Assessment of interobserver reproducibility in quantitative [¹⁸F]-fluoro-2-deoxy-glucose positron emission tomography/computed tomography measurements in non-small cell lung cancer

[¹⁸F]-fluoro-2-deoxy-glucose (FDG) positron emission tomography (PET) is widely used in both clinical practice and researches for non-small cell lung cancer (NSCLC). Various metabolic parameters are used in semiquantitative PET assessment and establishing the degree of reproducibility is important for their use as reliable factors. The purpose of this study was to evaluate interobserver reproducibility in quantitative PET measurements in NSCLC patients. PET images of 29 NSCLC patients were included. Three blinded readers assessed the maximum and peak standardized uptake values (SUVmax and SUVpeak) and metabolic tumor volumes (MTVs) of malignant lesions on PET images. Four cut-offs for MTVs were applied: mean liver SUV+2 SDs (MTVliver), a fixed value of 2.5 (MTV2.5), 50% of SUVmax (MTVmax50), and 50% of SUVpeak (MTVpeak50). Interobserver reproducibility of parameters was described by intraclass correlation coefficients (ICCs) and estimates of variance. Mean SUVs and MTVs were not significantly different among the three readers. The SUVmax and SUVpeak measurements were almost perfectly reproducible. The MTVmax50 and MTVpeak50 measurements showed excellent reproducibility (ICCs, 0.936-0.974). The MTVliver and MTV2.5 measurements were substantially reproducible (ICCs, 0.763-0.800). There was almost perfect interobserver reproducibility for SUVs from FDG PET measurements of NSCLC patients. MTVmax50 or MTVpeak50 may be appropriate to use for clinical guidance, because they yielded higher ICCs and narrower precision than MTVliver or MTV2.5. Although interobserver agreements for MTVmax50 and MTVpeak50 were excellent, careful consideration should still be given to determining cut-offs for response criteria or prognosis stratification because some variability does exist.