Endoscopic ultrasound-guided ablation of small pancreatic neuroendocrine neoplasms: Bridging the gap between surveillance and surgery

Abstract

The growing use of cross-sectional imaging and endoscopic ultrasound (EUS) has led to more frequent detection of small pancreatic neuroendocrine neoplasms, particularly non-functioning lesions ≤ 2 cm. Active surveillance is appropriate for many low-risk tumors, whereas pancreatic surgery may be associated with substantial morbidity. EUS-guided ablation has therefore emerged as a pancreas-preserving option between surveillance and surgical resection. We conducted a narrative review of EUS-guided treatments for small pancreatic neuroendocrine neoplasms, focusing on pretreatment assessment, technical approaches, patient selection, radiological outcomes, safety, and emerging indications. Ethanol ablation and radiofrequency ablation are the most extensively studied techniques. Technical success approaches 100% in most series. Complete radiological response has generally ranged from approximately 50%-70% after ethanol ablation and 80%-90% after radiofrequency ablation, often after more than one treatment session. Clinical efficacy is particularly high in insulinomas, with symptom resolution reported in more than 90% of patients. Comparative studies suggest lower short-term morbidity and shorter hospitalization than surgery, although local recurrence may occur and long-term oncological outcomes remain insufficiently defined. EUS-guided ablation is promising for carefully selected patients, particularly those with insulinomas, but should not yet be considered a routine alternative to surgery or active surveillance, as prospective comparative studies and standardized outcome definitions are still needed.

Key Words: Pancreatic neuroendocrine neoplasms; Endoscopic ultrasound; Radiofrequency ablation; Ethanol ablation; Insulinoma; Pancreas-preserving treatment; Therapeutic endoscopy

Core Tip: Small pancreatic neuroendocrine neoplasms pose a therapeutic dilemma because surveillance may expose selected patients to tumor progression and psychological burden, whereas pancreatic surgery carries substantial morbidity. Endoscopic ultrasound-guided ethanol injection and radiofrequency ablation offer a pancreas-preserving intermediate strategy. This review summarizes current techniques, clinical outcomes, safety considerations, and proposed selection criteria, emphasizing that endoscopic ultrasound-guided ablation should currently be reserved for carefully selected patients treated in expert multidisciplinary centers.

INTRODUCTION

Pancreatic neuroendocrine neoplasms (panNENs) are a heterogeneous group of neoplasms arising from neuroendocrine cells of the pancreas[1-5]. With the increasing use of high-resolution cross-sectional imaging and endoscopic ultrasound (EUS), small, asymptomatic panNENs measuring ≤ 2 cm are being detected with increasing frequency[6-9]. These lesions are often well differentiated, non-functioning (NF), and biologically indolent[10-12], leading current guidelines and expert recommendations to support active surveillance rather than immediate intervention in appropriately selected patients[13-16]. However, the natural history of small panNENs is variable, and the decision between surveillance and treatment remains complex, requiring a balance between oncological safety and the risks of overtreatment[10,17,18].

The prospective multicenter ASPEN study has provided important insights into the natural history and management of incidentally detected, sporadic, NF-panNENs measuring ≤ 2 cm[19,20]. Its interim analysis showed that most lesions remained radiologically stable during follow-up, supporting a conservative approach in selected patients[20]. Nevertheless, tumor size alone cannot reliably identify all lesions with potentially aggressive biological features, underscoring the importance of individualized risk stratification and continued evaluation of minimally invasive treatment alternatives[17,19-21].

Surgical resection has traditionally represented the standard treatment for localized panNENs requiring active treatment. However, pancreatic surgery is associated with substantial short-term and long-term morbidity[22], particularly when performed in patients with significant comorbidities or for tumors that might otherwise follow an indolent clinical course[23-27]. In this context, EUS has evolved from a predominantly diagnostic and staging modality into a potential therapeutic platform. The two most extensively investigated EUS-guided ablative techniques for panNENs are ethanol ablation (EA) and radiofrequency ablation (RFA)[25], which may provide local tumor control while avoiding the morbidity of surgical resection[28-32]. Other EUS-guided technologies, including hybrid cryothermal and laser ablation, remain at an earlier stage of clinical development. By contrast, high-intensity focused ultrasound, which has also been investigated for pancreatic tumors, should not be classified as an EUS-guided technique.

Growing international research activity reflects the increasing interest in EUS-guided treatment of panNENs[33-37]. In particular, a multicenter randomized controlled trial comparing EUS-guided RFA (EUS-RFA) with surgical resection for small sporadic pancreatic insulinomas has been initiated to determine whether EUS-RFA may represent an effective and less invasive treatment option in this setting[38].

This narrative review aims to summarize the current evidence on EUS-guided ablative treatment of small panNENs, with particular emphasis on technical aspects, clinical and radiological outcomes, safety, patient selection, and the potential role of these procedures as an intermediate treatment strategy between active surveillance and surgical resection.

METHODOLOGY

A narrative literature search was conducted in PubMed/MEDLINE from database inception through July 2026. The search combined controlled vocabulary and free-text terms related to panNENs and EUS-guided treatment, including: (“Pancreatic neuroendocrine tumor” OR “pancreatic neuroendocrine neoplasm” OR “panNEN” OR “insulinoma”) AND (“endoscopic ultrasound” OR “EUS”) AND (“radiofrequency ablation” OR “RFA” OR “ethanol ablation” OR “ethanol injection” OR “local ablation” OR “locoregional treatment” OR “thermal ablation”). Additional searches were performed for active surveillance, pancreatic surgery, treatment selection, treatment-related adverse events, and emerging ablative technologies.

Articles were considered relevant when they reported clinical outcomes, technical aspects, patient selection, safety, radiological response, symptom control, recurrence, or follow-up after EUS-guided ablation of panNENs. Priority was given to prospective studies, multicenter cohorts, comparative studies, systematic reviews, meta-analyses, international guidelines, and larger retrospective series. Case reports and small case series were considered when they described the initial application of a technique or addressed areas in which higher-level evidence was unavailable. Preclinical studies and studies involving other pancreatic neoplasms were included only in the discussion of emerging technologies.

Reference lists of relevant articles and reviews were manually screened to identify additional publications. When multiple reports originated from the same or potentially overlapping cohort, the most recent or most comprehensive publication was preferentially used, while earlier reports were retained only when they provided distinct technical or longitudinal information. No formal risk-of-bias assessment was performed because of the narrative design of the review. Owing to marked heterogeneity in study populations, tumor functionality, ablative techniques, procedural protocols, definitions of treatment success, and follow-up duration, no quantitative synthesis was undertaken.

