Prognostic value of 18 F-FDG PET/CT radiomics for survival and recurrence in non-small cell lung cancer: a systematic review of 3,180 patients

Abstract

BackgroundNon-small cell lung cancer (NSCLC) is a leading cause of cancer-related mortality, necessitating reliable prognostic tools to guide personalized treatment. 18 F-FDG positron emission tomography/computed tomography (PET/CT) radiomics, which extracts quantitative features from imaging, has emerged as a promising approach for predicting survival and recurrence outcomes. This systematic review evaluates the prognostic value of 18 F-FDG PET/CT radiomics in NSCLC, focusing on its predictive accuracy and clinical utility.MethodsA systematic search was conducted across PubMed, Embase, Web of Science, Scopus, and Cochrane Library, adhering to PRISMA guidelines (PROSPERO: CRD420251117214). Studies evaluating 18 F-FDG PET/CT radiomics for survival (overall survival OS, disease-free survival DFS, progression-free survival PFS) or recurrence in NSCLC were included. Data were extracted on study characteristics, radiomic features, prognostic models, and performance metrics. Methodological quality was assessed using the QUADAS-2 tool.ResultsFrom 1,245 records, 13 studies (n = 3,180 patients) met inclusion criteria. Radiomic features, including texture (e.g., Entropy, LowGrayLevelEmphasis), shape (e.g., Volume), and intensity (e.g., SUVmax), were consistently associated with survival and recurrence. Across studies, AUCs for OS prediction ranged from 0.76 to 0.968, with HRs of 2.0-3.9 indicating a 2- to 3.9-fold increased mortality risk with adverse radiomic signatures. Combined radiomic-clinical models outperformed single-modality models, though variability in feature extraction and validation strategies limited generalizability. Most studies exhibited moderate to high risk of bias, primarily due to retrospective designs and limited external validation.Conclusion18 F-FDG PET/CT radiomics shows preliminary prognostic potential in NSCLC, with reported predictive accuracies for survival outcomes ranging from moderate to high. However, methodological heterogeneity and validation limitations hinder clinical adoption. Standardized imaging protocols and prospective, multicenter studies are needed to enhance reproducibility and facilitate integration into personalized oncology.