We developed and validated a new staging system that includes metabolic information from pretreatment [18F]Fluorodeoxyglucose ([18F]FDG) positron emission tomography/computed tomography (PET/CT) for predicting disease-specific survival (DSS) in gastric cancer (GC) patients. Overall, 731 GC patients undergoing preoperative [18F]FDG PET/CT were enrolled and divided into the training (n = 543) and validation (n = 188) cohorts. A metabolic score (MS) was developed by combining the maximum standardized uptake value (SUVmax) of the primary tumor (T_SUVmax) and metastatic lymph node (N_SUVmax). A new staging system incorporating the MS and tumor-node-metastasis (TNM) stage was developed using conditional inference tree analysis. The MS was stratified as follows: score 1 (T_SUVmax ≤ 4.5 and N_SUVmax ≤ 1.9), score 2 (T_SUVmax > 4.5 and N_SUVmax ≤ 1.9), score 3 (T_SUVmax ≤ 4.5 and N_SUVmax > 1.9), and score 4 (T_SUVmax > 4.5 and N_SUVmax > 1.9) in the training cohort. The new staging system yielded five risk categories: category I (TNM I, II and MS 1), category II (TNM I, II and MS 2), category III (TNM I, II and MS ≥ 3), category IV (TNM III, IV and MS ≤ 3), and category V (TNM III, IV and MS 4) in the training cohort. DSS differed significantly between both staging systems; the new staging system showed better prognostic performance in both training and validation cohorts. The MS was an independent prognostic factor for DSS, and discriminatory power of the new staging system for DSS was better than that of the conventional TNM staging system alone.
© 2022. The Author(s).