Article Text
PDF
Abstract
Objective Endoscopic full-thickness resection (EFTR) is a novel treatment of colorectal lesions not amenable to conventional endoscopic resection. The aim of this prospective multicentre study was to assess the efficacy and safety of the full-thickness resection device.
Design 181 patients were recruited in 9 centres with the indication of difficult adenomas (non-lifting and/or at difficult locations), early cancers and subepithelial tumours (SET). Primary endpoint was complete en bloc and R0 resection.
Results EFTR was technically successful in 89.5%, R0 resection rate was 76.9%. In 127 patients with difficult adenomas and benign histology, R0 resection rate was 77.7%. In 14 cases, lesions harboured unsuspected cancer, another 15 lesions were primarily known as cancers. Of these 29 cases, R0 resection was achieved in 72.4%; 8 further cases had deep submucosal infiltration >1000 µm. Therefore, curative resection could only be achieved in 13/29 (44.8%). In the subgroup with SET (n=23), R0 resection rate was 87.0%. In general, R0 resection rate was higher with lesions ≤2 cm vs >2 cm (81.2% vs 58.1%, p=0.0038). Adverse event rate was 9.9% with a 2.2% rate of emergency surgery. Three-month follow-up was available from 154 cases and recurrent/residual tumour was evident in 15.3%.
Conclusion EFTR has a reasonable technical efficacy especially in lesions ≤2 cm with acceptable complication rates. Curative resection rate for early cancers was too low to recommend its primary use in this indication. Further comparative studies have to show the clinical value and long-term outcome of EFTR in benign colorectal lesions.
Trial registration number NCT02362126; Results.
- EFTR
- endoscopic full-thickness resection
- FTRD
- non-lifting adenomas
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Conventional endoscopic resection techniques like polypectomy, endoscopic mucosal resection and endoscopic submucosal dissection are highly effective but limited to the mucosa and submucosa.
Endoscopic resection of colorectal non-lifting lesions, early carcinomas and subepithelial tumours is difficult and may not even be possible in some cases.
Clip-assisted endoscopic full-thickness resection (EFTR) has been shown to be feasible for such lesions in small retrospective studies.
What are the new findings?
This is the first prospective multicentre study demonstrating that EFTR with the full-thickness resection device is effective for difficult-to-resect colorectal lesions, especially for lesions ≤2 cm.
Complication rates are higher than suggested by former retrospective data but still seem acceptable.
How might it impact on clinical practice in the foreseeable future?
This novel and relatively simple endoscopic technique allows effective treatment of ‘difficult’ colorectal lesions and may become an alternative to surgery in selected patients.
Introduction and background
Endoscopic mucosal resection (EMR) and endoscopic submucosal dissection (ESD) are highly effective techniques for resection of colorectal neoplasms.1–4 Both are based on separation of the mucosa from the muscular layer by submucosal (sm) injection. In ‘non-lifting’ lesions, the mucosa cannot be separated for the muscular layer due to sm scaring or sm tumour invasion. Conventional endoscopic resection of such lesions is technically demanding with inherent risk of perforation.5–9 ‘Conventional’ techniques also have limitations for resection of adenomas at difficult anatomic locations (such as lesions at a diverticulum and at the appendiceal orifice) and also for resection subepithelial tumours. Endoscopic full-thickness resection (EFTR) using a clip-assisted non-exposure technique has been described to be feasible in those ‘difficult-to-resect’ lesions and may be a less invasive alternative to surgical resection.10 11 The full-thickness resection device (FTRD, Ovesco Endoscopy, Tuebingen, Germany) is an over-the-scope system, which allows single-step EFTR after placement of a modified over-the-scope clip (OTSC). The device is CE marked for colorectal EFTR since September 2014 and has been described to be feasible and effective in a case series and two small retrospective clinical studies.12–14 We present the results of the first prospective multicentre trial investigating efficacy and safety of the device for EFTR of ‘difficult-to-resect’ colorectal lesions.
