed any Endocuff losses during withdrawal as previously reported by our study group
The use of EC is safe, feasible and significantly improves the adenoma detection rate. As previously discussed, improving the ADR correlates with a decrease in colon cancer risk. Future follow-up studies are necessary to show the impact of EC assisted colonoscopy on colon cancer morbidity and mortality, before broad use of the EC device in colon cancer prevention programs can be recommended.
There has been further development of the Endocuff device with even longer arms (Endocuff vision) that have the potential to improve ADR. The first pilot study on this device showed promising results
The impact of our findings on colon cancer mortality prevention remains uncertain, thus follow-up studies are necessary.
Due to technical reasons such as the occasional visibility of the rubber arms of the endocuff and the small, but detectable resistence during the endoscopy proceding into the colon, this study was not blinded.
(DOC)
(DOCX)
We thank the endoscopy teams of the endoscopy units at HELIOS Siegburg, Northeim, Helmstedt and University of Göttingen Hospitals for their excellent technical assistance and Mrs. Karly Conrads (Oklahoma City, OK, USA) for her review as a native speaker. We thank Dr. Markus Neef, member of the endoscopy unit at HELIOS Medical Center Siegburg, for his valuable study support. Parts of the study have been selected for an oral presentation at the United European Gastroenterology Week 2014 in Vienna, Austria.