Surveillance after Endoscopic Resection for Colorectal Tumors: A Comprehensive Review

The National Polyp Study (NPS) estimated that screening colonoscopy followed by polypectomy and subsequent surveillance can reduce the incidence of colorectal cancer (CRC) by 76–90% [1]. In addition, results from a large retrospective cohort study in the UK showed that surveillance endoscopy is more effective, especially in high-risk polyp (proximal polyps, high grade or large adenomas) and baseline suboptimal colonoscopy (incomplete or poor bowel preparation) [2]. The goal of surveillance after the endoscopic resection of colorectal tumors is to reduce CRC incidence and mortality. Considering the effective use of the limited endoscopic capacity and the cost of surveillance, it is desirable to develop a surveillance program that is as minimal as possible. In Europe and the USA, after the results of the NPS were made available, guidelines were developed that stratified risk by initial endoscopy and proposed surveillance programs for each risk group, which have been revised several times since then [3, 4].

In Japan, an effective surveillance program similar to the European and US guidelines did not exist because total excision of adenomatous polyps (clean colon) was not commonly performed in daily practice, and follow-up after polypectomy was usually performed once a year until about 10 years ago [5]. As a result, it was assumed that the valuable endoscopic capacity was occupied by surveillance after polypectomy. Increasing endoscopic screening of new patients is important to reduce incidence and mortality of CRC. If we consider the introduction of total colonoscopy (TCS) as an initial screening test in the future, it will be necessary to increase the number of colonoscopic examinations in new patients. Therefore, there is an urgent need to develop a surveillance program based on risk stratification after colorectal polypectomies in Japan. The “colonoscopy screening and surveillance guidelines” (Fig. 1) by the Japan Gastroenterological Endoscopy Society (JGES) were developed with the basic principle of “aiming for zero CRC deaths during surveillance, bowel preservation, and emphasis on patient quality of life” as the guideline principles [6].

The NPS was launched in 1980, and the results were published in 1993 [7]. Eligibility criteria included patients who had at least one adenoma, and all polyps were completely resected. Of the 2,623 eligible patients, 1,418 were allocated to 2 groups of 699 patients for two examinations at 1 year and 3 years and 799 patients for one examination at 3 years. The results showed that the detection rate of tumors with advanced pathological features (tumor diameter >10 mm, high-grade dysplasia [HGD], invasive cancer) was the same in both groups {3.3%, relative risk (95% confidence interval [CI]) of 1.0 [0.5–2.2]}. Moreover, a 76–90% reduction in the cumulative incidence of CRC could be expected after colorectal polypectomy, based on comparisons with previous cohorts of patients who did not have colorectal polyps removed and a cross-sectional study in a cancer registry. The long-term follow-up results of the NPS, with a median follow-up of 15.8 years, showed that colorectal adenoma resection and surveillance were effective in reducing CRC mortality by 53% [8].

In the UK, a randomized controlled trial (RCT) was initiated in 1984 to evaluate a surveillance program after colorectal polypectomy, and the results were published in 2001 [9]. A total of 776 patients were divided into two groups based on the nature of the polyps and family history: 307 in the high-risk group and 469 in the low-risk group. The high-risk group was randomized to (1) flexible sigmoidoscopy (FS) annually, (2) FS every 2 years, and (3) TCS every 2 years, while the low-risk group was randomized to (4) FS every 2 years, (5) FS every 5 years, and (6) TCS every 5 years. The high-risk group had a significantly higher adenoma detection rate than the low-risk group (relative risk rate, 1.82); however, there was no significant difference in the adenoma detection rates between the randomized groups. In addition, both groups had a lower incidence of adenomas than that predicted by the historical database. Based on these results, endoscopic surveillance was recommended for the high-risk group, and endoscopy (FS or TCS) every 5 years was recommended considering cost and compliance.

