Purastat Fails to Lower Delayed Bleeding After Large-Field EMR
A self-assembling haemostatic gel did not reduce delayed bleeding after endoscopic mucosal resection of large flat lesions in the duodenum and colorectum, despite a biologically plausible rationale and broad multicentre use. In the PURPLE trial, prophylactic gel application after EMR produced no significant improvement over no prophylaxis, and the study was stopped early for futility after interim review.
Delayed hemorrhage after EMR remains one of the most vexing postprocedural complications in therapeutic endoscopy. The bleeding risk does not end when the snare comes away; it often emerges days later, after thermal injury, vessel slough, and ongoing mucosal repair have had time to unfold. A topical agent that could reliably coat the resection bed and stabilize surface clot formation would have obvious appeal, especially in large, flat lesions where the exposed submucosal bed can be extensive.
Trial Design in High-Risk Duodenal and Colorectal Lesions
This prospective randomized trial enrolled patients undergoing hot-snare EMR of flat lesions in the duodenum measuring at least 10 mm and in the colorectum measuring at least 20 mm across 15 German centers. Patients were assigned after the procedure to receive Purastat or no prophylaxis. Prophylactic clip closure was not permitted, although selective clipping or coagulation could be used before randomization to control intraprocedural bleeding or treat visible vessels.
The primary endpoint was delayed bleeding within 30 days. The study design is clinically important because it tested the gel in a setting where bleeding risk is real, relevant, and common enough to challenge any prophylactic intervention. The enrolled cohort also reflects daily endoscopic practice: a mix of colorectal and duodenal resections, with lesion sizes large enough to matter and with hemostatic maneuvers already in use before the randomized intervention was even assigned.
Baseline characteristics were balanced between groups, including age, sex, comorbidities, and lesion features. Preventive measures such as selective clipping or coagulation were used before randomization in 51.9% of cases, with no difference between study arms. In other words, the comparison was not distorted by a lopsided distribution of obvious bleeding risks.
Delayed Bleeding Rose Despite the Gel
The trial’s primary endpoint was analyzed in 232 patients, including 208 colorectal and 26 duodenal lesions. Delayed bleeding occurred in 14 patients, or 11.7%, in the Purastat group and in 7 patients, or 6.3%, in the control group. The difference did not reach statistical significance, with a p value of 0.227.
No separation emerged in the colorectal or duodenal subgroups. The signal was flat across the anatomical settings most likely to have justified a prophylactic approach in the first place.
The direction of the numerical difference is hard to ignore. A product intended to reduce post-EMR hemorrhage did not produce a lower bleeding rate and, on the crude counts, bleeding was actually numerically more frequent in the gel arm. The finding does not prove harm, but it does erase the expectation that surface application alone can overcome the mechanical and vascular forces driving delayed postresection bleeding.
Why the Biology Did Not Translate Into Clinical Protection
The mechanism behind a self-assembling haemostatic gel is attractive: rapid adherence to tissue, formation of a physical barrier, and local support of clot stabilization at the resection margin. In theory, this could be especially useful in friable mucosa or on difficult duodenal beds where thermal injury and peristalsis make durable hemostasis difficult. In practice, delayed bleeding after EMR is often caused by vessels that reopen after the immediate procedure, not just by superficial oozing at the end of the case.
A topical gel may cover the surface without fully protecting deeper submucosal vessels that later become exposed as necrotic tissue sloughs. The result is a mismatch between what the product can reasonably do at the bedside and what delayed bleeding requires biologically. Thermal injury, vessel caliber, lesion size, location, and the underlying vascular anatomy likely outweigh the short-lived surface effect of a gel alone.
This is particularly relevant in the duodenum, where the wall is thin, access is constrained, and postprocedural bleeding can be difficult to manage. Colorectal resections, meanwhile, often involve larger lesions and broader raw surfaces, creating a different but equally challenging hemostatic environment. A single prophylactic strategy has to overcome both settings, and this trial suggests the gel does not do so.
Early Stopping for Futility Limits the Promise of Expansion
The trial was stopped early after interim analysis for futility. That decision gives the results practical weight: once an intervention fails to show a meaningful likelihood of benefit, continuing enrollment can expose additional patients without strong justification. Early termination also means the data set is smaller than originally planned, which can reduce precision, but the observed event rates still failed to favor the active intervention.
The final analysis therefore leaves a clear bedside message for endoscopists performing EMR of large flat lesions in these locations. Routine prophylactic application of this haemostatic gel after resection should not be expected to reduce delayed bleeding. Selective hemostasis during the procedure remains important, but the post-resection coating strategy did not add measurable protection in this randomized setting.
The practical implication is not that hemostasis after EMR is solved, but that the answer is not a simple topical sealant applied at the end of the case. Better strategies may need to target deeper vessel control, more refined lesion selection, or anatomically tailored approaches rather than assuming a universal surface barrier can suppress delayed hemorrhage.
What This Means for Therapeutic Endoscopy Practice
For endoscopists, the PURPLE trial narrows the field of plausible prophylaxis after EMR of large flat duodenal and colorectal lesions. A product with a strong mechanistic story did not change the clinically meaningful endpoint that matters most to patients and procedural teams: delayed bleeding within 30 days. The absence of benefit across both major anatomical subgroups strengthens the message.
The study also reflects the reality that bleeding prevention in advanced endoscopy is often multifactorial. Selective clipping and coagulation were already being used in more than half of cases before randomization, yet delayed bleeding still occurred. Surface hemostatic gels may have a role in other contexts, but this trial does not support routine use for delayed bleeding prevention after these EMR procedures.
The question now shifts away from whether the gel works in principle and toward which lesion types, if any, might still benefit from different prophylactic methods. For the population studied here, the randomized data do not support adding the gel as a standard step after EMR.