Role of endoscopic ultrasound in the treatment of pancreatic neuroendocrine tumors: Lights and shadows of endoscopic ultrasound-guided radiofrequency ablation

Abstract

Pancreatic neuroendocrine tumors (pNETs) are a heterogeneous group of pancreatic neoplasms that originate from the endocrine cells of the pancreas, whose prevalence and incidence are constantly increasing worldwide. Based on current knowledge of their natural history, pNETs can be divided into functioning pNET and nonfunctioning pNET tumors, characterized by hormone hypersecretion, which results distinct clinical presentations. Treatment options include observation, medical or surgical therapy, and the choice depends on various factors such as staging and grading of the pancreatic lesion and the presence of a specific hormonal syndrome. Surgical resection has long been considered the gold standard for treatment, with related risks of morbidity and mortality. Endoscopic ultrasound (EUS)-guided radiofrequency ablation (RFA) plays a crucial role as minimally invasive procedure for loco-regional treatment of pNETs in selected patients, showing promising results in terms of clinical outcome. EUS-RFA causes a coagulative necrosis with minimal damage to surrounding tissue, allowing for local ablation. This review summarizes the most recent evidences on the use of EUS-RFA as local ablation therapy describing the main endoscopic steps and providing a critical overview of patient selection criteria, side effects, and long-term outcomes.

Key Words: Pancreatic neuroendocrine tumor; Endoscopic ultrasound; Radiofrequency ablation; Loco-regional ablation; Guided radiofrequency ablation

Core Tip: In recent years, endoscopic ultrasound (EUS)-guided radiofrequency ablation (RFA) has become a subject of great interest in Digestive Endoscopy as locoregional treatment for pancreatic neoplastic lesions, particularly for pancreatic neoplastic cysts, pancreatic functional and non-functional neuroendocrine neoplasms and pancreatic ductal adenocarcinoma. The EUS-RFA procedure can potentially ablate these lesions locally in a minimally invasive manner with a very good safety profile. Our focus was to highlight the role of EUS-RFA in the treatment of pancreatic neuroendocrine tumors, based on available evidence, and describe the current potential benefits and limitations of this approach.

INTRODUCTION

Pancreatic neuroendocrine tumors (pNETs) are a subgroup of pancreatic neoplasms that arise from the endocrine cells of the pancreas and express neuroendocrine markers such as synaptophysin and chromogranin A, with an approximate incidence of 0.5/100000 inhabitants[1,2]. Based on their natural history, pNETs can be divided into two main groups: Functioning pNET (F-pNET) and non-functioning pNET (NF-pNET). The majority of pNETs are non-functioning, ranging from 50% to 85%, with an incidence that is rising, as recently reported at 0.48/100000 new cases per year[3]. NF-pNETs are usually asymptomatic because they often secrete inactive hormones in the absence of symptoms and, therefore, could be diagnosed at later stages with a more advanced disease[1,3]. Interestingly, due to the widespread availability of various diagnostic modalities and advancements in abdominal imaging techniques, such as endoscopic ultrasound (EUS), early detection of NF-pNETs can be achieved. Otherwise, F-pNETs are less common ranging between 10% and 30% of all pancreatic neuroendocrine neoplasms[4]. They secrete elevated levels of bioactive hormones, resulting in various clinical syndromes, including insulinoma, gastrinoma as well as the rare glucagonoma, vasoactive intestinal polypeptide-secreting tumor, adrenocorticotropic hormone-secreting tumor, somatostatinoma and pNET, which cause carcinoid syndrome.

The management of pNETs depends on grading (grade 1- grade 2- grade 3), clinical staging of the pancreatic lesion, and presence of clinical manifestations related to hormonal hypersecretion. The main therapeutic option for larger and more aggressive tumors has long been represented by surgical resection, although unfortunately, it is associated with high rate of morbidity and mortality. On the other hand, for small and asymptomatic neuroendocrine tumors with a low tumor grade (grade 1) a surveillance strategy is sometimes suggested, due to an indolent course of disease with a low risk of progression[3,5].

