Colonoscopy is routinely performed for colorectal cancer (CRC) screening, follow-up of other abnormal screening tests, work up of signs and symptoms of gastrointestinal disease, and surveillance after CRC and polyp removal. Post procedure, colonoscopists are expected to provide follow-up recommendations to patients and referring physicians. Recommendations for follow-up after normal colonoscopy among individuals age-eligible for screening, and post-polypectomy among all individuals with polyps are among the most common clinical scenarios requiring guidance 1 .
Risk of metachronous advanced neoplasia is associated with findings on prior colonoscopy. After high quality colonoscopy, patients with no neoplasia detected are at the lowest risk, and those with polyps are risk stratified based on the histology, number, location, and size of polyps detected. Since release of the last US Multi-Society Task Force (Task Force) recommendations for post colonoscopy follow-up and polyp surveillance in 2012 2 , a number of papers have been published on risk of CRC based on colonoscopy findings and patient characteristics, as well as the potential impact of screening and surveillance colonoscopy on outcomes such as incident CRC and polyps. Further, recent studies increasingly reflect the modern era of colonoscopy with more awareness of the importance of quality factors (e.g. adequate bowel preparation, cecal intubation, adequate adenoma detection, complete polyp resection), and utilization of state of the art technologies (e.g. high definition colonoscopes). Higher quality colonoscopy could impact the importance of previously identified risk factors. Our aim was to review newly available evidence and update recommendations for follow-up after colonoscopy with or without polypectomy.
To identify issues of greatest importance for the current revision, we developed Patient, Intervention, Comparison, and Outcome (PICO) questions (Appendix A– SG and DL, with input from TK). In consultation with a certified medical librarian (KH), literature searches were performed in PubMed, Embase, and CINAHL with a combination of controlled vocabulary and keyword terms for colonoscopy, polyps, and polypectomy surveillance (see Appendix B for search terms). English-language articles since 01/01/2012 were retrieved. Searches were run on March 30, 2017, and identified a total of 1904 unique articles.
Criteria used for inclusion/exclusion of titles, abstracts, and manuscripts are outlined in Table 1 . All titles were reviewed by a single author (SG), with potentially relevant titles selected for abstract review. All abstracts were reviewed by two authors (SG and DL), with potentially relevant abstracts selected for full manuscript review. Included manuscripts were reviewed in detail by the same two authors. The final list of papers selected for review was supplemented by repeating the literature search through 9/2018 to identify papers published since time of the literature search, as well as through opportunistic identification of additional relevant papers. References directly relevant to final recommendations were identified through joint consensus (SG and DL). Based on prior findings and the current literature review, post colonoscopy management recommendations were developed by two authors (SG and DL) and refined through consensus discussion with all authors after circulating both draft recommendations and a table summarizing key findings of papers that were included for manuscript review. For each recommendation, the quality of evidence ( Table 2 ) and strength of recommendation was rated, using our previously described approach 3 . Strong recommendations mean that most informed patients would choose the recommended management and that clinicians can structure their interactions with patients accordingly. Weak recommendations mean that patients’ choices will vary according to their values and preferences, and clinicians must ensure that patients’ care is in keeping with their values and preferences.
Criteria for inclusion/exclusion of titles, abstracts, and manuscripts
| Review Phase (reviewer) | Inclusion/exclusion criteria |
|---|---|
| Title (SG) | Goal: Identify paper(s) that might examine the relationship between baseline colonoscopy exam and subsequent neoplasia on follow up Exclusion Criteria Title clearly not relevant Review papers except other guidelines Focus on high risk conditions such as inflammatory bowel disease, prior history of CRC, or hereditary CRC syndromes Focus on children |
| Abstract (SG and DL) | Goal: Identify paper(s) that might examine relationship between the baseline colonoscopy exam and subsequent neoplasia on follow up Exclusion Criteria Narrative review or editorial Guidelines Focus on high risk conditions such as inflammatory bowel disease, prior history of CRC, or hereditary CRC syndromes Focus on children Abstract only; no associated manuscript Focus exclusively on endoscopic resection method or immediate completeness of resection Focus other than on post polypectomy surveillance or normal colonoscopy outcomes |
| Manuscript (SG and DL) | Goal: Identify paper(s) that might examine relationship between baseline colonoscopy exam and subsequent neoplasia on follow up, relevant to PICO questions Inclusion Criteria Relevant to one or more PICO questions Examined relationship between baseline colonoscopy exam findings and detection of CRC or advanced adenoma on follow up Examined relationship between surveillance vs. no surveillance for individuals who have undergone baseline polypectomy Exclusion Criteria * Methods insufficiently described to enable interpretation of study outcomes Narrative review or editorial Guidelines Focus on high risk conditions such as inflammatory bowel disease, prior history of CRC, or hereditary CRC syndromes Focus on children Abstract only; no associated manuscript Focus exclusively on endoscopic resection method or immediate completeness of resection Focus other than on post polypectomy surveillance or normal colonoscopy outcomes |
Grading of Recommendations Assessment, Development, and Evaluation Ratings of Evidence
| Rating of evidence | Definition |
|---|---|
| A: High quality | Further research is very unlikely to change our confidence in the estimate of effect |
| B: Moderate quality | Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate |
| C: Low quality | Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate |
| D: Very low quality | Any estimate of effect is very uncertain |
This document does not include recommendations for follow-up for individuals with hereditary colorectal cancer syndromes (e.g. Lynch syndrome and Familial Adenomatous Polyposis), inflammatory bowel disease, a personal history of CRC (including malignant polyps), family history of CRC or colorectal neoplasia, or Serrated Polyposis Syndrome. As such, our recommendations for follow up after colonoscopy and polypectomy do not apply to these groups except in cases where polyp findings would result in a shorter colonoscopy interval than indicated based on the status of these clinical conditions. Further, recommendations for polypectomy technique were outside the scope of this document. Notably, the Task Force has recently issued recommendations for follow-up colonoscopy for individuals with Lynch syndrome 4 and a personal history of CRC 3, 5, 6 . Recommendations for follow-up of Serrated Polyposis Syndrome, management of patients with a malignant polyp, as well as optimal polypectomy technique will be covered in subsequent Task Force documents.
The primary goals of colonoscopy screening and post-polypectomy surveillance are to reduce CRC incidence and mortality. We provide a review of the available evidence on the impact of surveillance on these outcomes. Next, we provide recommendations for follow-up strategies, with a summary of new evidence, including an overall assessment of the quality of evidence and strength of recommendations. This is followed by a summary of key limitations of existing evidence, future research opportunities, and best practices for research in the field. Given the large amount of data on post colonoscopy follow-up, we primarily focus on new publications since the Task Force recommendations in 2012.
