2289. Low-Energy Images of Contrast-Enhanced Mammogram: Are They Better Than Conventional 2D Digital Mammography in Assessing for Cancer?
Authors * Denotes Presenting Author
  1. Christina Konstantopoulos *; Beth Israel Deaconess
  2. Tejas Mehta; Beth Israel Deaconess
  3. Alexander Brook; Beth Israel Deaconess
  4. Vandana Dialani; Beth Israel Deaconess
  5. Rashmi Mehta; Beth Israel Deaconess
  6. Valerie Fein-Zachary; Beth Israel Deaconess
  7. Jordana Phillips; Beth Israel Deaconess
The low-energy (LE) images acquired during contrast enhanced mammography (CEM) have been shown to be non-inferior to conventional 2D digital mammography (DM). Anecdotally, however, breast malignancies seem to be better appreciated on LE images as compared to DM images. Our purpose was to determine whether LE images are superior to DM at demonstrating breast malignancies.

Materials and Methods:
A HIPAA compliant IRB approved reader study was performed that included 40 consecutive biopsy-proven malignancy CEM exams from February 17th 2016 to November 8th 2018. Inclusion criteria were: prior DM within 6 months, cancer visible on CC and MLO views of both CEM LE and DM, and imaging performed pre-biopsy. Bilateral CC LE images and DM case sets were presented side-by-side in random order to 7 breast imagers (either fellowship-trained or >30 years of experience) who were blinded to the study type. MLO projections were then shown in similar fashion. The location of the cancer was indicated with a circle. Readers provided interpretations of tissue density and rated each LE and DM set of CC and MLO images for cancer visibility, confidence in margins, and conspicuity of the finding as compared with tissue density using a 5-point Likert scale. Analysis was performed using 3-way ANOVA.

There was fair inter-observer agreement for cancer visibility, confidence in margins, and cancer conspicuity. For all readers combined, LE graded higher than DM for cancer visibility, 4.0 vs 3.5 for CC view (p<0.001) and 3.7 vs 3.5 for MLO (p=0.01), for confidence in margins, 3.2 vs 2.8 for CC view (p<0.001) and 3.1 vs 2.9 for MLO (p=0.008), and for cancer conspicuity, 3.6 vs 3.2 for CC view (p<0.001) and for MLO, 3.5 vs 3.2 (p<0.001). For individual readers, LE projections graded higher than DM for all criteria, but not all the differences were statistically significant.

In our study, all radiologists preferred CEM LE images over DM in evaluating visibility, margins, and conspicuity of cancer. This suggests CEM may be superior to 2D DM rather than non-inferior, as prior studies indicate. Additional study is warranted.