Introduction

Endoscopic resection of colon neoplasms is a topical issue of world health care within the current screening programs for colorectal cancer and reduction of cancer-related mortality [1, 2]. Endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD) are the most common approaches for endoscopic excision of colon neoplasms. The use of certain technology depends on surgeon’s preferences, surgical school and availability of these methods [3, 4].

Endoscopic mucosal resection implies excision of bowel wall up to submucosa using a diathermic loop [4].

However, technical difficulties arising in EMR of tumors over 20 mm became an impetus for development of a new technique — endoscopic submucosal dissection [5]. In 1999, Gotoda et al. [6] first reported ESD for colon tumor resection. This technique became popular among endoscopic surgeons in the Asia-Pacific region, especially in Japan, within a short period. Later, this approach was approved in European countries [7, 8].

Nevertheless, ESD is a long and energy-intensive manipulation followed by high incidence of complications [9].

Thus, advisability of a particular technology for resection of large tumors can be determined only in comparative studies. However, modern literature data are presented by few comparative trials devoted to EMR and ESD. The main limitation factors are retrospective design and small sample size [10—17]. In this regard, we performed a systematic review and meta-analysis of trials devoted to early (resection quality, morbidity, surgery time) and long-term (recurrence rate) outcomes of EMR and ESD.

Material and methods

Searching for scientific data was carried out in the MEDLINE e-database according to the PRISMA criteria regardless the date of publication. We used the following keywords: "endoscopic submucosal dissection", "endoscopic mucosal resection", "colorectal" [18]. Primary screening revealed 1018 manuscripts. The following articles were excluded: 254 reviews, 518 non-comparative studies devoted to one of the methods, 86 case reports, 152 manuscripts devoted to tumors of upper gastrointestinal tract (Fig. 1).

Fig. 1. Flowchart of the study.

The study included 8 studies including 6 retrospective ones, 1 case-control study and only 1 prospective trial (Tables 1—3). We found no randomized trials comparing ESD and EMR. Inclusion criteria for all studies were tumor dimension over 20 mm and epithelial colon tumor.

Table 1. Characteristics of studies

Table 2. Characteristics of studies (Continuation)

Table 3. Characteristics of studies (Continuation)

All studies summarized endoscopic resection of 1989 colon tumors (EMR – 748, ESD – 1241). We analyzed tumor dimensions, incidence of en-bloc and R0 resections, complications (perforations and delayed bleeding), as well as long-term treatment outcomes (mean follow-up period and recurrence rate). Statistical analysis was carried out using the Review Manager 5.3 program.

Results

Data on mean dimensions of resected tumors are presented in all studies. Nevertheless, there were no significant between-group differences (mean difference was only 3.48 mm (95% CI -7.9-0.9; p = 0.13)). This parameter emphasizes comparability of both groups subjected to different treatment strategies (Fig. 2A).

Mean surgery time (data are available in 7 studies) was significantly higher by 57 min for ESD compared to EMR (95% CI -87 – -28; p = 0.0001) (Fig. 2B).

Fig. 2. Forest plot for tumor dimensions (a) and surgery time (b).

Data on en bloc resection as the main indicator characterizing resection quality were available in 7 reports. We found significantly higher probability of en bloc resection for ESD (91.3 vs. 48 %, OR 0.13; 95% CI 0.03 — 0.49; p = 0.003) (Fig. 3).

Fig. 3. Forest plot for en-bloc resection rate.

Incidence of R0-resection was reported only in 3 out of 8 studies. This value was 84.7% for ESD that was more than 2 times higher than in EMR (40%) (OR 0.23; 95% CI 0.05 1.02; p = 0.05) (Fig. 4).

Fig. 4. Forest plot for R0 resection rate.

Incidence of delayed bleeding was similar in both groups (2.5% for EMR and 1.9% for ESD, 95% CI 0.44 — 1.87; p = 0.79) (Fig. 5C). However, risk of intraoperative perforation was 4 times lower in endoscopic mucosectomy (0.9%) compared to ESD (4.8%) (OR 0.25; 95% CI 0.08 — 0.81; p = 0.02) (Fig. 5A).

Fig. 5. Forest plot for incidence of perforation (a) and delayed bleeding (b).

We found no differences in the incidence of additional resections in accordance with morphological characteristics of tumor (data are available in 3 studies). However, the ESD group was characterized by higher risk of salvage surgery (9.6%) compared to EMR (5.8%) (OR 0.48; 95% CI 0.22 — 1,03; p = 0.02) (Fig. 6).