Terms and definitions *
Adenoma ≥10 mm in size Adenoma with tubulovillous/villous histology Adenoma with high grade dysplasia Adenoma ≥10 mm in size Adenoma with tubulovillous/villous histology Adenoma with high grade dysplasia Advanced neoplasia 3 or more adenoma* We propose moving forward that rather than using categories such as “high risk adenoma” or “low risk adenoma” that research papers specify the individual criteria being captured by the category (e.g. use 1–2 adenomas
CRC, colorectal cancer; ADR, adenoma detection rate
For the purposes of this review, we have defined high quality based on colonoscopist performance such as adequate adenoma detection rate (ADR), and exam specific characteristics, such as exam complete to cecum, attention to complete polypectomy, and adequate bowel preparation to reliably detect lesions >5mm. Benchmarks for ADR (ADR >30% in men; >20% in women), proportion of exams with adequate prep (>85%), and proportion of exams complete to cecum (>95%) should be universally and routinely monitored as colonoscopy quality metrics in practice 7 . Colonoscopists who are measuring quality metrics, but not meeting them, need to take steps to improve their exam quality, and document this improvement. Polyp size is a major factor in our scenario specific recommendations. Given importance of polyp size for informing surveillance intervals, documentation of a polyp ≥10 mm within a report should be accompanied by an endoscopic photo of the polyp with comparison to an open snare or open biopsy forceps. Such documentation is important for lesions such as HPs, where small size (
A cohort study of 304,774 individuals with normal colonoscopy, versus 980,154 individuals with no lower endoscopy, showed a reduced risk for incident CRC on long term follow-up [Hazard Ratio (HR) 0.44; 95%CI: 0.38–0.52]. The risk was persistently decreased across a range of years since last normal colonoscopy, ranging from a HR of 0.35 for ≤3 years to 0.65 at ≥15 years. Normal colonoscopy was also associated with reduced risk for fatal CRC (HR 0.32; 95%CI: 0.24–0.45) over 300,000 person-years of follow up 8 . A cohort study comparing 131,349 individuals who had normal colonoscopy to the general population in Utah showed the standardized incidence ratio (SIR) for CRC was 0.26 (95%CI: 0.19–0.32) through 5 years and 0.60 (95%CI: 0.44–0.76) for 7 to 10 years follow-up 9 . A 70% relative risk reduction was observed through the 10 year follow-up period (SIR 0.28; 95%CI: 0.24–0.33). Most recently, a cohort study of 1,251,318 adults at average risk for colorectal cancer served by a large health plan in the United States reported a 46% relative reduced risk for incident, and a 88% relative reduced risk for fatal CRC among 99,166 who had a normal screening colonoscopy through the traditionally recommended 10 year follow-up period for these individuals (HR 0.54; 95%CI: 0.31–0.94 for incident and HR 0.12; 95%CI: 0.02–0.82 for fatal CRC) 10 . Notably, reduced risk was noted even up to 12 years post normal screening colonoscopy. A strength of this study was use of a validated approach to identifying screening colonoscopy procedures. A potential limitation was unmeasured differences between plan members who elected for screening colonoscopy versus stool based testing or sigmoidoscopy, including a potential healthy user bias. A modeling study informed by age-specific rates of adenoma, advanced adenoma, and colorectal cancer observed among 4.3 million individuals who underwent screening colonoscopy suggested that a normal colonoscopy was associated with a less than 0.5% 10-year risk of subsequent CRC 11 . Since the 2012 review, we could identify no new data on risk of advanced neoplasia associated with small recto-sigmoid HPs. Prior literature has suggested that such patients have a similar risk of metachronous advanced neoplasia as patients with a normal exam, and recommendations for 10-year repeat exam remain unchanged 2 .
While we found no direct evidence to support the incremental effectiveness of repeat colonoscopy after 10 years, prior modeling studies have suggested that repeat colonoscopy in those with a baseline normal exam does confer additional benefit 12–14 . Knudsen et al. estimated rescreening after initial normal colonoscopy resulted in a reduction from 31.3 lifetime CRC cases per 1000 persons with no further screening to as low as 7.7 cases per 1000 persons with repeat screening 14 . Based upon current available evidence, our recommendation for repeat colonoscopy 10 years after a normal colonoscopy remains unchanged.
Five recent studies have shown that individuals with adenoma, despite adenoma removal, may have increased risk for CRC compared to the general population. An Irish cohort study of 6,972 patients with adenomas identified between 2000–2005 found a 2.9-fold increased risk for incident CRC compared to the general population (SIR 2.85; 95%CI: 2.61–3.25) 15 . Annual reported risk of CRC was 0.43% per year, and cumulative rate of CRC was less than 5% for men, and less than 3.5% for women with up to 10 years follow-up. This study was limited by lack of information on polyp size in the registry, limited information on type of follow-up patients received, and incomplete colonoscopy at baseline in some individuals. A French cohort study of 5,779 patients diagnosed with any adenoma 1990–1999, followed through 2003, found risk of CRC increased 1.3-fold after first adenoma removal compared to the general population (SIR 1.26; 95%CI 1.01--1.56) 16 . Stratifying based on adenoma risk category (advanced adenoma and non-advanced adenoma) showed baseline advanced adenoma was associated with 2.2-fold increased CRC risk compared to the general population (SIR 2.23; 95%CI 1.67-;-2.92), while baseline non-advanced adenoma was associated with reduced CRC risk (SIR 0.68 (95%CI 0.44--0.99). The 10-year cumulative probability of CRC in patients with advanced adenomas was 2.05% (95%CI 1.14--3.64%) with and 6.22% (95%CI 4.26--9.02%) without exposure to subsequent surveillance colonoscopy. A Norwegian cohort study of 40,826 patients with adenomas removed during years 1993–2007 and followed through 2011 found risk for fatal CRC was similar compared to the general population 17 . Risk was decreased by 25% for those with low risk adenoma [defined by single adenoma without advanced histology; Standardized Mortality Ratio (SMR) 0.75; 95%CI: 0.63-- 0.88], but increased 1.2-fold for those with high risk adenoma (defined by ≥ 2 adenomas, villous histology or high-grade dysplasia; SMR 1.16; 95%CI: 1.02--1.31). A limitation of this analysis was the inability to account for polyp size in the definition of high risk adenoma. Among 15,935 participants in a US trial of sigmoidoscopy screening who completed subsequent colonoscopy, compared to those with no adenoma, risk for incident and fatal CRC was increased among participants with advanced adenoma [RR 2.7 (95%CI: 1.9–3.7) for incident; RR 2.6 (95%CI: 1.2–5.7) for fatal], but similar among participants with non-advanced adenoma [RR 1.2 (95%CI: 0.8–1.7) for incident CRC; RR 1.2 (95%CI, 0.5–2.7) for fatal CRC] 18 . Notably, 11.3% of the non-advanced adenoma group had 3 or more adenomas, while 88.7% had 1 to 2 adenomas; none had villous features or high grade dysplasia, and all were smaller than 10 mm. On median 12.9 years follow-up, cumulative CRC incidence was 2.9% for the advanced adenoma group, 1.4% for the non-advanced adenoma group, and 1.2% in the no adenoma group. Caution is warranted in interpreting the incident CRC outcomes for the non-advanced versus no adenoma groups, as the non-advanced group had greater exposure to subsequent colonoscopy follow-up, perhaps introducing detection bias; cumulative colonoscopy exposure after baseline examination was 53.0% versus 36.9% at 5 years, and 78.1% versus 69.9% at 9 years follow-up for the non-advanced vs. no adenoma groups, respectively. Among 122,899 participants in either the Nurses Health Study or the Health Professionals Follow Up Study with median 10 years follow up, compared to those with no polyps at baseline lower endoscopy, risk for incident CRC was increased among participants with advanced adenoma (HR 4.07; 95% CI, 2.89–5.72), with cumulative CRC incidence of 0.6% at 5 years, and 1.7% at 10 years 19 . In contrast, risk for CRC was not increased among individuals with non-advanced adenoma, with cumulative CRC incidence of 0.1% at 5 years, and 0.3% at 10 years, compared to 0.2% at 5 years and 0.4% at 10 years for those with no polyps at baseline. This study shares limitations with the previously mentioned US flexible sigmoidoscopy cohort, in that cumulative exposure to surveillance colonoscopy was high across all groups, such that the favorable outcomes observed could be in part be attributable to exposure to surveillance.
