MATERIALS AND METHODS
Study design and patients
This retrospective study was conducted from January 2005 to June 2021 at the Samsung Medical Center, a tertiary institution in Seoul, South Korea. We reviewed the electronic medical records to obtain the variables for each patient. We enrolled 1096 patients with rectal NETs who underwent endoscopic or surgical treatment between 2005 and 2021 (Figure 1). After excluding patients with a pathological size > 1.5 cm of NET, missing data, or a history of rectal NET or multiple NETs (defined as synchronous or multifocal lesions identified at diagnosis), 1056 patients were included in the study. A total of 932 patients underwent endoscopic resection, and 124 patients underwent surgical resection. In the endoscopic resection group, there were 621 patients with NETs having the pathologic size 0 cm ≤ size < 0.5 cm, 284 patients with 0.5 cm ≤ size < 1 cm, and 27 patients with 1 cm ≤ size < 1.5 cm. In this study, propensity score matching was conducted at 1:10, with 27 patients in 1 cm ≤ size < 1.5 cm group, and 225 patients in the 0 cm ≤ size < 1 cm group.
Figure 1 Flowchart of the patients included in the study.
NET: Neuroendocrine tumor.
Clinicopathological factors including age, sex, date of cancer diagnosis, date of surgical or endoscopic resection, tumor location, pathological data of the specimen [longest size, depth, World Health Organization (WHO) grade, resection margin (RM), mitosis count, lymphovascular invasion (LVI), and perineural invasion], death, latest outpatient clinic date, and computed tomography (CT) findings after resection were assessed. The variables were based on the pathology of surgical resection in the primary surgery, post-endoscopic, and surgery combination groups, and the pathology of endoscopic resection in the post-endoscopic specimens. Information, including surgery date, endoscopic resection date, recurrence date, recurrence type, death date, survival status, and cause of death, was obtained from hospital records and test results. The follow-up duration was defined as the duration between the dates of the first and last outpatient visits. These data were extracted from the clinical data warehouse Darwin-C of Samsung Medical Center. The study protocol was reviewed and approved by the Institutional Review Board of Samsung Medical Center (Approval No. SMC 2024-10-073-001). The study was conducted in accordance with the principles of the Declaration of Helsinki. Information was used only if the patient consented to the electronic medical record access. The board waived the requirement for informed consent because all data were analyzed anonymously.
Diagnostic workup and procedure
All patients underwent an endoscopic evaluation, including colonoscopy, before the intervention. Additionally, endoscopic ultrasound was performed in patients if the tumor was quite large endoscopically or if it was not proven in the biopsy to assess the depth of the tumor before resection. The CT scan was performed to identify distant organ or lymph node metastases. Removal methods for rectal NETs are determined by the tumor size, depth, and patient factors. Endoscopic excision was performed for NETs < 10 mm without muscularis propria involvement. With NETs ranging from 10 mm to 20 mm, endoscopic resection is preferred when the tumor is considered localized and the patient is reluctant to undergo surgery. The endoscopic procedure was performed by a skilled endoscopist with extensive experience in colonoscopy. Before excision, a mixture of saline, indigo carmine, and diluted epinephrine (1:10000) was injected to elevate the tumor from the rectal muscle layer. The endoscopist selected the most suitable excision technique based on variations in standard EMR, such as EMR with a precut, EMR with a ligation, or ESD. This strategy aimed to provide a tailored, minimally invasive treatment for rectal NETs, considering the tumor characteristics and patient preferences. Transanal endoscopic microsurgery (TEM) transanal rectal resection, and lower anterior resection were performed. The RM was classified as follows: Negative, defined as a margin that clearly contained no tumor cells; positive, defined as a margin with definite tumor cell involvement; and indeterminate, defined as a margin that could not be assessed because of fragmented specimens or electrocautery artifacts. The invasion depths were classified as the muscularis mucosa, submucosa, and propria muscularis.
