Résumé
OBJECTIVE: The adjacent vessel sign (AVS) is a descriptor for differentiating malignant from benign breast lesions on breast MRI (bMRI). This investigation was designed to verify the previous reports on the diagnostic accuracy of AVS and to assess correlation between AVS, histopathological diagnosis, lesion size and lesion grade. MATERIALS AND METHODS: This study was approved by the local ethical committee. Experienced radiologists evaluated 1,084 lesions. The exclusion criteria were no histological verification after bMRI and breast interventions that were done up to one year before bMRI (surgery, core biopsy, chemo- or radiation therapy). The native and dynamic contrast-enhanced T1-weighted series were acquired using standardized protocols. The AVS was rated positive if a vessel leading to a lesion could be visualized. Prevalence of an AVS was correlated with the lesions' size, grade and histology using Chi-square-tests. RESULTS: The AVS was significantly associated with malignancy (p 2 cm more often presented with an AVS than did those malignant lesions < 2 cm (p < 0.0001; sensitivity: 65%, PPV: 90%). There was no correlation of the AVS with the tumor grade. The prevalence of an AVS didn't significantly differ between invasive lobular carcinomas versus ductal carcinomas. In situ cancers were less frequently associated with an AVS (p < 0.001). CONCLUSION: The adjacent vessel sign was significantly associated with malignancy. Thus, it can be used to accurately assess breast lesions on bMRI. In this study, the AVS was particularly associated with advanced and invasive carcinomas.