Composite gangliocytoma/neuroma and neuroendocrine tumor (CoGNET) is a rare tumor composed of neuroendocrine epithelial cells, Schwannian spindle cells, and ganglion cells. Previously, this entity was referred to as gangliocytic paraganglioma; however, in the 2022 World Health Organization classification of digestive system tumors, it was reclassified and renamed CoGNET. This change reflects its characteristic triphasic histology composed of neuroendocrine epithelial cells, Schwannian spindle cells, and ganglion cells, distinguishing it from true paragangliomas in terms of cellular composition and lineage [1-3]. Most CoGNETs are located in the second portion of the duodenum or around the ampulla, and clinically they often present with abdominal pain, gastrointestinal bleeding, or are found incidentally [1,4].

CoGNET is a rare entity, and CoGNET arising in the ampullary/periampullary region is particularly uncommon [5,6]. Due to the limited number of reported cases in this location, there is little known about optimal treatment and surveillance strategies, and no clear standard has yet been established [1]. Recent multicenter retrospective studies have reported that patients with CoGNET resected via endoscopic papillectomy, transduodenal ampullectomy, or pancreatoduodenectomy had excellent outcomes, and even in cases with lymph node metastasis, observation without recurrence has been reported [7-9].

Meanwhile, underwater endoscopic mucosal resection (UEMR) has attracted attention as a promising treatment that ensures both safety and efficacy in the resection of duodenal lesions [10]. UEMR is a technique that allows lesion resection under water without submucosal injection. Its potential safety advantage is thought to arise from the heat-sink effect of the water-filled lumen, which may help reduce thermal injury to the deeper mural layers, particularly the proper muscle layer, especially in anatomically challenging areas such as the duodenum [11]. Especially in areas with thin walls and complex anatomy, such as the duodenum, UEMR is considered a safer and more efficient alternative compared to conventional endoscopic mucosal resection or endoscopic submucosal dissection [12].

Herein, we report a case of CoGNET in the second portion of the duodenum that was diagnosed by endoscopic ultrasound and imaging studies and safely resected en bloc using UEMR, with final pathological confirmation and follow-up. This case suggests that endoscopic resection of CoGNET using UEMR is feasible in clinical practice and may help establish treatment strategies for similar lesions in the future.

A 53-year-old man was referred to our hospital for evaluation of a subepithelial tumor located in the second portion of the duodenum, adjacent to but not directly involving the ampulla, which was first identified during a screening endoscopy (Fig. 1A). No prior endoscopic records were available; therefore, interval growth or the exact duration of the lesion could not be assessed. Endoscopic ultrasound revealed a hypoechoic, homogeneous solid mass measuring approximately 13×10 mm located in the submucosal layer (third layer) with clear borders and no mucosal defect (Fig. 1B). Computed tomography (CT) showed a 1.5-cm enhancing polypoid lesion in the second portion of the duodenum. Laboratory tests prior to the procedure showed hemoglobin 16.0 g/dL, white blood cell count 3510/μL, platelet count 172000/μL, aspartate aminotransferase 21 IU/L, alanine aminotransferase 29 IU/L, and total bilirubin 0.8 mg/dL.

The UEMR procedure was performed under conscious sedation. Midazolam 5 mg, algiron 5 mg, and pethidine 25 mg were administered as premedication. An electrosurgical unit (VIO 300D; ERBE Elektromedizin) was used in Endocut Q mode (effect 2, duration 2, interval 3), and a 15-mm oval snare was applied for lesion resection. And the patient was not taking any thrombolytics or antiplatelet agents. Using the underwater technique, the lesion was resected en bloc without submucosal injection by filling the lumen with water (Fig. 2). The total procedure time was approximately 12 min. Immediately after resection, exposed vessels and minor bleeding were observed on the resection surface and were treated with electrocoagulation using coagulation forceps. Subsequently, endo-scopic clip closure was performed to achieve hemostasis and secure the mucosal defect. No perforation or residual lesion was observed. The patient was kept nil per os and hospitalized for observation for one day post-procedure. No complications such as abdominal pain or fever occurred, and the patient was discharged after resuming oral intake.

