432 chapter 9 as post-test probabilities, including predictive values and benign call rates, not only depend on test characteristics (i.e., sensitivity and specificity) but also pre-test probability (i.e., RoM) which varied widely from 4.2-50% among studies included. Pooled NPV was 99% for PET-only imaging, but varied from 74%-91% for PET/CT with rather robust but limited positive predictive values similar to previous work (34-37%) [658, 659] and similar to the RoM of [18F]FDG-incidentalomas in random patients [451, 660]. Also, the prevalence of a negative test, i.e., the benign call rate, varied between 37-92% between studies [658, 659]. The variation in RoM clearly reflects heterogeneity of the study populations including whether or not to include FTUMP and NIFTP as benign lesions, the variation in sensitivities reflect large differences in methodology. Drilling down in the three studies with strikingly high false-negative rates of 8%-19% [308, 502, 520], suggest that at least the inclusion of incidental, clinically irrelevant carcinomas in the millimetre range contributed in one study [502]. Qichang et al. tried to explain the apparent difference in performance of PET-only versus hybrid PETCT by speculating that atypical [18F]FDG-uptake in the neck (tonsils, pharynx, cervical lymph nodes and other thyroid nodules) may have been falsely attributed to a nearby false-negative thyroid nodule under study on PET-only imaging, which would not have been the case if hybrid PET-CT was performed. Furthermore, they argue that the definition of a positive PET was less strict in the earlier PET-only studies, causing this misinterpretation to occur more often [659]. The call for large, prospective, multicentre studies with unified image interpretation protocols, incorporating the evolving TBSRTC and uniform appraisal of lesion such as FTUMP and NIFTP encouraged us to design and execute a nationwide randomised-controlled trial in 15 academic and non-academic centres (“Efficacy of [18F]FDG-PET in Evaluation of Cytological indeterminate Thyroid nodules prior to Surgery (EfFECTS)”, NCT02208544). The main results were published earlier this year in EJNMMI [661-663]. We randomised 132 patients with indeterminate (Bethesda III and IV) cytology to either the experimental arm (diagnostic hemithyroidectomy only when the nodule was FDG-positive) or standard of care (diagnostic hemithyroidectomy regardless the result of the [18F] FDG-PET/CT). Patients in the experimental arm who were managed without surgery (i.e., negative [18F]FDG-PET/CT) were followed-up by their endocrinologists according to the risk of a benign nodule, including an ultrasonography after one year. The [18F]FDG-PET/CT-driven approach indeed did reduce the number of futile surgeries by 40% (48% in non-Hürthle cell nodules). No malignant or borderline tumours were observed in patients under surveillance. Sensitivity, specificity, negative and positive predictive value, and benign call rate of [18F]FDG-PET/CT were 94.1%, 39.8%, 95.1%, 35.2% and 31.1%, respectively [661]. This was fully in line of our 2011 meta-analysis [304] and the observed high NPV fits the American Thyroid Association 2015-guideline statement on an ideal “rule-out test” [17]. In the group of patients with a negative FDG-PET/CT, very few patients crossed over in the study from watchful waiting to surgery for fear of missed diagnosis or persistent obstructive complains of a nodule. None of these had a malignancy. The 2/132 [18F]FDG-PET/CT scans considered a false-
RkJQdWJsaXNoZXIy MTk4NDMw