ESD Technique 2026: DualKnife vs IT Knife, Traction, Pocket-Creation

Confirm ESD is indicated for this organ before you pick a knife. Esophageal squamous and most gastric lesions are ESD rooms. Most colorectal lesions are still EMR. Once ESD is the job, DualKnife versus IT knife is anatomy, not the 2026 time driver: adding ITknife nano to DualKnife J did not shorten colorectal ESD. Get a view with traction or a pocket. Do not snare the last bit of a possible T1 and call it hybrid ESD.

Experienced teaching points

Clinical Pearls

NANO 2025: adding ITknife nano to DualKnife J did not shorten colorectal ESD (111.8 vs 114.3 min). Rectum was faster with DualKnife J alone.
Takeuchi 2010: FlushKnife water-jet cut colorectal ESD time versus Flexknife (61 vs 87 min) by skipping injector swaps. The jet is the lever, not the logo.
Colorectal clip-and-thread: 40 vs 70 min (Yamasaki). Esophageal CONNECT-E: 44.5 vs 60.5 min. Gastric CONNECT-G overall time did not move; perforation fell, and greater-curve upper/middle time halved.
Yamashina PCM: you finish with the method you started 93% vs 73% of the time. Hybrid ESD is not a shortcut (Fuccio hybrid R0 60.6%).
Confirm the organ first. ESGE still wants EMR for most colorectal lesions. Gastric and esophageal squamous are different rooms.

Clinical Bottom Line

Problem in the room	2026 practical answer
Which knife tonight?	A needle-type knife you can mark, incise, and dissect with (DualKnife J or FlushKnife class). Adding ITknife nano to DualKnife J did not shorten colorectal ESD [6].
The case is slow because you keep swapping to an injector	Use a water-jet knife. Takeuchi: FlushKnife 61 vs Flexknife 87 minutes, fewer device changes [7].
Colorectal flap keeps falling into the lumen	Clip-and-thread traction: 40 vs 70 minutes, 95% traction success [10]. Expert DCT series: en bloc 95.7%, 39 mm2/min [12].
Gastric lesion on the greater curve, upper/middle third	Dental-floss clip traction. CONNECT-G overall time did not move, but this subgroup fell from 104 to 57 minutes, and perforation fell from 2.2% to 0.3% [9].
Large esophageal squamous lesion	Traction-assisted ESD: 44.5 vs 60.5 minutes, no adverse events in the traction arm [11].
Floppy or fibrotic colon, poor scope stability	Pocket-creation. Assigned-method completion 93% vs 73% [14]. Add a single clip if pocket opening stalls [16].
Tempted to snare the last island	Do not call that Japanese ESD. Hybrid R0 60.6% [18].
Not sure ESD is even the case	Colorectal default is still EMR. See EMR vs ESD for large colon polyps.

What do I do tonight?

Decide the organ first. Esophageal squamous and most gastric lesions: ESD is the ESGE default. Most colorectal lesions: EMR unless you need en bloc for suspected limited invasion or a snare that will not finish.
Photograph, classify, and mark. The knife follows optical diagnosis.
Inject a durable cushion. Re-lift through a water-jet knife when the plane fades, instead of parking the knife for a separate injector.
Incise with a needle-type knife you already mark with.
Get a view before you get tired: clip-and-thread or double-clip rubber-band traction, or enter a pocket if the colon will not hold a flap.
Dissect in the submucosa you can see. An insulated-tip is a lateral sweep tool if that is your trained vector, not a mandatory second device.
If fibrosis kills the plane, stop adding pieces. That is salvage ESD, UEMR, or surgery, not another hopeful snare. See the UEMR article.

Three ESD levers: water-jet needle-type knife for injection without swaps, traction for a falling flap, pocket-creation for floppy or fibrotic colon. Hybrid snaring is not a shortcut. — Three technique levers after the indication is settled. Knife geometry is real. Traction and a pocket are what the RCTs moved.

What do the societies say?

Does DualKnife versus IT knife change the clock?

Needle-type knives (DualKnife, DualKnife J, FlushKnife) use a short retractable tip for marking, mucosal incision, and fine dissection. Insulated-tip knives (ITKnife2, ITknife nano) put a ceramic ball at the end so the cutting current is lateral. That anatomy is real. Treating it as the 2026 outcome driver is not.

NANO randomized 30-50 mm colorectal neoplasms to DualKnife J alone or DualKnife J plus ITknife nano [6]. Procedure time 111.8 vs 114.3 minutes. Dissection speed 22.9 vs 22.1 mm2/min. En bloc, R0, and adverse events were comparable. The rectum was faster with DualKnife J alone (24.7 vs 16.2 mm2/min). Adding the insulated-tip did not buy the hour the catalog implies.

What did move time is injecting through the knife. Takeuchi randomized colorectal ESD to Flexknife versus FlushKnife [7]. Mean time 87.3 vs 61.0 minutes. En bloc 88% vs 100%. Fewer device changes, fewer injections, less hyaluronate. DualKnife J already has a water-jet; that is why NANO could run DualKnife J as a single-device arm. Toyonaga's ball-tipped FlushKnife-BT is a hemostasis iteration of the same family, not a DualKnife vs IT trial [8].

