Ricerc@Sapienza

Purpose: The Italian consensus to classify thyroid cytology has provided a standardized reporting scheme, including the subdivision of indeterminate for malignancy TIR-3 category into TIR-3A (low-risk) and TIR-3B (high-risk). We aimed to present our experience on this subclassification by evaluating risks of malignancy and the validity in sorting nodules with dissimilar risks. Another aim was to compare our performance against the Bethesda system. Methods: Fine-needle aspirates of 290 TIR-3 that underwent thyroid surgery at our hospital (2008–2013) were reviewed and divided into TIR-3A or TIR-3B, and AUS/FLUS or FN/SFN. Cytological diagnoses were then correlated to histology. Results were evaluated using univariate analysis. Results: The subclassification into TIR-3A and TIR-3B differentiated hyperplastic nodules (p = 0.000) but not adenomas (p = 0.090). Rates of malignancy were significantly different between TIR-3A (10.2%) and TIR-3B (43.8%); TIR-3B malignancies were often papillary carcinomas (83%). TIR-3A/TIR-3B accounted for high sensitivity (84.5%; CI 79.7–88.4%), accuracy (64.1%; CI 58.6–69.6%) and NPV (89.8%; CI 85.6–93.0%) as opposed to modest specificity (55.8%; CI 49.9–61.6%) and PPV (43.8%; CI 38.1–49.8%). The rate of malignancy in AUS-FLUS was higher than in TIR-3A (p = 0.007), whereas it was not different between FN/SFN and TIR-3B (p = 0.337). Sensitivity of the Bethesda system was significantly lower respect to the Italian system. Conclusions: The study supports the Italian consensus showing a different risk of malignancy for TIR-3A as compared to TIR-3B. TIR-3A/TIR-3B subclassification is valid to sort out benign nodules (high NPV) and malignancies (high sensitivity) but not adenomas (modest specificity, low PPV). In our experience, sensitivity is the main difference between Italian and Bethesda systems.