**原文标题**: 2020 WGO建议：结直肠息肉切除术后监测评估原则 --- Principles for Evaluation of Surveillance After Removal of Colorectal Polyps: Recommendations From the World Endoscopy Organization C olorectal polyps such as ade- nomas and serrated polyps are precursors for colorectal cancer (CRC). Therefore, removal of such polyps re- duces CRC risk.1 Patients who had ad- enomas or serrated polyps removed at colonoscopy are believed to be at increased risk of developing more polyps later in life and eventually CRC.2,3 Thus, colonoscopy surveillance after polyp removal is currently recommended.4 Case-control and cohort studies have indicated that surveillance after removal of advanced adenomas is associated with CRC incidence reduc- tion,5,6 although some modelling studies are reporting only a marginal benefit of surveillance.7 No random- ized trials have comparing the efficacy of surveillance in reduction of CRC incidence or mortality are available Owing to increasing colonoscopy screening activity around the world, more and more individuals are diagnosed with polyps and, there- fore, surveillance has become one of the most frequent indications for colonoscopy, requiring large amounts of resources and creating capacity problems in many coun- tries.8,9 Colonoscopy is an inconve- nient, invasive, and expensive procedure with a risk of complica- tions. Thus, surveillance colonoscopy for patients after polyp removal should be targeted at patients who are most likely to benefit, and rec- ommended at the minimum fre- quency required for decreasing the risk of cancer.10–12 The World Endoscopy Organization (WEO) requested this position statement to guide decision makers, clinicians, and researchers covering 4 areas of colonoscopy surveillance after polyp removal: (1) general principles and definitions in surveillance after polyp removal, (2) definitions of ex- posures and outcomes, (3) relevant comparators for surveillance studies, and (4) relevant thresholds to define cancer risk and surveillance efficacy. Methods This position statement and its rec- ommendations were developed based on a modified Delphi process.11 For the purpose of the project, the WEO appointed a project steering committee of 4 members of its working group on surveillance after detection of colorectal neoplasia (M.R., M.B., C.H., R.J.). The WEO invited a multidisciplinary group of 20 experts (Appendix 1) in gastroenterology, gastrointestinal endos- copy, epidemiology, and public health, including the 4 steering committee members, from across the world to participate in an expert panel to discuss the initial questionnaire and to vote in the subsequent successive rounds of the Delphi process. The panelists were cho- sen based on their expertise in colonos- copy, epidemiology, surveillance practice, and/or research, or participation in sur- veillance guidelines development, all of them with multiple relevant publications on these topics. Based on the initial questionnaire and the feedback from the panel, the steering committee developed a series of structured statements for voting in the following 4 areas of colonoscopy surveillance after polyp removal: (1) general surveillance principles, (2) surveillance outcome measures, (3) comparators in surveillance studies, and (4) thresholds for defining benefit of surveillance. The panel members were asked to indicate their agreement with each state- ment using a Likert scale with 5 possible answers (strongly disagree, disagree, neither agree nor disagree, agree, or strongly agree). In a free-text field, the panel members were allowed to comment on each statement, if desired. Two voting rounds were performed. Primary consensus was defined as
80% of participants agreeing or strongly agreeing to a statement. Statements with no consensus in the first round were either discarded (when consensus was <40%), or modified, merged, or split according to comments and revoted on in a second round. Secondary consensus was defined as
50% agreeing and <20% disagreeing. Participants received feedback about the anonymized results after each voting round. A flowchart with number of statements at the different rounds can be seen in Supplementary Figure 1. The discussion round and the 2 subsequent voting rounds were per- formed between January 2018 and December 2018. Recommendations The original questionnaire included 42 statements. After 2 rounds of voting, 39 statements achieved pri- mary or secondary consensus (“accepted statements”) and are included in this report (Supplementary Figure 1). Supplementary Tables 1, 2, and 3 show all statements and their voting results. Supplementary Table 4 shows the statements that did not reach consensus. The following outlines the most important statements in the 4 areas (panel recommendations in Italic, explaining text in Roman). General Surveillance