ORIGINAL ARTICLE: Clinical Endoscopy

Transcription

1 ORIGINAL ARTICLE: Clinical Endoscopy A meta-analysis evaluating the accuracy of colon capsule endoscopy in detecting colon polyps Theodore Rokkas, MD, PhD, Konstantinos Papaxoinis, MD, Konstantinos Triantafyllou, MD, Spiros D. Ladas, MD Athens, Greece Background: Colon capsule endoscopy (CCE) is a new, noninvasive method for examining the entire colon. The reported yield of CCE in detecting colorectal polyps has shown variable results. bjective: To assess the accuracy of CCE by pooling data of existing trials. Design: Meta-analysis. The fixed-effects or random-effects model was used as appropriate, based on whether homogeneity or heterogeneity, respectively, was indicated by the Cochran Q test. Setting: Studies that estimated the accuracy of CCE were identified. Two investigators independently conducted the search and data extraction. Patients: A total of 626 individuals were included in this meta-analysis. Intervention: Each patient underwent CCE and conventional colonoscopy. Main Outcome Measurements: Per-patient sensitivity and specificity, with 95% confidence intervals (CI). Results: Findings were categorized as significant polyps, that is, a report of a polyp 6 mm in size or 3 or more polyps of any size, or any polyp, that is, a report of any polyp found, independent of size. Pooled data on sensitivity and specificity with a 95% CI were estimated. For any polyp found, the pooled data showed per-patient CCE sensitivity of 73% (95% CI, 68%-77%) and specificity of 89% (95% CI, 81%-94%). For significant polyps, the respective values were 69% (95% CI, 62%-75%) and 86% (95% CI, 82%-90%). Limitations: A small number of studies met inclusion criteria. Conclusion: CCE is a reasonable method for screening asymptomatic individuals for colorectal polyps. It may be particularly useful for patients with incomplete colonoscopy, those with contraindications for conventional colonoscopy, and those unwilling to undergo colonoscopy because of its perceived inconvenience and discomfort. (Gastrointest Endosc 2010;71:792-8.) Colorectal cancer (CRC) is a major cause of morbidity and mortality of the general population in the Western world. The high cost of treatment also causes a substantial economic burden on health care systems worldwide. 1 Although identification and removal of precancerous adenomatous polyps during colonoscopy has been shown to be highly effective in preventing CRC, as shown in large cohort and case-control studies, 2-4 the application of CRC screening is low, especially when compared with the high rate of attendance for breast, cervical, and prostate gland screening programs. 5,6 For this reason, more attractive, noninvasive options have Abbreviations: AUC, area under the curve; CCE, colon capsule endoscopy; CRC, colorectal cancer; ROC, receiver operating characteristic; SROC, summary ROC. DISCLOSURE: All authors disclosed no financial relationships relevant to this publication. Copyright 2010 by the American Society for Gastrointestinal Endoscopy /$36.00 doi: /j.gie Received July 28, Accepted October 22, Current affiliations: Gastroenterology Clinic (T.R.), Henry Dunant Hospital, Hepato-Gastroenterology Unit (K.P., S.D.L.), 1st Department of Internal Medicine-Propaedeutic, Medical School, Athens University, Laikon University General Hospital, Gastroenterology Unit, (K.T.), 2nd Department of Internal Medicine-Propaedutic, Medical School, Athens University, Attikon University General Hospital, Athens, Greece. Presented in part at Digestive Disease Week, May 30-June 4, 2009, Chicago, Illinois (Gastroenterology 2009;136.) Reprint requests: S.D. Ladas, 1st Department of Internal Medicine- Propaedeutic, Medical School, Athens University, Laiko General Hospital of Athens, 17 Ag. Thoma Street, Athens 11527, Greece. 792 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 4 :