The Japan Polyp Study (JPS) consists of 11 centers in Japan and began an RCT in 2003. Patients aged 40–69 years who had at least one adenoma underwent two rounds of TCS at 1-year intervals before allocation, and all colorectal neoplasms detected were endoscopically removed. Of the 2166 eligible patients, 1,087 were allocated to the two-test group (one and 3 years) and 1,079 to the one-test group (3 years). The primary endpoint was the proportion of advanced neoplasia (AN) detected, and the design aimed to demonstrate the non-inferiority of the one-test group to the two-test group. There was no difference in the AN between the two groups, indicating non-inferiority (1.7% vs. 2.1%, p = 0.001). Regarding ANs discovered after allocation, the laterally spreading tumor (LST) non-granular type of the right colon was more common [10]. After a median follow-up of 6.1 years, the results showed an 86% reduction in CRC incidence compared with the Osaka Cancer Registry [11].

Based on the NPS results, 3 years was considered the standard interval after colorectal polypectomy [7]. Subsequently, several studies stratifying risk by number of adenomas, size, pathology, and other factors were reported [12‒14]. It has been reported that the examination interval could be extended beyond 3 years in low-risk adenoma groups other than three adenomas or more, excluding AN [13, 15]. In addition, a multicenter prospective study of 3,121 screening colonoscopies in the USA reported that the relative risk (95% CI) of AN occurring 5 years after the initial examination was 1.92 (0.83–4.42) for two or fewer adenomas <10 mm compared with 1.0 for the no adenoma group. The number of adenomas significantly affected the risk of AN occurrence in the group with three or more adenomas (5.01 [2.1–11.96]) [14]. The results of a prospective 5-year post-screening colonoscopy surveillance of 2,452 patients in Korea showed that the hazard ratio of cumulative AN occurrence in the low-risk adenoma group with two or fewer adenomas <10 mm was 1.14, which was similar to that in the group with no adenomas [12].

The results of a long-term cohort of post-sigmoidoscopy surveillance-naive patients in the UK showed that the standardized incidence ratios (95% CI) for rectal and colon cancer incidence were 0.4 (0.0–1.0) and 0.5 (0.1–1.3), for the low-risk group with only left-sided non-advanced adenomas (AAs), respectively, which were lower than those in the general population without surveillance [16]. Results from a large Norwegian cohort of post-colorectal polypectomy patients showed that the low-risk (1, non-AN) group had a standardized mortality rate (95% CI) of 0.75 (0.63–0.88) for CRC, which was lower than the general population [17]. Although the definition of low risk varies among reports, the results of studies using CRC incidence and mortality as endpoints suggest that surveillance may not be necessary in the low-risk adenoma group. Moreover, an analysis of cost-effectiveness of post-polypectomy surveillance based on UK large-scale database revealed that incremental costs per CRC prevented and quality-adjusted life-year gained with surveillance were higher for the low-risk groups [18].

Currently, US guidelines (United States Multi-Society Task Force [USMSTF] published in 2020) recommend TCS surveillance at 7–10 years for patients with 1–2 tubular adenomas <10 mm (Table 1) [4]. European guidelines (European Society of Gastrointestinal Endoscopy [ESGE], 2020) recommend a return to a screening program or a repeat TCS after 10 years (Table 1) [3].

Some studies in Japanese patients have reported that a certain percentage of ANs also occur during surveillance for adenomas <10 mm, suggesting a higher risk of CRC incidence compared to the group without adenomas [19, 20]. On the other hand, a German population-based case-control study (3,148 vs. 3,274 patients) showed that the odds ratio (95% CI) of CRC was 2.96 (1.7–5.16) for surveillance after 6 years and 3.73 (2.11–6.6) for uncompleted resection of all colorectal polyps. Both have been reported to be significant risk factors for post-colonoscopy CRC [21]. In a separate analysis, the low-risk group was 0.2 (0.1–0.2) for TCS surveillance of less than 3 years, 0.4 (0.2–0.6) for 3–5 years, and 0.8 (0.4–1.5) for 6–10 years, with the effect of the low-risk group in reducing CRC incidence disappearing after 6 years [22]. In addition to the above findings, it is assumed that non-polypoid colorectal neoplasms are easily missed and develop into invasive cancer within a short period after their occurrence; therefore, an interval of 5 years or longer is not acceptable in Japan. The Japanese guideline recommends the use of colonoscopy three to 5 years after the first colonoscopy for patients with adenomas (2 or less, non-AA) that have been resected (Table 1) [6]. However, compared to US or European guidelines, shorter surveillance intervals may result in excessive cost increases from the perspective of reducing CRC deaths, and the development of guidelines from the perspective of cost-effectiveness is an important issue for the future.

