Four-step formula for the Siewert classification of adenocarcinomas of the esophagogastric junction: a computed tomography-based quantitative model.

Objective: This study aimed to develop a novel quantitative methodology to classify the Siewert type on multiplanar reconstructed (MPR) computed tomography (CT) and subsequently guide the diagnosis of adenocarcinoma of the esophagogastric junction.

Methods: Patients in the retrospective development and prospective validation cohort 1 were recruited from Peking University Cancer Hospital. Patients in prospective validation cohort 2 were recruited from Changzhi People's Hospital. A streamlined method determined axial/coronal CT slice numbers for the esophagogastric junction (EGJ) and tumor center; then using these slices formulated a Siewert classification system. The CT-based four-step formula for Siewert type (CT-FSFS) is: STEP 1: determining the axial level of the EGJ; STEP 2: distinguishing Type III from Type I/II; STEP 3: predicting Type III misclassified as II; STEP 4: correcting the misclassification using the deflection angle.

Results: The CT-FSFS method demonstrated robust accuracy across all three cohorts: 89.5% in the development cohort, 91.0% in validation cohort 1 and 89.7% in validation cohort 2. In differentiating Type I/II from Type III, the formula demonstrated high specificity (96.9%) and positive predictive value (97.1%) in the development cohort.

Conclusions: This study presents a reproducible, quantitative Siewert classification method on MPR CT, it may aid preoperative surgical planning.