Follow-up protocol and outcomes
Upon initial diagnosis, a CT scan was performed to assess lymph node and distant metastases. Following endoscopic resection, patients underwent follow-up examinations, including abdominopelvic and chest CT, along with sigmoidoscopy or colonoscopy. These were initially performed at 6-12 months intervals and then annually according to the investigators’ protocol. In cases where pathological examination revealed positive RMs, salvage treatments such as additional endoscopic resection or surgery were recommended and performed for some patients. Patients who declined salvage treatment underwent an additional sigmoidoscopic examination one-month post-initial endoscopic resection. If no recurrence or residual disease was detected during the follow-up endoscopy, subsequent endoscopy and CT were scheduled six months later. The term synchronous rectal NETs denote rectal NETs identified in a single patient within six months. Metachronous tumors were defined as NETs diagnosed at different sites at least six months after the initial resection.
Statistical analysis
Baseline characteristics were analyzed using descriptive and frequency analyses. Categorical variables were analyzed using the χ2 test or Fisher’s exact test. Continuous variables were compared using the Kruskal-Wallis rank-sum test or Wilcoxon signed-rank test. Statistical significance was set at P < 0.05. All statistical analyses were performed using R version 4.4.0. A 1:10 propensity score matching was performed using the nearest neighbor method without replacement, with a caliper width of 0.2 times the standard deviation of the logit-transformed propensity score. Kaplan-Meier curves were used to compare the recurrence curves between the two size groups.
DISCUSSION
There has been controversy over the treatment method for 1 cm ≤ size < 1.5 cm rectal NETs. To the best of our knowledge, this is the first study to compare the long-term clinical outcomes of endoscopic resection of rectal NETs measuring 1, < 1.5, 0, and < 1 cm using a matched study. A propensity score analysis was used to generate precisely matched patient. No statistically significant intergroup differences were observed in recurrence-free survival. Increasing size, the presence of LVI, WHO grade, depth, and Ki-67 expression are known risk factors for lymph node metastasis in NET[11-14]. Conventionally, LVI is thought to increase the risk of micrometastasis in locoregional cancers[15]. Moreover, because rectal NETs are highly vascular, LVI is a potential high-risk factor for metastasis[16]. There was no patient with LVI among patients with size 1 cm ≤ size < 1.5 cm. In our data, there was no recurrence in the 1 cm ≤ size < 1.5 cm group after propensity score matching. Considering the results, it may be worthwhile to try endoscopic resection in the 1 cm ≤ size < 1.5 cm group of rectal NET.
In this study, positive RMs after endoscopic resection was observed in 2.4% of cases. Among them, there were only two positive RM patients in the 1 cm ≤ size < 1.5 cm group. The likelihood of a positive RM was also not significantly different between 0 cm (size < 1 cm) and 1 cm (size < 1.5 cm). These patients did not undergo salvage treatments. Patients with positive RM of 1 cm ≤ size < 1.5 cm were also followed up at 5.3 years and 4.9 years, respectively, with no recurrence. In patients with 1 cm or less, one patient with positive RM was treated with EMR with a precut, one with cold polypectomy with a snare, and the other two patients with no further management. All patients with 1 cm or less with positive RM also had no recurrence. Patients with positive RMs on actual histology can be divided into two groups: Those with truly partially resected and those with endoscopically complete resection but positive RMs with cautery artifacts[17-19]. Therefore, if endoscopic RM is clear, physicians should consider careful and regular colonoscopy surveillance without additional intervention. However, if the submucosal layer is not fully resected or if deep RM is positive, physicians may consider additional EMR. Our study showed no local recurrence in patients with positive endoscopic RMs or residual lesions on endoscopy. The low rate of positive RM in the 1 cm ≤ size < 1.5 cm group and the lack of recurrence in this group suggest that endoscopic resection may be considered as the primary treatment option.
Three patients experienced recurrence in the entire group before propensity score matching, and all three patients had a median follow-up of 4.5 months for metachronous recurrence. All the patients with metachronous recurrence underwent endoscopic resection. After propensity score matching, only one patient experienced recurrence. This patient experienced metachronous recurrence that occurred one year after ESD. The patient had no recurrence even though five years had passed since the second treatment. Chung et al[20] metachronous rectal NETs were diagnosed in 2.7% (n = 9) of the 333 patients, and metachronous lesions were associated with the number of synchronous lesions at initial diagnosis. In our study, the overall rate of metachronous rectal NETs was 0.28% (n = 3) in 1057 patients, which may be due to the exclusion of patients with pre-existing synchronous lesions from the analysis. Previous studies have reported that synchronous adenomas may contribute to metachronous neoplasia[21,22]. Therefore, the surrounding rectal mucosa must be carefully examined in patients who have undergone endoscopic resection of rectal NETs. The National Comprehensive Cancer Network guidelines for rectal NETs do not recommend routine follow-up for small rectal NETs measuring less than 10 mm. Similarly, the European Neuroendocrine Tumor Society guidelines do not recommend regular follow-up for completely treated tumors. However, the current study showed different results, which questions whether there is a need for a follow-up. At the study institution, patients with rectal NETs routinely undergo annual follow-ups with either sigmoidoscopy or colonoscopy, along with abdominal CT imaging. Although the current data are insufficient to establish formal guidelines, continued surveillance and future data collection will hopefully contribute to the development of evidence-based surveillance protocols. We plan to report outcomes once we accumulate more patients and longer follow-up durations, which we believe will provide valuable insights into long-term prognosis and guide future surveillance strategies.