Methods
Study design
This prospective, investigator-initiated, non-randomised clinical trial (‘WALL RESECT,’ clinicaltrials.gov: NCT02362126) was conducted at nine referral centres in Germany. The study protocol was approved by the institutional review board at each centre. The aim of the study was to confirm results of the pilot retrospective study12 regarding efficacy and safety of colorectal EFTR with the FTRD in a prospective multicentre setting.
We included patients with colorectal lesions which were difficult or not possible to resect with conventional endoscopic methods like polypectomy, EMR or ESD. Those included: colorectal adenoma with negative lifting sign (recurrent, incompletely resected or untreated), adenoma involving the appendiceal orifice or a diverticulum, T1 carcinoma with indication for endoscopic (re)resection and subepithelial colorectal tumour. Exclusion criteria were: maximum estimated diameter of lesion >3 cm, T1 carcinoma with known ‘high-risk features’ (defined as sm infiltration >1000 µm or invasion of lymphatic vessels or undifferentiated (G3) tumour); patients younger than 18 years, pregnant patients, patients unable to sign informed consent and patients on dual platelet inhibition. All patients had undergone complete colonoscopy before inclusion into the study.
Primary endpoints were:
Technical success: resection en bloc (in one piece) and macroscopically complete (no macroscopic evidence of residual lesion, judged by the endoscopist)
R0 resection: histologically complete resection, defined as tumour-free lateral and deep resection margins
Secondary endpoints were:
Histologically confirmed full-thickness resection (visibility of all layers of the colonic wall including serosa within the resection specimen), procedure-related adverse events, procedure time, necessity of surgical therapy and evidence of residual/recurrent lesion after 3 months. In case of carcinomas, curative resection was defined as lateral and deep R0 resection and lack of histological ‘high risk’ features,’ see above definition.
Sample size calculation
Based on the results of the pilot study,12 we aimed to show that technical success rate of full-thickness resection in the entire cohort would be ≥80%. To accept this hypothesis with an accuracy of at least 95%, a sample size of at least 160 patients was calculated. The sample size calculation was performed by an independent statistical institute (York Hilger Institute for Statistics, Freiburg, Germany).
Data management and statistical analysis
Study data were collected and analysed at the coordinating study centre in Ludwigsburg. The database was created with Microsoft Excel, data entry was done by a trained study nurse at the Department of Gastroenterology, Klinikum Ludwigsburg. Data entry was independently verified by the coordinating physician (AS) and a second trained study nurse. The source data at the participating centres were reviewed by the coordinating centre (one trained study nurse and the coordinating physician) after completion of patient enrolment. However, no systematic monitoring of the endoscopy database for selection bias and rate of all eligible patients with the respective indication who were included and excluded was available, so inclusion of patients depended on the participating centres.
Descriptive statistics were expressed as median, SD and range. χ2 test or Fisher’s exact test was used to compare qualitative data. A p value <0.05 was considered statistically significant. The statistical calculations were done with WinSTAT (Fitch Software) and Statistical Online Computational Resource (http://www.socr.ucla.edu/htmls).
All authors had access to the study data, reviewed and approved the final manuscript.
Full-thickness resection device
The FTRD is an over-the-scope system which can be mounted over standard colonoscopes (recommended diameter 11.5–13.2 mm). It consists of a transparent cap with a modified 14 mm OTSC. Compared with a conventional OTSC System, the cap is longer (23 mm measured from the tip of the endoscope) and clip design is slightly modified. The tip of the cap harbours a 13 mm snare. The handle of the snare runs on the outer surface of the endoscope underneath a transparent plastic sheath (figure 1). The device has gained CE mark for EFTR in the lower GI tract in September 2014.
The FTRD (Ovesco Endoscopy, Tuebingen, Germany). (A) Tip of the endoscope with mounted cap and clip. The snare is integrated in the tip of the cap. A grasping forceps can be advanced through the working channel of the endoscope. (B) The FTRD mounted on standard endoscope. The handle of the integrated snare runs underneath a plastic sheath on the outer side of the endoscope. (C) Schematic illustration of the resection process. FTRD, full-thickness resection device.
