甲状腺乳头状癌患者术前血清TK1表达水平联合甲状腺超声特征构建中央区淋巴结转移预测模型及验证

Construction and Validation of Prediction Model of Central Lymph Node Metastasis in Patients with Papillary Thyroid Carcinoma Based on Preoperative Serum TK1 Expression Level and Thyroid Ultrasound Features

目的探讨血清胸苷激酶1(serum thymidine kinase 1,sTK1)联合甲状腺超声特征预测甲状腺乳头状癌(papillary thyroid carcinoma,PTC)患者中央区淋巴结转移(central lymph node metastasis,CLNM)的风险。方法选取2020年10月~2021年12月陕西省人民医院确诊的CLNM的PTC患者110例为观察组,同期无CLNM的PTC患者104例为对照组。所有患者行甲状腺超声检查,检测sTK1等指标水平,采用t检验比较两组间sTK1等指标的水平差异,卡方检验分析颈部超声结果的差异,Logistic回归分析CLNM的独立危险因素,构建列线图预测模型,并选取2022年1~5月陕西省人民医院确诊的80例PTC患者对模型的预测准确度进行外部验证。结果观察组sTK1水平高于对照组(2.06±0.75pmol/L vs 1.59±0.66pmol/L),差异有统计学意义(t=4.75,P<0.001),而血清TSH,TG,TGAb水平与是否发生CLNM无关,差异无统计学意义(t=0.74,0.75,0.61,均P>0.05)。sTK1预测PTC患者CLNM的曲线下面积(AUC)为0.678,截断值为1.50 pmol/L,灵敏度和特异度分别为79.1%,61.0%。单因素分析结果显示,s TK1>1.5pmol/L,边界不清、结节微钙化、肿瘤数目、肿瘤直径>1cm和淋巴结明显血流信号是PTC患者CLNM的独立风险因素(χ^(2)=5.24~26.72,均P<0.05),而性别、年龄、低回声、纵横比>1、肿瘤位置与CLNM无关(χ^(2)=0.27~7.16,均P>0.05)。基于上述危险因素构建预测模型并进行准确度验证,采用建模原始数据进行内部验证AUC为0.826,验证队列进行外部验证AUC为0.809,表明该模型具有一定的预测准确度。结论PTC患者术前sTK1表达水平联合甲状腺超声特征构建CLNM预测模型具有一定的临床应用价值,当sTK1>1.5 pmol/L,边界不清、结节微钙化、肿瘤多灶、肿瘤直径>1cm,淋巴结有明显血流信号时,发生CLNM的概率较高,建议行预防性中央区淋巴结清扫术。

Objective To explore the prediction of the risk of central lymph node metastasis(CLNM)in patients with papillary thyroid carcinoma(PTC)by serum thymidine kinas 1(sTK1)combined with thyroid ultrasound features.Methods From October 2020 to December 2021,110 PTC patients with CLNM diagnosed by Shaanxi Provincial People’s Hospital were selected as the observation group,and 104 PTC patients without lymph node metastasis in the central region were selected as the control group.All patients were examined by thyroid ultrasound and the level of sTK1 and other indicators were detected.The difference of 104 PTC patients without lymph node metastasis in the ceatral region were selected as the control group,and other indicators between the two groups was compared by t-test,the difference of neck ultrasound results was analyzed by Chi square test(χ^(2)test),and the independent risk factors of CLNM were analyzed by logistic regression.A nomogram prediction model was constructed.80 PTC patients diagnosed by Shaanxi Provincial People’s Hospital from January 2022 to May 2022 were selected for external verification of the prediction accuracy of the model.Results The level of sTK1 in the observation group was higher than that in the control group(2.06±0.75pmol/L vs 1.59±0.66pmol/L),and the difference was statistically significant(t=4.75,P<0.001).The levels of serum TSH,TG and TGAb were not related to CLNM,and the differences were no statistically significant(t=0.74,0.75,0.61,all P>0.05).The area under the curve(AUC)predicted by sTK1 for CLNM in PTC patients was 0.678,the cut-off value was 1.50 pmol/L,and the sensitivity and the specificity was 79.1%,61.0%,respectively.The results of univariate analysis showed that sTK1>1.5 pmol/L,unclear boundary,nodule microcalcifications,tumor number,tumor diameter>1cm and obvious blood flow signal of lymph nodes were independent risk factors of CLNM in PTC patients(χ^(2)=5.24~26.72,all P<0.05),while gender,age,low echo,aspect ratio>1,and tumor location were not related to CLNM(χ^(2)=0.27~7.16,all P>0.05).Based on the above risk factors,a prediction model was built and its accuracy was verified.The AUC of the internal verification using the modeling raw data was 0.826,and the AUC of the external verification using the verification queue was 0.809,which indicates that the model had a certain prediction accuracy.Conclusion The expression of sTK1 in PTC patients before operation combined with the characteristics of thyroid ultrasound to construcs a prediction modesl of CLNM has certain clinical application value.When sTK1>1.5 pmol/L,irregular edges,microcalcifications,multiple tumors,tumor diameter>1 cm,and lymph nodes had obvious blood flow signals,the probability of CLNM is much higher.It is recommended to perfrom preventive lymph node dissection the central region.