Research on predicting the methylation status of NMUR1 in breast cancer based on a multi-parameter MRI radiomics model
Objective To explore the clinical value of a predictive model constructed by multi-parameter magnetic resonance imaging(MRI)radiomics features in predicting the methylation status of neuromedin U receptor 1(NMUR1)gene in breast cancer.Methods A retrospective analysis was conducted on 258 patients with breast cancer who were treated at Dazhou integrated TCM & Western Medicine Hospital from January 2018 to January 2023.The patients were divided into a training set(n=172)and a validation set(n=86)in a ratio of 2:1.Breast cancer cells were obtained using magnetic separation techniques,and the methylation status of NMUR1 gene was detected by methylation-specific PCR.Based on the detection results,the patients in the training set were divided into NMUR1 methylation-positive group(n=93)and NMUR1 methyla-tion-negative group(n=79).Multi-parametric MRI scans including T2WI-FS,DWI and DCE were performed to obtain breast cancer lesion information.Multivariate Logistics regression analysis was used to determine the independent factors influenc-ing the NMUR1 methylation status in clinical data.Univariate analysis of variance and least absolute shrinkage and selection operator(LASSO)regression were used to reduce the dimensionality of the multi-parameter MRI radiomics features and a single sequence radiomics label was constructed.Multivariable regression analysis was performed on the radiomics features to build a combined sequence radiomics score model.The combined sequence radiomics features,together with clinical features,were used to construct a joint model.The predictive ability of the model was evaluated using receiver operating characteristic(ROC)curve,Hosmer-Lemeshow test and calibration curve.The clinical significance of the radiomics combined model was assessed using clinical decision curve(DCA).Results Family history of breast cancer,advanced tumor stage,lymph node metastasis and triple-negative breast cancer were independent factors influencing the methylation status of NMUR1 gene in clinical data(P＜0.05).After univariate analysis of variance and LASSO regression,58 radiomics features were ultimately obtained,including 17 from DCE,13 from T2WI-FS and 28 from DWI.Among the radiomics features,ClusterPromi-nence_anglel35_offset4.1,DifferenceEntropy.l,GLCMEntropy_angle0_offset1.2,DifferenceEntropy.1,ADC.Quantile95 and ADC.Minlntensity were independent factors influencing the methylation status of NMUR1 gene(P＜0.05).The combined sequence radiomics score model had higher discrimination power compared to the single sequence radiomics label(P＜0.05).The joint model had high discrimination power and accuracy and showed good predictive performance,demonstrating sig-nificant clinical value.Conclusion Family history of breast cancer,advanced tumor stage,lymph node metastasis and tri-ple-negative breast cancer are independent factors influencing the methylation status of NMUR1 gene in breast cancer.The joint model constructed with combined sequence radiomics features and clinical data has high clinical value in predicting the methylation status of NMUR1 gene in breast cancer.
breast cancermagnetic resonance imagingneuromedin U receptor 1methylationradiomics
NETL
NSTL
万方数据
