Predictive value of sonographic features on molecular subtypes of invasive breast carcinoma / 中华超声影像学杂志

ABSTRACT

Objective:To explore the predictive value of various specific sonographic features on molecular subtypes for invasive breast carcinoma(IBC).Methods:Sonographic and clinicopathological data were retrospectively reviewed for 500 IBC patients who accepted surgical therapy in Fudan University Shanghai Cancer Center from January 2014 to March 2016. All tumors were divided into 5 molecular subtypes. The relationships of sonographic variations associated with the molecular subtypes for IBC were analyzed by univariate and multivariate Logsitic regression analyses.Results:Specific sonographic features for triple-negative subtype included regular shape ( OR=2.06, P=0.018), no spiculated/angular margin ( OR=1.98, P=0.029), posterior acoustic enhancement ( OR=2.26, P=0.005), and no calcification ( OR=2.13, P=0.006). Specific sonographic feature for human epidermal growth factor receptor-2 positive (HER2) subtype was posterior acoustic enhancement ( OR=2.23, P=0.006). Specific sonographic features for Luminal A subtype included spiculated/angular margin ( OR=2.24, P=0.001), posterior acoustic shadow ( OR=1.84, P=0.026), and no calcification ( OR=1.89, P=0.016). There were no specific sonographic features for the Luminal B with HER2 negative subtype, while that for the Luminal B with HER2 positive subtype was calcification ( OR=3.61, P<0.001). However, when used these sonographic features to predict molecular subtypes of breast cancer, the sensitivity values were 8.4%-57.3%, and positive predictive values were 9.5%-53.3%. Conclusions:The variety of sonographic features is associated with molecular subtypes of IBC.However, due to the overlap of sonographic features between different subtypes, molecular subtypes of IBC cannot be predicted by sonographic features.