Incremental impact of surveillance colonoscopy after baseline removal of adenoma with low risk features (such as 1 to 2 adenomas
Taken together, new evidence suggests that adenoma bearing patients with identifiable high-risk characteristics remain at increased risk for CRC in the absence of surveillance 17 , and that exposure to surveillance is associated with reduced risk for some high risk groups defined by baseline quality of exam or polyp characteristics. Further, new evidence suggests that most adenoma patients (such as those with 1 to 2 small adenomas) are at lower than average risk for subsequent CRC than the general population after baseline polypectomy. The incremental benefit of subsequent surveillance is uncertain for all patients with polyps, but benefit among patients with higher risk features (size ≥20 mm) is suggested by two studies. These studies highlight the importance of additional research to identify patients most likely to benefit from surveillance, and careful clinical management pending further clarification of which patients are at highest risk, and which strategies will be most effective for reducing risk. Limitations of prior studies include retrospective nature, and subsequent inability to control for confounding factors that could be associated with CRC risk and likelihood of participation in surveillance, such as proclivity towards healthy behaviors and following medical recommendations for follow-up.
In a Danish case-control study of 2,045 CRC cases compared to 8,105 CRC-free controls nested within a cohort of individuals who received colonoscopy between 1977–2009, having an SSP was associated with 3-fold increased odds for CRC (OR 3.07, 95%CI: 2.30–4.10), while having SSP with dysplasia was associated with a nearly 5-fold increased odds for CRC (OR 4.76, 95%CI: 2.59–8.73), compared to having no polyp 21 . A limitation of this study is that it is unclear whether baseline polyps were excised or only biopsied because all SSP patients were identified based on pathology records but colonoscopy records were not reviewed. A cohort study of patients included in a sigmoidoscopy screening trial compared CRC risk among 81 patients with ≥10 mm serrated lesions (including an SSP, traditional serrated adenoma, HP, or unclassified serrated lesions) to risk among patients who had a non-advanced adenoma, normal sigmoidoscopy, or no screening 22 . Compared to the group with no screening, a 2.5-fold non-statistically significant increased risk for incident CRC was observed in individuals with large serrated polyps (HR 2.5; 95%CI: 0.8--7.8). Compared to the normal sigmoidoscopy group, a 4-fold increased risk for incident CRC was observed in individuals with large serrated polyps (HR 4.2; 95%CI: 1.3–13.3). Risk for incident CRC for individuals with advanced adenoma at baseline compared to those with no screening was increased 2-fold (HR 2.0; 95%CI: 1.3--2.9). On multivariable analyses adjusted for histology, size, and number of concomitant adenomas, having a large serrated polyp was associated with a 3.3-fold increased risk for incident CRC (OR 3.3; 95%CI: 1.3--8.6). Interestingly, very little progression (including no progression to cancer) was observed in 23 large serrated polyps left in situ over a median 11 years follow-up, perhaps suggesting that some serrated polyps may be a general biomarker of risk, rather than an intermediate high-risk lesion. This study is limited by the small sample size, and uncertainty regarding whether a group of patients ascertained as a result of a sigmoidoscopy trial are representative of patients routinely encountered with SSP at colonoscopy. The previously mentioned analysis of incident CRC risk among 122,899 patients enrolled in the Nurses Health Study or Health Professionals Follow Up Study included 566 individuals with large serrated polyps ≥10 mm, and 5,010 with serrated polyps ≤ 10mm. Compared to individuals with no polyps at baseline, risk for incident CRC was increased 3-fold among those with large serrated polyps (HR 3.35; 95%CI: 1.37–8.15), but not increased among those with small serrated polyps (1.25; 95% CI: 0.76–2.08) 19 . Cumulative incidence of CRC at 3 and 5 years follow up was 0.4 and 1.1% among individuals with large serrated polyps, and 0.1 and 0.4% among those with small serrated polyps. A limitation of this study was that the definition of serrated polyps included any hyperplastic polyp, TSA, or SSP, precluding ability to distinguish risk by histologic subtype. Overall, despite data suggesting that patients with SSP have increased risk for CRC, the magnitude and significance of risk associated with SSPs is uncertain given limitations of available studies.
Studies published since our last recommendations suggest the evidence to support a low risk for incident and fatal CRC after normal screening colonoscopy is stronger. There continues to be little evidence on the incremental effectiveness of a repeat screening colonoscopy at 10 years after normal colonoscopy, but modeling studies suggest benefit. Recent studies vary in estimates of risk for incident and fatal CRC after baseline adenoma removal, with some showing increased risk, and others showing decreased risk. New evidence suggests that exposure to surveillance colonoscopy after baseline adenoma removal may reduce CRC risk, but the magnitude of benefit associated with exposure to surveillance colonoscopy is unclear. Generally, individuals with more advanced findings at baseline (or colonoscopy with poor baseline quality) have higher risk for subsequent cancer relative to those with low risk findings (e.g. 1 to 2 small adenomas) and benefit of repeat surveillance colonoscopy is more demonstrable in the higher risk groups. Further, determining which groups are most likely to benefit, and whether surveillance reduces CRC mortality remains a challenge. Recent studies suggest patients with SSPs may have an increased risk for incident CRC, but magnitude and consistency of risk remains uncertain. Overall, more evidence is needed to understand which patients are at lowest and highest risk for incident and fatal CRC after initial colonoscopy, and whether surveillance can consistently improve outcomes. Nonetheless, pending generation of new evidence, we provide colonoscopy surveillance recommendations to guide patient care, given the prevailing conventional wisdom and available observational evidence suggesting that some patients remain at risk for CRC despite baseline polypectomy.
Recommendations for follow-up after colonoscopy and polypectomy
USMSTF recommendations for post colonoscopy follow up in average risk adults with normal colonoscopy or adenomas *
| Baseline Colonoscopy Finding | Recommended Interval for Surveillance Colonoscopy | Strength of Recommendation | Quality of Evidence |
|---|---|---|---|
| Normal * | 10 years # | Strong | High |
| 1 to 2 tubular adenomas < 10 mm | 7 to 10 years ### | Strong | Moderate |
| 3 to 4 tubular adenomas < 10 mm | 3 to 5 years | Weak | Very Low |
| 5 to 10 tubular adenomas < 10 mm | 3 years | Strong | Moderate |
| Adenoma ≥10mm | 3 years | Strong | High |
| Adenoma with tubulovillous or villous histology | 3 years ** | Strong | Moderate |
| Adenoma with high grade dysplasia | 3 years ** | Strong | Moderate |
| >10 adenomas on single exam *** | 1 year | Weak | Very low |
| Piecemeal resection of adenoma ≥20 mm | 6 months | Strong | Moderate ## |
* All recommendations assume exam complete to cecum with bowel preparation adequate to detect lesions >5mm in size; recommendations do not apply to individuals with a hereditary CRC syndrome, personal history of inflammatory bowel disease, personal history of hereditary cancer syndrome, serrated polyposis, malignant polyp, personal history of CRC, or family history of CRC, and must be judiciously applied to such individuals, favoring the shortest indicated interval based on either history or polyp findings
** Assumes high confidence of complete resection # Follow up may be with colonoscopy or other screening modality for average risk individuals*** patients with >10 adenomas or lifetime >10 cumulative adenomas may need to be considered for genetic testing based on absolute/cumulative adenoma number, patient age, and other factors such as family history of CRC (see text)
## See USMSTF recommendations for endoscopic removal of colorectal lesions 67### Patients with recommendations issued prior to 2020 for shorter than 7 to 10 year follow up after diagnosis of 1 to 2 tubular adenomas may follow original recommendations. If feasible, physicians may re-evaluate patients previously recommended an interval shorter than 10 years and reasonably choose to provide an updated recommendation for 7 to 10 year follow up, taking into account factors such as quality of baseline exam, prior polyp history, and patient preferences.
