本研究旨在评估基于超声 (US) RSS、结节大小和细胞学子类别的综合风险分层系统 (RSS) 的性能，用于诊断最初在细针抽吸上被确定为 Bethesda III 类的甲状腺结节恶性肿瘤。研究在两个机构进行，包括连续患有 Bethesda III 类结节的患者，并通过重复活检或手术确认最终诊断。共纳入 309 名患者（223 名女性和 86 名男性；平均年龄，50.9 ± 12.0 岁）的 320 个 Bethesda III 类结节（≥1 cm）。根据 US RSS、结节大小和细胞学子类别评估 Bethesda III 类结节的恶性风险和恶性肿瘤的预测因子。比较了 US 大小细胞学 (USC) RSS 和单独 US RSS 对恶性肿瘤的诊断性能。中间或高度怀疑的 US 类别独立增加了所有 US RSS 中的恶性肿瘤风险 (P ≤ 0.001)。大结节尺寸（≥3 cm）独立增加低度或中度可疑美国类别结节的恶性风险。此外，在大多数美国 RSS 中，未确定意义的细胞学子类别的异型性独立增加了低怀疑美国类别结节的恶性风险。 USC RSS 的受试者工作特征曲线下面积大于单独的 US RSS 的受试者工作特征曲线下面积 (P < 0.048)。在 USC RSS 的极低风险类别中未发现恶性肿瘤。在 Bethesda III 类结节中，USC RSS 对恶性肿瘤的诊断性能优于单独使用 US RSS。 USC RSS 的极低风险类别中可以排除恶性肿瘤。版权所有 © 2024 韩国放射学会。

This study aimed to evaluate the performance of an integrated risk stratification system (RSS) based on ultrasound (US) RSSs, nodule size, and cytology subcategory for diagnosing malignancy in thyroid nodules initially identified as Bethesda category III on fine-needle aspiration.This retrospective study was conducted at two institutions and included consecutive patients with Bethesda category III nodules, and final diagnoses confirmed by repeat biopsy or surgery. A total of 320 Bethesda category III nodules (≥1 cm) from 309 patients (223 female and 86 male; mean age, 50.9 ± 12.0 years) were included. The malignancy risk of Bethesda category III nodules and predictors of malignancy were assessed according to US RSSs, nodule size, and cytology subcategory. The diagnostic performances of US-size cytology (USC) RSS and US RSS alone for malignancy were compared.The intermediate or high suspicion US category independently increased the malignancy risk in all US RSSs (P ≤ 0.001). Large nodule size (≥3 cm) independently increased the malignancy risk of low- or intermediate suspicion US category nodules. Additionally, the atypia of undetermined significance cytology subcategory independently increased the malignancy risk of low suspicion US category nodules in most US RSSs. The area under the receiver operating characteristic curve of the USC RSSs was greater than that of the US RSSs alone (P < 0.048). Malignancy was not found in the very low risk category of USC RSS.The diagnostic performance of USC RSS for malignancy was superior to that of US RSS alone in Bethesda category III nodules. Malignancy can be ruled out in the very low-risk category of USC RSS.Copyright © 2024 The Korean Society of Radiology.