Aliment Pharmacol Ther 2010; 32: 908–915
Background Endoscopic submucosal dissection (ESD) using short needle knives is safe and effective, but bleeding is a problem due to low haemostatic capability.
Aim To assess the performance of a novel ball-tipped needle knife (Flush knife-BT) for ESD with particular emphasis on haemostasis.
Methods A case–control study to compare the performance for ESD of 30 pairs of consecutive early gastrointestinal lesions (oesophagus: 12, stomach: 32, colorectum: 16) with standard Flush knife (F) vs. Flush knife-BT (BT). Primary outcome was efficacy of intraprocedure haemostasis. Secondary outcomes included procedure time, procedure speed (dividing procedure time into the area of resected specimen), en bloc resection rate and recurrence rate.
Results Median intraoperative bleeding points and bleeding points requiring haemostatic forceps were smaller in the BT group than in the F group (4 vs. 8, P < 0.0001, 0 vs. 3, P < 0.0001). There was no difference between groups for procedure time; however, procedure speed was shorter in the BT group (P = 0.0078). En bloc and en bloc R0 resection rates were 100%, with no perforation or post-operative bleeding. No recurrence was observed in either group at follow-up 1 year postprocedure.
Conclusions Ball-tipped Flush knife (Flush knife-BT) appears to improve haemostatic efficacy and dissection speed compared with standard Flush knife.
The use of endoscopic submucosal dissection (ESD) enables en bloc resection of lesions that are difficult to remove by conventional endoscopic mucosal resection (EMR).1–15 Challenges to overcome include the long procedure time; the high level of technical difficulty; and the high incidence of complications such as bleeding and perforation.16–22 To conduct ESD with greater facility, safety and efficiency, various kinds of knives have been developed, including endo-knives with water-jet functions. The Flush knife, developed by us, is one of the water-jet emitting short needle knives23–26 with which more than 1000 cases have been treated, yielding good results;27–29 however, needle knives have low haemostatic capability. Therefore, we have developed a new ball-tipped Flush knife (Flush knife-BT) to improve the haemostatic capabilities of the standard Flush knife. The aim of this study was to assess the results of treatment with the use of the Flush knife-BT compared with the standard Flush knife for ESD of early gastrointestinal neoplasms using a case–control study design, with particular emphasis on haemostasis.
Materials and methods
We conducted a preliminary study to compare the results of endoscopic submucosal dissection (ESD) of 60 consecutive early gastrointestinal neoplasms with the standard Flush knife vs. a new ball-tipped Flush knife (Flush knife-BT), at Kishiwada Tokushukai Hospital between March and June 2008. This pilot study necessarily used small numbers, potentially allowing a non-balanced randomization by chance. As the difficulty of ESD varies substantially according to the location of gastric lesions and the macroscopic type of colorectal lesions, we assigned one of a pair lesions of similar size and location, in the same organ (and of the same macroscopic type in the colorectum) to the Flush knife (F) group and the other lesion to the Flush knife-BT (BT) group in a case–control fashion. A total of 60 lesions in 52 patients (Table 1: Oesophagus: 12 lesions in 11 patients, Stomach: 32 lesions in 27 patients, Colorectum: 16 lesions in 14 patients) were treated between March and June 2008. Patients with intramucosal or slightly invaded submucosal cancer were considered eligible; those with ulcer scars or previously treated lesions were excluded, as were those with severe organ failure, or who had undergone anticoagulant/antiplatelet therapy.
|Flush knife||Flush knife BT||P|
|No. of lesions||30||30|
|Organ, location, macroscopic type|
|Age||69||70||P = 0.26|
|Gender(F:M)||21:9||16:14||P = 0.18|
|Median tumour diameter (cm)||2.0 |
|P = 0.42|
|Median resected specimen diameter (cm)||4.0|
|P = 0.16|
|Low-grade intraepithelial neoplasia||1||0|
All procedures were carried out by one skilled endoscopist (T.T.) with long-term experience of more than 2000 ESD cases performed over a period of 5 years. The study protocol was approved by the Ethics Committee of Kishiwada Tokushukai Hospital. This study has been registered in the University hospital Medical Information Network Clinical Trials Registry (UMIN-CTR) as number UMIN000003037. Written informed consent was obtained from all patients.
Endoscopic surveillance was performed at 3 months and annually thereafter in the oesophageal and gastric cases, and annually in colorectal cases.
