细胞核评分联合超声特点对术前甲状腺穿刺不确定性结节的分类

Classification of indeterminate thyroid nodules in preoperative fine needle aspiration by the combination of nuclear score and ultrasound features

目的探讨细胞核形态评分联合超声特点,用于细针穿刺细胞学诊断不确定性结节分类的可行性。方法收集119例有组织病理学随访结果,术前细针穿刺(fine needle aspiration,FNA)细胞学诊断为TBSRTCⅢ~Ⅴ级的病例。由2名超声科医师和2名细胞病理学医师分别针对超声图像特征及细胞形态学特征进行观察及评分,当评分不一致时协商达成共识。利用受试者工作特征(receiver operating characteristic curve,ROC)曲线和ROC曲线下面积评估细胞核形态及超声影像学评分诊断甲状腺恶性肿瘤的最佳临界值。通过交叉列联表评价各指标的特异性、敏感性、阳性预测值和阴性预测值,Logistic回归及ROC曲线分析评价简化核形态联合超声特征评分的诊断准确性。结果核沟、核内包涵体及核淡染更易出现于恶性/低危甲状腺结节(P=0.001、0.013、0.001)。细胞学简化核形态评分的阳性阈值设定为2分,诊断甲状腺恶性/低危肿瘤的敏感性高(99.1%)。纵横比>1、晕环征消失、形态不规则和结节内无血流信号的超声特征更易出现于恶性/低危甲状腺结节(P=0.003、0.007、0.003、0.004)。超声学评分阳性阈值设定为5分,诊断甲状腺恶性/低危肿瘤的阳性预测值为95.9%,特异性为78.6%。ROC曲线分析显示:两种评分联合诊断的准确性(0.873,95%CI=0.777~0.970)高于单独使用核形态评分(0.817,95%CI=0.687~0.948,P=0.092),但差异无统计学意义;亦高于单独超声学特征评分(0.758,95%CI=0.637~0.879,P=0.047),差异有统计学意义。结论对FNA细胞学诊断不确定的病例,结合简化核形态及超声特征评分可提高细胞学诊断甲状腺恶性/低危肿瘤的准确性。

Purpose To assess the feasibility of nuclear score combined with ultrasound features on indeterminate thyroid nodules in fine needle aspiration cytological diagnosis.Methods A consecutive cohort of 119 thyroid fine-needle aspiration(FNA)specimens with indeterminate diagnosis(TBSRTC categoryⅢ-Ⅴ)and available histopathologic follow-up was evaluated.Original sonographic images and cytological slides of each nodule were reviewed by two experienced sonographers and two experienced cytopathologists respectively,and evaluated jointly to obtain consensus.The identification of the optimal cut-off points of simplified nuclear score and ultrasound features score for the diagnosis of malignancy or low-risk neoplasm were evaluated using the receiver operating characteristic(ROC)curves and the assessment of the area under the ROC curve(AUC).The specificity,sensitivity,positive predictive value(PPV)and negative predictive value(NPV)of nuclear score and US score were evaluated from crosstabs based on Cut-off points and significance were calculated.The diagnostic accuracy of simplified nuclear score combined with ultrasound features was estimated by Logistic regression and ROC curve analysis.Results Nuclear grooves,intra-nuclear inclusions and chromatin clearing were more common in malignancy/low-risk neoplasms(P=0.001,0.013 and 0.001 respectively).A Cut-off point of≥2 for the simplified nuclear score was sensitive for defining malignancy/low-risk neoplasm,and the sensitivity was 99.1%.Nodule height greater than width,absent halo sign,irregular margin and absent intranodular vascularization were more common in malignancy/low-risk neoplasms(P=0.003,0.007,0.003 and 0.004 respectively).When the Cut-off point was set as≥5,the PPV and specificity for defining malignancy/low-risk neoplasm were 95.9% and 78.6% respectively.ROC curve analysis showed that the diagnostic accuracy of nuclear score combined with US score(0.873,95%CI=0.777-0.970)was higher than that of nuclear score alone(0.817,95%CI=0.687-0.948,P=0.092)or US score alone(0.758,95%CI 0.637-0.879,P=0.047).But no significant difference was detected between the combined score and the nuclear score.Conclusion The addition of simplified nuclear score and US score to FNA cytology can increase the diagnostic accuracy of cytology in indeterminate categories.