The goal of the treatment for both F-pNETs and NF-pNETs is complete tumor ablation. For F-pNETs, the primary aim is to achieve complete resolution of symptoms, inducing necrosis of the secreting cells to reduce the hormonal hypersecretion[6].

International neuroendocrine tumor community identifies a “grey zone”[5], which includes patients with NF-pNETs < 20 mm and grading tumor of grade 1-grade 2, where the choice of treatment is not standardized, and it can be influenced sometimes by the patient’s preference. In this context, EUS-guided radiofrequency ablation (RFA) appears to be a promising alternative for minimally invasive treatment in selected cases, with high efficacy and an acceptable safety profile, particularly in patients unfit for surgery[7]. Unlike NF-pNETs, this technique showed good results for the small functioning pNETs (< 20 mm)[3].

The choice of optimal patient selection remains a crucial topic, with careful consideration of tumor size, location, functional status and proximity to vital structures. Multidisciplinary evaluation is often essential to tailor treatment plans and ensure favorable outcomes. Moreover, long-term data regarding the effectiveness and safety of EUS-RFA are still emerging, necessitating ongoing research and prospective studies to better define its role into the therapeutic algorithm for p-NETs.

EUS-RFA: TECHNIQUE AND DEVICE

RFA is a technique that causes an indirect tissue damage with a high frequency alternating current. This current produces a high thermal energy between 60 °C and 100 °C inducing protein denaturation of the cells, followed by coagulative necrosis[8].

Patient is placed in left lateral decubitus, performing the procedure under general anesthesia or deep sedation. A diagnostic EUS is performed to localize the target lesion and to determine the optimal operative window. Color Doppler imaging is necessary to exclude the interposed vascular structures; thus, the device should avoid vessels or ductal structures during the procedure[6,9]. According to the size of the target tumor, an RFA device is selected and connected to a radiofrequency generator that regulates power and impedance. The device consists of a 19 gauge (G) monopolar electrode with an insulated trocar in various lengths of 5 mm, 7 mm, and 10 mm. A connected water pump enables the circulation of a saline solution into the electrode, thereby protecting adjacent tissue[6].

The device is inserted into the lesion trans-gastric or trans-duodenal, based on the lesion’s site. After setting up the generator, the ablation is performed under EUS control. Treatment success is assessed by the evidence of the hyperechoic bubbling effect within the tumor lesion. Often, the device is repositioned one or more times into another part of the lesion to achieve a complete tumor ablation. Furthermore, the use of an ultrasound contrast agent is indicated to evaluate, after the procedure, the residual vital tissue. Three types of RFA devices have been used in clinical practice but really one of them is no longer available on the market (Habib^TM EUS-RFA catheter)[9,10].

EUSRA^TM needle

It is a 19 G monopolar electrode for RFA; its upper end is connected via a wire to radiofrequency generator (VIVA Combo^TM RF Generator; STARmed, TaeWoong Medical, South Korea). An important component of the radio frequency generator system is an integrated cooling system (VIVA pump) that, by pumping cold saline solution into the needle, allows continuously cooling and prevents tissue carbonization[11,12].

Habib^TM EUS RFA

It is a flexible catheter with monopolar electrode in its distal end that can be inserted within fine needle aspiration needle (19 G or 22 G). The Habib probe is connected to an electrosurgical generator but without an internal cooling system. After needle is placed inside the target lesion under endosonographic guidance, the stylet is removed, and the probe is advanced into the needle until contact with the tissue is indicated by tasted resistance. Thus, to expose the tip of the probe, the endoscopists briefly retracts. Habib™ EUS RFA catheter is no longer available; therefore, it cannot be purchased[13].

Hybrid-Therm^® Probe HTP

Hybrid-Therm^® Probe HTP (ERBE, Tübingen, Germany) is a bipolar probe for RFA connected to a current generator (VIO 300D, ERBE Elektromedizin GmbH, Germany) with an integrated carbon dioxide cooling system. The innovation is the integration of cytotechnology, which enables the reduction of generator power, thereby decreasing the risk of tissue damage and making RFA a more effective procedure[14,15].