EFTR procedure and periprocedural management
After written informed consent, patients were scheduled for endoscopy including the full-thickness resection procedure. All procedures were done in an inpatient setting under deep sedation with propofol +/− midazolam. Patients received a single dose of intravenous antibiotics (ciprofloxacin and metronidazole) peri-interventionally. Patients on acetylsalicylic acid were advised to continue the medication, all other anticoagulants (clopidogrel, heparin, warfarin or direct oral anticoagulants) were discontinued.
All resections were performed by endoscopists (one to two physicians at each centre) with thorough expertise in colonoscopy, EMR and OTSC placement. All participating endoscopists additionally had undergone a 1-day training in EFTR with FTRD, which included hands-on training on ex vivo pig models within 3 months prior to the start of the study. All participants were instructed in the definitions and procedures of the study protocol.
All endoscopic procedures were done with CO2 insufflation. Before resection, diagnostic endoscopy was performed to identify the lesion. The size of the lesions was measured endoscopically using an opened forceps (7 mm) or a snare (10, 15, 25 mm). ‘Trimming’ of the lesion with a snare prior to EFTR was not allowed prior to EFTR.
Lateral margins of the lesion were circumferentially marked with coagulation using a high-frequency probe (Ovesco Endoscopy). The prOVE Cap (Ovesco Endoscopy), which is a ‘dummy’ cap designed to assess if the lesion can be reached and incorporated into the cap, was not used for the procedures during the study. The endoscope was then equipped with the FTRD System. Thereafter, the endoscope with the mounted FTRD was again advanced to the lesion. A grasping forceps or an anchor device (Ovesco Endoscopy) was advanced through the working channel. The anchor device was preferred in case of subepithelial tumours, in all other cases, the forceps was used to grasp the lesion. The lesion was then gently pulled into the cap until lateral markings were visible in the cap. The clip was then deployed, and tissue above the clip was immediately resected with the snare. The resection specimen was subsequently harvested, the resection site was endoscopically inspected for resection completeness, and signs of perforation or bleeding (for a clinical example, see figure 2).
Endoscopic full-thickness resection of a non-lifting recurrent adenoma. (A) Endoscopic image of the non-lifting adenoma in the ascending colon. The lesion has a diameter of approximately 25 mm and is flat elevated (Paris classification 0-IIa). (B) Virtual chromoendoscopy with narrow band imaging. (C) Endoscopic view with the mounted FTRD System. The lesion is pulled into the cap with the grasping forceps. (D) Resection site with FTRD clip securing colonic wall patency. (E) Resection specimen. (F) Histological image of the resection specimen (H&E staining) with the lateral resection margin of the tubular adenoma with high-grade dysplasia. Resection was deemed R0 for lateral and deep margins. FTRD, full-thickness resection device.
Procedure time was measured with a stopwatch by a second physician or a study nurse. Total procedure time was measured from first introduction of the colonoscope until extraction of the endoscope after reinspection of the resection site. Time for advancing the FTRD was measured from introduction of the FTRD until the target lesion was reached. Time for resection was measured from reaching the lesion with the FTRD until resection was achieved.
The resection specimen was pinned down on cork before immersion in formalin. Histopathological evaluation was done by the local pathologist in each centre. Patients were put on clear liquids at the same day if there was no clinical evidence of peritonitis, and received regular diet the next day. All patients stayed in the hospital until at least postoperative day 1 and were monitored clinically for signs of bleeding or perforation.
Follow-up
Patients were scheduled for follow-up (F/U) endoscopy 3 months after EFTR. The resection site was inspected for macroscopic evidence of residual or recurrent lesion. Biopsies were taken if there was evidence of residual/recurrent lesion. If the resection site showed no evidence of recurrence or residual tumour, it was left to the decision of the endoscopist whether to take biopsies. If still in place, the FTRD clip was removed only in patients who primarily had a histologically incomplete (‘R1 or Rx’) EFTR. For removal, the clip was cut with a bipolar cutting device (remOVE System, Ovesco Endoscopy) and the fragments were extracted.15 In patients with initial R0 resection and no evidence of residual tumour on F/U, the clip was not removed. In case of residual lesion, decision for further therapy was left to the judgement of the treating physician.
