Although NETs of the colon and rectum are relatively rare compared to colorectal adenocarcinomas, a growing number of colorectal NETs are diagnosed at an early stage thanks in part to greater investigation with colonoscopy. These tumors tend to progress indolently and have a satisfying prognosis[25]. The median survival duration in patients with localized NETs of the colon and rectum is 261 and 290 months, respectively[3]. Nevertheless, colorectal NETs, especially in T1N0M0 lesions, remains an elusive disease with unclear and inconsistent guidelines for selecting the optimal treatment. Currently, there are two main modalities including endoscopic and surgical procedures used in the treatment of T1N0M0 colorectal NETs, and ET provides a minimally invasive method for the removal of these lesions. Compared with surgery, ET has a stronger correlation with reduced mortality[26]. Moreover, data from previous studies have shown that ET offered a high complete resection rate (80.6% − 96.7%), a limited adverse event rate, and a low recurrence rate for colorectal NETs smaller than 20 mm in size, which further evidence that ET is a safe and effective approach for resecting T1N0M0 colorectal NETs with malignant potential[10, 27].
In the present study, we compared long-term (5-year and 10-year) outcomes of ET and surgery and identified independent prognostic factors of survival. According to our findings, there were significant differences in clinical characteristics between ET and surgery patients. ET was used more commonly in younger and married patients. Besides, the patients treated endoscopically were more likely to demonstrate smaller tumors, rectal lesions, no submucosal involvement, and well-differentiated histology. As in previous research, the prognosis was linked to age[28], marital status[29], tumor size[2], tumor site[2], and differentiation grade[1]. Since these covariates might lead to bias that disturbs the comparison of treatment strategies, PSM was employed in this study. OS and CSS were comparable at both 5 and 10 years between the 2 therapies. Similar results were noted in subgroup analyses of patients with different tumor sizes. In patients with colonic disease, the OS rate was higher in the ET group at both 5 and 10 years of follow-up in comparison to the surgery group, whereas patients undergoing ET did not differ from those receiving surgical resection regarding the 5-year and 10-year CSS. Limiting to patients with rectal lesions, there were no differences between the two groups in the 5-year and 10-year OS and CSS. Besides, we did not observe a significant difference in OS and CSS between treatment groups in the Cox proportional hazards regression models. Similar results were seen when modeling was based on different tumor size or site.
Thus far, there remains limited data comparing survival outcomes of ET and surgery for T1N0M0 colorectal NETs. Previously, one study using the SEER database[21] compared the long-term survival of 618 patients with T1N0M0 rectal carcinoid tumors diagnosed between 1998 through 2012 and discovered comparable CSS between local excision and radical surgery. However, this study did not include colonic disease. Accordingly, the same outcome that ET was related to equivalent OS and CSS was acquired in our large population-based study with a PSM cohort, and also found that therapeutic modality was not an important prognostic indicator by multivariate analysis. We also found that increasing age, higher tumor grade, and tumor size between 10–20 mm were predictive of poorer OS and CSS. Yet, gender, as well as race and marital status, was independently associated with OS.
The status of regional lymph nodes is an important factor in the choice of endoscopic or surgical treatment. Studies have shown that colorectal NETs smaller than 10 mm had a low prevalence of metastasis (colon: 4%; rectum: <3%), which demonstrated that ET was sufficient for these small tumors[20, 21]. This is in accordance with our result that ET and surgery were associated with similar survival in patients with tumors less than 10 mm. However, lymph node metastasis occurs more frequently in colorectal NETs varying from 10 mm to 20mm[30]. Additionally, Konishi et al.[17] reported that if patients with colorectal NETs had lymph node or distant metastases, their survival rate was similar to those with colorectal adenocarcinomas. Tumors of intermediate size (10–20 mm) should be treated as adenocarcinomas by aggressive surgical resection with regional lymphadenopathy[21, 31]. On the contrary, some research suggested that endoscopic resection was reasonable to remove these larger lesions as they were considered indolent in nature[10, 32, 33], which is consistent with our finding that survival rates following ET and surgery were equivalent in patients with tumors of 10–20 mm in size. Although our data favor safe ET for T1N0M0 colorectal NETs regardless of tumor size, endoscopic ultrasonography (EUS) should be conducted to evaluate the tumor diameter, the depth of intestinal wall invasion and local lymph node status prior to ET.
The current NCCN guidelines indicate that T1N0M0 rectal disease can be safely removed through endoscopy[15], which is supported by our study that no significant difference was observed in survival between rectal tumor patients who underwent ET and those who received surgery. Of note, patients with colonic NETs who treated surgically did not confer extra survival benefits. Our finding appears to be inconsistent with the recommendations of management guidelines for colonic NETs[15]. For example, a partial colectomy with regional lymphadenectomy is usually advised for localized colonic NETs on the basis of the NCCN guidelines. However, all of the patients included in this study were at stage T1N0M0, the earliest stage of the disease with the lowest degree of malignancy. What’s more, in a report by Landry et al.[8, 9], the survival rate was similar between early NETs of the colon and rectum, further revealing that the malignant degree of early tumors in both sites has little difference. Although it has not become a formal proposal in western countries, T1N0M0 colonic NETs can be managed with ET according to the consensus of Chinese experts[19]. Thus, we have reason to believe that ET is appropriate for T1N0M0 colonic NETs, similar to the treatment of T1N0M0 rectal NETs.
Tumor size is not only a major predictor of regional spread but also a predictive factor of survival[2]. In line with the previous reports[1, 20], our multivariate analysis disclosed that increasing tumor size was an independent predictor connected with the incremental risk of overall and cancer-specific mortality. Although it is generally acknowledged that the depth of wall penetration is associated with prognosis[30], no correlation between them was found in our study, which may be the result of unknown infiltration extent in 338 (30.0%) cases. Thus, further studies that evaluate the prognostic value of invasive depth are warranted.
There are several limitations to our study that deserve discussion. On account of its retrospective nature, there is affected by selection bias and confounding factors. Data on patient comorbidities is not available in the SEER database. In general, patients with increased comorbidities tend to undergo ET as a result of its less invasiveness. It is worth noting that although higher comorbidity can potentially lead to a bias against ET, there was no difference in CSS between these 2 treatments. Therefore, the absence of comorbidity information should not influence our overall results. Second, the SEER database does not provide data regarding several important prognostic factors (lymphovascular invasion and resection margin), treatment-related complications, means of endoscopic resection. Randomized, controlled trials can adequately address this limitation. Moreover, the use of targeted therapy is not captured by this database. However, patients with T1N0M0 colorectal NETs scarcely receive this treatment, which most likely does not impact our results. In addition, the SEER database records only original therapy and does not document patients in whom ET failed. Lastly, there was also no information on disease recurrence. Despite these issues, the SEER database is a large-scale database with detailed and accurate data recording. We made an attempt to reduce potential bias by performing a PSM and adjusting the HR for the influence of ET on survival.