F-PNET SYNDROMES

Insulinomas and gastrinomas represent the most common functioning pancreatic neuroendocrine tumors. Based on current knowledge, surgical resection has long been considered the gold standard for treatment of localized pancreatic insulinomas and gastrinomas[16].

According to recent clinical guidelines of the European Neuroendocrine Tumor Society, EUS-RFA could be performed for localized insulinomas smaller than 2 cm in selected patients such as those considered unfit for surgery or in patients who prefer a minimally invasive approach[4]. The efficacy of EUS-RFA in treating pancreatic gastrinomas, unlike insulinoma, is limited to a small number of case reports, which demonstrate a poor clinical success rate probably due to a high rate of nodal metastasis. Moreover, due to the high risk of perforation, this procedure is not recommended in duodenal gastrinomas (Figure 1)[4].

Figure 1 Algorithm for the management of functioning-pancreatic neuroendocrine tumors. pNET: Functioning pancreatic neuroendocrine tumor; RFA: Radiofrequency ablation.

Twenty-nine studies, including 230 patients with a total of 235 lesions, diagnosed as F-pNET assessed the role of EUS-RFA for the treatment of F-pNET between 2016 and 2025 (Table 1). The endpoints evaluated in the selected studies were technical and clinical success of EUS-RFA in F-pNET and the procedure-related adverse events. Authors defined technical success as successful and accurate placement of a dedicated needle into the pancreatic neoplasm, allowing for local ablation, while clinical success was based on the remission or improvement of clinical manifestations after the last EUS-RFA session. The recurrence rate was defined as a reappearance of an enhancing lesion and/or clinical symptom during the follow-up of a patient who had achieved a complete response. Another endpoint was to assess the peri-procedural and post-procedural safety of EUS-RFA reporting both major and minor adverse events.

Table 1 Studies of endoscopic ultrasound-guided radiofrequency ablation on functioning-pancreatic neuroendocrine tumors.

Ref.	Year	Study design	Number of patients/F-pNET	Tumor Size (mean, mm)	Location	Follow-up (months)1	Clinical success (%)1	Adverse events	Recurrence
Lakhtakia et al[36]	2016	Case series	3/3	17.3	Head: 3	12	100	0
Waung et al[37]	2016	Case report	1/1	18	Head	10	100	0
Bas-Cutrina et al[38]	2017	Case report	1/1	10	Body	10	100	NR
Goyal et al[39]	2017	Case series	1/1	7.5	Body	NR	100	0
Thosani et al[40]	2018	Retrospective, multicenter	3/3	23	NR	5	100	Not specified2
de Mussy et al[41]	2018	Case report	1/1	12	Body	2	100	NR
Choi et al[18]	2018	Prospective, monocenter	1/1	12	Head	13	100	0
Saurabh et al[42]	2019	Case report	1/1	17	Head	6	100	NR
Oleinikov et al[22]	2019	Retrospective, multicenter	7/9	14.8	Head: 7; body: 2	9.7	100	0
Choi et al[23]	2020	Prospective	1/1	12	Head	NR	100	0	1 (at 19 months)
Borwn et al[43]	2020	Case report	1/1	18	Head	8	100	0
Jonica and Wagh[44]	2020	Case report	1/1	22	Body	6	100	0
Furnica et al[45]	2020	Retrospective, monocenter	4/4	12	Head: 2; body: 1; tail: 1	22	100	2 acute pancreatitis
Kluz et al[46]	2020	Case report	1/1	9	Head	NR	NR	0
Lakhtakia et al[47]	2020	Prospective, monocenter	10/13	11.5	NR	10-64	100	1 acute pancreatitis; 1 abdominal pain
de Nucci et al[30]	2020	Prospective, monocenter	5/5	12.8	Body: 3; tail: 2	12	100	0
Younis et al[25]	2022	Prospective, monocenter	1/1	8	Tail	7	100	NR
Rossi et al[32]	2023	Prospective, monocenter	3/3	16	Head: 3; body: 3; tail: 4	22	100	1 intra-procedural bleeding
Chang et al[48]	2022	Case report	1/1	12	Head	18	100	1 abdominal pain
Marx et al[34]	2022	Prospective, multicenter	7/7	13.3	Head: 1; body: 6	21	85.7	3 NR; 1 death
Figueiredo Ferreira et al[27]	2022	Prospective, multicenter	13/13	14.4	NR	12	100	Not specified2
Rizzatti et al[29]	2023	Prospective, multicenter	30/30	12.1	Head: 15; body: 10; tail: 5	12	100	NR
Gugger et al[49]	2023	Case report	1/1	9	Tail	10	100	0
Borrelli de Andreis et al[50]	2023	Retrospective, monocenter	10/10	11.9	Head: 3; body: 3; tail: 4	19.5	100	2 abdominal pain
Crinò et al[16]	2023	Retrospective, multicenter	89/89	13.4	Head: 34; body: 39; tail: 16	23	95.5	8 acute pancreatitis, 5 abdominal pain, 1 spleen hematoma, 1 diabetes	16.9% at 9.5 momths
Napoléon et al[17]	2023	Retrospective, multicenter	16/16	15	NR	13	87.5	Not specified2
Biermann et al[7]	2024	Case series	3/3	14	Head: 2; tail: 1	12.7	100	1 case of melena
Lesmana[28]	2024	Case series	3/3	27.7	Head: 2; body: 1	6-12	100	0
Okasha et al[19]	2025	Prospective, multicenter	11/11	17.4	Head: NR; body: NR; tail: NR	12 ± 12.3	90.9	4 acute pancreatitis