ROLE OF EUS IN PRETREATMENT ASSESSMENT

Current European guidelines recommend a multimodal assessment of panNENs, including contrast-enhanced (CE) computed tomography (CT), magnetic resonance imaging (MRI), somatostatin receptor imaging, preferably with positron emission tomography/CT, and EUS[15]. The relative contribution of each modality should, however, be tailored to the clinical setting, tumor characteristics, and therapeutic strategy. EUS is particularly valuable in patients with small panNENs because it provides high-resolution assessment of tumor size, location, margins, internal features, vascular relationships, and distance from the main pancreatic duct (MPD). These features are essential when considering EUS-guided ablation. EUS may also detect lesions not identified by conventional cross-sectional imaging. In a 2016 series, CT failed to identify 68.4% of lesions measuring < 10 mm and 15% of lesions measuring ≤ 20 mm that were detected by EUS[39]. Similarly, a meta-analysis estimated that the addition of EUS increased the detection of panNENs by approximately 26% compared with CT alone[40].

On conventional B-mode imaging, well-differentiated, low-grade panNENs generally appear as homogeneous, hypoechoic, well-demarcated lesions[41,42]. By contrast, larger or biologically aggressive neoplasms may show irregular margins, heterogeneous echogenicity, necrotic or cystic components, and an infiltrative growth pattern[42].

Nevertheless, EUS morphology alone is not sufficient to establish the diagnosis. Pancreatic metastases, solid serous neoplasms, intrapancreatic accessory spleen, acinar cell carcinoma, and pancreatic ductal adenocarcinoma may occasionally mimic a panNEN, reinforcing the importance of tissue confirmation before local ablative treatment.

CE-EUS may improve lesion characterization and help assess tumor vascularity. Well-differentiated panNENs typically show early arterial hyperenhancement, reflecting their rich microvascular network, whereas reduced or heterogeneous enhancement has been associated with higher tumor grade and more aggressive biological features[43-45]. In high-grade tumors, the reduced microvessel density results in hypoenhancement during the arterial phase[45]. Importantly, a hypovascular pattern has been shown to predict malignancy with a sensitivity of 92%-96% and an accuracy of 82%-95%[46,47]. CE-EUS is also useful during and after ablation because the persistence of enhancing tissue may indicate incomplete treatment. Conversely, the absence of residual enhancement indicates a radiological response but cannot exclude microscopic viable tumor and should not be interpreted as evidence of complete oncological eradication. However, enhancement patterns cannot replace histological diagnosis or formal tumor grading.

Elastography may provide complementary information on tissue stiffness, but its clinical contribution in panNENs remains limited by operator dependence, interobserver variability, and overlap among different pancreatic lesions[48-52]. At present, shear wave elastography should be regarded as a promising complementary tool rather than a standalone diagnostic modality[53].

Similarly, newer microvascular imaging techniques, including eFLOW and detective flow imaging, remain investigational and cannot currently be incorporated into routine treatment selection[54].

A major advantage of EUS is the ability to obtain tissue for diagnostic confirmation and grading. This is particularly relevant before ablative treatment, which does not provide a surgical specimen for definitive pathological assessment. EUS-guided tissue acquisition has a reported sensitivity of approximately 80%-90%, a specificity approaching 100%, and sample adequacy rates of 83%-93%[49,55]. Nevertheless, grading based on EUS-acquired specimens is not infallible. A meta-analysis reported an overall concordance of 80.3% between preoperative EUS-based grading and surgical histology, with undergrading occurring more frequently than overgrading (14.7% vs 3.5%, P < 0.001)[56].

Fine-needle biopsy (FNB) is generally preferred over fine-needle aspiration when tumor grading is required because it provides histological cores and usually allows more reliable assessment of the Ki-67 proliferative index[56]. Franseen and fork-tip needles have shown high diagnostic performance and adequate tissue yield in solid pancreatic lesions[57-59]. Available studies suggest that FNB may achieve greater sensitivity than FNA and a stronger correlation between Ki-67 assessment on the biopsy specimen and the final surgical specimen[48,60,61]. However, intratumoral heterogeneity and limited sampling remain relevant limitations, particularly in small lesions and tumors close to the G1-G2 threshold.

Rapid on-site evaluation may improve sample adequacy when available, although its added value is likely to be less pronounced with modern FNB needles[62]. Molecular characterization, including DAXX/ATRX expression and alternative lengthening of telomeres, may further improve pretreatment risk stratification, but these analyses remain investigational and are not currently required for clinical decision-making[63]. In parallel, artificial intelligence is being explored for automated interpretation of EUS images, with the potential to enhance diagnostic accuracy and predict tumor behavior in real time[64].

Overall, pretreatment EUS should not be regarded merely as a diagnostic examination. Its principal role is to confirm technical feasibility, define the relationship between the lesion and the pancreatic duct and adjacent vessels, obtain adequate tissue for diagnosis and grading, and exclude features that would favor surgery or systemic oncological management rather than local ablation[48,49,65].

EUS-GUIDED EA

EUS-guided EA (EUS-EA) was the first local ablative technique applied to panNENs[25,28-31] and remains one of the most extensively investigated approaches[32,37,66,67]. It is relatively inexpensive and does not require a dedicated generator or ablation probe. Injection of ethanol at concentrations ranging from 80% to 100% induces coagulative necrosis of the tumor by causing protein dehydration and subsequent vascular occlusion (Table 1)[36-38,66].

Table 1 Comparison of endoscopic ultrasound-guided ablative techniques for small pancreatic neuroendocrine neoplasms.

Parameter	EUS-guided ethanol injection	EUS-RFA	Emerging EUS-guided techniques
Mechanism of action	Chemical coagulative necrosis through protein denaturation and vascular thrombosis	Thermal coagulative necrosis induced by high-frequency alternating current	Hybrid cryothermal or laser-induced thermal ablation
Technical complexity	Low	Moderate to high	High; experimental
Equipment	Standard 22-gauge or 25-gauge EUS needle	Dedicated needle-based or catheter-based RFA system and generator	Dedicated probes or optical fibers with limited availability
Technical success	Approximately 90%-100%	Approximately 95%-100%	Insufficient panNEN-specific data
Complete radiological response	Approximately 50%-70%; may increase with repeat treatment	Approximately 80%-90%	Unknown
Clinical success in functioning tumors	High, particularly in insulinomas	Very high, generally > 90% in insulinomas	Unknown
Number of sessions	Frequently more than one	Usually, one or two	Not standardized
Main adverse events	Pancreatitis, abdominal pain, ethanol leakage, pancreatic tissue injury	Pancreatitis, abdominal pain, pancreatic duct injury or stenosis	Safety profile not established in panNENs
Main procedural risk factor	Excessive or extratumoral ethanol diffusion	Close proximity to the MPD, particularly ≤ 1 mm	Not defined
Control of ablation volume	Limited and diffusion-dependent	More predictable and energy-controlled	Potentially precise, but not clinically validated
Most established clinical setting	Small insulinomas and selected small NF-panNENs	Small, well-differentiated insulinomas and selected NF-panNENs	Investigational
Main limitation	Unpredictable ethanol distribution and frequent need for repeat treatment	Requirement for dedicated equipment, expertise, and careful ductal assessment	Lack of dedicated clinical evidence in panNENs

EUS: Endoscopic ultrasound; panNENs: Pancreatic neuroendocrine neoplasms; NF-panNENs: Non-functioning pancreatic neuroendocrine neoplasms; EUS-RFA: Endoscopic ultrasound-guided radiofrequency ablation; MPD: Main pancreatic duct.

