The Value Of ¹⁸F-FDG PET/CT-based Radiomics Nomogram In Predicting Axillary Lymph Node Metastasis In Breast Cancer Patients

ObjectiveTo investigate the value of ¹⁸F-FDG PET/CT-based radiomics nomogram in predicting axillary lymph node（ALN）metastasis in breast cancer patients.MethodsA total of 175 breast cancer patients（140 in the training cohort and 35 in the validation cohort）who underwent ¹⁸F-FDG PET/CT scans before receiving treatment and had pathological response results were retrospectively analyzed.It was divided into lymph node metastasis group and non-lymph node metastasis group according to the pathological results.To establish a radiomics nomogram for predicting ALN metastasis in the training cohort of 140 cases and validate in the validation cohort of 35 cases.According to the PET/CT examination time,it was divided into training cohort of 140cases（examination time:from March 2012 to December 2020）and validation cohort of35 cases（examination time:from January 2021 to September 2021）.The CT images of each patient were exported in DICOM format,PET images were exported to the workstation,and segmentation of the tumor three-dimensional region of interest（ROI）was delineated by ITK-SNAP software and LIFEx software,and the radiomic features were extracted.Then,LASSO regression was used to select the optimal radiomics features and to construct the radiomics signature,and the radiomics score of each patient was calculated at the same time.At the same time,univariate and multivariate Logistic regression analysis was performed on the clinical data,immunohistochemical information and PET/CT metabolic parameters of the breast cancer patients,and the independent risk factors for predicting lymph node metastasis of breast cancer were screened out,and clinical feature model was constructed.Then,the radiomics nomogram combining radiomics features and independent risk factors was constructed by multivariate Logistic regression analysis.Receiver operating characteristic（ROC）curves were used to assess the performance of three models,and the difference was assessed with Delong test,and calibration curves were used to assess the fitness of the models.Finally,the decision curve analysis（DCA）was performed to evaluate the net benefit of the radiomics nomogram for predicting ALN metastasis in breast cancer patients.ResultFinally,14 best radiomics features were screened for constructing the radiomics signature.The AUC of the model for predicting breast cancer ALN metastasis in the training and validation cohorts were 0.83（95%CI:0.76-0.90）and 0.80（95%CI:0.64-0.96）.At the same time,two independent risk factors（including T stage and SUVmean）were screened out to construct the clinical feature model.The AUC of the clinical feature model constructed by independent risk factors in the training and validation cohorts for predicting breast cancer ALN metastasis were 0.78（95%CI:0.69-0.87）and 0.74（95%CI:0.54-0.95）.The AUC of the radiomics nomogram combining two independent risk factors and radiomic features in the training and validation cohorts for predicting breast cancer ALN metastasis were 0.85（95%CI:0.78-0.92）and 0.80（95%CI:0.65～0.95）.At the same time,the calibration curve also showed that the radiomics nomogram had a good fit for predicting ALN metastasis in breast cancer patients.The results of Delong test showed that the difference between the clinical feature model and the radiomics nomogram was statistically significant（Z=2.243,P<0.05）in the training cohort.The DCA indicated that the radiomics nomogram based on¹⁸F-FDG PET/CT radiomics had high overall net benefit in predicting ALN metastasis in breast cancer patients.ConclusionIn this study,the radiomics nomogram based on ¹⁸F-FDG PET/CT radiomics features,clinical features and metabolic parameters has high predictive value in predicting ALN metastasis in breast cancer patients,it illustrates the important gain value of radiomics for clinical models,and it can be used as an auxiliary diagnosis tool for predicting lymph node metastasis in breast cancer patients,which provides a reference for clinical decision-making and is helpful to guide individualized diagnosis and treatment of breast cancer patients.