Endo-knives (Video clip S1)
Flush knife. The Flush knife (Fujinon Optical Co, LTD, Tokyo, Japan)23–26, 28, 29 (Figure 1a) is a typical endo-knife with a needle 0.5 mm in diameter and lengths ranging between 1, 1.5, 2, 2.5 and 3 mm. Using the tip of the sheath as a guard, the knife is easily and safely handled to incise and dissect tissue with slight tension under direct observation. The diameter (0.5 mm) of the Flush knife is larger than that of a conventional needle knife (0.4 mm). Its tip is dull, with a low electrical current density and is therefore less likely to cause bleeding and more efficient in stopping oozing of blood compared with a standard, sharp, fine, needle knife;30 nevertheless, haemostatic forceps are often concurrently needed to stop pulsating bleeding. A water jet emitted from the tip of the sheath enables lavage of the operating field, as well as submucosal local injection without replacement of the knife, thus affording efficient treatment as reported.27
Theoretically, the current density in the Flush knife-BT is decreased by enlarging the tip. During preliminary studies in the developmental stage, the haemostatic capability was improved with the spark emanating radially from the ball-shaped tip. This provided a more suitable output for haemostasis over a wider area, and allowed catching the blood vessel between the back of the ball-shaped tip and the arm of the knife, leading to improved precoagulation capability. In the two groups of patients, the 2.5 mm long type was used for the stomach, 2 mm for the oesophagus, and 1.5 mm for the colorectum.
Other technical devices
Endoscopes. A single channel endoscope (GIF Q 240I,Olympus Optical Co., Ltd, Tokyo, Japan, for oesophagus and stomach, and CF 240I,Olympus Optical Co., Ltd., for colorectum) was used with a 4 mm long transparent hood to keep a clear operating field.
Electrosurgical unit. The Electrosurgical unit (ESU), VIO300D (ERBE Elektromedizin GmbH, Tubingen, Germany) was used at the same setting for both groups and as conventionally used with the standard Flush knife.24–26
ESD technique (Figure 2a–j)
Marking. The tip of the knife was used for marking the margin for resection, except in the colorectum, where the lesion edge is clearly defined (Forced coagulation mode; effect 3, 20W).
Local injection. With a 25-gauge endoscopic needle (Top Corporation, Tokyo, Japan), a solution was injected into the submucosal layer to provide mucosal elevation: saline for the stomach and diluted sodium hyaluronate solution (MucoUp; Johnson & Johnson K.K., Tokyo, Japan)31 for the oesophagus and the colorectum.
Mucosal incision. Each knife was used for continuously incising along the circumference of the lesion with the tip of the sheath as a guard (Endo-cut I mode; effect 4, duration 3, interval 3 in the oesophagus: effect 2, duration 3, interval 2 in the stomach: effect 2, duration 3, interval 3 in the colorectum).
Additional local injection. Additional saline was locally injected by the water-jet emitting function powered by an electric pump (Figure 1d).
Submucosal dissection. The submucosal layer was, after an additional local injection, dissected (Forced coagulation mode; effect 3, 45W in the stomach: effect 2, 40W in the oesophagus and colorectum). The precise procedure has been described elsewhere.23–26
Haemostasis and vessel coagulation. The vessels were isolated, both sides precoagulated (Forced coagulation mode; effect 3, 45W in the stomach: effect 2, 40W in the oesophagus and colorectum) and then dissected in the forced coagulation mode. Large vessels that were not coagulated to white were, after pre-coagulation, coped with haemostatic forceps (Soft coagulation mode; effect6, 100W in the stomach: effect 5, 100W in the oesophagus and colorectum). In case of bleeding, haemostasis with the knife itself was tried first (Forced coagulation mode; effect 3, 45W in the stomach: effect 2, 40W in the oesophagus and colorectum). If unsuccessful after 2–3 attempts, Coagulasper forceps (oesophagus and colorectum; FD-410LR; Olympus) or Hot biopsy forceps, Radial Jaw 3 (stomach; REF1550; Boston scientific Co, Massacheusetts, USA) were used (Soft coagulation mode effect 6, 100W in the stomach: effect 5, 100W in the oesophagus and the colorectum).
In either group, post-operative preventive soft coagulation by haemostatic forceps was performed to all visible vessels in stomach and to pulsating visible vessels in oesophagus and colorectum.
Histopathological assessment (Figure 2h,i)
The resected specimens were stretched and fixed onto a rubber plate, then immersed in formalin and sectioned serially at 2-mm intervals and subjected to histopathological examination. Resection of lesions in one piece with the lateral and vertical margins free of the tumour was defined as en bloc R0 resection. Patients diagnosed with sm2 or deeper invasion were referred for surgical resection.
