Since the influential study by van Tinteren et al. published in The Lancet in 2002, there have been an increasing number of diagnostic randomized controlled trials (RCTs) investigating the benefit of PET. If they provide valid and useful information on the benefit, these studies can play an important role in informing guideline developers and policy makers. Our aim was to investigate how far the nuclear medicine community has come on its way from accuracy studies to RCTs and which issues we have to take into account in planning future studies. METHODS: We conducted a systematic review of diagnostic randomized trials, in which PET was applied in only one arm. We covered published studies as well as registered unpublished and planned studies. We considered 3 quality indicators related to the usefulness of a trial to generate evidence for a clinical benefit: use of patient-important outcome, sufficient sample size, and current standard as comparator. RESULTS: Fourteen published and 15 planned studies were identified. Five of the published studies and 12 of the planned studies did not use a patient-important outcome. Sample sizes were often so small that a significant result could be expected only under the assumption of a substantial reduction in the event rate. Comparators typically reflected the current standard. CONCLUSION: If we consider the traditional areas of primary diagnosis, staging, and follow-up, then the number and quality of RCTs on PET is currently not sufficient to provide a major source for evidence-based decisions on the clinical benefit of PET. There will also be a future need in these traditional areas to deduce the clinical benefit of PET from the results of accuracy studies. The situation may be more favorable for the areas of treatment planning and response evaluation. Choice of patient-important outcomes and sufficient sample sizes are crucial issues in planning RCTs to demonstrate the clinical benefit of using PET.
Journal of Nuclear Medicine, 2014, Vol 55, Issue 8, p. 1228-1234