Objective: To evaluate the utility of the 9-gene panel as a differential diagnostic method for thyroid nodules within determinate cytological diagnosis and as a parallel diagnostic method for thyroid fine-needle aspiration (FNA) cytology. Methods: 579 liquid-based cytology samples from 544 patients were collected after thyroid FNA diagnosis in our hospital from December 2014 to April 2021. Mutations at any site of 9 genes, namely, BRAF, NRAS, HRAS, KRAS, GNAS, RET, TERT, TP53, and PIK3CA as recorded by the Catalogue of Somatic Mutations in Cancer (COSMIC), were analyzed by next-generation sequencing. Taking postoperative histopathology and cytology results with definite benign or malignant diagnosis as the gold standard, the diagnostic efficacy of the 9-gene panel as a reclassified method for thyroid nodules with indeterminate cytological diagnosis and as a parallel diagnostic method for thyroid FNA cytology were evaluated and compared with that of the BRAF V600E single-gene detection method. Results: Of the 579 thyroid nodules, 196 (33.85%) were Bethesda Ⅱ, 11 (1.90%) were Bethesda Ⅲ, 31 (5.35%) were Bethesda Ⅳ, 27 (4.66%) were Bethesda Ⅴ, and 314 (54.23%) were Bethesda Ⅵ, as diagnosed by thyroid FNA cytology. Among these 579 thyroid nodules, 275 were tested positive for 9-gene mutations, with a mutation rate of 47.5%. Of the 329 thyroid nodules surgically removed, 30 (9.12%) were benign, 5 (1.52%) were borderline, and 294 (89.36%) were malignant. Regarding borderline nodules as malignant nodules, the mutation rates of the 9 genes in the 299 malignant thyroid nodules from high to low were BRAF 62.21% (186/299), NRAS 5.02% (15/299), HRAS 1.00% (3/299), PIK3CA 0.67% (2/299), GNAS 0.67% (2/299), KRAS 0.33% (1/299), TP53 0.33% (1/299), TERT 0.33% (1/299) and RET 0.00% (0/299). The malignant risks of the 9 genes from high to low were BRAF 100% (186/186), PIK3CA 100.00% (2/2), GNAS 100.00% (2/2), TERT 100.00% (1/1), TP53 100.00% (1/1), NRAS 78.95% (15/19), HRAS 75.00% (3/4), and KRAS 50.00% (1/2). For thyroid nodules of Bethesda Ⅲ-Ⅳ (indeterminate diagnosis), the sensitivity (SN) of the 9-gene panel in diagnosing thyroid cancer is 34.48% (10/29), the specificity (SP) is 61.54% (8/13), and the accuracy is 42.86% (18/42); whereas the SN of the BRAF V600E detection method is 0%. Therefore, the diagnostic efficiency of the 9-gene panel is significantly better than that of BRAF V600E single gene detection. For thyroid nodules of Bethesda Ⅱ-Ⅵ, the SN of the 9-gene panel in diagnosing thyroid cancer was 68.83% (254/369), the SP was 90.00% (189/210), the accuracy was 76.51% (443/579), and the area under the curve (AUC) was 0.79; whereas the SN of BRAF V600E single-gene detection in diagnosing thyroid cancer was 63.69% (235/369), the SP was 99.52% (209/210), the accuracy was 76.68% (444/579), and the AUC was 0.82. The SP of BRAF V600E detection is higher than that of the 9-gene panel (P＜0.01), but there is no significant difference in SN, accuracy (both P＞0.05), and AUC (Z=0.85, P=0.396) between them. Gene mutations indicating poor prognosis were detected in 4 nodules of papillary thyroid carcinoma and 1 nodules of follicular thyroid carcinoma, including 2 nodules with TERT and BRAF V600E co-mutations, 1 nodule with TP53 mutation, and 2 nodules with PIK3CA mutation. Conclusions: As a reclassified method for thyroid lesions with indeterminate cytological diagnosis, the 9-gene panel is better than BRAF V600E single gene detection. As a parallel diagnostic method of thyroid FNA cytology, the 9-gene panel has similar diagnostic efficacy as BRAF V600E single-gene detection. The 9-gene panel can detect individual cases with gene mutations indicating poor prognosis. The identification of patients with these special gene mutations has certain implications for the clinical management of them.
