Logistic Regression Analysis To Predict The Nature Of Thyroid Nodules

Objectives To analyze the differences of preoperative serum thyroid stimulating hormone（TSH）,anti-thyroid peroxidase antibody（TPOAb）,peripheral blood neutrophil to lymphocyte ratio（NLR）,body mass index（BMI）and ultrasound grading of thyroid between patients with thyroid papillary carcinoma（PTC）and benign thyroid nodule.Establish the preoperative PTC risk prediction equation.For hospitals without thyroid fine needle aspiration biopsy（FNAB）technology or when the cytological pathological results cannot determine the nature,it provides the diagnostic basis for clinicians to predict the benign and malignant thyroid nodules.Methods A total of 510 patients with thyroid nodules（included 271 patients with PTC and239 patients with benign nodules）who admitted to the Department of Thyroid&Breast of Hebei General Hospital from September 2018 to December 2019 were retrospectively analyzed.The clinical data collected and analyzed include:1.General patient information:including the patient’s hospital number,name,gender,age,BMI;2.Preoperative thyroid function test results:thyroid stimulating hormone（TSH）,anti-Thyroid peroxidase antibody（anti-thyroid peroxidase antibody,TPOAb）value;3.The number of neutrophils and lymphocytes in the blood routine results within 1 week before surgery,and the ratio is calculated;4.Preoperative thyroid ultrasound results:nodule diameter,number of nodules,thyroid imaging reporting and data system（TI-RADS）classification in color Doppler ultrasound;grouping according to postoperative pathological properties;clinical data are statistically analyzed by IBM SPSS Statistics version 26.0 software,Measurement data use Two-Sample t-test or Mann-Whitney U-test,count data use Chi-Square test or Fisher’s exact probability method Perform statistical analysis,and use receiver operating characteristic curve（ROC）to determine the best cut-off value of statistically significant variables to predict PTC.When the test standard is P<0.05,the difference is statistically significant.With postoperative pathological properties as the dependent variable,Binary logistic regression was used to analyze and clarify the independent predictors of PTC,and a diagnostic prediction model was constructed.ROC curves were drawn and the accuracy of the area under the curve test equation was calculated.From September 2018 to the present,patients with thyroid nodules whose preoperative cytology tests have not determined the nature and have a definite pathological diagnosis after surgical treatment are used to verify the predictive ability of the model and thyroid color ultrasound.Results Among the 510 patients with thyroid nodules included in this study,217 patients were pathologically diagnosed as PTC after surgery,and the malignant rate was 53.14%.In the Logistic regression analysis,four variables of age,BMI,TSH,and TI-RADS grade entered the regression model:age<55years（P<0.01,B=0.71,OR=2.04,95%CI:1.19-3.48）;TSH≥1.5 m IU/L（P<0.01,B=0.917,OR=2.50,95%CI:1.45-4.32）;BMI（P<0.01,B=0.66,OR=1.94,95%CI:1.36-2.75）;ultrasound classification（P<0.01,B=3.00,OR=20.12,95%CI:12.02-3.67）.Age<55 years,TSH≥1.5m IU/L,BMI,and TI-RADS classification are independent risk factors for PTC.The Logistic regression model established on this basis corresponds to an area under the ROC curve of 0.92,and the model has high accuracy.Cytological testing could not confirm the nature.26 patients with thyroid nodules who had a definite pathological diagnosis after surgical treatment verified the model and the predictive ability of thyroid color Doppler ultrasound.The specificity of the prediction model is 41.67%,the sensitivity is 100%,the positive predictive value is 66.67%,the negative predictive value is 100%,and the total accuracy rate is 73.77%,which are all higher than the predictive ability of ultrasound alone.Conclusions 1 In patients with thyroid nodules,independent risk factors for papillary thyroid carcinoma are:age,TSH,BMI,TI-RADS grade.The assessment of benign and malignant nodules has important value.2 The equation for predicting PTC risk before surgery is:P=1/(1+e^-y),y=0.89×X₁+0.85×X₂+1.17×X₃+2.31×X₄-8.72,X₁ represents age group,X₂ represents the BMI group,X₃ represents the TSH group,and X₄ represents the TI-RADS group.3 This predictive model can provide a reference for clinicians to evaluate the nature of thyroid nodules when the FNAB cytological pathological results are uncertain or there is no FNAB condition.Figure[5];Table[10];Reference[175]...