A Step Toward a Global Consensus on Gastric Cancer Resectability Integrating Artificial Intelligence-Based Consensus...

Area of Science:

• Surgical Oncology
• Artificial Intelligence in Medicine
• Gastric Cancer Management

Background:

• Surgical resection is key for curative treatment of locally advanced gastric cancer (GC).
• Global variability exists in defining resectability, especially with multivisceral invasion, positive peritoneal cytology (CY1), or oligometastatic disease.
• The Intercontinental Criteria of Resectability for Gastric Cancer (ICRGC) project was initiated to address these challenges.

Purpose of the Study:

• To develop standardized resectability criteria for gastric cancer (GC) using expert surgical input and artificial intelligence (AI).
• To assess the concordance between expert surgeons and AI-driven recommendations in complex GC resectability scenarios.
• To evaluate AI's potential to harmonize global surgical practices in GC management.

Main Methods:

• A two-stage prospective survey was conducted among 58 surgical oncologists at the 2024 European Gastric Cancer Association (EGCA) meeting.
• Participants completed a questionnaire on resectability, strategy, and quality metrics, followed by a review of AI-generated responses.
• Concordance between human and AI responses was analyzed, and changes in surgeon opinions post-AI exposure were statistically evaluated.

Main Results:

• High agreement was found between surgeons and AI on distinguishing technical from oncological resectability (79%).
• Significant alignment was observed in complex cases like cT4b (61% support for high-volume centers), CY1 (54%), and N3 nodal disease (63%).
• AI exposure led to guideline-consistent shifts in surgeon decisions, with 27.1% changing answers, notably in surgical margin definition (p=0.015) and anatomical criteria (p<0.05).

Conclusions:

• AI-driven consensus modeling effectively replicates expert reasoning in complex surgical oncology cases.
• The ICRGC project demonstrates AI's potential to serve as a catalyst for harmonizing global surgical practices in gastric cancer.
• AI-supported consensus modeling can complement expert judgment, enhancing decision-making consistency in ambiguous or controversial GC resectability scenarios.