Surgical resection (e.g., transanal excision, TEM, or low anterior resection) provides the advantage of complete local excision with regional lymph node dissection, which may be appropriate for cases with high-risk features, such as LVI, poor differentiation, or deep submucosal invasion. In our study, patients who underwent surgery did not experience any recurrence during the follow-up period. However, surgical resection is associated with higher morbidity, longer hospitalization, and a higher procedural burden, which may not be justified in cases with low metastatic risk. In a large French study, the overall in-hospital mortality rate of colorectal resections, including rectal surgeries, was 3.4%, with a morbidity rate of 35%[23]. Although the proportion of complications in rectal surgery may be lower than that in general colorectal resection, wound complications, clinical anastomotic dehiscence, postoperative hemorrhage, and urinary tract complications can occur. For endoscopic resection of rectal NETs, complication rates are generally low, with bleeding occurring in 0%-8.8% of cases, and perforation is rare across multiple studies[24,25]. Although these rates can vary based on factors such as tumor size, stage, grade, and specific surgical approach, endoscopic resection is associated with lower complication rates than more extensive surgical procedures. In our study, there were no life-threatening complications such as death, need for intensive care, excessive bleeding, or urgent surgery due to perforation in the endoscopic resection group. In the endoscopic group, the procedure was often outpatient-based, and the procedure time was significantly shorter than that in the surgical group, making endoscopic procedures more cost-efficient and time-efficient. From a cost-effectiveness perspective, endoscopic resection is typically performed in an outpatient setting, involves minimal anesthesia, and allows faster return to daily activities. These factors translate into lower direct medical costs (procedural fees and hospital stay) and indirect costs (loss of productivity) compared to surgery. Although endoscopic resection may require close surveillance and occasional retreatment in cases with positive margins, it remains a more cost-efficient option for carefully selected patients without high-risk features. Treatment decisions should be based on a comprehensive evaluation of tumor size, histologic features, patient comorbidities, and patient preferences. As the prognosis in our study shows that endoscopic resection of 1 cm ≤ size < 1.5 cm is favorable, patients would likely prefer to undergo endoscopic treatment rather than surgery.
This study has some limitations. First, being conducted at a single center may limit the generalizability of the findings to all rectal NET patient populations. Our institution is one of the largest referral centers for rectal NETs in Korea, receiving patients from a wide range of geographic regions across the country. Due to Korea’s well-developed national transportation infrastructure and centralized healthcare system, our patient population is not limited to a specific region but represents a diverse and heterogeneous group. Moreover, clinical practices and management protocols for rectal NETs at our institution closely follow national guidelines, which are consistently implemented throughout Korea. Demographic characteristics, including age, sex, and the proportion of patients undergoing ESD or EMR, were comparable to those reported in previous single-center and multicenter studies conducted in Korea. Although outcomes at high-volume institutions may be somewhat more favorable than those at smaller centers, this is likely attributable to the greater procedural experience of endoscopists at larger hospitals. Therefore, while this is a single-center study, we consider our findings to be broadly representative of the overall Korean rectal NET patient population. Nonetheless, we acknowledge that further validation through multicenter studies would enhance the generalizability of our results. Second, the sample size of the 1-1.5 cm group was small, which may not have provided sufficient statistical power to detect small but clinically meaningful differences. NETs have an incidence of approximately 0.86 per 100000 individuals, with a higher prevalence among Black males and Asian populations[26,27]. Rectal and gastrointestinal NETs account for 18% of total NETs and 27% of gastrointestinal NETs[27]. Given the notable prevalence of rectal NETs, this study had the largest sample size among single study groups. Even in previous studies and those cited by the European Neuroendocrine Tumor Society 2023 guidelines, the prevalence of rectal NETs measuring 1-1.9 cm remains low[27,28]. In most studies, this size category accounts for approximately 8% of all rectal NET cases[29]. Naturally, the number of tumors measuring 1-1.5 cm was even smaller. Therefore, although the number of 1-1.5 cm NETs in our study