¹In case of multiple types of pancreatic lesions, the overall data were indicated.

²It was not possible to verify in which patients the reported adverse events occurred.

F-pNET: Functioning pancreatic neuroendocrine tumor; NR: Not reported.

The technical success rate was achieved in 100% of cases across all studies; only a retrospective multicenter study conducted by Napoléon et al[17], including 16 F-pNET, showed 97% of technical success rate. Twenty-four out of 28 studies described a 100% clinical success rate, while the remaining four studies showed a clinical success rate ranging from 85% to 96%. Only one study that evaluated the highest number of patients (89 patients) reported a clinical success rate of 95.5%[16]. Adverse events occurred in 29/229 (0.13%), with a single case of death.

Unfortunately, data from ten papers, including different types of lesions (e.g., pancreatic cystic lesion, ductal adenocarcinoma and pNET), were ruled out (81 patients) because it was not possible to verify in which patients the reported adverse events occurred. Two studies reported recurrence after RFA treatment; Choi et al[18] described one case which occurred 19 months later, after EUS-RFA, while Crinò et al[16] found a 16.9% recurrence within 1 year after the procedure.

Furthermore, the number of RFA sessions for any pancreatic neoplasm was evaluated. In most cases, a single RFA session was performed for each lesion. In 7 studies, some lesions were subjected to more than one session, and in only one case, multiple lesions in the same patient were treated in the same session. The number of RFA sessions was not reported in two studies. The time required for applications, to achieve a complete ablation of the mass was expressed in terms of seconds ranging, in some studies, from 5 seconds to 50 seconds. Eleven studies reported the mean application time, with a mean of 40.7 seconds. The shortest average time was 10 seconds, while the longest average time was 120 seconds. In 13 studies, it was not expressed.

In a recent prospective multicenter study conducted by Okasha et al[19], 11 patients with pancreatic insulinomas underwent EUS-RFA. The mean size was equal to 17.4 mm. A complete clinical success was observed in 10 out of 11 patients (91%). All patients with insulinoma achieved a complete cure (8 patients in 1 session, 2 in 2 sessions, and 1 in 3 sessions). No major post-procedural adverse events occurred, demonstrating once again that the EUS-RFA treatment was safe and feasible.

NF-PNET

NF-pNETs represent the majority of pNETs, ranging from 50% to 85%. They are most commonly identified in the pancreatic head, and they tend to present more advanced disease at diagnosis than F-pNETs due to the higher risk of aggressive behavior, local invasion, lymph node involvement and liver metastasis. The management of NF-pNETs depends on various factors including tumor size, grading, lymph node involvement, patient’s comorbidities and preferences. Currently, according to recent clinical guidelines from the European Neuroendocrine Tumor Society, surgical resection is consistently recommended for all NF-pNETs larger than 2 cm, in the case of main pancreatic duct involvement with dilation and pancreatic lesions with elevated Ki-67 proliferation index (grade 2-grade 3)[3].