Principles (for Statements, see Supplementary Table 1) a. The primary aim is to reduce CRC incidence. b. The secondary goal is to reduce CRC mortality. c. CRC mortality reduction by sur- veillance is achieved firstly by reducing CRC incidence and sec- ondly by detecting early stage CRC at surveillance. Endoscopic surveillance should aim to reduce CRC incidence by polyp removal rather than mortality by downstaging of diagnosis of invasive cancer. Patients receiving surveillance have already undergone a “clearing” Gastroenterology 2020;158:1529–1533 COMMENTARIES colonoscopy; therefore, surveillance should aim to prevent rather than detecting CRC. d. Surveillance colonoscopy should only be offered to individuals with a sufficiently high risk to expect a clinically significant benefit from surveillance. e. The impact of surveillance on CRC incidence must be balanced against its harms (colonoscopy complications and psychological distress), patient burden, and costs. Colonoscopy capacity is limited and the number of patients with removed polyps is higher, owing to rapidly increasing CRC screening activity and improvements in tech- nique and technology in detecting polyps during colonoscopy. The benefit of colonoscopy surveillance must be balanced with its harms and burden. The potential benefit should outweigh the potential harm from psychological or physical adverse events at colonoscopy, and the burden that is directly related to the prevalence of each type of polyp at colonoscopy, as well as the fre- quency at which surveillance is recommended. Outcome Measures after Polyp Removal and for Evaluation of the Effectiveness of Surveillance (Statements on This Topic Can Be Seen in Supplementary Table 2) a. CRC incidence is the preferred outcome measure. b. CRC mortality is an acceptable outcome measure. Once agreed that the primary aim of surveillance is CRC incidence reduction (see above), the presurveillance and postsurveillance estimate of CRC inci- dence becomes the top-ranking outcome for assessing the baseline risk and the efficacy of surveillance, respectively. Because CRC mortality reduction has been defined as a secondary aim of surveillance, postsurveillance CRC mor- tality risk is also an acceptable outcome. Both of these outcome measures require long-term follow-up owing to a long time from the initial polyp removal to the outcome. The choice between CRC inci- dence and mortality as the primary outcome also entails methodological differences. Despite being agreed upon as the primary goal of sur- veillance, estimates of CRC incidence reduction are prone to completeness of follow-up, lead time bias, and overdiagnosis bias.13 When such bias cannot be properly addressed, CRC mortality represents a more unbiased outcome.14 If available, the panel recommends reporting of both outcomes. c. The surrogate measures of “any adenomas,” “any serrated polyps,” and “any polyps” at surveillance colonoscopy should no longer be used in studies aiming at identifying patients at risk for future CRC or for eval- uation of the effectiveness of surveillance, whereas “advanced colorectal polyps” is considered acceptable (although imperfect). Although widely used, the panel recommends against the use of non- advanced polyps as a surrogate mea- sure for presurveillance or postsurveillance CRC risk. This recommendation is based on the epidemiology of colorectal polyps as compared with invasive CRC; although the panel believes that most CRCs arise from colorectal polyps, the panel rec- ognizes that the vast majority of colo- rectal polyps do not progress to cancer. Thus, these surrogate measures are not reliable predictors of future CRC risk. Owing to the inherent limitations in estimating CRC incidence or mortality, the panel recognized the need for possible surrogate measures in sur- veillance studies. In this regard, the rate of “advanced colorectal poly- ps”—defined as an advanced adenoma or advanced serrated lesion (Supplementary Table 1)—represents an acceptable surrogate outcome, are considered a target of CRC screening.15 However, as the transition rate and time of progression from advanced adenomas to cancer is unknown, this surrogate measure is of lower validity compared with CRC incidence and mortality. d. If the term “advanced colorectal neoplasia” (summating advanced colorectal polyps and CRC) is used as an outcome measure, the panel recommends always also displaying advanced colorectal polyps and CRC separately. As adopted by most studies, the panel recommends that the use of “advanced colorectal neoplasia” is an acceptable outcome only if the rate of advanced polyps and already invasive lesions are reported separately. These 2 entities entail completely different treatments, risks of harms, and prognoses. e. Absolute values of CRC risk and absolute