Under EUS guidance, a 22-gauge or 25-gauge needle is advanced into the lesion, and small aliquots of ethanol are injected until intratumoral diffusion is achieved. Careful assessment of the MPD, adjacent vessels, and gastrointestinal wall is essential to minimize extratumoral leakage and chemical injury[68,69]. Uniform hyperechogenicity is commonly used as an immediate procedural endpoint, although it does not confirm complete ablation. Repeat treatment may be required when residual enhancing tissue persists.

Initial reports focused mainly on insulinomas, in which symptom control represents a clinically meaningful endpoint[67]. Jürgensen et al[67] first described EUS-EA of a 13-mm insulinoma located in the pancreatic tail, reporting successful resolution of hypoglycemic symptoms. In the pilot study by Park et al[70], complete radiological response was achieved in 61.5% of evaluable tumors after repeat treatment, with resolution of hypoglycemia in both insulinoma patients; three patients developed mild pancreatitis. In the prospective study by Choi et al[71], complete ablation was achieved in 24/40 tumors (60%), with a median follow-up of 42 months. Two episodes of acute pancreatitis occurred, including one case complicated by pancreatic duct stricture requiring stenting. Importantly, three lesions considered completely ablated on imaging were subsequently found to contain residual tumor on pathological reassessment, highlighting the limitations of radiological endpoints[71]. More recently, Matsumoto et al[72] conducted a prospective multicenter study of 25 patients with pathologically confirmed G1 NF-panNENs ≤ 15 mm. Complete radiological ablation was achieved in 22/25 patients (88%) at both 1 month and 6 months; however, 8/25 patients required an additional early treatment session. Acute pancreatitis occurred in 5/25 patients, including one moderate case, whereas severe adverse events were uncommon. The short 6-month follow-up precluded assessment of late recurrence and long-term oncological efficacy[72,73]. Efficacy appears to decline as lesion size increases, because homogeneous ethanol distribution becomes more difficult and residual viable tissue is more likely. In this study, complete ablation was numerically higher for tumors < 10 mm than for those measuring 10-15 mm, and nearly one-third of patients required an additional early treatment session[72,73].

Available evidence suggests that EUS-EA is technically feasible and can effectively control symptoms in insulinomas. It also shows high short-term radiological response rates in carefully selected small NF-panNENs. However, its main limitations include unpredictable ethanol diffusion, a frequent need for retreatment, procedure-related pancreatitis, and the inability of imaging to exclude microscopic residual disease. The long-term oncological efficacy of this approach remains uncertain.

EUS-RFA

EUS-RFA should be performed in high-volume tertiary referral centers with expertise in interventional EUS and multidisciplinary management of panNENs. RFA produces coagulative necrosis through the delivery of high-frequency alternating current within the target lesion[37]. Dedicated internally cooled needle systems are currently the most commonly used devices. Under EUS guidance, the electrode is positioned within the lesion, and one or more short energy applications are delivered. CE-EUS may help identify residual viable tissue immediately after treatment, whereas CT or MRI remains necessary for staging and longitudinal surveillance[74].

Tumor proximity to the MPD is a major safety consideration. A distance of ≤ 1 mm was independently associated with adverse events in the French RAFPAN study[75], and some prospective protocols excluded lesions located within 2 mm of the duct[73]. Accordingly, lesions within 1-2 mm of the MPD should generally be considered anatomically unfavorable and excluded from ablation, except in exceptional cases managed at highly experienced centers. Prophylactic pancreatic duct stenting should not be routinely performed because supporting evidence is insufficient; it may be considered only in exceptional cases involving tumors in direct proximity to the duct[76,77].

Rectal NSAIDs and periprocedural intravenous hydration may be considered as low-risk preventive measures extrapolated from post-ERCP practice, but their benefit in EUS-RFA has not been demonstrated. Routine antibiotic prophylaxis cannot currently be recommended for ablation of solid panNENs in the absence of specific infectious risk factors. These preventive strategies should therefore remain center-specific until prospective evidence or consensus recommendations become available.

The largest study to date evaluating the safety and efficacy of EUS-RFA for pancreatic neoplasms was published in 2023 by Napoléon et al[75]. In this multicenter French study, 100 patients with 104 pancreatic neoplasms underwent EUS-RFA; 64 lesions were panNENs, including 48 G1 tumors and 16 functioning tumors. The median tumor size was 15 mm (range, 6-39 mm). Technical success was 97%. A combined clinical and radiological response was achieved in 60.2% of patients. On multivariate analysis, tumor size < 20 mm (odds ratio = 5.26; 95%CI: 2.17-14.29; P < 0.001) and neuroendocrine neoplasm histology (odds ratio = 7.95; 95%CI: 1.66-51.79; P < 0.001) remained independently associated with complete response. The adverse-event rate was approximately 18%, and proximity to the MPD was the only independent predictor of complications[75].

The most recent and comprehensive meta-analysis evaluating the efficacy and safety of EUS-RFA in the management of panNENs was published in 2024 by Khoury et al[76], including 11 studies with a total of 292 patients with panNENs, and reported pooled technical and complete radiological response rates of 99.2% and 87.1%, respectively. The pooled clinical response rate in functioning tumors was 94.9%. Adverse events occurred in 20.0%, but severe events were uncommon at 0.9%. The most frequently reported complications were transient abdominal pain, observed in 19 patients (6.5%), and mild-to-moderate pancreatitis, occurring in 23 patients (7.9%). Importantly, no treatment-related mortality was reported[76].

The strongest comparative evidence concerns insulinomas. In the propensity-matched study by Crinò et al[36], EUS-RFA was associated with fewer overall and severe adverse events than surgery, shorter hospitalization, and similar short-term clinical efficacy. Notably, no severe AEs were observed after EUS-RFA, whereas 15.7% of patients undergoing surgery experienced severe complications (P < 0.0001). Clinical efficacy was achieved in 100% of patients after surgery and in 95.5% after EUS-RFA (P = 0.160). During follow-up, 15 lesions (16.9%) recurred after EUS-RFA. Among these, 11 were successfully managed with repeat EUS-RFA, whereas four patients ultimately required surgery[36].

Two prospective cohort studies are currently ongoing and registered on ClinicalTrials.gov to further evaluate the safety and efficacy of EUS-RFA in NF-panNENs: (1) The RAPNEN trial (NCT03834701); and (2) The RFANET trial (NCT04520932).

