Estimation of esophageal involvement length in esophagogastric junction tumors using PET-CT

BACKGROUND: Esophageal involvement length (EIL) is crucial to determining the surgical strategy for esophagogastric junction (EGJ) tumors. This study developed a positron emission tomography-computed tomography (PET-CT)-based method using anatomical landmarks to estimate EIL and evaluated the clinical utility of the model. METHODS: We enrolled 50 patients with EGJ adenocarcinoma who underwent surgical resection. Among patients who underwent upfront surgery (Cohort 1, n = 24), craniocaudal distances were measured from the right and left diaphragms and vena cava foramen (VCF) to the proximal tumor margin on PET-CT images and correlated with histological EIL to establish the best-performing model. Predictive ability was validated in patients who underwent surgery after neoadjuvant chemotherapy (Cohort 2, n = 26). RESULTS: In Cohort 1, among the three anatomical landmarks, VCF most strongly correlated with histological EIL (r = 0.81), providing the following formula: Predicted EIL (mm) = 22 + 0.3 × [Craniocaudal distance from VCF to proximal tumor margin (mm)]. The median absolute difference between the histological and predicted EILs was 4.6 mm (interquartile range, 1.6-6.3 mm), with 96% of cases (23/24) within 10 mm. Evaluation of the model in Cohort 2 found r = 0.81, a median absolute difference of 3.3 mm (interquartile range, 1.1-7.2 mm), and 92% of cases (24/26) within 10 mm. Predictive errors were significantly smaller in both cohorts than in endoscopic findings (Cohort 1, P = 0.004; Cohort 2, P = 0.002). CONCLUSIONS: A PET-CT-based approach using the craniocaudal distance from the VCF to the proximal margin could estimate EIL with sufficient accuracy to support surgical decision-making in clinical practice.