estimated risk re- ductions through surveillance rather than relative effects should be used for scientific studies and for guideline rec- ommendations in polyp surveillance. To make informed choices, pa- tients and caregivers need to be able to value the benefits of surveillance colonoscopy after polyp removal in the context of their absolute risk of future disease, that is, CRC. For instance, this would simplify the comparison between the expected benefit of surveillance against the competing cause of general mortality that should marginalize the need of surveillance in patients with severe comorbidities or elderly age. Comparator Groups and Thresholds (Statements on This Topic Can Be Seen in Supplementary Table 3) a. Surveillance effectiveness is best assessed using an appropriate comparator group and considering the risk of bias of these comparators. COMMENTARIES 1530 b. When assessing postpolypectomy CRC risk, we consider an age- and gender-matched general population comparator to be the optimal comparator group. c. When assessing surveillance effec- tiveness, we consider a same-risk nonsurveillance comparator to be the optimal comparator group. The panel established 2 main comparator situations to define the need for surveillance, specially in observational studies. The first is the general population. If there is an equivalent CRC risk between a post- polypectomy cohort with no surveil- lance and the general population (unscreened), under the basic assumption that if patients are not at increased risk of CRC incidence or mortality, screening rather than surveillance is recommended. The latter is comparison between post- polypectomy patients receiving and not surveillance. If there is an equiva- lence of CRC risk between 2 cohorts of patients who had polyps remov- ed—one with surveillance and one without surveillance—irrespective of the baseline risk, if surveillance is not reducing the risk, it should not be recommended. In contrast, if the group undergoing surveillance has lower CRC incidence than the equivalent cohort non receiving surveillance, in this case, surveillance should be warranted (Table 1). It should be noted that although the same-risk cohort that warranted surveillance colonoscopy but did not undergo surveillance serves as a comparator group, it is subject to bias, for example owing to non-attenders potentially being a less health-aware group, thus perhaps making other unhealthy lifestyle choices (eg, smoking). These confounders are likely to exaggerate the necessity for and the benefit of surveillance. When studying short-term surro- gate outcomes (surveillance colonos- copy yield), it is not possible to have a no-surveillance comparator. Useful in- formation can be gained from analyzing the advanced colonic polyp (ACP) yield separately from the CRC yield, and by using a general popula- tion comparator (which can be sur- mised from general population screening colonoscopy datasets).2 Appropriate use of these comparators is shown in Table 2. There may be occasions where other comparator groups are helpful to Table 1.Interpretation of Findings on Long-term Follow-up With Comparator Group Comparator: General population Comparator: Equivalent cohort of patients not undergoing surveillance Cohort of patients at increased CRC risk undergoing surveillance If the surveillance cohort’s CRC risk is similar to the general population, surveillance had successfully decreased that increased risk. If the surveillance cohort’s CRC risk is similar to an equivalent nonsurveillance cohort, surveillance had no impact on CRC risk, and is not justified. If the group undergoing surveillance has lower incidence than the equivalent cohort not undergoing surveillance, surveillance should be warranted. Cohort of patients at increased CRC risk not undergoing surveillance If the nonsurveillance cohort’s CRC risk is similar to the general population, that cohort was not at increased risk of CRC. — CRC, colorectal cancer. Table 2.Surveillance Colonoscopy Yield, CRC and Comparator Groups Low CRC yield (including interval cancers) High CRC yield (including interval cancers) Low advanced colorectal polyp yield Indicates that the cohort is not at increased risk and therefore the procedure is not warranted Remains possible that cohort is at increased risk but the colonoscopy interval is too short The cohort is at increased risk The current surveillance strategy is ineffective Possibly indicates that CRC has arisen through alternative pathway from adenoma-CRC sequence Possibly indicates that the colonoscopy interval is too long High advanced colorectal polyp yield Probably the ideal scenario Cohort probably at risk Interval not too short; does not exclude the possibility that interval could be extended Cohort is at risk Either current surveillance is