In patients with NF-pNETs whose size is between 10 mm and 20 mm, surgery is typically indicated, especially with main pancreatic duct dilation and /or high Ki67 proliferation index. In contrast, in the cases of low-grade NF-pNETs (Ki-67 < 5%) with no main pancreatic duct dilation, an active clinical and radiological surveillance could be an option. A conservative approach through active surveillance is generally still recommended for small and indolent non-functioning pancreatic lesions (10 mm). In this context, for all NF-pNETs < 2 cm, EUS-RFA may be considered, after multidisciplinary discussion, in selected patients such as those elderly or unfit for surgery with significant comorbidities; however, in the literature, real data demonstrate the effectiveness of the RFA in the treatment of these neoplasms are still lacking, limited to a small number of cases[3,11]. In NF-pNETs less than 2 cm with low level of Ki-67 index and /or no main pancreatic duct dilation surgery could be considered depends on lesion localization and patients’ comorbidities (Figure 2).

Figure 2 Algorithm for the management of non-functioning-pancreatic neuroendocrine tumors. NF-pNET: Non-functioning-pancreatic neuroendocrine tumors; RFA: Radiofrequency ablation; MDT: Multidisciplinary team; MPD: Main pancreatic duct; G: Grade.

Thirteen studies, including 177 lesions in 120 patients, assessed the role of EUS-RFA for the treatment of NF-pNET between 2016 and 2025 (Table 2). Only in a retrospective study, the number of patients with NF-pNETs was not reported[17]. As previously described, the endpoints evaluated were the technical success rate, radiological response and adverse events rate.

Table 2 Studies of endoscopic ultrasound-guided radiofrequency ablation on non-functioning-pancreatic neuroendocrine tumors.

Ref.	Year	Study design	Number of patients/NF-pNET	Tumor size (mean, mm)	Location	Follow-up (months)1	Radiologic response	Adverse events
Rossi et al[20]	2014	Prospective monocenter	1/1	9	Head	34	CR	NR
Armellini et al[51]	2015	Case report	1/1	20	Tail	NR	CR	0
Pai et al[6]	2015	Prospective, multicenter	2/2	27.5	Head; body	3-6	NR	0
Choi et al[18,23]	2018, 2020	Prospective, monocenter	13/13	18.1	Head: 5; body: 7; tail: 1	28	CR: 9/13 (69.2%); PR: 3/13 (23.1%); NoR: 1/13 (7.7%)	Not specified2
Oleinikov et al[22]	2019	Retrospective, multicenter	11/16	14.2	Head: 8; body: 6; tail: 2	8.9	CR: 8/11 (72.2%); PR: 1/11 (9.1%); 2 missing data	2 acute pancreatitis
Barthet et al[21,26]	2021, 2019	Prospective, multicenter	12/14	13.4	Head: 3; body: 6; tail: 5	45.6	CR: 12/14; (85.7%); NoR: 2/14 (14.3%)	Not specified2; 1 late recurrence
de Nucci et al[30]	2020	Prospective, monocenter	5/6	16	Head: 3; body: 2; tail: 2	12	CR: 5/5 (100%)	2 abdominal pain
Lesmana[28]	2024	Case report	1/1	35	Head	36	CR only of the solid part, no CR of the cystic part	0
Younis et al[25]	2022	Prospective, monocenter	6/6	8.9	Head: 2; body: 4	7	CR: 4/6 (66.7%); NoR: 2/6 (33.3%)	Not specified2
Marx et al[34]	2022	Prospective, multicenter	27/27	14	Head: 8; body: 8; tail: 11	15.7	CR: 25/27 (92.6%); NoR: 2/27 (7.4%)	4 acute pancreatitis
Figueiredo Ferreira et al[27]	2022	Prospective, multicenter	9/10	14.4	NR	9.6	CR: 6/11 (54.6%); PR: 3/11 (27.2%); NoR: 2/11 (18.2%); F-pNETs + NF-pNETs	Not specified2
Rizzatti et al[29]	2023	Prospective, multicenter	32/32	16.8	Head: 8; body: 15; tail: 9	12	CR: 24/24 (100%); 8 missing data	Not specified2
Napoléon et al[17]	2023	Retrospective, multicenter	NR/48	15	NR	13	CR: 33/48 (71.7%); PR: 12/48 (26.1%); NoR: 1/48 (2.2%); 2 missing data	Not specified2

¹In case of multiple types of pancreatic lesions, the overall data were indicated.

