Predictive value of 18F-FDG PET conventional parameters and radiomics features for invasive breast cancer patients with uncertain HER2 expression by immunohistochemistry / 中华核医学与分子影像杂志

ABSTRACT

Objective@#To explore the predictive value of 18F-fluorodeoxyglucose (FDG) PET conventional parameters and radiomics features for human epidermal growth factor receptor 2 (HER2) expression which was uncertain by immunohistochemistry (IHC) detection in invasive breast cancer.@*Methods@#From April 2012 to December 2017, 76 patients (all were females, age (50.8±10.9) years) with invasive breast cancer and with uncertain HER2 expression by IHC in Tianjin Medical University Cancer Institute and Hospital were enrolled retrospectively. The 18F-FDG PET/CT images before treatment were reviewed and the expression of HER2 were confirmed by fluorescence in situ hybridization (FISH). The tumor lesions were manually outlined, and the radiomics features from PET images were extracted. Wilcoxon test was used to determine whether there was difference in PET conventional metabolic parameters (maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG)) and radiomics features between HER2-negative and HER2-positive groups. The receiver operating characteristic (ROC) curve analysis was used to compare the predictive efficacy of PET radiomics features for HER2 expression.@*Results@#There were 41 HER2-positive patients and 35 HER2-negative patients. No significant differences in PET conventional metabolic parameters between different HER2 expression groups were observed (U values from -1.598 to 1.551, all P>0.05). A total of 38 PET radiomics features were extracted, and there were significant differences in gray mean, correlation, contrast, inertia, and inverse different moments between 2 groups(U values from -2.413 to 2.527, all P<0.05). The area under the curve of the above 5 parameters for prediction of HER2 expression were 0.643, 0.638, 0.647, 0.644 and 0.643, respectively, and the contrast was the best parameter.@*Conclusions@#PET radiomics features can effectively identify HER2 expression in patients with invasive breast cancer to some extent, and the contrast may be the best. Conventional metabolic parameters have limited predictive value.