The primary end point for the study was efficacy of procedural haemostasis, defined as the number of intraoperative bleeding points and the number of points necessitating the use of haemostatic forceps.
Secondary end points were the number of episodes of pre-coagulation by haemostatic forceps, procedure time, procedure speed, en bloc resection rate, en bloc R0 resection rate, perforation rate and post-operative bleeding rate. The procedure time was counted from the beginning of the local injection to the end of the procedure (when en bloc resection was achieved), measured using a stop watch. The procedure speed, the area of mucosa dissected per unit time (cm2/min), was calculated by dividing the procedure time into the area of the resected specimen. The approximate oval area (cm2) of the resected specimen was calculated as follows: 3.14 × 0.25 × long axis × minor axis. All procedures were recorded on videotape for the evaluation of parameters.
Data are presented as medians. Independent continuous variables were compared by the Mann–Whitney test, and categorical variables were compared by the χ2 (chi-squared) test or Fisher’s exact test using Statview version 5.0. All P values were two sided, P values <0.05 was considered significant. As no correction has been made for multiple testing of the data, with the exception of the primary end point, results should be considered exploratory or hypothesis generating.
There was no significant difference between the two groups in age, gender, histological diagnosis, median diameter of the tumour (P = 0.42) or the median diameter of the resected specimen (P = 0.16) (Table 1).
The number of intraoperative bleeding points and bleeding points necessitating the use of haemostatic forceps was significantly smaller in the BT group than in the F group (median 4 vs. 8, P < 0.0001; and 0 vs. 3, P < 0.0001) (Table 2). The number of pre-coagulation by haemostatic forceps was median 0 (range: 0–2) in the BT group and 0 (range: 0–3) in the F group, P = 0.094. In the examination by organ, the bleeding points were significantly fewer in the BT group (Table 3). Also, in the three locations of the stomach, these points were fewer, although this did not always reach statistical significance, particularly for the lower body (Table 4). There was no significant difference between the groups in procedure time (P = 0.13), except that it tended to be shorter in the BT group; however, the median procedure speed was significantly shorter in the BT group (P = 0.0078) (Table 2). The en bloc and the en bloc R0 resection rates were both 100%, with no perforation or post-operative bleeding (Table 2). At follow-up endoscopy, the ulcers at the resection sites had healed completely to become scars. No recurrence was observed in either group at follow-up 1 year after the procedure (Table 2), and there has been no difference in the degree of deformation or stricture of the lumen during the ulcer healing process.
|Flush knife||Flush knife BT||P|
|Procedure time (min)||32.0 (14.5–83.0)||28.7 (11.8–58.0)||P = 0.13|
|Procedure speed (cm2/min)||1.20 (0.61–3.11)||1.59 (0.72–3.44)||P = 0.0078|
|Intraoperative bleeding points||8 (2–20)||4 (1–11)||P < 0.0001|
|Bleeding points requiring haemostatic forceps||3 (0–10)||0 (0–2)||P < 0.0001|
|En bloc resection rate||100% (30/30)||100.0% (30/30)|
|En bloc R0 resection||100% (30/30)||100% (30/30)|
|Perforation rate||0% (0/30)||0% (0/30)|
|Post-operative bleeding||0% (0/30)||0% (0/30)|
|Recurrence rate (1 year postprocedure)||0% (0/30)||0% (0/30)|
|Flush knife||Flush knife BT||P|
|Intraoperative bleeding points||8 (6–12)||4 (3–10)||P = 0.029|
|Bleeding points requiring haemostatic forceps||3 (0–6)||0 (0–0)||P = 0.0071|
|Intraoperative bleeding points||7 (2–15)||4 (1–11)||P = 0.012|
|Bleeding points requiring haemostatic forceps||3 (0–10)||0.5 (0–2)||P = 0.0044|
|Intraoperative bleeding points||7 (3–20)||4.5 (1–7)||P = 0.064|
|Bleeding points requiring haemostatic forceps||3 (0–10)||0 (0–1)||P = 0.0057|
|Stomach||Flush knife||Flush knife BT||P|
|Upper body (n = 4)|
|Intraoperative bleeding points||9.5 (6–15)||4 (2–6)||P = 0.028|
|Bleeding points requiring haemostatic forceps||5 (2–10)||0.5 (0–1)||P = 0.019|
|Middle body (n = 7)|
|Intraoperative bleeding points||8 (4–15)||4 (1–11)||P = 0.14|
|Bleeding points requiring haemostatic forceps||3 (0–6)||1 (0–2)||P = 0.019|
|Lower body (n = 5)|
|Intraoperative bleeding points||5 (2–7)||2 (1–7)||P = 0.44|
|Bleeding points requiring haemostatic forceps||0 (0–3)||0 (0–2)||P = 0.52|
In the procedure of ESD, prevention of intraoperative bleeding and effective haemostasis are essential for safe operation in a clear surgical field.25, 26
