Comparison of two region of interest definition methods for metabolic response evaluation with [18F]FDG-PET

AIM: In therapy response monitoring by [18F]2-fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET), different tumor delineations are used, resulting in different values for change in glucose metabolic rate (DMRglu). We propose a technique to compare metabolic rates in a region of interest (ROI) based on fixed volumes rather than on fixed thresholds. This method involves change in lesion size. METHODS: In 49 patients with colorectal carcinoma (CRC) and 50 patients with non-small cell lung carcinoma (NSCLC) scheduled for chemotherapy, FDG-PET was performed at baseline and during chemotherapy. A ROIfixed thresholds was determined by using a 50% threshold on both baseline and follow-up FDG-PET. A ROIfixed volumes was determined by using a 50% threshold, determined on the series with the largest tumor volume. This ROIfixed volumes is used on consecutive scans. Predictive effects of both methods were investigated by survival analysis for overall and progression free survival. RESULTS: In CRC, only ROIfixed volumes based DMRglu showed significant predictive ability. In NSCLC, both techniques showed significant predictive ability. During multivariate analysis, ROIfixed volumes determined DMRglu was an independent predictor for both overall and progression free survival in NSCLC whereas ROIfixed thresholds determined MRglu was not. After dichotomization at the median DMRglu, median survival ratio was higher in ROIfixed volumes than ROIfixed thresholds for CRC (overall survival: 1.78 vs 1.25, progression free survival: 1.57 vs 1.21) and NSCLC (overall survival: 2.01 vs 2.01, progression free survival: 2.93 vs 2.13). CONCLUSION: ROIfixed volumes based DMRglu shows better correlation with survival than DMRglu determined from a ROIfixed thresholds.