USMSTF recommendations for post colonoscopy follow up in average risk adults with serrated polyps
| Baseline Colonoscopy Finding | Recommended Interval for Surveillance Colonoscopy | Strength of Recommendation | Quality of Evidence |
|---|---|---|---|
| ≤ 20 Hyperplastic polyps in rectum or sigmoid colon < 10mm | 10 years ### | Strong | Moderate |
| ≤ 20 Hyperplastic polyps proximal to sigmoid colon < 10mm | 10 years | Weak | Very low |
| 1 to 2 SSPs < 10 mm | 5 to 10 years | Weak | Very low |
| 3 to 4 SSPs < 10 mm | 3 to 5 years | Weak | Very low |
| 5 to 10 SSPs < 10 mm | 3 years | Weak | Very low |
| SSP ≥ 10 mm | 3 years | Weak | Very low |
| SSP with dysplasia | 3 years | Weak | Very low |
| Hyperplastic polyp ≥ 10 mm | 3 to 5 years # | Weak | Very low |
| Traditional serrated adenoma | 3 years | Weak | Very low |
| Piecemeal resection of SSP ≥20 mm | 6 months | Strong | Moderate ## |
* All recommendations assume exam complete to cecum with bowel preparation adequate to detect lesions >5mm in size; recommendations do not apply to individuals with a hereditary CRC syndrome, personal history of inflammatory bowel disease, personal history of hereditary cancer syndrome, serrated polyposis, or malignant polyp, personal history of CRC, or family history of CRC, and must be judiciously applied to individuals with a personal or family history of CRC, favoring the shortest indicated interval based on either history or polyp findings
** Assumes high confidence of complete resection# A 3 year follow up interval is favored if concern about consistency in distinction between SSP and HP locally, bowel preparation, or complete excision, whereas a 5 year interval is favored if low concerns for consistency in distinction between SSP and HP locally, adequate bowel preparation, and confident complete excision.
## See USMSTF recommendations for endoscopic removal of colorectal lesions 67 ### Follow up may be with colonoscopy or other screening modality for average risk individualsSSP; sessile serrated polyp
New observational and modeling studies of colonoscopy confirm and strengthen the evidence base to support the conclusion that individuals with normal colonoscopy are at lower than average risk for CRC, as mentioned previously 8–11 . Based on this reduced risk, we recommend CRC screening in average risk individuals be repeated 10 years after a normal exam complete to the cecum with bowel preparation adequate to detect polyps > 5 mm in size. Future studies may clarify whether lengthening the interval beyond 10 years may be possible. A 10 year follow up after normal colonoscopy is recommended regardless of indication for the colonoscopy, except for individuals at increased risk for CRC such as those with history of a hereditary CRC syndrome, personal history of inflammatory bowel disease, personal history of hereditary cancer syndrome, Serrated Polyposis Syndrome, malignant polyp, personal history of CRC, or family history of CRC.
Metachronous advanced neoplasia risk among individuals with no adenoma, 1 to 2 < 10mm adenomas, and high risk adenomas at baseline.
The Task Force recognizes that many patients with 1 to 2 non-advanced adenomas smaller than 10 mm will have had a prior documented recommendation for a 5-year exam or other interval shorter than 7 to 10 years, consistent with 2012 recommendations. Patients with recommendations prior to this publication for shorter than 7 to 10 year follow up after diagnosis of 1 to 2 tubular adenomas < 10 years can reasonably follow original recommendations. Based on the new evidence presented and our current recommendation for 7 to 10-year follow-up, if feasible, we suggest that physicians may re-evaluate patients previously recommended an interval shorter than 7 to 10 years and reasonably choose to provide an updated recommendation for follow up between 7 and 10 years after the prior exam which diagnosed 1 to 2 adenomas
Since the 2012 recommendations, additional studies have confirmed and extended the evidence supporting identification of 1 or more adenomas ≥10 mm size as a high risk feature 26–28, 31, 32 . A study of 2,990 patients from the Netherlands diagnosed with adenoma 1988–2002 and followed through 2008 found size ≥10 mm was independently associated with 1.7-fold increased risk for metachronous advanced neoplasia (OR 1.7; 95% CI: 1.2–2.3) 31 . A cohort study of 3,300 patients diagnosed with adenomas at a large integrated US healthcare system that found size ≥10 mm was independently associated with 3.6-fold increased risk for advanced adenoma (OR 3.6, 95%CI: 2.8–4.5), and 5.2-fold increased risk for CRC on follow-up (OR 5.2, 95% CI: 1.8–15.1) 27 . An Australian cohort study of 5141 patients found having advanced neoplasia (defined by villous histology, size >9 mm, serrated histology, high grade dysplasia, or >2 adenomas) was associated with increased risk for advanced neoplasia on follow-up, but risk associated with size >9mm, villous histology, or high grade dysplasia alone was not specifically examined. An additional limitation of this study is that half of enrolled patients had a family history of CRC 28 . As previously mentioned, a US cohort study found individuals with advanced adenoma had increased risk for incident and fatal CRC compared to those with no adenoma, and the cumulative rate of advanced adenoma removal at up to 9 years follow-up was 13.0% 18 . Though the study did not specifically report outcomes for individuals with adenoma ≥ 10 mm or larger, adenoma with high grade dysplasia or villous histology, the majority of individuals followed in the advanced adenoma group met the increased size criteria. As such, this study also supports closer follow-up for individuals with adenoma ≥ 10 mm. Data from the Nurses Health Study/Health Professional Follow up study found a 3-fold increased risk for incident CRC among individuals with adenoma ≥ 10 mm or larger, further supporting this as an increased risk group (HR 3.40; 95% CI: 1.86–6.24) 19 . The Task Force acknowledges the importance of accurate polyp size estimation for this recommendation and recommends photo-documentation verifying polyp size ≥ 10mm relative to an open forceps or open snare of known size.
Studies published since the 2012 recommendations continue to support villous histology as a potential risk factor for advanced neoplasia on follow-up. These studies include the aforementioned large cohort studies from a large US healthcare system and the Netherlands 19, 27, 28, 31 .
Multiple recent cohort studies have confirmed and extended evidence to support high-grade dysplasia as a risk factor for metachronous advanced neoplasia 27, 28, 35 and CRC 19, 27 . However, the Netherlands cohort of 2,990 patients did not find baseline high-grade dysplasia to be an independent predictor of risk 31 . Studying high-grade dysplasia as a risk factor is a major challenge since this finding is rare at baseline, perhaps accounting for some of the variability in risk observed across studies. The 3-year recommendation assumes that there was complete resection of neoplasia, including HGD at the baseline exam.
Since 2012, we found a single cohort study of 214 Korean patients with > 10 adenomas in which risk for metachronous advanced adenoma was evaluated. On median 4.3 years follow-up, 26.6% had metachronous advanced adenoma 36 . Patients with more than 10 adenomas may be at increased risk for having a hereditary polyposis syndrome such as Familial Adenomatous Polyposis or MYH-associated polyposis 37 , and multiple groups have recommended patients with >10 cumulative lifetime adenomas be considered for genetic testing 38, 39 . Decision to perform genetic testing may be based on absolute or cumulative adenoma number, patient age, as well as other factors such as family history of CRC and/or personal history of features associated with polyposis such as desmoid tumor, hepatoblastoma, cribriform morular variant of papillary thyroid cancer, or multifocal/bilateral congenital hypertrophy of the retinal pigment epithelium (CHRPE) 39 .
