A systematic review of the prognostic value of texture analysis in 18F-FDG PET in lung cancer.

Abstract

The aim of this study was to perform a systematic review of the prognostic value of texture parameters derived by 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) in patients with lung cancer. PubMed and EMBASE databases were searched up to March 12, 2018, for original articles involving texture analysis for the prediction of prognosis in patients with lung cancer. Risk of bias in the studies was critically assessed using the QUIPS tool. The results of survival analysis in the included studies were compared. Of the 446 articles retrieved, 17 studies were eligible for inclusion. Our review suggests that the prognostic value of texture parameters in lung cancer remains unproven. Most studies had a moderate to high risk of bias. Texture parameters that described prognosis were not replicated across studies. Conflicting results on hazard ratios were found among the studies. This discrepancy is partly explained by false-positive findings originating from statistical error and variability caused by different methodologies used for image acquisition and processing in the included studies. Based on currently available evidence, there is insufficient evidence to support the prognostic value of texture analysis in 18F-FDG PET in lung cancer. Further studies implementing well-established methodologies and statistical evidence are warranted for proper validation of these promising imaging biomarkers.