Analysis Of Risk Factor For TI-RADS 3 Thyroid Cancer And Construction Of Prediction Model

Background: The atypical features of benign and malignant TI-RADS 3 thyroid nodules overlap on ultrasound imaging.In clinical work,it was found that missed diagnosis and misdiagnosis of TI-RADS 3 malignant thyroid nodules were not uncommon,and some malignant patients had cervical lymph node metastasis when hospitalized.Therefore,it is particularly important to confirm the diagnosis of TI-RADS 3 malignant thyroid nodules as soon as possible.Currently,there is no noninvasive and economical preoperative test to make a definitive diagnosis in these patients.The purpose of this study was to evaluate the risk factors of TI-RADS 3thyroid cancer and construct a prediction model for TI-RADS 3 thyroid cancer.Methods: This study retrospectively analyzed the clinicopathological data of286 patients with TI-RADS 3 thyroid nodule who underwent surgical treatment for the first time from January 2018 to September 2021 in the Department of Breast and Thyroid Tumor Surgery,Affiliated Hospital of Qing Hai University(single center),and divided them into benign group(186 cases)and malignant group(100 cases)according to postoperative pathology.Univariate and multivariate logistic regression analysis was used to evaluate the risk factors of TI-RADS 3 thyroid cancer,and the prediction model for TI-RADS 3 thyroid cancer was constructed according to the results of multivariate analysis.Results: The subjects enrolled in this study were all patients with TI-RADS 3thyroid nodule who had undergone surgical treatment for the first time.According to univariate logistic regression analysis,the age,iso/low/very low echo,echo light spot,diameter of nodule,blood supply in nodule,PTH,TPO-Ab were closely associated with TI-RADS 3 thyroid cancer.Multivariate logistic regression analysis showed that young age,iso/low/very low echo,echo light spot,nodule diameter(<30mm),blood supply in nodule and low level of PTH were independent risk factors for TI-RADS 3thyroid cancer.According to the above results of multivariate analysis,a prediction model for TI-RADS 3 thyroid cancer was constructed.Logistic(P)= 2.084-0.032 ×age +0.767× echo intensity +1.36×echo light spot-2.048×diameter of nodule +0.892×blood supply-0.018× PTH.The area under ROC curve(AUC)of age,echo intensity,echo light spot,nodule diameter,blood supply in nodule and PTH were 0.610(95%CI,0.539 ～ 0.681,P=0.002)and 0.627(95%CI,0.560 ～ 0.695,P < 0.001),0.629(95%CI,0.558 ～ 0.699,P < 0.001),0.721(95%CI,0.654 ～ 0.787,P < 0.001),0.573(95%CI,0.504 ～ 0.641,P =0.04),0.599(95%CI,0.528-0.671,P =0.009),respectively.The cut-off values of age and PTH were 46.5 years and 61.15pg/ m L,respectively.The AUC of the prediction model was 0.824(95%CI,0.766 ～ 0.882,P < 0.001),the maximum Youden index was 0.535,the cut-off value was 0.457,the sensitivity was62.4%,and the specificity was 91.2%.Conclusion: Young age,iso/low/very low echo,echo light spot,nodule diameter(<30mm),blood supply in nodule and low level of PTH are all independent risk factors for TI-RADS 3 thyroid cancer.Patients with age<46.5 years and PTH<61.15pg/ m L had a higher risk of malignancy in type 3 thyroid nodules.When the predictive model value reached 0.457,it was highly likely that the patient’s TI-RADS 3 thyroid nodule had undergone malignant transformation.