The largest study to date has been a cohort study of 5,433 individuals with baseline colonoscopy and at least 1 surveillance colonoscopy ≥1 year after initial exam. Baseline categories included presence of normal colonoscopy, low risk adenoma, high risk adenoma, and/or SSP (defined by histologic SSP or TSA) 40 . Primary outcomes assessed on follow-up included risk for metachronous conventional high risk adenoma as well as large serrated polyp (HP, SSP, or TSA) ≥10 mm. Findings are summarized in Table 6 . Rate of high risk adenoma among patients with SSP but no synchronous high risk adenoma was just 2.9%, much lower than the observed rate for individuals with isolated high risk adenoma at baseline of 18.2%. Rate of high risk adenoma was markedly higher in patients with both SSP and high risk adenoma at baseline, estimated at 46.4%. Rate of serrated polyp ≥10 mm (HP, SSP, or TSA) on follow-up was substantially higher among patients with isolated SSP versus high risk adenoma at baseline (9.6% vs. 1.0%). Among patients with low risk adenoma plus SSP at baseline, the rate of metachronous high risk adenoma was 18.4% (9/49), and metachronous SSP ≥10mm was 8.2% (4/49; unpublished data obtained through personal communication 3/14/18 with Anderson JC, Butterly LF, Robinson CM). These findings suggest that patients with isolated SSP have low rates of metachronous conventional high risk adenoma unless they have synchronous conventional adenomas at baseline. However, patients with SSP at baseline appear to be at increased risk for metachronous large serrated polyps ≥10 mm (HP, SSP, or TSA), irrespective of whether concurrent conventional adenomas are present. While this is the largest study to date of metachronous findings among patients with and without SSPs, a limitation is that the risk estimates remain imprecise owing to relatively small number of patients with SSP at baseline available for evaluation in the various risk strata. In contrast to the aforementioned even smaller studies, however, it is interesting to note that patients with isolated SSP of any size as well as HPs ≥ 10mm were not found to have increased risk for conventional high risk adenoma on follow-up.
Risk for high risk adenoma and large SP stratified by baseline colonoscopy findings in the New Hampshire Colonoscopy Registry *
| Baseline Finding | Surveillance Colonoscopy Finding | |
|---|---|---|
| HRA, % (n) | SP ≥ 10 mm, % (n) | |
| No adenoma | 4.8 (116/2396) | 0.7 (18/2396) |
| LRA ** | 9.7 (96/991) | 0.5 (5/991) |
| HRA | 18.2 (11/603) | 1.0 (6/603) |
| LRA + SSP ** | 18.4 (9/49) | 8.2 (4/49) |
| HRA + SSP | 46.4 (13/28) | 3.6 (1/28) |
| SSA/P | 2.9 (3/104) | 9.6 (10/104) |
| SP ≥10 mm | 3.1 (2/65) | 12.3 (8/65) |
** previously unpublished data provided through personal communication 3/14/18 from Anderson JC, Butterly LF, Robinson CM, with permission.
** Included TSA in SSA/P group; LRA=1 to 2 adenomas <10 mm in size; SP = hyperplastic, SP, or TSA; HRA= advanced neoplasia or >2 adenomasWe found little new evidence to guide the follow-up recommendation for patients with TSA. A cross sectional study compared risk for advanced neoplasia and/or ≥ 3 adenomas at surveillance colonoscopy for patients with prior isolated TSA (n=186) versus a group of age/sex matched patients with prior conventional adenoma (n=372). Proportion with metachronous high risk adenoma was higher in the TSA versus conventional adenoma group (47.3 vs. 32.0%), and associated with higher risk on adjusted analyses (OR high risk adenoma = 2.37; 95% CI: 1.55–3.63), 52 supporting our recommendation for repeat colonoscopy in 3 years after TSA diagnosis.
We were unable to identify published papers that specifically examined risk for metachronous neoplasia in patients with 3 to 10 SSPs, or any combination of 3 to 10 SSPs and conventional adenomas. The previously mentioned unpublished data on 49 patients with a combination of LRA and SSP at baseline with unknown total number suggests increased risk for metachronous advanced neoplasia and for large SSP. In absence of additional data, we have chosen to recommend 3 to 5 year repeat colonoscopy for individuals with 3 to 4 SSPs
For patients with HP >=10mm, repeat colonoscopy in 3 to 5 years. A 3 year follow-up interval is favored if concern about pathologist consistency in distinguishing SSPs from HPs, quality of bowel preparation, or complete polyp excision, whereas a 5 year interval is favored if low concerns for consistency in distinguishing between SSP and HP by the pathologist, adequate bowel preparation, and confident complete polyp excision. (Weak recommendations, very low quality of evidence).
We found little new evidence to guide management of patients with SSP ≥ 10mm or HP ≥10mm. In the previously cited New Hampshire Colonoscopy registry study, among 65 patients with large SP (HP, SSP, or TSA), 3.1% had high risk adenoma on follow-up, compared to 4.8% among 2,396 patients with no adenoma at index colonoscopy 40 . However, having any serrated polyp ≥ 10mm in size was associated with increased risk for large serrated polyp (≥10mm SSP, TSA, or HP), ranging from an absolute risk of 12.3% (8/65) for no concurrent conventional adenoma to 11.2 % (2/18) for concurrent high risk adenoma, compared to an absolute risk of 0.7% (18/2,396) for those without adenoma or any serrated polyp. Thus, based on this new evidence, the implications for having a large serrated polyp on risk for subsequent conventional high risk adenoma are uncertain. However, having a large serrated polyp at baseline does appear to be associated with risk for subsequent large serrated polyps. Additionally, as brought up previously, the Nurses Heath Study/Health Professionals follow up study reported that the risk for incident CRC among 566 individuals with large serrated polyps (≥10mm SSP, TSA, or HP) was increased 3-fold relative to individuals with no polyps. Cumulative CRC incidence at 10 years follow up was 0.4% among patients with no polyps, and 1.1% among subjects with large serrated polyps at baseline 19 . A challenge in interpreting available literature is a lack of data separating outcomes for those with >10 mm SSP, TSA, and HP. Because of variation in consistent distinction by pathologists between SSPs and HPs in usual care 41–46 , a conservative approach might be to assume all HPs >10mm are SSPs. However, this may subject some patients (especially if consultant pathology expertise in distinguishing SSPs from HPs is high) to over diagnosis and more aggressive surveillance than necessary if rates of advanced neoplasia or large serrated polyp on follow-up among individuals with large SSPs versus large HPs differ. An added problem in making recommendations for large serrated polyps is the potential challenge of resection of SSPs ≥ 10mm. For example, Pohl et al. reported 47% of SSPs 10 to 20 mm had evidence of incomplete resection 51 . Given uncertainties regarding implications of having serrated polyp ≥10 mm and whether outcomes differ for those with SSP versus HP ≥10mm, as well as observed variation in ability of pathologists to distinguish SSPs from HPs, and the known challenge of resection of ≥10mm SSPs, the Task Force recommends 3 year follow-up for individuals with SSP ≥ 10mm in size, and 3 to 5 year follow-up for individuals with HP≥10mm. For HP≥10mm, a 3 year follow-up interval is favored if concern about consistency in distinction by the consult pathologist between SSP and HP, adequacy of bowel preparation, or complete excision, whereas a 5 year interval is favored if there are limited concerns about consult pathologist ability to distinguish SSP from HP, adequacy of bowel preparation, or complete polyp excision. The Task Force acknowledges the importance of accurate polyp size estimation for this recommendation and recommends photo documentation verifying polyp size relative to an open forceps or open snare of known size.