²It was not possible to verify in which patients the reported adverse events occurred.

NF-pNET: Nonfunctioning pancreatic neuroendocrine tumor; CR: Complete response; PR: Partial response; NoR: No response; F-pNET: Functioning pancreatic neuroendocrine tumor; NR: Not reported.

We observed an excellent technical success rate of EUS-RFA for NF-pNETs in 99.7% of all studies assessed (120 patients). Most pancreatic lesions were localized more frequently at the level of the pancreatic head, and their mean tumor size was less than 2 cm, except for two cases, where the lesions were greater than 2 cm, respectively 3 cm and 4 cm, due to high surgical risk and refusal of surgical intervention. The technical success rate was 100%, with no adverse events reported Table 2.

The first experience with EUS-RFA in NF-pNETs was conducted by Rossi et al[20] in 2014. They published a prospective single-center study enrolling 10 patients with histologically diagnosed pNETs (3 F-pNETs and 7 NF-pNETs). Only a case of non-functioning pancreatic lesion (tumor size 9 mm) underwent EUS-RFA endoscopically (7 percutaneously and 2 intraoperatively) with complete ablation defined as the absence of enhancing tissue on the tumor site. No recurrences occurred during the follow-up except for one case described by Barthet et al[21] after 3 years from the procedure.

Radiologic response was defined in a different manner among authors. The “complete response” was defined clearly as the absence of the pancreatic lesion or complete necrosis with no signs of residual vital tissue after the treatment showed by an imaging technique, while “partial response” and “no response” were reported with different definitions by the authors, restricting the comparability of the results.

According to Choi et al[18], Oleinikov et al[22], Choi et al[23], and Fritsch et al[24], the radiologic response was simply differentiated into complete or incomplete based on the disappearance or persistence of residual vital tissue at the tumor site. In contrast, other authors gave a quantitative definition of the radiological response. According to Napoléon et al[17], no response was defined as a volume reduction of pancreatic neoplasms < 75% while partial response was a volume reduction between 75% and 95%. Younis et al[25], Barthet et al[21,26], and Figueiredo Ferreira et al[27] explained that a reduction in the lesion volume < 50% corresponded to a no response and a reduction of between 50% and 100% was related to a partial response.

Some studies described that complete response rate was variable ranging from 54.6% to 92.6%[17,18,21-26], while others reported a 100% of complete response[20,28-30]. To date, the incidence of adverse events closely associated with NF-pNETs, has been evaluated only in some papers; data from seven studies, including a total of 73 patients, were excluded because it was not known which kind of lesion was related to adverse events. However, on a total number of residual 47 NF-pNETs patients, 8 adverse events occurred with no cases of mortality reported.

DISCUSSION

EUS-RFA is a promising minimally invasive technique for small and low-grade pancreatic neoplasms, whose spread has been increasing over the last few years, compared to surgical treatment and active surveillance. Our review aims to assess the current potential benefits and limitations of the EUS-guided loco-regional ablation technique in treating pancreatic neuroendocrine neoplasms based on available literature and clinical practice, highlighting the “lights and shadows”.

Many studies reported a very high technical and clinical success rate for the treatment of small F-pNET (insulinoma < 2 cm), demonstrating that EUS-RFA is a potentially curative option. According to European Neuroendocrine Tumor Society guidelines, small localized insulinoma and gastrinomas (< 2 cm) should be treated differently. In this context, recent prospective series and meta-analyses have demonstrated a high clinical success rate for patients with localized insulinomas who underwent RFA (86%-96%)[16,31]. On the other hand, there are very few reported cases of localized pancreatic gastrinomas with a limited clinical success rate. Therefore currently, this technique is not indicated for these patients and mostly in cases of duodenal gastrinomas due to the high risk of perforation. Therefore, EUS-RFA cannot be considered a standard curative option for all F-pNETs smaller than 2 cm[4]. Thus, it still represents a shadow in the treatment of gastrinomas.