No randomized evidence favors EA over RFA, or vice versa, and treatment selection therefore remains dependent on local expertise, device availability, tumor characteristics, and anatomy[35,70-72,76]. EA is simple and inexpensive but has less predictable tissue diffusion, whereas RFA provides a more controlled ablation zone but requires dedicated equipment and carries a risk of thermal injury[35,70-72,75]. Either technique may provide rapid symptom control in small insulinomas[36,70-72,75]. In NF-panNENs, RFA may offer greater procedural reproducibility, although superiority in long-term oncological control has not been demonstrated[35,72,75].

Lesions within 1-2 mm of the MPD require particular caution and should generally be avoided or treated only in expert centers. Repeat ablation may be considered for small, localized residual disease, whereas surgery should be reconsidered after persistent failure, regrowth, upgrading, ductal or vascular involvement, or nodal or distant progression[36,70-72,75].

HYBRID CRYOTHERMAL ABLATION AND OTHER EMERGING TECHNIQUES

Hybrid cryothermal ablation combines bipolar radiofrequency energy with internal cryogenic cooling. The cryogenic cooling component is intended to reduce tissue carbonization and permit a more homogeneous delivery of thermal energy, rather than to induce tissue destruction through conventional freeze-thaw cycles[31,78,79].

Clinical experience with this approach has been obtained almost exclusively in patients with locally advanced pancreatic ductal adenocarcinoma, mainly using the HybridTherm Probe. These devices were developed to achieve partial ablation within relatively large pancreatic masses and may not be suitable for the complete treatment of small, well-circumscribed panNENs. To date, no clinical studies have specifically evaluated hybrid cryothermal ablation in panNENs. Its role in this setting should therefore be considered investigational.

Laser ablation represents another potential EUS-guided thermal technique. It involves the insertion of an optical fiber into the target lesion through an EUS needle, followed by delivery of laser energy to induce focal coagulative necrosis. However, available evidence is limited to preclinical studies and very small clinical series involving solid pancreatic lesions, with no meaningful data in panNENs[31,78,79]. Laser ablation may theoretically offer precise energy delivery and could be attractive for very small lesions, particularly those < 10 mm, whereas hybrid cryothermal ablation may reduce thermal injury through tissue cooling and could potentially be advantageous near vascular structures. However, these potential benefits remain speculative and are supported mainly by preclinical or proof-of-concept evidence. Their use in panNENs should therefore remain confined to early-phase clinical studies.

High-intensity focused ultrasound has also been investigated for the local treatment of pancreatic tumors. Unlike the other techniques discussed in this review, HIFU is generally delivered extracorporeally under transabdominal ultrasound or magnetic resonance guidance and should not be considered an EUS-guided modality. Moreover, clinical experience relates almost exclusively to locally advanced pancreatic adenocarcinoma, with no evidence supporting its use in small panNENs.

Overall, hybrid cryothermal ablation, laser ablation, and HIFU remain experimental in panNENs. Ethanol injection and RFA are currently the only EUS-guided ablative techniques supported by substantive clinical experience in this setting.

PATIENT SELECTION

Although interest in EUS-guided ablation has increased, it is not currently established as a standard treatment for panNENs in international guidelines. Its use should therefore be restricted to carefully selected patients after multidisciplinary assessment of tumor biology, anatomical feasibility, surgical risk, and patient preference[14,15,80-82]. Recent ENETS guidance acknowledges the potential role of minimally invasive approaches in highly selected patients, while emphasizing that treatment decisions should remain individualized and discussed within a multidisciplinary team[15]. Tumor size remains an important component of risk stratification, although it should never be considered in isolation. Small, NF-panNENs measuring ≤ 2 cm frequently follow an indolent clinical course and may be managed conservatively in appropriately selected patients[19,20,80]. Lesions measuring 15-20 mm fall within a particularly relevant gray zone where optimal management remains controversial and should be individualized. Nevertheless, tumor diameter alone cannot reliably distinguish indolent from potentially aggressive lesions. Tumor differentiation, proliferative grade, growth during follow-up, functional status, anatomical relationships, and the presence of nodal or distant disease must all be considered. EUS-guided ablation should primarily be considered for localized, well-differentiated, low-grade tumors without nodal or distant disease. Most available evidence concerns G1 lesions. Selected low-proliferation G2 tumors, pragmatically defined as Ki-67 ≤ 5%, may occasionally be considered[83]; however, this threshold has not been formally validated for EUS-guided ablation, and treatment should remain exceptional because of sampling limitations and the risk of undergrading[56,81].

Functional status is central to selection. Insulinoma represents the most established indication because local treatment can provide rapid control of hormone-related symptoms. Comparative studies in patients with insulinomas suggest that EUS-guided ablation may achieve high clinical efficacy with lower morbidity and shorter hospitalization than surgery[36,68]. These findings are particularly relevant for elderly patients, patients with significant comorbidities, and those at increased surgical risk. The clinical rationale for EUS-guided ablation is substantially stronger for insulinomas than for NF-panNENs. Insulinomas are generally benign and, when surgery is performed, often require only parenchyma-sparing enucleation rather than formal oncological pancreatic resection. Accordingly, rapid resolution of hypoglycemia represents an immediate and clinically meaningful endpoint, and complete histological tumor eradication may not always be necessary to achieve durable symptom control. By contrast, in NF-panNENs, treatment is undertaken primarily to achieve long-term oncological control. This requires confidence regarding complete tumor eradication, biological behavior, nodal status, and the absence of microscopic residual disease, outcomes that cannot be reliably established by post-ablation imaging alone. Moreover, evidence for NF-panNENs is substantially weaker. Available studies of NF-panNENs report high radiological response rates, but most are small, non-randomized[70-72], imaging-based series with heterogeneous follow-up[75,84,85]. In these NF-panNENs, ablation should not be proposed merely because a lesion is present, as active surveillance remains appropriate for many small, asymptomatic, low-risk tumors[15,19,20]. EUS-guided ablation may be considered only in carefully selected patients who do not accept surveillance and in whom surgery is declined, contraindicated, or associated with disproportionate morbidity.

Features suggesting aggressive or advanced disease should prompt surgical or oncological evaluation rather than local treatment.

Anatomical feasibility is equally important and requires safe access from the stomach or duodenum, a favorable relationship with the MPD and adjacent vessels, and the absence of features suggesting invasive disease. Patient age, comorbidity, previous pancreatic surgery, life expectancy, surgical risk, willingness to undergo surgery, and ability to comply with structured long-term surveillance should also be considered. The proposed selection framework is summarized in Table 2, while situations in which ablation should generally be avoided and its positioning relative to surveillance and surgery are discussed below.

Table 2 Proposed criteria for patient selection for endoscopic ultrasound-guided ablative treatment in small pancreatic neuroendocrine neoplasms.

