Risk of malignancy in follicular neoplasms without nuclear atypia: statistical analysis of 397 thyroidectomies.

OBJECTIVE: To determine how to triage patients with a follicular neoplasm (FN), without nuclear atypia reported by fine-needle aspiration, on the basis of risk factors. METHODS: The age, sex, tumor size, and cell type of 397 patients who underwent thyroidectomy for follicular carcinoma (FC) or follicular adenoma between 1991 and 2001 were analyzed statistically. The likelihood ratio and probability of FC for various combinations of tumor size, sex, and cell type were estimated with use of Bayes' theorem. RESULTS: FC was significantly associated with tumor size >2.1 cm (P = 0.048), male sex (P = 0.0007), and Hürthle cell type (P<0.0001). The mean size of minimally invasive FC was significantly smaller (2.9 cm versus 4.8 cm; P = 0.004) and the mean patient age was significantly younger (47.6 years versus 61.0 years; P = 0.003) than for widely invasive FC. The lowest probability (0.31%) for FC was in female patients with a small (< or = 2.1 cm) micro-follicular FN reported by a cytopathology practice with 10% accuracy of true FN at surgical intervention, whereas the highest probability (29.5%) for FC was in male patients with a large (>2.1 cm) Hürthle cell neoplasm reported by a cytopathology practice with 70% accuracy of true FN at surgical intervention. CONCLUSION: Although an estimate of probability for FC based on age, sex, thyroid nodule size, and cell type is provided in this study for patients diagnosed with FN without nuclear atypia on fine-needle aspiration, the variability of the accuracy in cytopathology practice makes it difficult to change the current treatment paradigm, which requires carefully planned prospective studies with long-term follow-up.