Similarly, the technical success rate for all small NF-pNETs was high, approximately 100%. NF-pNETs (< 2 cm) treated with radiofrequency have shown promising results; However, there are few cases reported in the literature up to now. Therefore, EUS-RFA may be considered a treatment option in selected patients (e.g., elderly or unfit for surgery), although real-world data about the effectiveness of RFA are still lacking[3,11].

In the case of F-pNETs, the response to EUS-RFA is generally related to the disappearance of the clinical syndrome. On the contrary, the timing of evaluating the radiologic response of NF-pNETs after the treatment represents a relevant shadow because, unfortunately, a standardized protocol for a long-term follow-up is not yet defined. As previously reported, the authors presented a different concept of “radiologic response” and a variable follow-up of imaging control, restricting the comparison of the results. As suggested by Rossi et al[32], at 24-72 hours after the procedure, a close imaging control should be performed to assess the size of coagulative necrosis achieved and to evidence any post-procedural complications. Furthermore, at 6-12 months later, a computed tomography or magnetic resonance imaging should be planned in order to exclude residues of the lesion and late complications such as late fluid collections and pancreatic duct stenosis. Furthermore, the use of contrast-enhanced harmonic (CEH) EUS can be useful for assessing pre-ablation vascularization of the neoplasm, accurately identifying hypervascular sites within lesions and also in post-ablation examination, evaluating the early treatment response after EUS-RFA[23,33].

Choi et al[23] published a study on the utility of CEH-EUS for guidance and monitoring of ablation therapy, performing CEH-EUS-assisted RFA in 19 patients with solid abdominal neoplasms. Seven treated patients showed the absence of intralesional enhancement on CEH-EUS, while twelve patients manifested an incomplete ablation after the first RFA session; thus, the latter underwent additional RFA under CEH-EUS guidance. Thirteen patients (68.4%) achieved a radiologic complete response. During the median follow-up of 28 months, the local recurrence rate was 7.7%. The authors concluded that the use of CEH-EUS for RFA could be helpful in detecting early treatment response after RFA and also targeting residual tumors[23].

Several studies have reported positive results regarding patients’ safety, suggesting that EUS-RFA is an effective therapy with a good safety profile. The most common side effects reported are mild events such as abdominal pain and mild acute pancreatitis. In some studies, we found rare cases of severe acute pancreatitis, main pancreatic duct leak and main pancreatic duct stricture with chronic pancreatitis; only one case of death has been described[34].

In a recent systematic review and meta-analysis, Khoury et al[35] evaluated the efficacy and safety of EUS-RFA for pNETs, in particular the primary objective was to assess the complete radiological response defined as absence of the pancreatic neoplasm or presence of complete colliquative necrosis on imaging control while the secondary aims were the safety (adverse events rate), technical success rate based on the procedure and clinical success rate related to subsequent disappearance of a specific hormonal syndrome. A total of 292 patients from 11 studies were included in the meta-analysis. They reported a complete radiological response of 87.1% [95% confidence interval (CI): 80.1-92.8], a technical success rate of 99.2% (95%CI: 97.9-99.9), and a clinical success rate of 94.9% (95%CI: 90.7-97.9). Regarding patient safety, the rate of adverse events was 20% (95%CI: 14.0-26.7) while the rate of severe adverse events was 0.9% (95%CI: 0.2-2.3). The majority of adverse events were mild-moderate, such as mild abdominal pain, mild-moderate pancreatitis, mild peri-procedural bleeding, severe pancreatitis, pancreatic duct stenosis, diabetes mellitus and spleen hematoma. No case of mortality was reported.