Results
Patient characteristics and indication for EFTR
Between February 2015 and April 2016, a total of 205 patients were screened for eligibility at 9 centres in Germany; 24 patients were not included in the study, reasons for non-enrolment are shown in figure 3. One hundred eighty-one patients finally met the inclusion criteria and were enrolled. Patient and lesion characteristics are shown in table 1.
Flow chart showing enrolment and main results. Sixteen patients were not included due to positive lifting sign. Those lesions had primarily been designated as ‘non lifting’ by referring physicians but showed complete lifting after submucosal injection upon endoscopy at the study centres. *Three patients were not referred to surgery due to medical comorbidities. Two patients were scheduled for surgery but then refused operation and underwent further F/U. Three-month endoscopic F/U was available in all five patients without evidence of residual/recurrent tumour. **Endoscopic F/U was available in two patients (one before and one after surgical resection). ***In one patient, there was microscopic evidence of residual adenoma but no carcinoma, patient was scheduled for further F/U. EFTR, endoscopic full-thickness resection; EMR, endoscopic mucosal resection; ESD, endoscopic submucosal dissection; F/U, follow-up; SET, subepithelial tumour. Submucosal infiltration study centres: A, Augsburg; D, Düsseldorf; DO, Dortmund; E, Essen; FR, Freiburg; KR, Krefeld; LB, Ludwigsburg; TÜ, Tübingen; U, Ulm.
Patient and lesion characteristics (entire cohort)
Procedural data and technical success
The target lesion was successfully reached with the FTRD in all patients; no patients were excluded due to unsuccessful advancement of the endoscope. Procedural data are shown in table 2. Resection was technically successful (macroscopically complete and en bloc) in 162/181 patients (89.5%, see table 2). No significant differences in resection success were observed between the participating centres. Median duration of hospital stay was 4 days (range 1–11), prolonged hospital stay was due to adverse events (see table 3) or medical comorbidities.
Procedural data (entire cohort)
Subgroup analysis concerning technical success and R0 resection
Resection was reported to be difficult in a total of 28 procedures. In 13 cases, this was due to dysfunction of the snare and in 15 cases difficult incorporation of the lesion into the cap was reported. In case of snare dysfunction, the FTRD was extracted after clip deployment and resection could be completed with a conventional snare in 12/13 cases, 1 patient was sent to surgery. In patients with difficult tissue incorporation into the cap, 14/15 resections were completed, 1 patient was referred to surgery. Technical success and R0 resection rates for this subgroup are shown in online supplementary table 1. Detailed information about the patients with difficult tissue incorporation is shown in online supplementary table 2. Of note, most of those patients (75%) had undergone prior endoscopic therapy.
Histology/R0 resection
Total cohort
In the total cohort, histologically complete resection (R0) was achieved in 139 patients (76.9%). A detailed subgroup analysis concerning resection success is shown in table 3. R0 resection rate dropped with increasing lesion size. R0 resection rate was significantly lower for lesions >20 mm (58.1%) versus lesions ≤20 mm (81.2%; p=0.0038). We did not observe any improvement in resection success during the course of the study.
Full-thickness resection was histologically confirmed in 81% of cases. Full-thickness resection rate in the rectum was lower than in the colon (66.7% vs 83.9%, p=0.056).
Subgroup of difficult adenomas (n=127)
The subgroup of patients with difficult (non-lifting) adenomas included 104 cases with non-lifting adenomas (recurrent adenomas n=53, incompletely resected adenomas n=19, non-lifting adenomas without prior treatment n=32), 34 cases with adenomas at the appendiceal orifice and 5 cases with adenomas at a diverticulum (see table 1). Fourteen patients who had been classified as non-lifting adenomas had unsuspected cancer on histology and are dealt with below (see also flow chart in figure 3). Final histology of the EFTR specimen in the remaining 129 cases was: adenoma with low-grade dysplasia (n=86), adenoma with high-grade dysplasia (n=21), metastasis of an oesophageal adenocarcinoma (n=1), hyperplastic or scar tissue (n=18, see online supplementary table 3). In two cases, histology was not obtained due to unsuccessful snare resection. In those benign lesions, R0 resection rate was 77.7% (98/127).