2 Rokkas et al Colon capsule endoscopy been developed for people who are noncompliant to current screening procedures. Along these lines, wireless colon capsule endoscopy (CCE) (PillCam Colon; Given Imaging Ltd, Yoqneam, Israel), which is able to visualize the colonic mucosa, has been successfully pioneered. So far, only scattered, small-sized trials exist evaluating its efficacy in detecting colon polyps, compared with established colon diagnostic methods, that is, colonoscopy. The aim of the present study therefore was to meta-analyze the data of existing CCE trials to help determine the yield and miss rate of this new diagnostic modality. MATERIAL AND METHODS Data sources Extensive English-language, computer-aided medical literature searches of the PubMed/MEDLINE and Embase databases for human studies were performed to find relevant publications on the diagnostic performance of CCE in detecting colonic polyps, in comparison to established colon diagnostic methods, that is, colonoscopy. The following search terms were used: colon and capsule and endoscopy or wireless endoscopy. No beginning date limit was used. The search was updated until the end of July To expand our search, we screened the articles that remained after the selection process for potentially suitable references. In addition, we sought abstracts presented from 2004 to 2009 at the proceedings of the International Conference on Capsule Endoscopy, Digestive Disease Week, the annual meeting of the British Society of Gastroenterology, and the United European Gastroenterology Week. Two authors (K.P. and K.T.) extracted data from each study independently by using a predefined form, and disagreements were resolved by discussion and consensus by a third reviewer (T.R. or S.L.). Study selection Inclusion and exclusion criteria were delineated before the commencement of the literature search. Thus, eligible studies published as full articles or abstracts were included in this meta-analysis if they met the following criteria: (1) written in the English language and (2) providing sufficient data for the authors to construct a 2 2 contingency table to calculate sensitivity and specificity. When two articles reported the same study, the publication that was more informative was selected. In studies reporting data according to level of bowel preparation, data referring to the best level were taken into account. Capsule Summary What is already known on this topic The reported yield of colon capsule endoscopy in detecting colorectal polyps has shown variable results. What this study adds to our knowledge Pooled data from 626 patients showed colon capsule endoscopy to have a per-patient sensitivity of 73% and a specificity of 89% for any finding. Characteristics of studies For each included study, information was collected concerning year of publication, country of origin, number of investigated patients, and number and type of endoscopic findings. The latter were categorized as any polyp, that is, a report of any polyp found, regardless of size, or significant polyps, that is, a report of a polyp 6 mm in size or 3 or more polyps of any size. Statistical analysis Agreement on the selection of studies between the two reviewers was evaluated by the k coefficient. Sensitivity and specificity of CCE in comparison to colonoscopy (with corresponding 95% confidence intervals [CIs]) were calculated from the original numbers given in the included studies by constructing 2 2 contingency tables. Pooled results (with corresponding 95% CIs) were derived by using the fixed effects model (Mantel and Haenszel method 7 ), unless significant heterogeneity was present, in which case the random effects model was applied (DerSimonian and Laird method 8 ). Forest plots were constructed for visual display of individual studies and pooled results. In addition, the results of the individual studies were displayed in a receiver operating characteristic (ROC) space to illustrate the distribution of sensitivities and specificities. A weighted symmetric summary ROC (sroc) curve was computed by using the Moses-Shapiro-Littenberg method, 9 and the area under the curve (AUC) was calculated, with perfect tests having an AUC of 1 and poor tests having an AUC close to Heterogeneity between studies was evaluated with the Cochran Q test and the inconsistency index (I 2 ), which has been proposed as a measure for quantifying the amount of heterogeneity, 11 and it was considered to be present if the Q test provided a P value of Statistical analyses were executed by using Statistical Package for the Social Sciences software version 14.0 (SPSS Inc, Chicago, IL), the Comprehensive Meta-Analysis software version 2 (Biostat Inc, Englewood, NJ), and Meta-DiSc statistical software version 1.4 (Unit of Clinical Biostatistics, Ramon y Cajal Hospital, Madrid, Spain) Volume 71, No. 4 : 2010 GASTROINTESTINAL ENDOSCOPY 793