may be limited, this reflects the inherently low frequency of such tumors in clinical practice. It would be appropriate to mention this context to clarify that the small sample size was not a result of inadequate recruitment but rather the rarity of this tumor subset. Recent multicenter studies have reported similarly low frequencies of intermediate-sized tumors[27]. Owing to the rarity of rectal NETs, this study is considered as a preliminary investigation that may serve as a valuable foundation for future large-scale prospective research. Third, there were missing or incomplete data in pathological reports. The WHO grading system for NETs was first established in 2010 and subsequently revised in 2017 using more detailed criteria. Since our study enrolled patients beginning in 2005, some pathological data were inevitably missing or inconsistently reported owing to the absence of standardized grading criteria during earlier years. Because this was a retrospective study conducted over a median period of 54 months, variability in the completeness and details of pathological reporting was inevitable. This exclusion may have introduced a selection bias because the excluded patients could have differed systematically from those with complete data. To maintain analytical rigor, we established predefined criteria requiring the presence of essential pathological data for inclusion in the final analysis. Patients with missing information for any of the key prognostic indicators were excluded from outcome-related analyses. This exclusion is explicitly stated in the methods section and allowed us to ensure that comparisons between the groups were based on consistent and reliable pathological profiles. In addition, endoscopically resected specimens were fixed in 10% formalin and cut into 2-3 mm thick sections, whereas surgically resected specimens were cut into 5-6 mm thick sections. Therefore, surgically resected specimens may undergo different pathological evaluations than endoscopically resected specimens. Fourth, the retrospective nature of the study introduced a potential selection bias and limited the ability to control for confounding factors that could influence the outcomes. Fifth, although the current study demonstrated no recurrence in the 1-1.5 cm group during a median follow-up period of 54 months, the risk of late recurrence, although low, cannot be completely excluded. Considering the slow-growing nature of NETs, long-term follow-up is essential to evaluate the oncological safety of endoscopic resection, especially for intermediate-sized rectal NETs. Continued surveillance and long-term observations are warranted to better evaluate the long-term prognosis of patients with intermediate-sized rectal NETs. Sixth, although propensity score matching is a robust method for minimizing confounding factors in observational studies, it has inherent limitations. Specifically, propensity score matching only accounts for the measured confounders included in the matching algorithm. In the study, we matched for age, sex, and endoscopic resection method; however, unmeasured or unavailable variables, such as tumor histological features, operator experience, and patient comorbidities, may still influence treatment outcomes and introduce residual bias. The study used multivariate regression analysis to reduce the confounding factors. However, given the small sample size of the 1-1.5 cm group and the absence of recurrence events in this subgroup, it was not feasible to adjust for a larger number of covariates or perform multivariable analyses. It is also noted that the 1-1.5 cm group had a relatively small sample size owing to the inherent rarity of intermediate-sized tumors. Therefore, a 1:10 matching ratio was employed to maximize statistical power while retaining all eligible cases in this group. Even after applying 1:10 propensity score matching, the covariate balance remained acceptable, with absolute standardized mean differences of less than 0.2 for all variables. This indicated that the groups were well-matched and comparable after adjustment. Future studies should employ complementary statistical techniques to strengthen causal inferences.
Our findings suggest that endoscopic resection for 1-1.5 cm grade 1 rectal NETs results in comparable recurrence-free survival to < 1 cm tumors, with no observed recurrence or LVI in the matched study. Given the low morbidity, favorable oncological outcomes, and minimally invasive nature of endoscopic treatment, endoscopic resection should be considered a viable treatment option for selected patients with 1 cm to 1.5 cm rectal NETs. However, there are limitations, such as the relatively small sample size and short median follow-up period of approximately 54 months, which may still be insufficient to detect late recurrences, particularly given the slow-growing nature of rectal NETs. Future prospective multicenter studies are warranted to confirm these findings and refine treatment guidelines.
