BARCELONA, Spain — Underwater endoscopic submucosal dissection (U-ESD), filling the lumen with water instead of conventional gas insufflation, was as effective for en bloc resection of colorectal neoplasia as gas ESD (G-ESD), results from a multicenter randomized trial showed.

It also shortened procedure time and accelerated dissection while reducing intraprocedural bleeding and smoke interference.

These findings position U-ESD as "a promising alternative to conventional gas-assisted ESD," combining comparable resection efficacy with "several procedural advantages," said Arianna Andreozzi, MD, a gastroenterology and digestive endoscopy resident at University Campus Bio-Medico, Rome. 

Andreozzi presented the study findings here at United European Gastroenterology (UEG) Week 2026. 

Taking ESD Underwater 

ESD is an advanced endoscopic technique that enables en bloc removal of superficial colorectal neoplasms for which resection using conventional techniques, such as endoscopic mucosal resection (EMR), is unlikely to succeed, Andreozzi explained. 

However, colorectal ESD remains technically demanding. Lesion location, size, and fibrosis can make dissection more difficult, and peristalsis and respiratory movements can affect knife stability. In addition, smoke produced during the procedure can obscure the operating field.

Gravity is another consideration with conventional gas-assisted ESD, where the patient is positioned "to place the lesion on the upper side of gravity," Andreozzi noted, adding that this can help open the mucosal flap and improve visualization of the dissection field. 

Filling the lumen with water instead of gas can address some of these challenges, she said, describing several advantages of using water, starting with "the floating effect of the lesion," whereby water provides natural traction and exposes the submucosal plane.

Andreozzi also highlighted improved visibility. "Water acts as a natural lens," she explained, increasing magnification during the procedure. The reduction in droplets and fumes also translates into "better visibility of the dissection field."

Buoyancy may additionally separate the layers of the colonic wall, reducing the need for repeated submucosal injections. Other proposed advantages include improved endoscope stabilization and a "heat-sink effect," whereby water dissipates thermal energy and may reduce thermal injury, she added.

Multicenter Randomized Comparison 

The prospective, open-label trial was conducted at five Italian tertiary referral centers between December 2024 and December 2025. Of 337 patients assessed for eligibility, 295 were randomized: 149 to U-ESD and 146 to conventional G-ESD.

Nine procedures, six in the U-ESD group and three in the G-ESD group, were interrupted because of suspected deep invasion. These cases were considered failures for the primary endpoint in the intention-to-treat analysis. The remaining 286 patients, 143 in each group, formed the per-protocol population.

The groups were generally comparable in age (67-68 years), sex distribution, lesion size (approximately 49 mm), morphology, and lesion location.

The primary endpoint was en bloc resection, with a prespecified noninferiority margin of 10 percentage points. Secondary endpoints included R0 resection, procedure time, dissection speed, visualization, intraprocedural bleeding, smoke interference, use of hemostatic forceps, and postprocedural adverse events. 

Noninferior Resection, Faster Procedure 

In the intention-to-treat analysis, en bloc resection was achieved in 95.3% of patients assigned to U-ESD compared with 97.3% assigned to G-ESD, an absolute risk difference of -2.0 percentage points (95% CI, -6.9 to 2.8), confirming noninferiority.

The finding was supported by the per-protocol analysis, in which en bloc resection was achieved in 99.3% of patients in both groups (risk difference, 0.0 percentage points; 95% CI, -3.2 to 3.2). R0 resection rates were also similar in the per-protocol analysis, at 95.1% with U-ESD and 95.8% with G-ESD.

The differences were more apparent in procedural performance. Among patients who completed their assigned procedures , median procedure time was 59 minutes with U-ESD compared with 70 minutes with G-ESD (P = .038). Median dissection speed was significantly faster, at 27.0 vs 21.4 mm²/min, respectively (P < .001). 

Less Bleeding, Less Smoke 

Major intraprocedural bleeding occurred in 19.6% of U-ESD procedures vs 35.0% of G-ESD procedures (P = .005); hemostatic forceps were required in 44.8% vs 61.5%, respectively (P = .006). 

One of the largest differences involved visibility. Smoke interference occurred in just 8.4% of underwater procedures compared with 48.3% of gas-assisted procedures (P < .001), "another important result" of the study according to Andreozzi. Vessel and cutting-plane visualization were also significantly better underwater. 

The procedural advantages did not appear to come at the expense of safety. Postprocedural bleeding occurred in 0.7% of patients after U-ESD and 4.2% after G-ESD, a difference that was not statistically significant. Delayed bleeding occurred in 0.7% in each group, and no perforations were reported in either group. Abdominal pain was similar at 15 and 30 minutes after the procedure but was significantly lower at 60 minutes following U-ESD. 

A Promising Alternative, but ESD Remains Difficult 

During the discussion, Jérémie Jacques, MD, of Limoges University Hospital, France, asked for further detail on the procedures.

"I congratulate you on conducting a randomized trial in a technically challenging field," he said, "but ask whether the conventional ESD procedures had incorporated techniques such as pocket creation, tunneling, or traction, which might influence procedural performance."

Andreozzi responded that these techniques could be used with either underwater or conventional ESD. "The real point is the procedural advantages," she said, emphasizing that ESD has "an important learning curve" and remains technically difficult regardless of whether water or gas is used.

She was also cautious about extending the findings beyond ESD. Asked whether the results could be extrapolated to other endoscopic resection procedures, including EMR, she replied: "I don't think so. Different procedures, different results."

Andreozzi reported no conflicts of interest related to the study. Jacques has reported consulting fees from Boston Scientific; training fees from Boston Scientific, Olympus, Erbe, Fujifilm, Pentax, and Lumendi; and research funding from Boston Scientific. He has also reported lecture fees from pharmaceutical and medical device companies.