Fig. 6. Forest plot for incidence of redo surgery ensuring adequate resection quality.

Follow-up period was reported in 4 out of 8 studies (mean 17 — 53 months). There were no significant differences in patient-years between the EMR and ESD groups (p = 0.25). Local recurrence (available data in 5 studies) was more common after endoscopic mucosectomy (15.2%) compared to ESD (0.9%) (OR 13.94; 95% CI 6.3 — 30,8; p = 0.00001) (Fig. 7—8).

Fig. 7. Forest plot for mean follow-up period.

Fig. 8. Forest plot for recurrence rate.

Conclusion

Endoscopic dissection of colon tumors was introduced into clinical practice in the late 1990s. This technique is characterized by more precise manipulations, deeper capture of submucosa layers and need for additional technical equipment compared to more traditional mucosectomy.

According to the results of our meta-analysis, incidence of en-bloc resection is 6 times higher for ESD compared to EMR (OR 0.13; 95% CI 0.03 — 0.49; p = 0.003). R0 resection rate was assessed only in 3 studies. However, this value was also significantly higher for ESD (OR 0.23; 95% CI 0.05 — 1.02; p = 0.05). As a result, we obtained 13-fold higher incidence of local recurrence after mucosectomy (OR 13.94; 95% CI 6.3 — 30.8; p = 0.00001).

Despite higher resection quality following ESD, there are also other important parameters of quality of surgical treatment favorably distinguishing EMR. Thus, duration of EMR is significantly less by 57 min compared to ESD (95% CI -87 – -28; p = 0.0001). Incidence of delayed bleeding was similar (0.91 95% CI 0.44 — 1.87; p = 0.79). However, ESD is associated with 4 times higher risk of colon wall perforation compared to EMR (OR 0,25; 95% CI 0.08 — 0.81; p = 0.02).

According to meta-analysis by Ceglie A. et al. [19], incidence of en-bloc resections following EMR is significantly lower (62.8%) compared to submucosal dissection (90.5%) (OR = 0.18; p <0.0001; 95% CI 0.16–0.2) that directly affects recurrence rate (OR = 8.19; 95% CI: 6.2–10.9; p <0.0001) [19]. The authors also found higher incidence of complications after ESD compared to EMR procedure (p <0.0001, OR = 0.19, 95% CI 0.15–0.24).

Our data are similar to the results of meta-analysis by Zhao H. et al. [20]. These authors pooled 12 English- and Chinese-language retrospective studies. According to their data, incidence of en-bloc resection was 95% in the ESD group. This value was significantly higher than in the EMR group (42.8%) (OR 0.07; 95% CI 0.02 — 0.07, p <0.00001). Higher recurrence rate was also obtained in the EMR group (15.9% versus 0.5%) (OR 23.06; 95% CI 11.11 — 47.85; p <0.00001). Moreover, the authors revealed significant between-group differences in the incidence of perforation. This complication was more common in case of submucosal dissection (2.4%) (OR 0.56; 95% CI 0.32 — 0.97; p = 0.04).

However, the results of this systematic review are limited by inclusion of not only full-text articles, but also abstracts and articles in Chinese. These features explain difference in the number of articles analyzed. Nevertheless, similar conclusions indicate reproducibility and regularity of data. It is also important that the authors did not adjust for tumor dimension as a main factor compromising the results of mucosectomy.

A common disadvantage of meta-analyses is no adjustment for dimension of neoplasms. Mucosectomy is limited by dimension of endoscopic loop and does not allow total resection of large colon neoplasms.

The limitations of our meta-analysis are small number of studies (retrospective as a rule) and no randomized trials. Another significant limitation is inclusion of the results of endoscopic treatment of patients with early colorectal cancer despite their even between-group distribution.

Conclusion

Thus, mucosectomy does not allow en-bloc resection of large colon tumors that is significant limitation of this method. In this case, endoscopic submucosal dissection is preferable regarding resection quality compared to mucosectomy. However, higher incidence of perforations, surgery time and technical features of ESD do not allow us to unambiguously interpret the results of our meta-analysis and determine the optimal surgical approach for large colon tumors. The limitations of our meta-analysis are small number of studies (retrospective as a rule) and no randomized trials. Thus, comparative studies including randomized trials are required.

The authors declare no conflicts of interest.