The efficacy of coagulation increases as the current density decreases, resulting in better haemostasis. Conversely, at a high current density, incision efficacy increases, whereas haemostasis is less effective. Theoretically, to decrease the density of the current and to raise the efficiency of haemostasis, the diameter of the knife should be increased. We, therefore, developed the Flush knife with a ball tip (Flush knife-BT), and as predicted, the density of the current at the tip of the knife decreased as compared with that of the Flush knife, and haemostatic efficiency improved (Figure 2d–f). The number of intraoperative bleeding points and the number of points necessitating the use of haemostatic forceps were significantly smaller in the BT group than in the F group. This effect was observed in every organ; in the stomach, the improvement tends to be most marked in the upper and middle gastric parts where large vessels are located at a higher density;25 however, there were relatively few data points for each location in the stomach and hence these comparisons should be treated with caution. Other than haemostatic efficiency, the ideal ESD knife affords easy operability, efficient incision, low risk of perforation and minimal effect of coagulation on tissue at the border of the resected specimen.
Operability would be improved and made effective by fashioning the tip in the shape of an anchor to move the target to a safe distance; however, improving haemostasis would be less attainable because the density of the current in the sharp edge would increase. As the efficiency of haemostasis was given priority, such construction was not adopted. The ball-shaped tip of the Flush knife-BT facilitated the scooping up of the incised and dissected tissue, potentially improving its operability (Figure 2b,c) (Figure 3); moreover, the utility of this knife increased with the additional benefit of haemostatic efficiency.
The risk of perforation needs to be addressed because the tip of the Flush knife-BT is not insulated, and perforation may occur during its use in the muscle layer. This knife is fundamentally different from the tip-insulated IT knife, even though both knives have similar forms.
The degree of coagulation at the border of the resected specimen with possible damage to the nearby tissue was of concern because of the high efficiency of haemostasis. Such damage was only to the extent that a thin layer of coagulated tissue formed at the border of the resected specimen, and the degree of coagulation was similar in the F group and BT group with no difference in the healing process of the ulcer (Figure 2g,i,j).
For the BT group, the procedure time tended to be shorter than for the F group, and the procedure speed was significantly faster, both attributable to the improved haemostatic efficiency and easy operability of the knife. There was concern that the water-jet emitting function of the knife would be impeded by the ball-shaped tip; however, no disruption occurred, although backflows tended to increase during the additional local injections (Figure 1d).
This study has some limitations. Data from case–control study designs are not as robust as those from randomized studies, but case–control design does facilitate lesion matching for relatively small sample sizes. Although the results of this study were encouraging, patients with ulcer scars or those who had undergone previous treatment were not enrolled. Procedures with the Flush knife-BT for lesions with severe fibrosis were thought to be difficult because the ball-shaped tip decreased incision capability. Such cases could be dealt with, in part, by using endo-cut mode, but a sharper-tipped needle knife (including the standard Flush knife) may also be needed.
As the results of this study were obtained by one skilled endoscopist and at one location, they may have less generalizability. Randomized, multi-site trials are ideally needed to confirm both the usefulness of the ball-tipped Flush knife (Flush knife-BT) and the outcomes reported in this study.
The ball-tipped Flush knife (Flush knife-BT) appears to have improved haemostatic efficacy and higher dissection speed compared with the standard Flush knife. The development of the Flush knife-BT may facilitate more rapid ESD with fewer complications.
We thank Dr James East, John Radcliffe Hospital, Oxford, United Kingdom for helpful comments on this manuscript. Dr Toyonaga invented the standard Flush knife and the ball-tipped Flush knife (Flush knife-BT) in conjunction with Fujifilm Inc., Tokyo, Japan and receives royalties from its sale. We registered the trial procedure for this manuscript with UMIN Clinical Trials Registry (UMIN-CTR). UMIN-CTR ID: UMIN000003037. The receipt number: R000003681. Title: Performance of a novel ball-tipped Flush knife for endoscopic submucosal dissection https://center.umin.ac.jp/cgi-open-bin/ctr/ctr.cgi?function=brows&action=brows&recptno=R000003681&type=summary&language=E. Declaration of personal and funding interests: None.