Diffusion Weighted Imaging of the Breast: Can it Decrease the Number of Unneeded Biopsies?

Objective: To date, breast MRI is analyzed according to morphologic criteria, enhancement kinetics and the T2 characteristic of the breast lesion (1, 2). However, all these criteria show an overlap between benign and malignant lesions and the characterization of the lesion can be challenging. Diffusion weighted imaging (DWI) and ADC values are highly influenced by the lesions’ cellularity and thus they can differentiate between benign and malignant lesions (3). The objective of our study is to determine whether ADC values would be a valuable tool in the characterization of breast lesions thus sparing the patients from unneeded biopsies. Materials and Methods: Using a 1.5T Philips Intera scanner with a 4-element phased array breast coil, 30 patients and 42 lesions with inconclusive mammographic and ultrasonographic evaluation [Breast Imaging Reporting and Data System (BIRADS) 3 or 4] were subjected to a routine breast MRI study including a dedicated dynamic post contrast study with dynamic time intensity curves plotted for each lesion. The DWI EPI STIR sequence was performed prior to the dynamic scan as the T1 relaxation due to the contrast agent would cause changes to the inversion of the tissue and thus can have a strong impact. The parameters included TR/TE/TI = 2750/68/180, FOV = 37x 27 cm, slice /gap = 3/0mm, voxel size = 5 mm, NEX=4. Prone positioning and oblique transverse phase-encoding direction minimized respiratory and cardiac motion, respectively. Diffusion-gradients were employed along 3 orthogonal directions (P, M, S) with b values of 500, 1000 s/mm2. And +/-36 axial slices were taken to cover the entire breast. Interpretations of DWI were done through correlating areas of restricted diffusion with areas of contrast enhancement. ADC maps were generated and ADC values were measured through ROI’s drawn at sites of maximum signal intensity (i.e. maximum restricted diffusion) and copied to the ADC maps. Conventional images including dynamic time intensity curves were interpreted by an experienced MR radiologists on one setting then re-read after addition of DWI’s and ADC values in another setting. Receiver operating characteristic curves (ROC) were generated. Results: Using the histopathological results as the gold diagnostic reference, there were 23 malignant and 19 benign lesions. The mean ADC values were significantly lower in malignant lesions compared to benign lesions (Figure).