No new evidence regarding outcomes of surveillance in individuals with isolated SSP containing dysplasia was identified. SSP with dysplasia is rare; in one series of 179,111 patients with polyps submitted for histologic exam, of 2,139 SSPs identified, 302 contained low or high grade dysplasia 53 . Dysplastic SSPs have more features consistent with CRC than SSPs without dysplasia. In absence of additional data on whether metachronous neoplasia risk differs for individuals with SSP and dysplasia compared to SSP without dysplasia, the Task Force recommends repeat colonoscopy in 3 years after SSP with dysplasia diagnosis, as long as a high confidence complete resection of the lesion was performed.
Recommendations for 2nd surveillance stratified by adenoma findings at baseline and 1 st surveillance
| Baseline Finding | Recommended interval for 1st surveillance | Finding at 1st surveillance | Recommended interval for next surveillance |
|---|---|---|---|
| 1 to 2 tubular adenomas < 10 mm | 7 to 10 years | Normal colonoscopy | 10 years |
| 1 to 2 tubular adenomas < 10 mm | 7 to 10 years | ||
| 3 to 4 tubular adenomas < 10 mm | 3 to 5 years | ||
| Adenoma ≥10 mm in size; or Adenoma with tubulovillous/villous histology; or Adenoma with high grade dysplasia; or 5 to 10 adenomas < 10 mm | 3 years | ||
| 3 to 4 tubular adenomas < 10 mm | 3 to 5 years | Normal colonoscopy | 10 years |
| 1 to 2 tubular adenomas < 10 mm | 7 to 10 years | ||
| 3 to 4 tubular adenomas < 10 mm | 3 to 5 years | ||
| Adenoma ≥10 mm in size; or Adenoma with tubulovillous/villous histology; or Adenoma with high grade dysplasia; or 5 to 10 adenomas < 10 mm | 3 years | ||
| Adenoma ≥10 mm in size; or Adenoma with tubulovillous/villous histology; or Adenoma with high grade dysplasia; or 5 to10 adenomas < 10 mm | 3 years | Normal colonoscopy | 5 years |
| 1 to 2 tubular adenomas < 10 mm | 5 years | ||
| 3 to 4 tubular adenomas < 10 mm | 3 to 5 years | ||
| Adenoma ≥10 mm in size; or Adenoma with tubulovillous/villous histology; or Adenoma with high grade dysplasia; or 5 to 10 adenomas < 10 mm | 3 years |
Normal colonoscopy is defined by colonoscopy where no adenoma, SSP, or CRC is found
We identified several studies on serial surveillance published since 2012 31, 54–58 . Findings from the largest of these studies 31, 54, 55 , as well as those considered as part of the 2012 recommendations are summarized in Table 8 . Across all studies, individuals with low risk adenoma at baseline, and no adenoma at 1 st surveillance had low rates of high risk adenoma on follow-up, ranging from 1 to 6.6%. Similarly, across all but one of the studies reviewed, individuals with high risk adenoma at both baseline and subsequent surveillance exam have >18% rate of metachronous high risk adenoma on follow-up, supporting our recommendation for follow-up colonoscopy in 3 years. However, the outcomes at 2 nd surveillance for other clinical scenarios of baseline and 1 st surveillance findings are more variable across studies. Our recommendations for 2 nd surveillance colonoscopy based on findings at baseline and 1 st surveillance are summarized in Table 7 . More evidence is needed to clarify the best intervals for surveillance in patients who have had baseline and repeat colonoscopy, particularly for those with low risk adenoma at baseline and follow-up. Also, new evidence is required to guide serial surveillance of individuals with SSPs and large HPs.
Risk for neoplasia at 2 nd surveillance stratified by findings at baseline and 1 st surveillance
| Baseline Finding | 1 st Surveillance Finding | High Risk Adenoma at 2 nd Surveillance, % | |||||
|---|---|---|---|---|---|---|---|
| Morelli M 2013 (n=965) 38 | Park HW 2015 (n=2087) 39 | van Heijningen EB 2013 (n=1482) * 22 | Pinsky P 2009 (n=1032) * 43 | Laiyemo A 2009 (n=1297) 44 | Robertson D 2009 (n=564) 45 | ||
| Low risk adenoma (LRA) | No adenoma | 6.6 | 6.0 | 1.0 | 3.9 | 2.8 | 4.9 |
| LRA | 13.8 | 10.6 | 1.0 | 5.7 | 4.7 | 9.5 | |
| HRA | 18.0 | 16.4 | 0.0 | 15.6 | 6.9 | 20.0 | |
| High risk adenoma (HRA) | No adenoma | 9.6 | 6.7 | 4.0 | 5.9 | 4.8 | 12.3 |
| LRA | 14.0 | 24.3 | 3.0 | 6.7 | 8.9 | 13.6 | |
| HRA | 22.0 | 38.2 | 4.0 | 19.3 | 30.6 | 18.2 | |
LRA=1–2 non-advanced adenomas; HRA=advanced adenoma or ≥ 3 adenomas
* risk and outcome characterized based on non-advanced and advanced adenomaMultiple models have been developed to stratify the risk of metachronous neoplasia and guide surveillance 28, 31, 57, 59–63 . Results are promising, but incremental value over current risk stratification recommendations informed by number, size and histology of polyps is unclear. For example, a comprehensive model including polyp size, villous histology, proximal location, and number of adenomas had a superior C statistic compared with the 2012 Task Force guidelines, but the magnitude of improvement was small (0.71 for the model vs. 0.66 for 2012 guidelines) 31 . An important limitation of current published work is that many of these studies have not included a test and independent validation set, raising concerns about generalizability 28, 31, 59, 60 . Additionally, the range of variables utilized as part of models varies considerably. Notably, models reviewed here suggest the best predictors of future risk for advanced neoplasia remain baseline colonoscopy polyp findings.
Piecemeal polyp resection contributes to risk for metachronous neoplasia. A meta-analysis by Belderbos et al. of 33 studies found risk for recurrent neoplasia was 20% for piecemeal versus just 3% for en bloc resection utilizing endoscopic mucosal resection (EMR) technique 65 . In the subgroup with EMR of polyps 10 to 20 mm in size, piecemeal resection was associated with an 18% risk for recurrent neoplasia, similar to the 19% rate observed for polyps 20 to 30 mm and > 30mm in size. Pohl et al studied rate of incomplete resection using biopsy immediately after assumed complete resection of 5 to 20 mm polyps, including patients with and without EMR 51 . Incomplete resection was more common with piecemeal (20%) versus en bloc resection (8.4%), but piecemeal resection was not an independent predictor of incomplete resection after adjusting for size and histology. For polyps >=20mm, additional papers since the Belderbos et al. meta-analysis have reported high risk for recurrent neoplasia associated with piecemeal vs en block resection 66, 67 . These findings suggest that colonoscopist must put an emphasis on complete polyp excision at baseline, and, particularly for polyps ≥20mm in size consider strategies for verifying complete excision. The evidence base to support management of patients with polyps ≥20 mm in size resected piecemeal has been reviewed in detail in the recent Task Force recommendations on endoscopic removal of colorectal lesions 68 . Based on the evidence reviewed, the Task Force recommended patients with polyps ≥ 20 mm resected piecemeal have first surveillance colonoscopy at approximately 6 months, second surveillance 1 year from first surveillance, and third surveillance 3 years from the 2 nd surveillance.