Parameter	More favorable candidate	Borderline candidate (case-by-case)	Generally unsuitable candidate
Tumor size	≤ 15 mm	15-20 mm	> 20 mm
Differentiation	Well-differentiated NET	Well-differentiated NET	Poorly differentiated NEC
Tumor grade	G1, Ki-67 < 3%	Selected low-G2 tumor (Ki-67 ≤ 5%), after multidisciplinary review	G3 or clearly high-proliferative G2 tumor
Functionality	Insulinoma	Other functioning tumors are considered only for symptom control in exceptional circumstances	Functioning tumor with multifocal, nodal, or metastatic disease
Nodal or distant disease	Absent	Indeterminate findings requiring further staging	Present
Behavior during surveillance	No clear evidence of aggressive behavior	Limited interval growth requiring repeat staging and reassessment	Rapid growth or other evidence of progression
Tumor margins and local invasion	Well-defined, confined lesion	Uncertain local relationships	Vascular invasion, perineural invasion, or infiltrative margins
EUS accessibility	Safe and straightforward access	Technically feasible but challenging	Unsafe or impossible access
Relationship with MPD	Clear separation from the duct	Close proximity requiring increased caution	Direct contact, duct involvement, or inability to preserve a safe margin
Pancreatic duct dilation	Absent	Equivocal or unrelated dilation requiring clarification	Dilation attributable to tumor obstruction
NF-panNENs	Active treatment considered appropriate despite disproportionate surgical morbidity	Uncertain balance between surveillance, ablation, and surgery	Stable low-risk lesion appropriately managed by surveillance
Surgical risk	High surgical risk or relevant comorbidities	Intermediate risk	Fit patient with a clear surgical indication
Patient preference	Declines surgery after informed counseling	Preference uncertain or evolving	Prefers surgery or does not accept structured follow-up
Treatment setting	Expert tertiary center with multidisciplinary NET expertise	Expert center after extended multidisciplinary discussion	Center without advanced EUS and pancreatic support
Post-treatment surveillance	Accepts structured long-term follow-up	Potential concerns regarding adherence	Unable or unwilling to undergo surveillance

This framework reflects the authors’ interpretation of the available evidence and should not be regarded as a validated guideline. Eligibility should be determined within an experienced multidisciplinary team. EUS: Endoscopic ultrasound; NF-panNENs: Non-functioning pancreatic neuroendocrine neoplasms; NET: Neuroendocrine tumour; NEC: Neuroendocrine carcinoma; MPD: Main pancreatic duct.

When EUS-guided ablation should not be used

EUS-guided ablation should generally not be considered when clinical, pathological, or radiological findings suggest biologically aggressive disease, or when surgical resection is required to provide definitive pathological staging and regional lymph-node assessment[15,25,29,32,37]. Features generally considered unsuitable for EUS-guided ablation include poorly differentiated neuroendocrine carcinoma, G3 tumors, clearly high-proliferative G2 tumors, nodal or distant metastases, vascular or perineural invasion, infiltrative margins, tumor-related MPD dilation[15-17], and rapid interval growth[20,25,80-82].

Lesions larger than 20 mm should generally not be treated with EUS-guided ablation outside exceptional circumstances, because increasing tumor size is associated with a greater risk of nodal involvement and aggressive behavior[15-17], while tumors < 20 mm have shown a significantly higher probability of complete local response after EUS-RFA[20,25,80-82].

EUS-guided ablation is also generally inappropriate for multifocal or hereditary disease, particularly multiple endocrine neoplasia type 1-associated tumors, because treatment of a single lesion may not provide meaningful disease control. In multiple endocrine neoplasia type 1, pancreatic and duodenal neuroendocrine tumors may be multiple and metachronous, and gastrinomas are frequently multifocal and arise predominantly in the duodenum rather than in the pancreas; occult nodal disease may also be present[86,87]. EUS-guided ablation in these settings should therefore be considered only in exceptional circumstances, mainly for symptom control in patients who are not candidates for surgery or who decline surgery after adequate counseling[14,80,85,88,89].

Among functioning tumors, indications for EUS-guided ablation beyond insulinoma are extremely limited. Gastrinomas, glucagonomas, VIPomas, and somatostatinomas frequently require a broader surgical or oncological strategy because they may present with multifocal disease, or with nodal and distant metastases. In these tumors, EUS-guided ablation should not generally be regarded as definitive oncological treatment. Its use should be restricted to exceptional circumstances, primarily as a palliative strategy for hormonal symptom control in patients who are not candidates for surgery. Surgery remains the cornerstone of treatment for localized resectable disease, whereas advanced or unresectable tumors usually require multimodal therapy aimed at both tumor and hormonal control[90,91].

Ablation should be avoided when the lesion cannot be safely accessed by EUS, when major vessels must be traversed, or when the relationship with the MPD does not permit an adequate safety margin. The procedure should also not be performed outside centers with expertise in therapeutic EUS and access to pancreatic surgery, interventional radiology[15-17], and multidisciplinary neuroendocrine tumor care[20,25,80-82].

Finally, EUS-guided ablation should not be offered when active surveillance remains the most appropriate strategy for a stable, asymptomatic, low-risk NF-panNEN.

Positioning EUS-guided ablation between surveillance and surgery

The management of small panNENs should not be based on tumor size alone, but rather on the balance between the expected risk of disease progression, the morbidity of treatment, and the degree of oncological certainty required (Figure 1). Active surveillance, EUS-guided ablation, and surgery should therefore be regarded as complementary rather than competing strategies.

Figure 1 Proposed decision-making algorithm for the management of small pancreatic neuroendocrine neoplasms. Endoscopic ultrasound-guided therapies should be considered only in carefully selected patients after assessment of tumor biology, anatomical feasibility, surgical risk, acceptability of surveillance, and patient preference. The algorithm is intended as a pragmatic framework for multidisciplinary discussion and should not be regarded as a validated clinical guideline. EUS: Endoscopic ultrasound; panNEN: Pancreatic neuroendocrine neoplasm; CT: Computed tomography; MRI: Magnetic resonance imaging; CE: Contrast-enhanced.

Active surveillance remains the preferred approach for many asymptomatic, sporadic, NF, well-differentiated panNENs measuring ≤ 2 cm, particularly when the lesion is stable, lacks suspicious radiological features, and the patient accepts structured follow-up[10,15,19,20]. Its main advantage is avoidance of procedure-related morbidity and preservation of pancreatic function. However, surveillance does not provide definitive pathological staging and may be associated with uncertainty, repeated imaging, and the need to reconsider treatment if tumor growth or other adverse features emerge.

Surgical resection remains the reference treatment when definitive oncological management is required. It provides a complete specimen for assessment of tumor differentiation and grade, resection margins, lymphovascular and perineural invasion, and, when indicated, regional lymph-node involvement. These advantages are particularly relevant in patients with tumors showing higher-risk biological or anatomical features. By contrast, EUS-guided ablation does not allow formal nodal staging or pathological confirmation of complete tumor eradication. The absence of residual enhancement on imaging cannot exclude microscopic viable disease. However, pancreatic surgery may be associated with substantial short-term morbidity, prolonged hospitalization, pancreatic fistula[22-24], and long-term endocrine or exocrine insufficiency[25-27].

