Letter to the Editor: Reconsidering relevance of endoscopic ultrasound-guided radiofrequency ablation in pancreatic cancer treatment: Future perspectives on artificial intelligence

Abstract

Despite advances in curative and technological approaches, pancreatic cancer (PC) remains a major global health challenge. In the majority of cases, PC is diagnosed at an advanced stage and/or in patients with significant comorbidities, which limits treatment options, including the feasibility of surgical resection. To address these challenges, endoscopic ultrasound-guided radiofrequency ablation (EUS-RFA) has emerged as a promising approach, although its outcomes remain debatable. Tong et al highlighted the role of EUS-RFA in PC treatment in their recent review. Furthermore, the existing literature lacks sufficient evidence regarding post-treatment outcomes, comparative assessments through multi-arm studies, and survival benefits. The exploration of artificial intelligence may potentially assist in optimizing radiofrequency ablation parameters, reducing operator dependency and inter-operator variability, thereby enhancing the efficacy of EUS-RFA procedures in future hypothetically.

Key Words: Pancreatic cancer; Endoscopic ultrasound-guided radiofrequency ablation; Artificial intelligence; Radiofrequency ablation; Endoscopy

Core Tip: Pancreatic cancer is a serious clinical challenge associated with high morbidity and mortality. Treatment options for non-resectable or high-risk patients are limited. Endoscopic ultrasound-guided radiofrequency ablation (EUS-RFA) is an emerging therapeutic approach, albeit with certain limitations. Tong et al highlighted the role of EUS-RFA in pancreatic cancer treatment in their recent review.

TO THE EDITOR

I read with great interest the article by Tong et al[1], titled “Endoscopicultrasound-guided radiofrequency ablation: A game changer in pancreatic cancertreatment”, published in the current issue of the World Journal of Gastrointestinal Oncology.

Tong et al[1] aimed to review the literature on a topic that represents one of the most challenging areas in pancreatic cancer (PC) treatment. The article discusses the role of endoscopic ultrasound-guided radiofrequency ablation (EUS-RFA) in PC and highlights existing controversies along with future directions.

However, several concerns merit attention: (1) The term “game changer” used in the title appears overstated, as EUS-RFA is still under evaluation and requires further validation through robust studies; (2) The manuscript is more descriptive than analytical, and there is a lack of structured data presentation from the included studies; and (3) There is limited inclusion of contemporary advancements, particularly in the domain of technological innovations in EUS-RFA.

CHALLENGES AND CONTEMPORARY DEVELOPMENTS

In EUS-RFA for PC, endoscopic ultrasound (EUS) enables the identification and characterization of lesions, while RFA facilitates targeted ablation. This combined approach leads to thermal tumor destruction, cytoreduction, and potential immunomodulatory effects[2]. Major technical challenges include the possibility of missing small lesions, operator dependency, and bias in determining optimal ablation parameters (Table 1).

Table 1 Outline of challenges, risks and suggestive remarks.

Challenges		Risk	Suggestive remarks
Technical	Lack of target specific optimized parameters for RFA	Complications including thermal injury	Development of AI-integrated target specific RFA modality
Procedure-related	Inter-operator variability, dependency on interventional endoscopist skills	Thermal injury, pancreatitis, collateral damage may cause bleeding, infection, perforation	Require highly skilled interventional endoscopist, AI-EUS integration for guidance[2], development of AI + EUS + RFA system
	Safety of vascular structures	Damage of vascular structures near ablation	High resolution colour Doppler guidance[3]
Precision	Probe FOV constrain	Irregular or small tumours may miss, suboptimal execution of procedure, chances of recurrence	Multiangle scanning, use of contrast-enhanced EUS for functional unambiguity and targeting of ROI, confirmations through multi-modalities such as MRI, CT. If needed, multiple episodes of procedure may execute
Investigative	Lack of standardized protocol for combination therapy	Suboptimal therapeutic response, recurrence	A large sample sized, RCT required, to assess comparative efficacy of treatment response and survival benefits, involving combination therapy e.g., immunotherapy and/or chemotherapy, with EUS-RFA procedure[1]
Financial	Costly	Less generalized reach for patients in under-resourced regions	Development of cost-effective system

AI: Artificial intelligence; RFA: Radiofrequency ablation; EUS: Endoscopic ultrasound; FOV: Field of view; ROI: Region of interest; RCT: Randomized controlled trial; MRI: Magnetic resonance imaging; CT: Computed tomography.

Recent technological advancements include the integration of artificial intelligence (AI). AI-integrated EUS has demonstrated promising performance for detection/characterization of pancreatic lesions, potentially reducing variability in lesion assessment. AI-enhanced EUS has demonstrated comparable outcomes to expert endosonographers (97.1% vs 100%) in detecting solid pancreatic lesions[3].

Additionally, systematic reviews and meta-analyses have demonstrated excellent diagnostic performance of AI-integrated EUS in PC, with pooled sensitivity ranging from 91% to 93%, specificity around 90%, and area under the curve values between 0.923 and 0.95[4-6].

Thus, AI integration with EUS offers a strategy for potentially supporting standardized pre-RFA lesion assessment, facilitating the overall EUS-RFA workflow. Furthermore, the development of future AI-assisted systems could theoretically support real-time optimizationof ablation parameters, however, whether this improves procedural or clinical outcomes remains to be determined.

However, diagnostic AI should not be equated with therapeutic AI. The ability of an AI system to detect or characterize a pancreatic lesion does not demonstrate that AI-guided RFA can achieve more complete tumour ablation, reduce complications, improve local control, or prolong survival. Thus, the current evidence supporting AI-assisted EUS provides a rationale for further investigation but does not establish the efficacy of AI-integrated EUS-RFA.

FUTURE DIRECTIONS

To establish clinical applicability and generate robust evidence, large-scale, multicenter randomized controlled trials are required, incorporating: (1) Patients with non-resectable or high-risk PC; (2) Standardized protocols for EUS-RFA; (3) Comprehensive follow-up evaluating short- and long-term outcomes, including adverse events, serious adverse events, and quality of life; and (4) Head-to-head comparisons between EUS-RFA and standard-of-care treatments in comparable patient cohorts to assess safety, efficacy, and survival benefits.

CONCLUSION

Furthermore, given that inter-operator variability remains a key limitation, the application of AI for automated optimization and delivery of ablation parameters could be a future possibility. This could eventually support investigation of semi-automated or robotic approaches, following appropriate validation. Therefore, the development and clinical validation of AI-integrated EUS-RFA systems—focusing on diagnostic accuracy, procedural consistency, and safety—should be considered a priority.

ACKNOWLEDGEMENTS

We acknowledge the authors of the referenced publications for their valuable contributions to the evolving field of EUS-guided radiofrequency ablation and artificial intelligence in PC.