Assessing FDG-PET diagnostic accuracy studies to develop recommendations for clinical use in dementia.

BACKGROUND: FDG-PET is frequently used as a marker of synaptic damage to diagnose dementing neurodegenerative disorders. We aimed to adapt the items of evidence quality to FDG-PET diagnostic studies, and assess the evidence available in current literature to assist Delphi decisions for European recommendations for clinical use.
METHODS: Based on acknowledged methodological guidance, we defined the domains, specific to FDG-PET, required to assess the quality of evidence in 21 literature searches addressing as many Population Intervention Comparison Outcome (PICO) questions. We ranked findings for each PICO and fed experts making Delphi decisions for recommending clinical use.
RESULTS: Among the 1435 retrieved studies, most lacked validated measures of test performance, an adequate gold standard, and head-to-head comparison of FDG-PET and clinical diagnosis, and only 58 entered detailed assessment. Only two studies assessed the accuracy of the comparator (clinical diagnosis) versus any kind of gold-/reference-standard. As to the index-test (FDG-PET-based diagnosis), an independent gold-standard was available in 24% of the examined papers; 38% used an acceptable reference-standard (clinical follow-up); and 38% compared FDG-PET-based diagnosis only to baseline clinical diagnosis. These methodological limitations did not allow for deriving recommendations from evidence. DISCUSSION: An incremental diagnostic value of FDG-PET versus clinical diagnosis or lack thereof cannot be derived from the current literature. Many of the observed limitations may easily be overcome, and we outlined them as research priorities to improve the quality of current evidence. Such improvement is necessary to outline evidence-based guidelines. The available data were anyway provided to expert clinicians who defined interim recommendations.