A shadow for the management of pNET remains the use of preventive measure for reducing the risk of acute pancreatitis. Some authors, in order to reduce this risk, have adopted preventive measures, including use of rectal nonsteroidal anti-inflammatory drugs (NSAIDs) and placement of pancreatic stenting if the lesion was closed to the pancreatic duct (< 1mm). The distance from the lesion to the main pancreatic duct has been described as an independent risk factor for acute pancreatitis. Although there is a debate about the minimum distance from the main pancreatic duct for RFA, it has been suggested that maintaining a distance greater than 1 mm is necessary to perform a safe procedure. Unfortunately, many studies have not adopted it, making a potential comparison difficult.

Barthet et al[21], due to two cases of severe acute pancreatitis occurring after RFA for p-NET, modified the original study protocol to add the use of rectal NSAIDs as a preventive measure. On the other hand, Napoléon et al[17], in a multicenter retrospective study including 116 EUS-RFA sessions, administered rectal NSAIDs in 88.8% of cases and they failed to prove their protective role. Unfortunately, the available evidence is not yet clear.

Another issue regards the minimum safe distance between the lesion and the main pancreatic duct to perform RFA. Some endoscopists suggest the placement of a prophylactic pancreatic stent several days prior to RFA when the lesion is in close proximity to the duct to prevent acute pancreatitis and pancreatic duct stenosis. Napoléon et al[17] showed that the only independent risk factor for adverse events is close proximity to the main pancreatic duct. Currently, there are no strong recommendations to place a pancreatic stent.

Unfortunately, few prospective studies have been conducted on the role of EUS-RFA for pancreatic neuroendocrine neoplasms, and this represents a limitation. Most of the included papers are case series, reports, and retrospective studies, whose nature suggests potential selection bias in terms of the underlying patients’ conditions and indications for EUS-RFA.

A large multicenter retrospective study by Crinò et al[16], which aimed to compare EUS-RFA and surgical resection for the treatment of pancreatic insulinoma, had several limitations. First, due to the retrospective design, there is a risk of selection bias related to the preference for EUS-RFA in older patients with high surgical risk. Second, the duration of follow up varied significantly between the two groups. Third, data come from experience centers and operator that might not be reproducible. Fourth, neither surgical nor the EUS-RFA technique was not standardized, resulting in significant heterogeneity.

A study by Napoléon et al[17], assessing the role of EUS-RFA for p-NETs, presents several limitations due to the retrospective nature with potential selection bias. Additionally, oncological outcome was not assessed as primary endpoint and there was significant difference in terms of experience and procedure volumes among different centers. Finally, there are no data about standardization of the EUS-RFA procedure in terms of settings and number of ablations per session.

A prospective study by Figueiredo Ferreira et al[27], evaluating safety, technical and clinical success of EUS-RFA on p-NETs, has some limitations including limited number of patients, relatively short follow up period and the lack for optimal method and time for response evaluation. A retrospective study by Oleinikov et al[22] described the safety and efficacy of EUS-RFA in patients with p-NETs. Retrospective nature, heterogeneous cohort and relatively short-length of follow-up were the drawbacks of the study. Barthet et al[21] examined the safety and efficacy of EUS-RFA in NF-pNETs in a prospective multicenter study. This article presents some limitations, including the limited number of patients, the absence of histological control of the efficacy and limited follow-up.

Unfortunately, in all studies available many relevant data are lacking, including recurrence rate, time to recurrence and need for surgery. This gap limits the possibility of assessing the real efficacy of the procedure. The follow up was reported in almost studies, unfortunately there was a significant heterogeneity in the length of follow up that have limited the possibility to extrapolate robust data to examine the efficacy of the procedure. Therefore, relevant evidence should be collected, leading to larger prospective studies at experienced centers with adequate long-term follow-up.

CONCLUSION

EUS-RFA could be considered an effective and relatively safe procedure for treating p-NETs in selected cases, and its indication should be carefully evaluated in a multidisciplinary team. The standardization of the procedure in terms of preventive measures and the radiological follow-up protocols for the management of RFA outcomes is necessary. Lack of relevant data including recurrence rate, time to recurrence and need for surgery limit the possibility to assess the efficacy of the technique. Large prospective studies with longer follow-up are needed to determine the long-term results and also to better choose the patients who would take advantage of the endoscopic procedure.