3 Colon capsule endoscopy Rokkas et al Figure 1. Flow diagram of the studies identified in this meta-analysis. TABLE 1. The main characteristics of studies selected for meta-analysis Study, year Country Type of publication Type of study No. of patients involved Any polyp reported Significant polyp reported Schoofs et al, Germany Full article Single center 36 Yes Yes Eliakim et al, Israel Full article Multicenter (3 centers) Lewis et al, USA Abstract Multicenter (3 centers) 84 Yes Yes 25 No Yes Van Gossum et al, Belgium, Spain, France, Germany, Italy, United Kingdom Full article Multicenter (8 centers) 320 Yes Yes Sieg et al, Germany Full article Single center 36 Yes No Sacher-Huvelin et al, France Abstract Multicenter* 105 No Yes Costamagna et al, Italy Abstract Single center 20 No Yes *Number of participating centers not mentioned in the abstract. RESULTS Descriptive assessment and study characteristics A flow chart describing the process of study selection is shown in Fig. 1. Out of 106 titles initially generated by the literature searches, 7 studies, of which were published as full articles and as abstracts, fulfilled the inclusion criteria and therefore were eligible for meta-analysis. Initial agreement between the reviewers for the selection of relevant articles was high (k 0.95; 95% CI, ). The main characteristics of the studies eligible for meta-analysis are shown in Table 1. The studies were conducted in different parts of the world 794 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 4 :

4 Rokkas et al Colon capsule endoscopy Figure 2. Forest plot (random effects model), concerning sensitivity (upper panel) and specificity (lower panel), with 95% confidence interval, for detecting any polyp, that is, a report of any polyp found, regardless of size. and included 626 patients. Three studies 14,15,17 reported data on both any polyp and significant polyps, 1 study 16 reported data on any polyp only, and the other 3 studies reported data on significant polyps only. Two studies 19,20 reported data for various levels of bowel preparation. In these two studies, data referring to the best level of bowel preparation were included in the meta-analysis. In all meta-analyzed studies, the same type of capsule, that is, PillCam Colon, was used. Capsules were expelled within 10 hours after ingestion in 85.3% (95% CI, ) of the individuals. Diagnostic performance of CCE Detection of any polyp found. Four studies contained data concerning polyps found of any size. The pooled data (random effects analysis) showed CCE sensitivity of 73% (95% CI, 68-77) and specificity of 89% (95% CI, 81-94). Heterogeneity in sensitivities was not observed among the studies (Cochran Q test 0.81, df 3, P.85, I 2 0%). However, specificities were heterogeneous (Q test 18.6, df 3, P.0003, I %). The forest plots of these results are presented in Fig. 2, and the corresponding ROC plot with sroc is displayed in Fig. 3. The AUC under the sroc was Detection of significant polyps. Six studies 14,15,17-20 contained data concerning significant polyps, that is, a polyp 6 mm in size or 3 or more polyps of any size. The pooled data (random effects analysis) showed CCE sensitivity of 69% (95% CI, 62-75) and specificity of 86% (95% Figure 3. SROC, with 95% confidence interval, for studies reporting any polyp, that is, a report of any polyp found, regardless of size. SROC, weighted symmetric summary receiver operating curve; AUC, area under the curve. CI, 82-90). Heterogeneity in sensitivities was not observed among the studies (Cochran Q test 2.76, df 5, P.74, I 2 0%). However, specificities were heterogeneous (Q test 32, df 5, P.000, I %). The forest plots of these results are presented in Fig. 4, and the corre- Volume 71, No. 4 : 2010 GASTROINTESTINAL ENDOSCOPY 795

