PET with the glucose analog (18)F-FDG is increasingly used to monitor tumor response to therapy. To use quantitative measurements of tumor (18)F-FDG uptake for assessment of tumor response, the repeatability of this quantitative metabolic imaging method needs to be established. Therefore, we determined the repeatability of different standardized uptake value (SUV) measurements using the available data.
Methods: A systematic literature search was performed to identify studies addressing (18)F-FDG repeatability in malignant tumors. The level of agreement between test and retest values of 2 PET uptake measures, maximum SUV (SUV(max)) and mean SUV (SUV(mean)), was assessed with the coefficient of repeatability using generalized linear mixed-effects models. In addition, the influence of tumor volume on repeatability was assessed. Principal component transformation was used to compare the reproducibility of the 2 different uptake measures.
Results: Five cohorts were identified for this metaanalysis. For SUV(max) and SUV(mean), datasets of 86 and 102 patients, respectively, were available. Percentage repeatability is a function of the level of uptake. SUV(mean) had the best repeatability characteristics; for serial PET scans, a threshold of a combination of 20% as well as 1.2 SUV(mean) units was most appropriate. After adjusting for uptake rate, tumor volume had minimal influence on repeatability.
Conclusion: SUV(mean) had better repeatability performance than SUV(max). Both measures showed poor repeatability for lesions with low (18)F-FDG uptake. We recommend the evaluation of biologic effects in PET by reporting a combination of minimal relative and absolute changes to account for test-retest variability.