Subgroup of carcinomas (n=29)
Fourteen patients who were initially classified as non-lifting adenomas harboured unsuspected adenocarcinoma, eight of them with deep sm infiltration (>1000 µm) on histology. Fifteen additional patients underwent EFTR for known T1 cancer, R0 resection rate for this subgroup was 72.4% (21/29), but curative resection could only be achieved in 45% (2/14 with unsuspected and 11/15 with known cancer). Detailed information about resection success in those malignant lesions is shown in online supplementary table 4.
Subepithelial tumours
In patients with subepithelial tumours, final histological examination showed neuroendocrine tumour (10, all well differentiated (G1)), gastrointestinal stromal tumor (GIST) (2, both with low mitotic index), granulosa cell tumour (2), fibroma (3), granuloma secondary to helminth infection (2), lipoma (1), angiolipoma (1), endometriosis (1), no evidence of residual tumour after endoscopic resection (1), no histology obtained due to unsuccessful EFTR (1). R0 resection was achieved in 20/22 patients (87.0%).
Adverse events
Procedure-related adverse events were observed in 18 patients (9.9%) (see table 3). There were no significant differences between the participating centres. In four patients, bleeding at the resection site was observed. All bleedings occurred delayed within postoperative days 1–3, patients were haemodynamically stable and blood transfusion was not required. None of those patients had received anticoagulants. Endoscopic haemostasis with standard clips, coagulation or injection of fibrin glue was successful in all cases.
In 6 of 181 patients (3.3%), perforation occurred. In five cases, perforation was observed immediately after resection because the clip was unintendedly not released from the cap and resection was carried out before clip closure. In four cases, perforation could be successfully closed in the same session endoscopically, one patient required surgery. Another patient required surgery due to secondary perforation at day 3 after resection. In this case, the integrated snare had not closed during resection and resection was completed with a conventional snare after extraction of the FTRD. Unintended entrapment of the clip into the snare may have caused thermal damage to the colonic wall (for detailed information, see online supplementary table 5).
Acute appendicitis occurred in three patients who had undergone resection of an adenoma involving the appendiceal orifice. In two cases, symptoms were mild and conservative treatment with intravenous antibiotics was successful. The other patient required laparoscopic appendectomy.
Three patients developed postpolypectomy syndrome after resection in the coecum or ascending colon, all patients were managed successfully with intravenous antibiotics; CT imaging did not show evidence of colonic perforation.
One patient complained about recurrent right-sided abdominal pain after resection of a subepithelial tumour in the right colonic flexure. Repeat endoscopy, abdominal ultrasound and CT scan did not show signs of perforation, local inflammation or involvement of surrounding organs. Due to the persistent symptoms, the clip was removed endoscopically after 6 weeks and symptoms gradually resolved after an additional 2 weeks.
One patient developed persistent severe diarrhoea after successful resection of an adenoma in the coecum. F/U endoscopy showed an enterocolonic fistula, the FTRD clip was not in place any more. The fistula probably had developed because of entrapment of small bowel into the clip during the resection procedure. The patient was referred to surgery. One patient developed an ischaemic cerebral stroke at day 8 after resection; this was considered very unlikely to be procedure related.
Follow-up/outcome (n=154)
Total cohort
As mentioned above, 3-month endoscopic F/U data could be obtained from 154 patients (55% male, median age 67 years, range 29–88). Of the remaining 27, fourteen patients were lost to F/U(1 had disabling stroke and was not able to come, 1 moved to another country and 12 refused further F/U despite repeated phone calls) and 13 patients did not undergo endoscopic F/U because they had undergone surgical resection. In 106/154 patients (68.8%), the clip had spontaneously detached from the colonic wall. In 10 cases, the clip was endoscopically removed using a bipolar direct current cutting device (remOVE System, Ovesco Endoscopy).