目的： 评价9基因突变检测作为细胞学诊断不明确甲状腺病变的鉴别诊断方法和甲状腺细针穿刺细胞学检查的平行诊断方法的意义。 方法： 2014年12月至2021年4月在中国医学科学院肿瘤医院行甲状腺细针穿刺细胞学检查的544例患者的579个甲状腺结节，收集其细胞学诊断后剩余的液基细胞学样本，采用下一代测序技术检测B-Raf原癌基因（BRAF）、成神经细胞瘤大鼠肉瘤病毒癌基因同源物（NRAS）、Harvey大鼠肉瘤病毒癌基因同源物（HRAS）、Kirsten大鼠肉瘤病毒癌基因同源物（KRAS）、肿瘤蛋白p53（TP53）、端粒酶逆转录酶（TERT）、磷脂酰肌醇-4，5-二磷酸3-激酶催化亚基α（PIK3CA）、鸟苷酸结合蛋白活性刺激肽（GNAS）和转染重排（RET）这9个基因任何癌症体细胞突变目录所收录位点的突变状态。以术后组织病理诊断和明确的细胞学诊断为金标准，比较9基因突变检测与BRAF V600E单基因检测的诊断效力。 结果： 579个甲状腺结节中，细针穿刺细胞学诊断196个为Bethesda Ⅱ级，11个为Bethesda Ⅲ级，31个为Bethesda Ⅳ级，27个为Bethesda Ⅴ级，314个为Bethesda Ⅵ级；9基因突变检测阳性275个，突变率为47.5%。手术切除的329个甲状腺结节中，术后病理诊断30个为良性，5个为交界性，294个为恶性。将交界性结节视为恶性结节时，299个恶性结节中9基因的突变率由高到低依次为BRAF 62.21%（186/299），NRAS 5.02%（15/299），HRAS 1.00%（3/299），PIK3CA 0.67%（2/299），GNAS 0.67%（2/299），KRAS 0.33%（1/299），TP53 0.33%（1/299），TERT 0.33%（1/299），RET 0.00%（0/299）。9基因突变的恶性风险由高到低依次为BRAF 100%（186/186），PIK3CA 100.00%（2/2），GNAS 100.00%（2/2），TERT 100.00%（1/1），TP53 100.00%（1/1），NRAS 78.95%（15/19），HRAS 75.00%（3/4），KRAS 50.00%（1/2）。对于细胞学诊断为Bethesda Ⅲ～Ⅳ级（诊断不明确）的甲状腺结节，9基因突变检测诊断甲状腺癌的灵敏度为34.48%（10/29），特异度为61.54%（8/13），准确率为42.86%（18/42），而BRAF V600E单基因检测诊断甲状腺癌的灵敏度为0。对于细胞学诊断为Bethesda Ⅱ～Ⅵ级的甲状腺结节，9基因突变检测诊断甲状腺癌的灵敏度为68.83%（254/369），特异度为90.00%（189/210），准确率为76.51%（443/579），受试者工作特征曲线的曲线下面积（AUC）为0.79；BRAF V600E单基因检测诊断甲状腺癌的灵敏度为63.69%（235/369），特异度为99.52%（209/210），准确率为76.68%（444/579），AUC为0.82。BRAF V600E单基因检测的特异度高于9基因突变检测（P＜0.01），而灵敏度、准确率和AUC与9基因突变检测差异无统计学意义（均P＞0.05）。有4个诊断为甲状腺乳头状癌的结节和1个诊断为甲状腺滤泡癌的结节检测到提示不良预后的基因突变，其中2个结节检测到TERT与BRAF V600E共突变，2个结节检测到PIK3CA突变，1个结节检测到TP53突变。 结论： 作为细胞学诊断不明确的甲状腺病变（Bethesda Ⅲ～Ⅳ级）的鉴别诊断方法时，9基因突变检测优于BRAF V600E单基因检测。作为甲状腺细针穿刺细胞学检查的平行诊断方法时，9基因突变检测的诊断效力与BRAF V600E单基因检测相当，并且9基因突变检测可以检测出个别提示不良预后的基因突变，对这些有特殊基因改变患者的临床管理有一定参考价值。.