5 Colon capsule endoscopy Rokkas et al Figure 4. Forest plot (random effects model), concerning sensitivity (upper panel) and specificity (lower panel), with 95% confidence interval, for detecting significant polyps, that is, a polyp 6 mm in size or 3 or more polyps of any size. sponding ROC plot with sroc is displayed in Fig. 5. The AUC for the sroc was DISCUSSION Figure 5. SROC, with 95% confidence interval, for studies reporting significant polyps, that is, a polyp 6 mm in size or 3 or more polyps of any size. SROC, weighted symmetric summary receiver operating curve; AUC, area under the curve. Small intestine capsule endoscopy is a well-evaluated method for noninvasive detection of small-bowel disease. However, CCE is a more recent development, and so far, only scattered trials evaluating its efficacy in comparison with colonoscopy exist. In this study, we meta-analyzed the data of existing CCE trials to help determine the yield and miss rate of this new diagnostic modality. The pooled data showed CCE per-patient sensitivity of 73% and specificity of 89% for any finding (ie, report of any polyp found, independent of size). The area under the sroc was , with perfect tests having an AUC of 1 and poor tests having an AUC close to 0.5. The respective sensitivity and specificity values for significant findings, that is, a report of a polyp 6 mm in size or 3 or more polyps of any size, were 69% and 86%. The area under the sroc was These figures mean that, although it is not as accurate as colonoscopy, CCE could be a reasonable alternative method of colon examination. 796 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 4 :

6 Rokkas et al Colon capsule endoscopy CRC represents a major cause of morbidity and mortality in the general population. Despite the fact that several effective tests, including colonoscopy, have been widely available in recent decades, the application of CRC screening is still much lower than that for other cancers such as breast, cervical, and prostate gland cancer. Colonoscopy is the preferred method of examining the colon. However, in certain cases, there are technical difficulties in completing the examination, that is, in visualizing the cecum. In addition, psychological barriers regarding the potential nuisance of a painful examination through the rectum discourage some people from accepting colonoscopy. In such cases, barium enema or virtual CT colonography are recommended to complete the colon evaluation. CCE, although not as accurate as colonoscopy, seems to be a reasonable alternative solution. It is administered per os, is painless, and, as this meta-analysis showed, it possesses reasonable sensitivity and specificity. It seems therefore, that CCE is feasible and safe, and it is expected to increase compliance among individuals for CRC screening. It may not be feasible for screening in high-volume centers, but as shown recently in a relevant study, 16 CCE is feasible in a gastroenterological practice setting. The best possible preparation for cleaning the large bowel is of crucial importance in CCE. In each of the studies included in this meta-analysis, the patients underwent the same preparation for cleaning the large bowel to assure uniformity and comparability of CCE and conventional colonoscopy. However, some additional fine tuning of the preparation might help to avoid the presence of opaque liquid that, in some cases, compromises complete visualization during CCE. Indeed, this matter was addressed in two of the studies, 19,20 and it was found that excellent preparation improved sensitivity. One interesting question is how CCE compares with the other friendly method for colon investigation, that is, virtual colonography performed in expert radiological centers. So far, only one small study 18 has addressed this matter. In that study, CCE was directly compared with virtual colonography and colonoscopy. It was found that CCE picked out 63.5% of significant findings, as compared with 54.5% for virtual colonography and 82% for colonoscopy. However, more well-designed studies are needed in order to answer this question fully. CCE has some limitations, in that it does not allow the physician to obtain tissue biopsy specimens or perform therapeutic maneuvers, and it is not possible to aspirate liquid luminal debris. Future studies on CCE should address the limitations of this technique. Technological advances may enable scientists to equip the capsule with capabilities of drug delivery, brushings and cytology, physiological measurements with imaging, optical biopsy, and immunological cancer recognition. Concerning the cost-effectiveness of CCE, recently, Hassan et al 21 showed that the cost-effectiveness of capsule endoscopy in CRC screening will mainly depend on its ability to improve compliance in the general population. If such compliance mirrors that of other cancer screenings, CCE will be an efficient strategy for CRC prevention. In conclusion, the