Biopsies from the resection site were taken in 128/154 patients (83.1%). In the total cohort, residual lesions were observed in 19/154 patients (12.3%). Of those 19 patients, 11 initially had histologically incomplete resection whereas 8 initially had R0 resection. Patients without evidence of residual lesion were scheduled for further F/U according to current German guidelines (usually 3 years). In the patients with residual lesions, median F/U was 162 days (range 38–301).
Patients with adenomas and final benign histology
In the subgroup of patients with difficult adenomas and final benign histology (n=127), F/U was available in 117 patients (9 patients lost to F/U, 1 operated). Ninety-nine of these 117 patients had biopsies taken. Residual lesions were seen in 18/117 patients (15.3%), 14 on endoscopy and 4 only on histology. Of those 18 patients, 15 were successfully retreated endoscopically (6 re-EFTR, 9 snare resection or removal with forceps +/− argon plasma coagulation) without evidence of recurrence so far. In one patient, the clip was deeply ingrown into the colonic wall and could not be removed. In another two patients, the lesion could not be removed endoscopically due to severe scarring and size of lesion. Those three patients were referred to surgical resection (see flow chart in figure 3).
Patients with carcinomas
In the subgroup of 13 patients with curative resection of carcinoma, F/U was available in 11 patients (2 refused re-endoscopy). There was no evidence of residual tumour in these patients (see flow chart in figure 1). Of 16 patients with incurative resection, 11 were referred to surgical resection. Three patients were not suitable for surgery due to medical comorbidities and two patients were scheduled for surgical resection but finally refused operation. All of those five patients underwent further endoscopic F/U after 3 months. There was no evidence of residual tumour in four cases, in one there was histological evidence of residual adenoma with low-grade dysplasia, invasive cancer was not seen (see flow chart in figure 3 and online supplementary table 4).
Patients with subepithelial tumours
In patients with subepithelial tumours, F/U was available in 19/23 patients, there was no evidence of residual or recurrent lesion. Biopsies were taken in 14 cases to confirm these results.
Surgery (n=181)
In total, 20 of 181 patients (11.0%) needed surgical therapy. In four patients (2.2%), surgery was required due to procedure-related adverse events (one patient with appendicitis, two patients with perforation, one patient with an enterocolonic fistula, see above). Eleven patients underwent oncological resection due to incurative carcinoma resection (see above and flow chart and table 4), two patients were referred to surgery due to unsuccessful EFTR and three patients due to residual/recurrent adenoma after EFTR which could not be retreated endoscopically.
Adverse events (entire cohort)
Discussion
This is the first prospective study on EFTR in the colorectum using the FTRD. The study mainly included colorectal lesions, which are difficult to resect with conventional endoscopic techniques. The study demonstrated a total technical success rate of 89.5% and confirms the results of the pilot trial concerning feasibility and efficacy in a multicentre setting.
R0 resection rate did not reach the high proportion of technical resection success (76.9%). However, the majority of lesions were considered to be ‘difficult’ and most of them would even have required surgical resection. In our study population, 11% of patients underwent surgical therapy due to non-curative endoscopic resection or adverse events. A recent meta-analysis by Hassan and colleagues investigating endoscopic resection (26% polypectomy, 56% EMR and 18% ESD) of large colorectal polyps showed similar results with pooled rate of 9% of patients undergoing surgical therapy.16 Another meta-analysis by Fujiya and colleagues reported R0 resection rates of 42.3% for EMR and 80.3% for ESD, 5.8% and 9.9% had to undergo additional surgical therapy.17 Although these numbers seem to be similar to those of our study, the patient cohort is different as we included mainly non-lifting and other ‘difficult-to-resect’ lesions.
