A systematic comparison of abbreviated and unenhanced breast MRI protocols in women with mammographically dense breasts compared to full multiparametric breast MR imaging.

内容摘要

While the feasibility of MRI as a supplemental screening tool for breast cancer has been established, there is no consensus regarding the optimal examination protocol. This reader study aims to provide a systematic comparison of abbreviated contrast-enhanced MRI, IV-contrast-free, and a full multiparametric protocol, to diagnose breast cancer in a population of women with mammographically dense breasts. This IRB-approved retrospective reader study was performed in 166 patients with mammographically dense breasts recruited from a tertiary care university hospital. MR images were acquired at 1.5 or 3&#xa0;T MRI units in line with international recommendations. Three blinded off-site readers evaluated the images in three different approaches in a region-wise analysis: (1) full multiparametric protocol, (2) abbreviated first pass MRI protocol, and (3) unenhanced DWI and T2w/STIR images. Histopathology and/or imaging follow-up of at least 24&#xa0;months served as a reference standard. Statistics included generalized estimating equation methodology based logistic regression for repeated measures. 1660 regions (166 women, mean age, 45 +/- 12&#xa0;years) with 41 histologically verified cancers were read. At a BI-RADS cutoff&#xa0;>&#xa0;3, sensitivity was significantly higher (p&#xa0;<&#xa0;0.001) using the full protocol (80.4-90.2%) followed by the abbreviated (70.7-78.1%) and lastly the unenhanced protocol (40.6-53.5%). Specificity was significantly lower (p&#xa0;<&#xa0;0.001) using the abbreviated protocol. Inter-reader agreement was fair to moderate. The full multiparametric protocol demonstrated superior sensitivity compared to the abbreviated and unenhanced protocol, while specificity of the full and unenhanced protocols were superior to the abbreviated protocol.