In a prospective observational study examining the risk of AN 5 years after baseline TCS, the relative risk (95% CI) of adenoma (≥3, other than AA) compared with no adenoma was 5.01 (2.01–11.96), and in another study the hazard ratio (95% CI) with adenoma (1–2 adenomas) was 3.06 (1.51–6.57); the risk is considered equivalent to having AN in the baseline [14]. However, US guidelines recommend TCS surveillance at three to 5 years for 3–4 non-AAs as part of conventional low-risk adenomas. In contrast, TCS surveillance at 3 years is recommended for more than 5–10 non-AA (Table 1) [4]. ESGE guidelines recommend that adenomas (3–4 non-AA) return to the screening program or TCS after 10 years, similar to those with adenomas (1–2 non-AA). In contrast, TCS surveillance at 3 years is recommended for more than five adenomas (non-AA) (Table 1) [3].

The JGES guidelines recommend TCS surveillance at 3 years for adenomas (3–9 non-AA) (Table 1) [6]. This is because these guidelines were developed with the goal of bowel preservation, which is stricter than the goal of CRC mortality control. In the JPS, several cases of non-polypoid colorectal neoplasia were found at the 3-year interval [10]. It is difficult to extend the interval beyond 3 years. The USMSTF guidelines recommend TCS surveillance after 1 year for 11 or more adenomas (Table 1) [4], ESGE guidelines recommend genetic consultation after 10 or more adenomas (Table 1) [3], and the JGES guidelines recommend TCS surveillance after 1 year for 10 or more adenomas (Table 1) [6].

The USMSTF and ESGE guidelines recommend TCS surveillance for 3 years (Table 1) [3, 4]. The JGES guidelines recommend surveillance after 3 years in principle but recommend surveillance after 1 year in cases of adenomas > 20 mm, intramucosal carcinomas (equivalent to HGD in the USA and Europe), and T1 carcinomas (Table 1) [6]. A recent study based on the Polish national screening program reported that the risk of CRC incidence and mortality after endoscopic resection of HGD and tumors > 20 mm were more than twice that of other AN [23]. In addition, a long-term cohort study (CREATE-J) of colorectal ESD cases performed in Japan reported that 15 (1.0%) cases of metachronous colorectal invasive cancer occurred during long-term follow-ups (median 46.0 months) in 1,437 patients (approximately 50% with pTis or pT1 cancer) measuring 20 mm or more (pTis or pT1 cancer). The median time to detection of the metachronous carcinomas was 26.8 months. Most cases were detected during the second surveillance after ESD, and 13 of the 15 cases were treated surgically [24]. Therefore, when metachronous CRC is detected, although infrequent after 3 years, it is likely that the disease is too advanced for endoscopic resection. Therefore, a surveillance interval of 1 year is appropriate after the initial endoscopic resection of early-stage CRC.