One reason for incomplete resection was dysfunction of the snare, which occurred in 13 patients after clip deployment. In those cases, the device had to be extracted and resection was completed with a conventional snare. Although feasible, resection in those cases is more difficult. Due to clip geometry, the snare often cannot be placed close enough to the pseudopolyp base which leads to incomplete or piece meal resection. In the meantime, the integrated snare of the device was modified by the company and in our experience, dysfunction did not occur any more. Incomplete resection can also occur, when the lesion cannot be fully incorporated into the cap. In experimental studies on healthy porcine colonic tissue, maximum size of resection specimen was up to 54 mm.18–20 In the FTRD pilot trial, we suggested a maximum lesion diameter of 30 mm.12 However, a subgroup analysis in the current study showed that R0 resection decreased to 58.1% for lesions >2 cm vs 81.2% for lesions ≤2 cm (p=0.0038). Successful incorporation of the lesion into the cap depends also on thickness and rigidity of the lesion and colonic wall. In 15 patients, difficult incorporation of the lesions into the cap was reported, which was likely due to severe scarring after prior therapy. In this subgroup, R0 resection rate was as low as 57%. Hence, better patient selection may increase R0 resection rate. In case of uncertainty, incorporation of the lesion may be evaluated prior resection using a novel ‘test-cap’ (prOVE CAP, Ovesco Endoscopy). In case the lesion cannot be sufficiently pulled into the cap, FTRD resection may not be attempted and an alternative resection strategy may be considered.
The study included different subgroups of patients and the pros and cons of EFTR with the FTRD shall therefore be discussed separately for each of these groups.
The majority of patients in our study had non-lifting colorectal adenomas, most of them being recurrent or incompletely resected lesions. This is potentially the widest indication for the device, but also alternative methods such as snare resection, ESD, ablation alone or in combination are available, probably requiring repeated colonoscopies. This issue clearly deserves a prospective randomised trial for this indication.
A recent Australian study by Moss and colleagues suggests that conventional endoscopic retreatment of residual or recurrent lesion is highly effective, reaching success rates of 88.2% and 94.5%, respectively.21 However, the patients in the study underwent a close F/U protocol and the majority of lesions were diminutive being resected with a small snare or were ablated with coagulation. In our cohort, lesion size was <5 mm in only two patients. Compliance with F/U may be a problem in these patients as recently shown by a retrospective study by Seidel and colleagues.22 This might result in larger residual/recurrent lesions which may impede retreatment. In our study population, non-lifting adenomas up to a size of 3 cm were included.
In addition, EFTR probably provides better histological specimens than mucosal resection techniques such as ESD23 which may in addition be more difficult due to sm fibrosis5–9 and has therefore not become routine at least in the Western world. Laparoscopic (or occasionally open surgery) may be another alternative which provides definitive treatment for any colonic lesion, is relatively easy to perform above the rectum, but probably has a slightly higher morbidity rate.4 24 Future studies will show which will be the preferred therapy.
We also included patients with adenomas at difficult anatomical locations, the majority of them involving the appendiceal orifice. Up to date, these lesions are usually not endoscopically resectable and require surgical therapy, for example, ileocoecal resection. In our cohort, the majority of such lesions could be resected endoscopically, R0 resection rate was 81.8% for this subgroup. We would like to stress that closure of the appendiceal orifice with the FTRD exhibits a risk of acute appendicitis. In our cohort, only 3 of 34 patients (8.8%) developed appendicitis, one of them (2.9%) requiring laparoscopic appendectomy. Currently, clip-assisted EFTR for those indications should only be done after thorough patient information, interdisciplinary discussion and within clinical studies.
In the subgroup of 29 patients with submucosal carcinomas, R0 resection rate was only 72.4% and curative resection even much lower. In the total subgroup, resection was deemed incurative in 55% of cases, mainly due to deep sm infiltration. Based on these (limited) data, EFTR for lesions with suspected or known carcinoma cannot be recommended for therapeutic purposes but it could be considered under special circumstances as diagnostic tool for tissue acquisition providing exact determination of sm infiltration depth, invasion of lymphatic vessels and tumour differentiation if required.
Finally, the subgroup of subepithelial tumours did not include 1 case of malignancy but 12 cases with lesions with malignant potential (NET, GIST); in this respect, FTDR can be regarded again as tissue acquisition method, but also with sufficient efficacy in complete removal.
