P1.169. Magnetic and ICG High Risk Lymph Node Mapping in Upper GI Cancer: A Safety, Feasibility and Diagnostic Performance Clinical Study

Abstract

Abstract Topic Esophageal Cancer: Surgical Treatment of Esophageal Cancer Background Accurate localisation of high-risk lymph node (HRLN) stations in upper gastrointestinal cancer could enable tailored resection and focused ultra-staging while avoiding unnecessary lymphadenectomy. Clinically practical tracers and workflows remain limited. Preliminary results are provided at submission, with planned update to the final dataset before abstract publication. Methods This multi-centre observational comparator study enrolled patients with gastric, gastro-oesophageal junction, or oesophageal cancer undergoing curative-intent surgery with lymphadenectomy (planned n=62). A SPION-based iron oxide tracer (FerroTrace) was injected endoscopically peri-tumourally before neoadjuvant therapy to minimise treatment-related lymphatic disruption. MRI identified and localised HRLN stations; treating teams remained blinded to MRI to maintain standard-of-care surgery. Indocyanine green fluorescence was used intraoperatively to confirm no fluorescent nodes remained after lymphadenectomy and on the back table to identify fluorescent nodes in the specimen. A handheld magnetometer was used on the back table to detect iron oxide signal and confirm retrieval of MRI-identified HRLNs for detailed pathology. HRLNs underwent serial sectioning with immunohistochemistry; all nodes received standard histopathology. Adverse events and serious adverse events were recorded and assessed for relatedness. Results Preliminary results (25 participants): Sixteen adverse events occurred, including two serious adverse events; all were unrelated to FerroTrace except one minor related MRI artefact from superficial tracer deposition. Fourteen patients had interpretable MRI with matched pathology; others did not proceed to surgery and/or had non-interpretable scans due to acquisition limitations (including incomplete regional coverage and some 1.5T scans). Fifty-nine HRLNs were identified (mean 4.3/patient). In 9/9 node-positive patients, MRI identified at least one pathologically positive HRLN station (patient-level FNR 0%). Node-negative patients had no malignant MRI-identified HRLNs (patient-level NPV 100%), with overall patient-level accuracy 100% in this interim cohort. Of resected HRLNs, 31.1% were positive overall, increasing to 41.5% in node-positive patients. Iron oxide signal remained detectable by back-table magnetometer nine months after injection in one patient. In 10/14 (71%) patients, at least one MRI-identified HRLN lay outside standard lymphadenectomy fields and was not resected due to blinding. Conclusion Preliminary results support feasibility and safety of MRI-guided HRLN localisation with specimen-based magnetic confirmation and adjunct fluorescence confirmation, with no related serious adverse events. Early patient-level performance showed no false negatives in node-positive patients and malignancy enrichment within MRI-identified HRLNs. Detectable signal at nine months suggests durable node retention that may enable mapping despite delayed surgery. These findings support standardised MRI acquisition and future evaluation of an individualised (unblinded) resection strategy to reduce potential under-staging.