When does traction actually help?

Traction is a view problem. Meta-analysis across organs: about 20 minutes shorter, slightly higher R0, fewer perforations, no bleeding difference [13]. Quote the organ, not the pooled headline.

Colorectum: Yamasaki's clip-and-thread RCT cut median time from 70 to 40 minutes and raised intermediate self-completion from 90% to 100% [10]. Bordillon's French DCT series (599 colonic lesions, mean 53 mm) reached en bloc 95.7% and R0 83.5% at 39.4 mm2/min, with 4.9% perforation [12]. That is a high-volume DCT shop, not a US community default.

Esophagus: CONNECT-E, tumors larger than 20 mm, traction 44.5 vs 60.5 minutes, no adverse events in the traction arm [11].

Stomach: CONNECT-G is the paper people misquote. Overall time 60.7 vs 58.1 minutes (not significant) [9]. Perforation 2.2% vs 0.3%. The win sat on the greater curvature of the upper or middle stomach (104 vs 57 minutes). Put floss-clip traction on the flap that falls into the lumen. Do not promise an hour off every antral ESD.

When do I make a pocket?

Pocket-creation keeps the scope in a submucosal tunnel so the colon's floppy wall cannot dump the flap. Yamashina's RCT did not make ESD faster. It made the assigned method finish: 93% vs 73% [14]. Takezawa's colonic series (rectum excluded) had higher en bloc and R0 with PCM and faster dissection speed [15]. A PCM meta found higher en bloc and R0, faster dissection, and less perforation, with the strongest signal in the colorectum [17].

Opening the pocket is still the grind. Morikawa added single-clip traction to PCM and raised dissection speed from 21.4 to 27.0 mm2/min without changing en bloc [16]. Pocket plus a clip is a finishing trick, not a new religion.

Why hybrid ESD is not the shortcut

Hybrid ESD (knife, then snare) looks like a way to skip the last hour. Fuccio's colorectal meta: hybrid R0 60.6% and en bloc 68.4%, versus standard ESD R0 82.9% overall and 71.3% outside Asia [18]. North American mixed-organ ESD already sits below Japanese long-term numbers (Draganov en bloc 91.5%, R0 84.2%, perforation 2.9% across organs; Ohata local recurrence 0.5% in expert Japan) [19] [20]. Adding a snare does not close that gap. If the lesion needed ESD for intact margins, finish ESD or stop and refer.

Pitfalls

Pitfall	Better move
Picking DualKnife vs IT knife as the strategy	NANO: adding ITknife nano did not shorten colorectal ESD. Learn a water-jet needle-type well.
Quoting CONNECT-G as a universal time win	Overall gastric time did not move. Quote perforation and the greater-curve subgroup.
Laundering Bordillon 39 mm2/min or Ohata 0.5% recurrence as your room	Those are expert DCT and expert Japanese follow-up. Draganov is the Western mixed-organ floor.
Hybrid snaring a possible T1 to "save time"	You destroy the staging specimen. Fuccio hybrid R0 60.6%.
Starting colorectal ESD because you own the knives	ESGE still defaults to EMR for most LNPCPs. Indication is the other article.
Circumferential incision into a floppy colon with no traction plan	Enter a pocket, or clip the flap, before the specimen folds over the plane.

Selected references

Questions doctors ask

Should I add an IT knife to DualKnife J for colorectal ESD?

Not for speed. NANO randomized DualKnife J alone versus DualKnife J plus ITknife nano for 30-50 mm colorectal lesions. Procedure time and dissection speed were the same. The rectum subgroup was faster without the IT knife. Learn one needle-type knife well, then add an insulated-tip only if that is your trained lateral vector.

Does traction always shorten gastric ESD?

No. CONNECT-G found overall time 60.7 versus 58.1 minutes. Traction did drop perforation from 2.2% to 0.3%, and it halved time on the greater curvature of the upper or middle stomach (104 vs 57 min). Use it where the flap falls into the lumen, not as a universal hour-saver.

When do I start a pocket instead of a full circumferential incision?

When the colon is floppy, fibrotic, or the scope position will not hold a flap. Yamashina: pocket-creation let operators finish with the assigned method 93% versus 73%. Takezawa: higher colonic en bloc and R0 versus conventional ESD in a large Japanese series. Opening the pocket is still the hard step; clip traction can help (Morikawa).

Can I switch to a snare at the end to save time?

Not if you needed ESD for staging. Fuccio: hybrid ESD R0 60.6% and en bloc 68.4%, versus standard ESD R0 82.9% (Western 71.3%). That is not Japanese R0 with a shortcut.

Is this the page that tells me to ESD every 20 mm colon polyp?

No. ESGE 2022 rec 12 and ESGE 2024 rec 5 still keep EMR as the colorectal default. This page is how to run ESD after that decision. Indication details live on the EMR vs ESD article.

Topics

For your teaching file

Save this article as a PDF

Drop your email and we'll open a print-ready version you can save as a PDF — and you'll start getting our weekly GI endoscopy newsletter.

← All articles