EUS-guided ablation may be considered an intermediate option when active treatment is clinically justified but the expected morbidity of surgery appears disproportionate to the anticipated benefit. This scenario is most clearly supported in small, localized insulinomas, particularly in elderly patients, patients with comorbidities, or individuals who decline surgery after informed counseling. In NF-panNENs, the indication is considerably less established and should be restricted to highly selected patients in whom surveillance is considered inappropriate or unacceptable, surgery is undesirable or high risk, and tumor biology and anatomy appear favorable for local treatment.

The choice among surveillance, ablation, and surgery should therefore integrate tumor functionality, size, grade, growth pattern, anatomical relationships, nodal or distant disease, patient age, comorbidity, surgical risk, life expectancy, and patient preference. Importantly, EUS-guided ablation should not be regarded as a substitute for surgery when oncological staging or definitive pathological assessment is required, nor as an alternative to surveillance when the latter remains appropriate and acceptable. The main characteristics, advantages, and limitations of these three management strategies are summarized in Table 3.

Table 3 Comparison of active surveillance, endoscopic ultrasound-guided ablation, and surgery for small pancreatic neuroendocrine neoplasms.

Parameter	Active surveillance	EUS-guided ablation	Surgery
Main clinical role	Observation of low-risk lesions	Minimally invasive local treatment in selected patients	Definitive treatment when oncological resection is indicated
Most appropriate setting	Small, asymptomatic, NF, well-differentiated, stable panNENs	Small insulinomas and selected localized low-grade NF-panNENs	Tumors with higher-risk biological or anatomical features
Pathological confirmation	Based on biopsy when performed	Pretreatment biopsy required; no resection specimen	Complete surgical specimen available
Assessment of margins	Not applicable	Not available	Available
Lymph-node assessment	Not available	Not available	Available when lymphadenectomy is performed
Oncological certainty	Limited during observation	Limited; imaging response does not prove eradication	Highest among the three strategies
Procedure-related morbidity	None	Generally lower than surgery, but pancreatitis, pain, and ductal injury may occur	Substantial, depending on procedure type and location
Hospital stay	None	Usually short	Usually longer
Pancreatic preservation	Complete	Usually preserved	Variable; risk of endocrine and exocrine insufficiency
Local residual or recurrent disease	Tumor remains untreated and monitored	Possible; repeat ablation may be required	Lower after complete resection, although recurrence remains possible
Need for repeat intervention	Possible if growth or progression occurs	Relatively common after incomplete response or recurrence	Usually not required for local control after complete resection
Main advantage	Avoids overtreatment and procedural morbidity	Pancreas-preserving local treatment with lower short-term burden	Definitive pathology and oncological resection
Main limitation	Persistent uncertainty and need for long-term follow-up	No nodal staging or confirmation of complete microscopic eradication	Operative morbidity and potential long-term pancreatic dysfunction
Evidence base	Supported by prospective observational data in selected NF-panNENs	Mainly retrospective series, small prospective cohorts, and limited comparative data	Established standard for tumors requiring active definitive treatment

EUS: Endoscopic ultrasound; NF-panNENs: Non-functioning pancreatic neuroendocrine neoplasms; NF: Non-functioning; panNENs: Pancreatic neuroendocrine neoplasms.

Clinical outcomes and evidence

The evidence supporting EUS-guided ablation of panNENs remains limited in both quantity and methodological quality. Most available studies are retrospective case series or small prospective cohorts conducted in high-volume referral centers[36,85] and involving highly selected patients treated by experienced operators[92-96]. Comparative studies remain few, and randomized outcome data are not yet available. These features introduce substantial risks of selection bias, referral-center bias, operator-related bias, and limited generalizability to routine clinical practice.

Published studies also differ considerably in tumor functionality, lesion size and grade, ablation technique, number of treatment sessions, imaging methods, follow-up duration, and definitions of treatment response[33,93,97,98]. Technical success, clinical response, complete ablation, residual disease, recurrence, and adverse events have not been uniformly defined, thereby limiting direct comparisons and pooled interpretation. Publication bias is also possible, because technically unsuccessful procedures and unfavorable outcomes may be underreported.

The available evidence supports technical feasibility and short-term local control more strongly than definitive long-term oncological efficacy. The main clinical studies evaluating EUS-EA and RFA in panNENs are summarized in Table 4[36,70-72,75,84,85,99,100]. The ongoing randomized trial comparing EUS-RFA with surgery for pancreatic insulinomas reflects both the growing clinical interest in this approach and the need for more robust comparative evidence[38].

Table 4 Main clinical studies of endoscopic ultrasound-guided ablation in pancreatic neuroendocrine neoplasms.