Since the 2012 recommendations, a number of studies have reported on risk factors for metachronous neoplasia. Smoking may be associated with risk for recurrent conventional adenoma as well as serrated polyps 69, 70 . Environmental factors such as rural versus urban residence may contribute to risk for cancer after advanced adenoma removal 71 . Metabolic syndrome 70, 72, 73 (as well as components of this diagnosis such as increased waist to hip ratio, increased hip circumference) and obesity 72–74 have been shown by a number of studies to be associated with increased risk for recurrent neoplasia. Race does not appear to modify risk for recurrent adenoma and metachronous advanced neoplasia. A retrospective cohort study of 246 Whites and 203 Blacks who had an adenoma at baseline and at least one surveillance colonoscopy found similar rates of recurrent adenoma and advanced neoplasia. 75 A cohort study of participants in the Polyp Prevention Trial compared risk for metachronous adenoma and advanced neoplasia among 1668 Whites and 153 Blacks with adenoma at baseline, all of whom received surveillance colonoscopy, found no difference in rate of metachronous adenoma or advanced neoplasia 76 . Thus, while there is evidence that Blacks have a higher age-adjusted incidence and mortality from CRC, and develop CRC at a younger age than other racial and ethnic groups in the US, once screened, there is no robust evidence that being Black modifies the risk for recurrent adenoma or advanced neoplasia. Having a flat adenoma may increase risk for recurrent neoplasia, but more data are needed to support differential management 77 . Diet might modify risk, but new evidence to support its impact is limited. One study found no clear association between fruit and vegetable intake and risk for adenoma recurrence 78 , and another pooled study of 1727 participants from 2 randomized trials did not identify a relationship between pro inflammatory diet and risk for adenoma, advanced adenoma, or 3 or more adenomas on follow-up colonoscopy after initial polypectomy 79 . Lifestyle factors such as increased sedentary behavior may increase risk for adenoma recurrence 80 , but it is unclear whether specifically modifying behavior will reduce risk.
Since 2012, several studies have been published on chemopreventive strategies for reducing risk for recurrent neoplasia. A large, well done RCT found that supplementation with calcium or vitamin D (alone or in combination) was not associated with reduced risk for recurrent neoplasia 81 , and a small study that included intervention with calcitriol, aspirin, and calcium also found no benefit on risk for recurrent neoplasia 82 . A prospective cohort study reported that dietary supplement use was not associated with reduced risk of metachronous neoplasia 83 . An observational study demonstrated that exposure to metformin was associated with reduced risk for finding adenoma at surveillance colonoscopy among diabetics 84 , and a pilot RCT of non-diabetics found that low dose metformin was associated with reduced risk for recurrent adenoma at one year 85 , suggesting metformin may be a promising chemopreventive agent warranting further study.
Newly published work has confirmed that aspirin and exposure to non-steroidal anti-inflammatory medications may reduce risk for adenoma recurrence, but optimal dose, mechanism of action, and characteristics of patients most likely to benefit has not been well established 86, 87 . While there is insufficient evidence to support routine recommendation of aspirin for cancer and adenoma prevention in patients with baseline adenoma, the overall impact of aspirin on cardiovascular disease (CVD) and CRC risk reduction might support recommending aspirin for some patients. Specifically, it should be noted that, for patients age 50–59 years who have ≥10% risk for cardiovascular disease CVD and life expectancy of ≥10 years, without increased risk for bleeding, the US Preventive Services Task Force has recommended use of aspirin 81 mg per day for primary prevention of both CVD and CRC (Grade B recommendation), and has recommended that aspirin could also be considered for 60 to 69 year old based on shared decision making taking into account potential harms and benefits (Grade C recommendation) 88 . Thus, for patients who inquire about strategies to reduce future CRC risk after polypectomy, an opportunity exists to recommend estimation of cardiovascular risk and to consider aspirin for both CVD and CRC risk reduction if these criteria are met.
In summary, there is little evidence that lifestyle factors such as diet, smoking, obesity, sedentary behavior increase the risk of metachronous neoplasia, or that modification of these behaviors, reduces the risk. Likewise, there is little new evidence that chemoprevention impacts the risk of metachronous advanced neoplasia in patients with adenoma. At this time there is insufficient evidence to recommend modification of surveillance intervals based on these factors. More work needs to be done to identify risk factors and chemopreventive strategies that can reduce risk for metachronous neoplasia and possibly allow for less frequent surveillance colonoscopy.
Given that risk for metachronous advanced neoplasia has been accepted thus far as a surrogate for risk for incident CRC, and the plethora of studies that have examined risk for metachronous advanced neoplasia among individuals with baseline polyps, the Task Force has provided updated recommendations for surveillance based the relationship of baseline findings to risk for metachronous advanced neoplasia. Key updates since the 2012 USMSTF recommendations are summarized in Table 9 . Recommendations for patients with advanced adenoma, including those with adenoma ≥10 mm, or containing high-grade dysplasia and/or villous features are unchanged, with evidence to support close surveillance in 3 years strengthened. One year, rather than a more general recommendation for less than 3 year follow-up colonoscopy for individuals with more than 10 adenomas at a single examination has been recommended to simplify follow-up, though the evidence base to support this strategy has not been markedly strengthened. Emerging evidence suggests that individuals with 3 to 4 adenomas < 10 mm are at low risk for metachronous neoplasia, supporting our recommendation for 3 to 5 interval rather than a strict 3 year follow-up colonoscopy for this group of patients. Another significant change from prior guidance is our recommendation for surveillance colonoscopy in 7 to 10, rather than 5 to 10 years for patients with 1 to 2 adenomas < 10mm, based on the growing body of evidence to support low risk for metachronous advanced neoplasia. In this population, the risk for metachronous advanced neoplasia is similar to that for individuals with no adenoma (Figure 2 ). Importantly, the observed risk for fatal CRC among individuals with 1 to 10 adenomas
Key updates since 2012 recommendations provided in the 2019 USMSTF recommendations for follow up after colonoscopy and polypectomy
7 to 10 year rather than 5 to 10 year follow up is recommended after removal of 1 to 2 tubular adenomas < 10 mm in size (Table 4 )
More detailed recommendations for follow up after removal of serrated polyps have been provided ( Table 5 )
Importance of high quality baseline examination has been emphasized 1 year rather than < 3 year follow up is recommended after removal of >10 adenomasOption to recommend 3 to 5 year instead of 3 year follow after removal of 3 to 4 adenomas < 10 mm in size
Our review highlights several opportunities for research to clarify risk stratification and management of patients post polypectomy. In order to optimize risk reduction strategies, the mechanisms driving metachronous advanced neoplasia after baseline polypectomy, and their relative frequency need to be better understood through studies that include large numbers of patients with interval cancers and/or advanced neoplasia after baseline polypectomy. Mechanisms may include new/incident growth, incomplete baseline resection, and missed neoplasia; each of these potential causes may require different interventions for improvement 89 . For example, if most interval cancers after polypectomy are attributable to missed neoplasia 90, 91 , redoubled focus on quality of baseline exam may be indicated. Indeed, quality factors, such as incomplete exam and poor bowel preparation, have been associated with risk for cancer after polypectomy 20, 31, 32 . Further, it is plausible that the adenoma detection rate (ADR) of a colonoscopist, which has been tied closely with risk for interval cancer after normal screening colonoscopy 92, 93 , might have a similar correlation with risk for interval cancer after polypectomy. If incomplete resection is the major cause of metachronous neoplasia after polypectomy 64 , focus on implementing strategies that improve polypectomy technique may be indicated. If the main driver is incident neoplasia, then strategies that optimize risk stratification and timing of colonoscopy (early for high risk, and deferred for low risk) might be most impactful. Interestingly, one study has found that the attributable fraction of risk for CRC after baseline polypectomy is highest for incomplete polyp removal and not having “on time” follow up colonoscopy, underscoring the importance of complete removal and appropriate follow-up intervals 32 . More work is needed to identify the key drivers of metachronous advanced neoplasia, particularly CRC. Application of precision medicine, such as offering chemoprevention to individuals with genotypes associated with response to therapy, may improve effectiveness of chemoprevention, but requires further study 94 . Biomarkers of adenoma recurrence also merit study. 95–99 Widespread promotion of colonoscopist ADR as a quality metric is likely to increase the frequency of diagnosing patients with multiple small adenomas. Since finding multiple small adenomas may be a marker of careful colonoscopy, patients with multiple (e.g. 1 to 4) small adenomas may be subject to a so-called “adenoma detector paradox,” in which they are currently recommended short interval (e.g. 3-year) colonoscopy despite potentially having very low risk for incident CRC secondary to having a very high quality exam. Though we have recommended 3 to 5 year follow-up for individuals with 3 to 4 small adenomas based on emerging evidence, understanding implications of having multiple small adenomas should be a key focus of future research. We found little data to guide management of individuals with isolated HPs
Beyond risk stratification, more fundamental research on the potential benefits of surveillance is needed. In particular, better evidence is needed to support whether exposure to surveillance colonoscopy, compared to no surveillance, reduces CRC incidence or mortality. Such evidence is needed given the increasing proportion of patients who are having adenomas detected as part of increased participation in CRC screening.