Procedure-related adverse events were observed in 18 patients (9.9%) and 2.2% required surgical therapy due to complications. In a recent meta-analysis including 50 studies on endoscopic resection (mainly polypectomy and EMR) of colorectal lesions ≥20 mm, 1% of patients required surgical therapy of complications.16 Reported perforation rates for colorectal ESD vary between 1.3% and 18%, strongly depending on local expertise, lesion location and characteristics.25–27
In view of these facts, this complication rate appears acceptable but has also to be seen in the light of the respective indication. Adequate experience with the device as well as with different colorectal resection and salvage technique is however mandatory as with all advanced endoscopic therapeutic techniques. We also assume that technical modification during or after the study will prevent inadvertent polypectomy without prior clip closure which contributed to the complication rate.
Although our study showed favourable results concerning feasibility and efficacy, the clip-assisted EFTR still has technical limitations. First, advancing of the FTRD to the target lesion can be difficult. The long cap with the loaded clip has an outer diameter of 21 mm, impairs endoscopic vision and flexibility of the endoscope tip. In the pilot FTRD study, advancing the scope was not possible in one case due to sigmoid stenosis,12 which still can happen in clinical practice, even if, in the current cohort, all lesions could be reached successfully, with more than 50% being located in the right hemicolon. Second, visibility during resection is impaired by the long cap. Usually, the lateral margins of the lesion cannot be seen circumferentially but only in the upper part of the cap (see figure 2). Third, the maximum size of the lesions, which can be successfully resected, is limited. As mentioned, lesions up to 30 mm in diameter may be successfully resected but resection success decreased significantly for lesions >20 mm in our study. For larger lesions, reduction of lesion size with EMR prior to same-session EFTR may be performed.28 and further studies are needed to evaluate the role of EFTR for larger lesions.
This study also has several limitations. First, it is a non-randomised study without a control group. It can be debated which randomised controlled trial (RCT) appears feasible, and an RCT comparing EFTR with surgical resection may be impaired by patients and physicians’ unwillingness of randomisation.
Second, several selection biases are possible: the number of screened patients was low and 3 of 9 participating centres recruited <10 patients. Although we aimed to capture data of all consecutive patients eligible for the study, we cannot exclude that a variable number of cases might have been missed and this might have biased the data either way.
Although R0 resection rates were similar between the three groups, it is important to note that resection success has different clinical implications depending on the nature of lesions. While a R0 resection rate of around 77% may be acceptable for recurrent adenomas, it is not acceptable for carcinomas. It has also to be mentioned that the lack of structured biopsy even of normally looking scars (no biopsies in 16%) and only one F/U examination could also introduce a positive bias towards EFTR. A study by Knabe and colleagues showed that in 7% of cases adenoma recurrency was detected in macroscopically unsuspicious scars.29 Hence, recurrency might have been missed in some patients and long-term outcome is not yet clear.
In conclusion, this first prospective multicentre study on EFTR with the FTRD System demonstrated good overall technical efficacy in benign lesions ≤2 cm with acceptable safety. The low curative resection rate in malignancy however was too low to recommend EFTR as primary therapy. Further studies are necessary to investigate the role for EFTR for lesions >2 cm and to evaluate long-term outcome.
Acknowledgments
We thank Ovesco Endoscopy for the technical support during the study.
References
Footnotes
Contributors AS wrote the study protocol, was responsible for conductance of the study at Klinikum Ludwigsburg, coordinated the study centers, analysed the data and wrote the manuscript draft. KC was the sponsor of the study, participated in writing the study protocol, analysing the data and writing the manuscript draft. All other authors enrolled and treated patients at the study sites, and critically reviewed the manuscript.
Competing interests The coordinating centre and the participating centres received financial support for case documentation and costs from Ovesco Endoscopy. AS and KC received lecture fees from Ovesco Endoscopy.
Ethics approval Ethics comittee/Landesärztekammer Nord-Württemberg.
Provenance and peer review Not commissioned; internally peer reviewed.