Ref.	Design	Population	Technique	Main efficacy and safety outcomes	Follow-up	Main limitations
Park et al[70], 2015	Retrospective, single center	11 patients; 14 panNENs < 20 mm; 10 NF-panNENs, 4 insulinomas; 2 MEN1; mean size 12.3 mm	EUS-EA; 18 sessions	Technical success 100%; complete radiological response 53.8% after one session and 61.5% after retreatment; symptom resolution in 2/2 insulinomas; pancreatitis in 3 patients and 1 duct stricture	Median 370 days	Small uncontrolled cohort; mixed tumors; variable ethanol dose and sessions; imaging-based response; short follow-up
Choi et al[71], 2018	Prospective, single center	33 patients; 40 panNENs < 20 mm; 39 NF-panNENs, 1 insulinoma; median size 11 mm; 3 hereditary cases	EUS ethanol-lipiodol ablation; 63 sessions	Technical success 100%; complete ablation 60%; pancreatitis in 2/63 sessions; radiological-pathological concordance 86% in 22 reassessed tumors	Median 42 months	No control group; single expert operator; repeat FNB available in only 22/40 tumors; 3 radiological complete responses reclassified as incomplete
Barthet et al[84], 2021	Prospective, multicenter, non-randomized	12 patients; 14 G1 NF-panNENs < 20 mm; mean size 13.1 mm; 1 MEN1	EUS-RFA	Complete disappearance/necrosis 85.7% at 1 year and final follow-up; 1 late recurrence, 1 delayed response, and 1 persistent failure followed by liver metastases after initial undergrading	Mean 45.6 months	Very small panNEN subgroup; mixed study cohort; no control group; protocol evolved during the study; response mainly imaging based
Marx et al[35], 2022	Retrospective, two centers	27 patients with sporadic G1 NF-panNENs; mean size 14.0 mm, range 7-25	EUS-RFA	Technical success 100%; complete radiological response 85.2% after one session and 92.6% overall; pancreatitis 14.8%; 3 grade IIIb complications; surgical specimen in 1 patient showed 3-mm residual tumor	Mean 15.7 months	Small selected cohort; no comparator; short and heterogeneous follow-up; imaging-based response; clinically relevant AE rate
Napoléon et al[75], 2023	National, multicenter, retrospective	100 patients; 104 pancreatic neoplasms, including 64 panNENs and 16 functioning tumors; median overall size 15 mm	EUS-RFA; 116 sessions	Technical success 97%; overall complete response 60.2%; panNEN histology and size < 20 mm predicted complete response; 22 AEs; MPD distance ≤ 1 mm independently predicted AEs	Imaging planned at 1 month, 6 months, and 12 months	Heterogeneous tumor types; panNEN-specific outcomes incompletely separated; variable protocols across 20 centers; non-standardized longer-term follow-up
Yan et al[99], 2023	Prospective, single center	9 high-risk or surgery-refusing patients with symptomatic insulinomas; 5 previously operated; 1 multifocal case	EUS-EA; 14 procedures	Initial clinical response 77.8%; relapse in 55.6%; 4 patients remained relapse-free long term; repeat-ablation response 40%; pancreatitis in 1/14 procedures	Median 994 days	Very small cohort; no control group; highly selected patients; heterogeneous tumor burden; clinical rather than oncological endpoint
Crinò et al[36], 2023	International, multicenter, retrospective propensity-matched study	178 patients with sporadic insulinomas: 89 EUS-RFA and 89 surgery; mean size 13.5 mm	EUS-RFA vs surgery	AEs 18.0% vs 61.8%; severe AEs 0% vs 15.7%; clinical efficacy 95.5% vs 100%; shorter hospital stay after RFA; recurrence after RFA 16.9%	Median 23 vs 37 months	Non-randomized design; residual confounding possible; shorter follow-up after RFA; clinical rather than histological efficacy endpoint
Matsumoto et al[72], 2025	Prospective, multicenter, single arm with historical surgical comparison	25 patients with G1 NF-panNENs ≤ 15 mm; median size 10.1 mm; MPD distance ≤ 2 mm excluded	Standardized EUS-EA with early reassessment and retreatment	Complete ablation 88% at 1 month and 6 months; 32% required early retreatment; severe AEs 4%; pancreatitis 20%; new or worsening diabetes 12%; no conversion to surgery	6 months	Small single-arm cohort; historical control; very short follow-up; imaging-based endpoint; highly selected tumors
Okasha et al[100], 2025	Prospective, multicenter cohort	27 patients: RFA group included 11 insulinomas and 2 PDACs; EA group included 13 insulinomas and 1 NF-panNEN	EUS-RFA or EUS-EA	RFA: Complete insulinoma response 10/11 and partial response 1/11, with no reported AEs. EA: Complete response 8/14, partial response 2/14, failure 4/14; pancreatitis in 4/14	Mean 12.0 months for RFA and 9.6 months for EA	Small heterogeneous cohort; non-comparative groups; unclear grading; clinical endpoints; short follow-up; internal reporting inconsistencies

AEs: Adverse events; EA: Ethanol ablation; EUS: Endoscopic ultrasound; MEN1: Multiple endocrine neoplasia type 1; MPD: Main pancreatic duct; NF-panNENs: Non-functioning pancreatic neuroendocrine neoplasms; panNENs: Pancreatic neuroendocrine neoplasms; PDACs: Pancreatic ductal adenocarcinomas; RFA: Radiofrequency ablation; EUS-RFA: Endoscopic ultrasound-guided radiofrequency ablation; EUS-EA: Endoscopic ultrasound-guided ethanol ablation; FNB: Fine-needle biopsy.

Despite these limitations, EUS-guided ablation has shown consistently high technical feasibility, with technical success approaching 100% in most published series[89,93]. Technical success, however, should not be equated with complete tumor eradication. Clinical and radiological outcomes depend on tumor type, lesion size, treatment modality, number of sessions, and the endpoint used to define success.

In functioning tumors, particularly insulinomas, symptomatic and biochemical control has generally exceeded 90% after EUS-RFA[68,99], often occurring shortly after the procedure[85,93,100]. Effective symptom control has also been reported after EA[30,68], although repeat sessions are more frequently required[93,99,101-103]. The strongest comparative evidence concerns insulinomas. In the propensity score-matched study by Crinò et al[36], EUS-RFA achieved clinical efficacy in 95.5% of patients and was associated with fewer adverse events and shorter hospitalization than surgery.

However, local recurrence occurred in approximately 17% of treated lesions, and some patients ultimately required surgery.

In NF-panNENs, radiological response is the principal endpoint. Reported complete response rates range from approximately 50%-80% after one or more sessions[68,99] and exceed 85% in several RFA series[84,89,93]. These estimates should be interpreted cautiously because radiological response may be defined by lesion disappearance, absence of enhancement, or apparent necrosis. None of these findings can confirm complete histological eradication or exclude microscopic viable tumor, occult nodal disease, or pretreatment undergrading. The available evidence therefore supports technical feasibility and short-term local control more strongly than definitive long-term oncological efficacy.

Tumor size appears to influence treatment success, with most favorable outcomes reported in lesions < 20 mm and particularly in very small tumors[85]. The prospective Japanese multicenter study of standardized small-volume ethanol injection achieved complete radiological ablation in 88% of G1 NF-panNENs ≤ 15 mm, although 32% required an additional early treatment session and follow-up was limited to 6 months[72,73]. Similarly, prospective RFA series have reported complete radiological response rates above 85% in selected small panNENs, but long-term comparisons between functioning and NF-panNENs[30,68] are difficult because treatment endpoints differ[93,99,101-103].

Safety profiles differ according to technique. The most frequent adverse events are abdominal pain and mild-to-moderate acute pancreatitis. EA carries a risk of uncontrolled chemical diffusion and ductal injury, whereas RFA may cause thermal damage to the MPD or adjacent structures, particularly when the lesion lies within 1-2 mm of the duct. Severe adverse events are uncommon but cannot be excluded, especially outside experienced centers[30,68,99].

No formal cost-effectiveness analyses have compared EUS-guided ablation with active surveillance or pancreatic surgery, and comparative data on patient-reported outcomes, quality of life, anxiety associated with surveillance, recovery time, and long-term functional consequences remain limited. These outcomes are particularly relevant when minimally invasive treatment is considered for otherwise indolent tumors and should be incorporated into future prospective studies.

Overall, current evidence suggests that EUS-guided ablation may achieve high rates of symptom control in insulinomas and encourage local response in selected small NF-panNENs. However, no head-to-head trial has established superiority of EA or RFA, and no randomized evidence has demonstrated oncological equivalence with surgery or superiority over active surveillance. The main clinical studies are summarized in Table 4[36,70-72,75,84,85,99,100].

