E1697. Tumor-to-Fibroglandular Enhancement Ratio on Breast DCE-MRI: Association with Breast Cancer Prognosis
  1. Jessica Rubino; Dartmouth Hitchcock Medical Center
  2. Brook Byrd; Thayer School of Engineering, Dartmouth College
  3. Roberta DiFlorio-Alexander; Dartmouth Hitchcock Medical Center
  4. Timothy Rooney; University of Virginia Health
  5. Venkataramanan Krishnaswamy; CairnSurgical, Inc
  6. Keith Paulsen; Geisel School of Medicine
The breast micro-environment plays an important role in the development and spread of breast cancer. The interaction between tumor biology and patient-specific breast parenchyma influences tumor progression and prognosis. Tumor kinetics are associated with cancer aggressiveness, and background parenchymal enhancement (BPE) is associated with breast cancer risk and treatment response. The purpose of this retrospective study was to quantitatively evaluate the relationship between poor prognosis breast cancer and tumor-to-background enhancement ratio (TFR) on breast MRI on early and markedly delayed post-contrast sequences.

Materials and Methods:
Dynamic contrast-enhanced breast MRI examinations (DCE-MRI) were retrospectively analyzed for 69 women with breast cancer undergoing an investigative pre-operative prone to supine breast MRI study for surgical planning. The breast cancers were manually segmented by a breast imaging radiologist on contiguous axial MRI slices of the first post-contrast sequence of the prone breast DCE-MRI examination, and on the markedly delayed supine sequence (obtained approximately 20 minutes after completion of the prone exam without additional contrast administration). The regions of breast tissue directly surrounding the segmented tumors were determined as a volume of breast tissue equal to the tumor volume. The tumor-to-fibroglandular contrast ratio (TFR) was calculated as the mean signal intensity of segmented tumor to surrounding fibro-glandular tissue on the T1-weighted gadolinium (Gd)-enhanced sequence of the prone and supine exams. Comparative statistics were used to compare TFR to poor prognosis breast cancer using four prognostic tools/definitions: Nottingham Prognostic Index (NPI), Oncotype DX (ODX), PREDICT, and Tomosynthesis Mammographic Imaging Screening Trial (TMIST) criteria for advanced cancers.

There was a significant association between TFR on the first post-contrast sequence and NPI score stratified by good prognosis (score = 3.4) vs poor prognosis (score > 3.4), (p = 0.007). There was also a significant correlation between early TFR and ODX (R2 = 0.45, p = 0.010, 95% CI = [0.116, 0.682]). There was a trend toward significance for the association between TMIST advanced cancer and TFR on the supine MRI with markedly delayed enhancement that reflects BPE rather than tumor kinetics due to the protracted time-point in which minimal residual contrast is identified in the tumor (p = 0.87). No significant association or correlation was seen with PREDICT scores and TFR.

Identifying imaging markers of prognosis is increasingly important in this era of tailored breast cancer treatment. TFR combines features of tumor biology and individualized breast physiology, and TFR was significantly associated with prognosis scores in our study. Future studies may evaluate the accuracy of incorporating TFR into prognostic scores to improve risk prediction. If successful, TFR combined with other imaging features may predict recurrence and chemotherapy benefits without the need for costly tumor genetic profiling.