ineffective, or the quality of the prior colonoscopy was inadequate Colonoscopy interval is too long NOTE. The cohort categories do not preclude there being subgroups within each category who do/do not benefit from surveillance. This categorization illustrates why, for this purpose, combining advanced colonic polyps and CRCs into one category (advanced colorectal neoplasia) is inappropriate. CRC, colorectal cancer. COMMENTARIES 1531 address specific questions. Potential comparators include: C Negative colonoscopy cohort— individuals who have undergone a colonoscopy where neither pre- malignant polyps nor cancer was detected. This cohort is likely to have a risk of CRC that is below that of the general population. C Low-risk (no surveillance) cohort—individuals who have undergone a colonoscopy where the outcome was that no sur- veillance was warranted. Again, this cohort is likely to have a risk below that of the general population. C Local population screening com- parator—in a resource- constrained health care system, there is an opportunity cost of performing colonoscopy with a low yield, because it potentially denies other higher risk patients from undergoing colonoscopy. Therefore, costs and availability of resources may represent additional factors to be consid- ered when setting a surveillance threshold. d. At surveillance colonoscopy, a high advanced colonic polyp and low CRC yield is the optimal combination, because it indicates that the cohort is probably at risk and that the surveillance interval is not too short. However, it does not exclude the possibility that the surveillance interval could be safely extended. Agreement was reached on the fact that a combination of a high yield of advanced colorectal polyps and low yield of CRC at surveillance supports the need and proper timing of sur- veillance. Different combinations be- tween advanced colonic polyp and CRC yield can be seen in Table 2. Thresholds for Surveillance (Statements on This Topic Can Be Seen in Supplementary Tables 3 and 4) The panel was asked to provide a threshold for an absolute risk of CRC for when surveillance would be worthwhile to recommend. Different alternatives were provided at the questionnaire regarding expected reduction on absolute and relative risk of CRC and advanced polyps (Supplementary Table 4). However, there was not a final agreement and, thus, the steering group decided not to include any state- ment on threshold for an increased risk to recommend surveillance. The definition of reduction of a high CRC risk cohort to the same risk of the general population as proxy for a favorable efficacy of surveillance did not reach consensus, as some of the experts voiced concerns about this metric as a satisfactory risk reduction (ie, a lower than general population risk should be pursued). Conclusion In this WEO recommendations, we have discussed the general principles of surveillance after polyp removal. The panel has defined new and groundbreaking definitions for the evaluation of benefits, harms, and burdens for surveillance, to be used in future studies and guide- lines. We recommend CRC incidence as the consensus primary outcome in surveillance studies, with CRC mortality as a secondary outcome. Also, we have defined recommenda- tions for situations that should not be reported routinely as meaningful outcomes for surveillance evaluation in the future. We have reached a consensus about the most appropriate com- parators for surveillance studies, which are the general population and a same risk cohort not receiving postcolonoscopy surveillance, and provided considerations about different scenarios where surveil- lance can be considered as relevant or irrelevant. Finally, we have tried to establish relevant thresholds to define baseline cancer risk and surveillance efficacy. However, no agreement was reached for these metrics and more work is needed to establish what we should envision as an adequate reduction in CRC incidence and mortality with surveillance. The aim of this position statement is that these recommendations may be used in future studies and guidelines about this important topic in colonos- copy activity and CRC prevention. MATTHEW D. RUTTER University Hospital of North Tees Stockton on Tees, UK and Northern Institute for Cancer Research Newcastle University Newcastle-upon-Tyne, United Kingdom MICHAEL BRETTHAUER University of Oslo and Oslo University Hospital Oslo, Norway CESARE HASSAN Nuovo Regina Margherita Hospital Rome, Italy RODRIGO JOVER Hospital General Universitario de Ali- cante Instituto de Investigación Sanitaria ISA- BIAL Alicante, Spain On behalf of the WEO Surveillance Working Group Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Gastroen- terology at www.gastrojournal.org, and at https://doi.org/10.1053/j.gastro. 2019.12.052.