results of this meta-analysis showed that colon CCE is a feasible alternative method for colon investigation, including screening for polyps and colorectal cancer, patients with incomplete colonoscopy, 22 those with contraindications for conventional colonoscopy, or those unwilling to undergo colonoscopy because of its perceived inconvenience and discomfort. Hopefully, this method will expand the portfolio of screening methods available. However, larger, multicenter, well-designed trials are needed to further establish the role of CCE in the evaluation of the large bowel in health and disease. REFERENCES 1. Ries LA, Wingo PA, Miller DS, et al. The annual report to the nation on the status of cancer, , with a special section on colorectal cancer. Cancer 2000;88: Atkin WS, Morson BC, Cuzick J. Long-term risk of colorectal cancer after excision of rectosigmoid adenomas. N Engl J Med 1992;326: Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. The National Polyp Study Workgroup. N Engl J Med 1993;329: Citarda F, Tomaselli G, Capocaccia R, et al. Efficacy in standard clinical practice of colonoscopic polypectomy in reducing colorectal cancer incidence. Gut 2001;48: Nadel MR, Blackman DK, Shapiro JA, et al. Are people being screened for colorectal cancer as recommended? Results from the National Health Interview Survey. Prev Med 2002;35: Boyle P, Autier P, Bartelink H, et al. European Code Against Cancer and scientific justification: third version (2003). Ann Oncol 2003;14: Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22: DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7: Moses LE, Shapiro D, Littenberg B. Combining independent studies of a diagnostic test into a summary ROC curve: data-analytic approaches and some additional considerations. Stat Med 1993;12: Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143: Cochran WG. The combination of estimates from different experiments. Biometrics 1954;8: Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002;21: Zamora J, Abraira V, Muriel A, et al. Meta-DiSc: a software for metaanalysis of test accuracy data. BMC Med Res Methodol 2006;6: Schoofs N, Deviere J, Van Gossum A. PillCam Colon capsule endoscopy compared with colonoscopy for colorectal tumor diagnosis: a prospective pilot study. Endoscopy 2006;38: Eliakim R, Fireman Z, Gralnek IM, et al. Evaluation of the PillCam Colon capsule in the detection of colonic pathology: results of the first multicenter, prospective, comparative study. Endoscopy 2006;38: Sieg A, Friedrich K, Sieg U. Is PillCam Colon capsule endoscopy ready for colorectal cancer screening? A prospective feasibility study in a Volume 71, No. 4 : 2010 GASTROINTESTINAL ENDOSCOPY 797

7 Colon capsule endoscopy Rokkas et al community gastroenterology practice. Am J Gastroenterol 2009; 104: Van Gossum A, Munoz Navas M, Fernandez-Urien I, et al. Capsule endoscopy versus colonoscopy for the detection of polyps and cancer. N Eng J Med 2009;361: Lewis B, Rex D, Leiberman D. Capsule colonoscopy: an interim report of a pilot 3 arm, blinded trial of capsule colonoscopy versus virtual colonoscopy and colonoscopy [abstract]. Am J Gastroenterol 2006; 101(suppl 1):A Sacher-Huvelin S, Le Roun M, Sebile V, et al. Wireless capsule endoscopy compared to conventional colonoscopy in patients at moderate or increased risk for colorectal cancer: interim analysis of a prospective multicenter study [abstract]. Gastroenterology 2009; 136(suppl 1):A Costamagna G, Spada C, Riccioni ME, et al. Evaluation of bowel preparation and procedure for PillCam Colon capsule: an interim analysis [abstract]. Gastroenterology 2009;136(suppl 1):A Hassan C, Zullo A, Winn S, et al. Cost-effectiveness of capsule endoscopy in screening for colorectal cancer. Endoscopy 2008;40: Triantafyllou K, Tsibouris P, Kalantzis C, et al. PillCam Colon capsule endoscopy does not always complement incomplete colonoscopy. Gastrointest Endosc 2009;69: Availability of Journal back issues As a service to our subscribers, copies of back issues of Gastrointestinal Endoscopy for the preceding 5 years are maintained and are available for purchase from Elsevier until inventory is depleted. Please write to Elsevier Inc., Subscription Customer Service, 3251 Riverport Lane, Maryland Heights, MO or call or for information on availability of particular issues and prices. 798 GASTROINTESTINAL ENDOSCOPY Volume 71, No. 4 :