Several areas not covered by our current recommendations also warrant investigation. We do not provide recommendations for management of young patients (under age 50 years) with incidentally detected adenoma, though evidence to guide management is emerging 100, 101 . At the other end of the age spectrum, more research is needed to determine whether the potential cancer prevention and early detection benefits of surveillance outweigh immediate procedure related risks for individuals older than age 75, or with multiple comorbidities. Cost-effectiveness of surveillance, as well as alternative strategies for surveillance (such as FIT or multi-target FIT-DNA) requires further study. Indeed, one modeling study has suggested that surveillance FIT (rather than colonoscopy) might be effective post-polypectomy 102 .
As a result of our review, we have several suggestions for best practices to improve the quality and comparability of future research on post-polypectomy surveillance. Studies vary in the definition of “high risk adenoma”. Ideally, when considering both predictors and outcomes, we suggest as a best practice reporting presence of individual findings (e.g. villous adenoma, SSP, HP≥10mm) in addition to several potentially clinical relevant summary categories, including: 1) advanced neoplasia; 2) advanced adenoma; and 3) large serrated polyp (HP or SSP≥10mm). Because our understanding of the risks and outcomes among patients with SSPs is still limited, we suggest it is particularly important to separate SSPs from aggregate predictor or outcome categories such as advanced neoplasia. Further, we suggest specifically reporting SSP, HPs, and TSAs separately as predictors and outcomes, and clearly defining any aggregate categories (such as serrated polyps ≥10 mm) precisely. Providing histology-specific data will allow for greater comparability across studies, and better assessment of whether outcomes differ by serrated polyp histology. For example, histology specific outcome data could help elucidate whether individuals with HP ≥10 mm have similar outcomes as patients with SSP ≥ 10 mm. More studies are needed which include patients that are racially and ethnically diverse. Most surveillance studies provide limited data on the quality of baseline colonoscopy, which could help in interpreting results. Additionally, we recommend that both relative and absolute risks for outcomes, such as metachronous advanced neoplasia, be provided in surveillance studies. Absolute risks are key to providing perspective to patients and physicians on the true risk associated with a given polyp finding scenario. Studies examining the potential benefit of exposure to surveillance versus no surveillance should seek to avoid several potential sources of bias. First, risk for cancer associated with adenoma is often compared to the general population, not to people who had normal colonoscopy. Comparing cancer risk among individuals with adenoma removal to a general population without ascertaining for presence of CRC or adenoma may bias towards underestimating risk reduction that can be gained by removing adenomas 8, 16, 17, 103 . Second, risk for cancer associated with surveillance is often compared to general population, not to people who had polypectomy but no surveillance; this may bias towards an overestimation of the benefit of surveillance 8, 15, 16 . Also, risk for cancer associated with surveillance often excludes cancers diagnosed within 1 year, which may bias towards overestimating benefit of surveillance, since in usual practice surveillance time frames are assigned based on initial results, not initial results plus clinical course within a year 16, 104 . Finally, some studies may compare outcomes among patients who did not receive surveillance to those who survived cancer free and received surveillance 20 . This is analogous to a per-protocol analysis of a randomized trial, may overestimate the benefit of surveillance, and may be considered a form of immortal time bias. Additionally, we note that very few randomized trials of surveillance strategies have been done. In the United States, the National Polyp Study is the only RCT of surveillance colonoscopy. This study was conducted in the 1980’s before availability of modern technology (e.g. high definition colonoscopies) and widespread awareness of importance of quality on outcomes, employed a highly aggressive baseline polyp clearing strategy, and compared a very short 1 versus 3 year follow-up interval among patients with baseline adenoma 105 . The European Polyp Surveillance Trial (EPoS) trial, which includes arms randomized to different surveillance intervals based on specific baseline polyp finding strata, is well underway, and will likely offer new insights to guide polyp surveillance 106 . Lack of randomized trials in the area of surveillance is quite remarkable given the frequency of surveillance colonoscopy in usual practice, and in context of the many trials that are available on CRC screening.
Several limitations may be considered in interpreting and applying our recommendations to practice. Our recommendations for surveillance intervals depend on the performance of a high quality exam (as evidenced by exam complete to the cecum with adequate bowel preparation and complete polyp resection) by a high-quality colonoscopist (based on adequate adenoma detection rate). This requires that colonoscopists continuously strive to improve quality, but also use caution in applying surveillance recommendations when concerns about quality exist. We focused on updating our recommendations based on a literature review of papers published since the prior recommendations were issued in 2012, and did not perform pooled or meta-analyses. A more comprehensive literature review of all papers published relevant to surveillance over a longer time period, as well as meta-analyses were beyond scope of this work. In many cases, our recommendations are based on very low or low quality evidence. Even where evidence was judged to be of moderate or high quality, few studies were randomized trials. Thus, future research has a high likelihood of producing evidence that may change recommendations, particularly those based on lower quality evidence. We recognize the challenge of applying new recommendations, such as a 7 to 10, rather than a 5 to 10 year follow up recommendation for patients with 1 to 2 adenomas < 10 mm, in practice. Patients, primary care physicians, and colonoscopists may have concerns about lengthening a previously recommended interval, and will need to engage in shared decision making regarding whether to lengthen the follow up interval based upon the guidance here or utilize the recommendation made at the time of the prior colonoscopy.
CRC incidence and mortality are decreasing secondary to improvements in risk factor exposures, screening, treatment, and perhaps exposure to surveillance among patients with polyps 107 . Given that some patients with polyps appear to have persistently increased risk for CRC, and many have increased risk for advanced neoplasia on follow-up, surveillance colonoscopy to attempt to reduce CRC risk is clinically rational and recommended. Evidence to support best practices for surveillance colonoscopy has strengthened, and has helped to support close follow-up for some groups, as well as less intense follow-up for others. More work is needed to fully understand which patient populations are most likely to benefit from surveillance, and the ideal surveillance interventions to apply for optimizing CRC prevention and early detection.
We thank Karen Heskett, medical librarian, UC San Diego, for her assistance with conduct of our literature review.
