Context: The decision on diagnostic lobectomy for follicular neoplasms (FN) is challenging. Objective: This meta-analysis investigates whether an appropriate size cutoff exists for recommending surgery for thyroid nodules diagnosed as FN by fine needle aspiration. Methods: The Ovid-Medline, EMBASE, Cochrane, and KoreaMed databases were searched for studies reporting the malignancy rate of FN/suspicious for FN (FN/SFN) according to tumor size, using search terms "fine needle aspiration,""follicular neoplasm,""lobectomy,""surgery,"and "thyroidectomy."Results: Fourteen observational studies comprising 2016 FN/SFN nodules with postsurgical pathologic reports were included, and 2 studies included malignancy rates with various tumor sizes. The pooled malignancy risk of FN/SFN nodules according to size was: odds ratio (OR) 2.29 (95% CI, 1.68-3.11) with cutoff of 4 cm (9 studies), OR 2.39 (95% CI, 1.45-3.95) with cutoff of 3 cm (3 studies), and OR 1.81 (95% CI, 0.94-3.50) with cutoff of 2 cm (5 studies). However, tumors ≥2 cm also showed a higher risk (OR 2.43; 95% CI, 1.54-3.82) based on the leave-one-out meta-analysis after removal of 1 influence study. When each cutoff size was evaluated by summary receiver operating characteristic (sROC) curves, the cutoff of 4 cm showed the highest summary area under the curve (sAUC, 0.645) compared to other cutoffs (sAUC, 0.58 with 2 cm, and 0.62 with 3 cm), although there was no significant difference. Conclusion: Although the risk of malignancy increases with increasing tumor size, the risk remains significant at all tumor sizes and no cutoff limit can be recommended as a decision-making parameter for diagnostic surgery in Bethesda IV thyroid nodules.
