Thirteen studies were included (n more than 490; one study reported the number of tumours only, n=59). Four studies were retrospective in design; it was unclear for the rest.
Seven studies investigated the performance of FDG-PET in distinguishing tumour from radiation necrosis in recurrent brain lesions. In two studies assessing the impact on therapeutic choice, FDG-PET did not appear to have a benefit over MRI. There were five diagnostic accuracy studies addressing this question. With the exception of one study with only one patient without recurrence, the sensitivity of FDG-PET ranged from 67 to 83% and the specificity from 50 to 62%. FDG-PET was compared with SPECT/SPET in three of these studies:
in one study, FDG-PET sensitivity and specificity were 50% and 80%, respectively, for the diagnosis of radiation necrosis and recurrence, while SPECT sensitivity and specificity were 50% and 75% for radiation necrosis and 75% and 50% for recurrence;
in the second study, FDG-PET had 67% sensitivity and 100% specificity while both sensitivity and specificity were 100% for SPET;
in the third study, sensitivity and specificity for recurrence were 76% and 100% (based on one patient), respectively, for FDG-PET and 70% and 100% (based on one patient) for SPET.
No studies addressing the performance of FDG-PET guided lesion biopsy compared with conventional imaging in patients with recurrent brain tumour were identified; nor were there any studies addressing the performance of FDG-PET in distinguishing between tumour grades compared with biopsy when a new brain tumour is deemed indeterminate by biopsy. Four studies of patients with a definite biopsy grade were reported, though these did not meet the inclusion criteria for the review.