In the present study, ESD achieved en bloc resection and histological complete resection rates of 94.3% and 89.4% for patients with advanced colorectal neoplasia, and the rate of major complications was only 2.3%. Hybrid ESD was an independent factor of piecemeal resection. Tumor location in the colon was associated with increased risk of ESD-related complications.
ESD offered a high en bloc and complete resection rate for early-stage CRC and HGD in the colorectum, even for lesions that were larger than 4.0 cm in diameter or located deeper in the submucosal layer [8, 9]. Additionally, the complete resection rate of ESD was relatively higher than that of endoscopic mucosal resection (EMR) . On the other hand, EMR is unlikely to offer a high complete resection rate for large colorectal epithelial lesions or for post-EMR tumor recurrences. A piecemeal EMR resection results in increased post-EMR recurrences and uncertainty in histological assessment of the complete resection . However, ESD of colorectal tumors is technically difficult. Therefore, hybrid ESD (a combined EMR and ESD technique) has been proposed to facilitate the ESD procedure and could be an option in certain situations [21, 22]. The multivariate analysis conducted in the present study indicated that the piecemeal resection rate was independently higher for tumors treated with hybrid ESD than for tumors treated with standard ESD. The risk of recurrence after piecemeal removal is relatively high, suggesting that standard ESD is preferred for patients with potential malignance .
Previous work indicated that ESD was the first-line treatment for tumors larger than 2.0 cm . In the present study, ESD was indicated for lesions requiring endoscopic en bloc resection for which it was difficult to use the snare technique. ESD was the preferred treatment for advanced colorectal neoplasia and tumors with higher malignant potential, such as non-granular LST (LST-NG) , irrespective of tumor size. Notably, tumor location in the colon contributed to piecemeal resection in the present study. Hayashi et al.  indicated that poor endoscopic operability was an independent predictor of histological incomplete resection and perforation. Hori et al.  studied predictive factors for technical difficulty in ESD of the colorectum and indicated that colon flexure location was an independent risk factor for piecemeal resection. Isomoto et al.  reported that a tumor located in the right-side colon was an independent risk factor for histological incomplete resection. Although multivariate analysis indicated that tumor location was not an independent risk factor of piecemeal resection in the present study, ESD for colon tumors requires more experience and attention due to the unique anatomical characteristics of the colon and its flexures.
Repici et al.  suggested that colorectal anatomic characteristics were the main contributors to the risk of ESD-related complications. We noted that colon tumor location was an independent risk factor for complications. Hori et al.  also indicated that tumor location at the colon flexure was an independent risk factor for ESD-related complications and that ESD for colon tumors requires more technical skills and, especially for tumors located at flexures. However, the importance of tumor location was influenced by the endoscopist’s experience.
Our results indicate an adequate safety profile for colorectal ESD. Major complications occurred in 14 (2.3%) patients: 10 had postoperative bleeding, and 4 had a perforation. Postoperative bleeding could be managed with endoscopy without emergency surgery, and patients then often recovered under conservative observation. In the present study, all cases of recognized perforation were successfully managed immediately by endoscopic closure using endoclips.
In the present large consecutive study with a median follow-up period of more than 4 years, neither ESD-related nor disease-specific death was observed. The overall local recurrence rate was low and occurred in patients with piecemeal resection of tumors. As discussed above, hybrid ESD is a significant and independent contributor to piecemeal resection, which is the most important risk factor for local recurrence after endoscopic resection for colorectal neoplasia . Consistently, in the present study, we found that the local recurrence rate of patients treated with hybrid ESD was higher than that of patients treated with standard ESD, but without significant difference, probably because the proportion of patients who underwent hybrid ESD was low. Thus, standard ESD is indicated for treating carcinomatous lesions because it features en bloc resection, which may decrease the risk of local recurrence.
The potential malignance of a lesion should be considered before choosing the endoscopic resection method. A regular endoscopic examination during follow-up is important, particularly for patients with piecemeal resection. Follow-up colonoscopy was recommended at 3–6 months after piecemeal resection, and it has been reported that most local recurrent lesions were detected within 12 months after the initial endoscopic resection [28, 29]. When local recurrence occurs, additional ESD was acceptable [30, 31].
For patients with invasive tumors, such as T1 carcinomas with deep submucosal invasion, surgical treatment is safer than ESD. It was reported that long-term outcomes were favorable if patients with non-R0 resection undergo appropriate additional surgical resection . According to the 2014 JSCCR Guidelines for the Treatment of Colorectal Cancer , patients with T1 CRC should be considered for additional colectomy with lymph node dissection. However, the probability of lymph node metastasis is extremely low if no other risk factors exist [10, 32, 33]. In the present study, 2 patients with T1 CRC had a histological complete resection. Although additional surgery was recommended, it was rejected by these 2 patients because of high surgical risk, and instead each patient accepted an intensive follow-up. No recurrence or lymph node metastasis occurred during the follow-up period.
The limitations of the present study included its retrospective design and single-center analysis, although we evaluated many consecutive case series. Thus, a prospective and multi-center study is warranted.