The resected specimen was a clearly defined solid mass measuring 2.0×1.5×1.2 cm confined to the mucosal and submucosal layers. Hematoxylin and eosin (H&E) staining revealed a triphasic composition of neuroendocrine epithelial cells, Schwannian spindle cells, and ganglion cells (Fig. 3). Immunohistochemical staining showed positivity for chromogranin A, synaptophysin, CD56, neuron-specific enolase (NES), and S-100, confirming the presence of neuroendocrine cells, spindle cells, and ganglion cells (Fig. 4). In particular, ganglion cells appeared as pleomorphic cells with prominent nucleoli and abundant cytoplasm on H&E staining. The Ki-67 labeling index was less than 1% on quantitative analysis, and the mitotic rate was 0/10 HPF. No lymphovascular or perineural invasion or necrosis was observed. Based on these histological and immunohistochemical findings, a final diagnosis of CoGNET was made.

Follow-up chest CT showed no pulmonary or lymph node metastasis, and follow-up endoscopy at 3 months revealed only a scar at the resection site without any specific abnormal findings. Biopsy also showed no residual lesion. Given the absence of established follow-up guidelines for CoGNET, annual endoscopic surveillance and abdominal CT for up to 5 years were planned based on individual risk assessment.

The average size of CoGNET is approximately 1–2 cm, and most are confined to the mucosa or submucosa [8]. Unlike typical NETs, CoGNETs are multi-lineage composite tumors consisting of neuroendocrine epithelial cells, Schwannian spindle cells, and ganglion cells, exhibiting heterogeneous cellular components. Each component shows specific immunohistochemical expression patterns [13-15]. Chromogranin A, synaptophysin, and CD56 indicate neuroendocrine epithelial cells, S-100 indicates Schwannian spindle cells, and neurofilament protein is used as a marker for ganglion cells [14,15]. Most CoGNETs have a very low Ki-67 labeling index and rarely show mitosis or necrosis, suggesting that they are benign or very lowgrade malignant tumors [13]. In this case, all three cell types were identified, leading to the diagnosis of CoGNET, with immunohistochemical positivity for chromogranin A, synaptophysin, CD56, NSE, and S-100, Ki-67 index <1%, and mitotic rate 0/10 HPF. These pathological characteristics of CoGNET are associated with a low metastatic rate and favorable prognosis, suggesting that local resection alone is sufficient for treatment, although surgical treatment may be considered depending on tumor location, depth of invasion, and lymph node metastasis status [7]. In this case, the tumor was confined to the mucosa and submucosa, with no lymphovascular invasion, perineural invasion, or involvement of the resection margins, and complete resection was achieved pathologically.

UEMR has recently attracted attention as a therapeutic endoscopic resection method [10]. UEMR allows for lesion resection using the buoyancy of water without saline injection, providing both clear visualization and safety [11]. This technique has been widely used to enhance the safety and completeness of EMR in anatomically challenging areas such as the duodenum, which has thin walls [12]. Literature reviews to date have reported numerous cases and studies applying UEMR mainly to superficial non-ampullary duodenal epithelial tumors, showing high resection success rates and low complication rates, especially for lesions ≤20 mm [10-12]. However, there have been no reports or studies of UEMR being applied to subepithelial tumors arising in the ampullary or periampullary region. This is because these areas are adjacent to critical pancreatobiliary structures such as the common bile duct, pancreatic duct, and ampulla, and due to their thin walls and difficult visualization, endoscopic resection in these areas is technically challenging and carries a high risk of complications.

Notably, previous literature has only reported treatment of ampullary CoGNET with endoscopic papillectomy, transduodenal ampullectomy, or pancreatoduodenectomy, and there have been no reports of periampullary CoGNET resected by UEMR [1]. Most previously reported underwater resections in the ampullary or periampullary region have focused on epithelial ampullary lesions requiring papillectomy, rather than subepithelial tumors originating from the submucosal layer. This case is the first report of en bloc resection of periampullary CoGNET using UEMR, suggesting its potential as a treatment option for CoGNET in this location. No recurrence was observed during follow-up, indicating that with accurate lesion evaluation, appropriate technique, and visualization strategy, UEMR can be a feasible treatment option even in such anatomically limited areas. Although no recurrence was observed during short-term follow-up, longer surveillance is planned. Given the absence of established follow-up guidelines for CoGNET, annual endoscopic surveillance and abdominal CT for up to 5 years are planned based on individual risk assessment. This case may serve as meaningful evidence supporting the technical safety and expandability of UEMR in similar lesions in the future, and additional prospective studies and long-term clinical outcome data are needed in this field.