The risk of local recurrence increases when endoscopic resection of early-stage CRC results in piecemeal resection or positive horizontal margins, regardless of the endoscopic resection method. In particular, local recurrence after piecemeal resection is 9.1–27.5% likely to occur and often occurs within 2 years [25‒29]. A multicenter RCT conducted in Japan and Taiwan to evaluate the optimal surveillance period after piecemeal resection concluded that 6 months was optimal because both local recurrent lesions detected by surveillance at three and 6 months were endoscopically salvageable and the recurrence rates at 24 months were similar [30]. The ESGE and Japanese guidelines for the treatment of CRC recommend that the next endoscopic follow-up after piecemeal resection is performed at approximately 6 months [31]. Unplanned piecemeal EMR, which does not take into account where the lesion is invading, risks dividing the invasive area and may underestimate the depth of invasion. Moreover, there is a possibility that factors that should be subjected to additional surgical resection may be overlooked. In fact, although not unplanned, there have been reports of cases where large LSTs diagnosed as intramucosal lesions were treated with piecemeal EMR, and although the pathological diagnosis was intramucosal carcinomas and negative for vascular invasion, local invasive carcinomas recurrence occurred during follow-up [32]. Even when piecemeal resection is anticipated, the accuracy of pathological diagnosis may be improved and local recurrence due to incomplete resection may be reduced by resecting without fragmenting the deepest part of the tumor and by ensuring that the non-neoplastic mucosa at the margins is included in the resection. Furthermore, there is a certain percentage of cases where it is difficult to diagnose whether or not the cancer is invasive before treatment, so in cases where there is a high possibility that the lesion is invasive cancer, such as LST non-granular type or lesions with depression, it is also considered appropriate to perform ESD with the aim of performing en bloc resection. In order to do this, it is important to train endoscopists who are able to perform endoscopic submucosal dissection.

The USMSTF, ESGE, and JGES surveillance guidelines share the recommendation for TCS surveillance at 3 years for sessile serrated lesions with dysplasia and traditional serrated adenoma [3, 4, 6]. For serrated lesions, evidence regarding the long-term effects of surveillance is lacking, which remains an issue for future research. Furthermore, sessile serrated polyposis is considered a high risk factor for CRC, and closer surveillance is recommended; however, there is no robust evidence but only small-size, retrospective reports [33, 34].

Recently, guidelines of post-polypectomy surveillance were based on evidences of primary screening colonoscopy. It is not yet known whether the surveillance intervals for fecal immunological test (FIT)-positive colonoscopy and primary screening colonoscopy can be the same. Large-scale cohort study based on population-based FIT screening program in Taiwan revealed that higher primary FIT concentration was associated with increased risk of incident CRC and subsequent FIT should be scheduled after negative colonoscopy to reduce the risk of incident CRC [35]. In FIT-based screening, the efficient combination of colonoscopy and FIT in surveillance programs is thought to be an issue for the future.

It is noteworthy that the ESGE guidelines include a statement on cessation of surveillance, recommending “cessation of surveillance at age 80 years or if the expected life expectancy is short due to comorbidities.” In Japan, where the life expectancy is one of the longest in the world, surveillance endoscopy continued even in the 80s. However, specifying the criteria for cessation in the guidelines may avoid exposing the elderly to unnecessary risks in clinical practice.

Data on multiple surveillance colonoscopies are very limited. For example, if the second surveillance endoscopy shows low-risk adenoma, the question is whether the third surveillance interval should be the same for a case with low-risk adenoma in the first case and a case with HGD or super-high-risk adenoma (>20 mm) in the first case.

In addition, we believe that the next revision of the JGES guidelines will provide guidelines for the second and subsequent rounds of surveillance, which will greatly increase opportunities for its use in clinical practice. The current guidelines are based on colonoscopy findings, but it is necessary to consider whether patient background factors such as a family history of CRC and metabolic syndrome [36], which are known to increase CRC risk, should be added to the factors determining surveillance intervals using big data.

Recently, technology of artificial intelligence was introduced to endoscopic practice. So, in the near future, the integration of artificial intelligence in post-polypectomy surveillance promises to enhance detection, improve risk stratification, standardize practices, and optimize surveillance intervals, ultimately leading to better patient outcomes and more efficient use of healthcare resources.

Long-awaited surveillance guidelines based on risk stratification were developed [6]. Guidelines are meaningful only when they are effectively utilized in clinical practice. They must also be revised based on new evidence. It is hoped that new knowledge will be accumulated, especially in Japan, on topics that are currently lacking.

Professional English editing was performed by Editage.

The authors have no conflicts of interest to declare.

There is no funding about this manuscript.

K.H.: conceptualization and writing – original draft. T.M., Y.S., T.F., and Y.S.: writing – review and editing.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.