EUS-guided ablation should therefore be regarded as a selectively applicable strategy rather than an established replacement for surgery or surveillance. Its definitive role will require standardized outcome definitions, prospective comparative studies, randomized trials, and long-term follow-up evaluating recurrence, metastatic progression, pancreatic function, quality of life, and overall survival.

Definitions of treatment response and post-ablation surveillance

The interpretation of outcomes after EUS-guided ablation is limited by the absence of standardized definitions across published studies[33,76,93,97,98]. Technical success should be defined as successful access to the target lesion, correct placement of the ablation needle or electrode, and completion of the intended treatment protocol[35,70-72,75]. In functioning tumors, clinical success should require complete resolution of hormone-related symptoms together with appropriate biochemical improvement[36,75,99].

Complete radiological ablation should be defined as the absence of residual enhancing tissue within the treated lesion on CE imaging[35,70-72,75]. Persistent enhancement after treatment should be classified as residual disease, whereas the reappearance of enhancing or measurable tumor within a previously non-enhancing ablation zone should be considered local recurrence[70,71,84]. These imaging-based definitions should not be interpreted as evidence of complete histological eradication, as residual viable tumor has occasionally been identified despite an apparently favorable radiological response[35,71].

No validated post-ablation surveillance schedule currently exists. Published protocols have included early CE imaging within 3-5 days to identify residual viable tissue and guide prompt retreatment[103], first formal assessment at approximately 1-3 months[35,70,71], subsequent imaging at 6 months and 12 months[35,71,72,75], and annual surveillance thereafter[70,75,84]. Accordingly, a pragmatic approach may include early CE imaging when immediate retreatment is being considered, followed by CE-CT, MRI, or CE-EUS at approximately 1-3 months, 6 months, 12 months, and annually thereafter. Long-term surveillance is particularly important for NF-panNENs because late local recurrence and delayed treatment failure have been reported more than 3 years after an apparently complete response[84].

CE-EUS may be particularly useful for identifying small foci of residual vascularized tissue when cross-sectional imaging is equivocal and for guiding targeted repeat ablation[70-72]. However, CT or MRI remains necessary to assess the entire pancreas and to detect regional nodal or distant progression[15,75].

Repeat ablation may be considered when residual or recurrent disease remains small, localized, biologically favorable, and safely accessible by EUS[35,36,70-72]. Surgical reassessment should be favored after repeated incomplete ablation, persistent viable tissue despite technically adequate treatment, rapid regrowth, increasing tumor size, ductal or vascular involvement, suspected nodal disease, upgrading, or other features suggesting aggressive biology[15,36,84]. In the comparative insulinoma study by Crinò et al[36], most local recurrences after EUS-RFA were successfully managed with repeat ablation, whereas a minority required surgery. Conversely, the long-term study by Barthet et al[84] supports surgical reconsideration when treatment failure persists, particularly when tumor biology may have been underestimated at pretreatment sampling.

LIMITATIONS AND FUTURE DIRECTIONS

Despite encouraging early results, EUS-guided treatment of small panNENs remains an evolving field with several important limitations. Most available evidence derives from retrospective series, often conducted in single high-volume centers, with substantial heterogeneity in patient selection, ablation techniques, procedural protocols, follow-up schedules, and outcome definitions. Long-term oncological efficacy, particularly with respect to local recurrence, disease progression, and metastatic spread, remains insufficiently characterized, especially for NF-panNENs.

Radiological response is commonly used as a surrogate endpoint, but histological confirmation of complete tumor necrosis is rarely available. The absence of contrast enhancement or the disappearance of the lesion on imaging cannot therefore exclude residual microscopic disease. This limitation is particularly relevant when local ablation is proposed as an alternative to surgical resection, which provides definitive pathological assessment. Technical challenges also remain. These include achieving homogeneous ablation in irregularly shaped lesions, limiting injury to the MPD, vessels, and surrounding pancreatic parenchyma, and treating tumors in difficult anatomical locations. Procedural outcomes are likely influenced by operator experience and institutional volume, supporting the concentration of these treatments in specialized referral centers.

The prospective Japanese multicenter study provided important data on the standardization and short-term outcomes of EUS-guided ethanol injection[72,73]. However, its single-arm design, historical surgical comparator, small sample size, and 6-month follow-up highlight the need for longer prospective and comparative studies. The ongoing randomized trial comparing EUS-RFA with surgery for insulinomas represents the most important source of forthcoming comparative evidence[38]. Priority areas for standardization include defining an adequate ablation endpoint, CE imaging at approximately 1 month and 6 months, uniform subsequent surveillance intervals, and consistent criteria distinguishing residual disease from local recurrence. Future research should also focus on direct comparisons between EA and RFA, standardized definitions of technical and clinical success, validated patient selection criteria, and structured long-term surveillance.

Moreover, no formal cost-effectiveness analyses have compared EUS-guided ablation with active surveillance or pancreatic surgery. Similarly, comparative data on patient-reported outcomes, quality of life, anxiety associated with surveillance, recovery time, and long-term functional consequences remain limited. Improved risk stratification will be essential. Tumor grade, growth rate, functional status, anatomical features, and advanced imaging findings should be integrated rather than considered separately. Additional biomarkers, including molecular alterations and a more reliable assessment of proliferative activity, may further help identify patients most likely to benefit from a minimally invasive approach.

Artificial intelligence, radiomics, molecular biomarkers, and emerging ablative technologies may eventually improve patient selection, procedural planning, and response assessment. However, their current clinical role remains investigational, and available data are largely derived from retrospective studies, small cohorts, preclinical models, or proof-of-concept analyses. Prospective validation, standardized endpoints, and demonstration of incremental clinical benefit are required before these approaches can be incorporated into routine management.

CONCLUSION

EUS-guided ablation represents a promising, minimally invasive, pancreas-preserving strategy for selected patients with small, well-differentiated panNENs. Its potential role lies between active surveillance and surgical resection, particularly when treatment is considered appropriate but the expected morbidity of pancreatic surgery appears excessive relative to the anticipated oncological benefit.

Surgical resection remains the reference treatment for medically fit patients requiring definitive therapy, although it may be associated with substantial short-term and long-term morbidity. Active surveillance remains appropriate for many small, asymptomatic, low-risk NF-panNENs, provided that structured follow-up is feasible.

The most convincing clinical application of EUS-guided ablation is in selected patients with insulinoma, particularly when surgery is declined, contraindicated, or associated with substantial morbidity. In NF-panNENs, its role remains investigational and should be limited to carefully selected cases discussed within a multidisciplinary team, with explicit acknowledgment of uncertain long-term oncological efficacy and the need for prolonged surveillance.

At present, EUS-guided ablation should not be regarded as an established standard treatment for small panNENs or as a routine alternative to surgery or surveillance. Prospective comparative studies with standardized definitions, long-term oncological outcomes, quality-of-life assessment, and cost-effectiveness analyses are required before broader clinical adoption can be recommended.