Ahmed Bilal Ashraf PhD, Abstract Co-Author: Nothing to Disclose

Sara C. Gavenonis MD, Abstract Co-Author: Research support, Hologic, Inc

Dania Daye BS, Presenter: Nothing to Disclose

Carolyn J. Mies MD, Abstract Co-Author: Nothing to Disclose

Michael D Feldman MD, PhD, Abstract Co-Author: Nothing to Disclose

Mark Alan Rosen MD, PhD, Abstract Co-Author: Consultant, General Electric Company

Despina Kontos PhD, Abstract Co-Author: Research Consultant, Hologic, Inc Software Evaluation Agreement, Hologic, Inc

To investigate the predictive value of breast DCE-MRI kinetic features for assessing the probability of breast cancer recurrence as determined by a validated tumor gene expression assay.

Breast DCE-MRI images from 60 women diagnosed with estrogen receptor positive (ER+), node negative breast cancer were collected and analyzed retrospectively under HIPAA and IRB approval. Women in our study had their primary tumor tissue samples analyzed as part of their clinical care (during 2007-2010) with a validated reverse-transcriptase-polymerase-chain-reaction (RT-PCR) assay (Oncotype DX, Genomic Health Inc.) that measures the expression of 21 genes in RNA from formalin-fixed paraffin-embedded (FFPE) tissue. The outcome of the assay is a continuous score that predicts the likelihood of breast cancer recurrence in 10 years after treatment (risk: low ≤17%, medium >18%,<30%, high ≥ 31%). Pixel-wise maps of validated kinetic features, including peak enhancement (PE), time-to-peak (TTP), wash-in-slope (WIS), and wash-out-slope (WOS) were computed from the DCE-MRI sequences. Pixels were clustered into groups according to the similarity of their kinetic behavior and within-group statistics for every feature map were calculated, resulting in a total of 21 features. Univariate ANOVA was performed to assess the variance of each feature within the three recurrence risk categories. Multiple linear regression was performed to predict the breast cancer recurrence scores based on the extracted DCE-MRI features. The Pearson correlation between the predicted and actual scores was evaluated to assess the quality of prediction.

ANOVA analysis showed that 10 (out of 21) DCE-MRI kinetic features are significantly different (p<0.05) within the recurrence risk categories. Multiple linear regression using the selected significant features yields a statistically significant correlation (r=0.78, R2=0.61, p<0.05) with the breast cancer recurrence scores estimated by the tumor gene expression assay (Fig).

DCE-MRI quantitative kinetic features are associated with gene expression signatures that can predict breast cancer recurrence.

DCE-MRI could potentially be used as a non-invasive imaging marker for breast cancer prognosis and aid in making treatment decisions. Larger studies are warranted to validate clinical utility.

Ashraf, A, Gavenonis, S, Daye, D, Mies, C, Feldman, M, Rosen, M, Kontos, D, Prediction of Prognostic Tumor Gene Expression Signatures via Kinetic Analysis of DCE-MRI for Assessing Breast Cancer Recurrence.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11007450.html