Quantitative MRI assessment of uninvolved breast tissue in patients with locally-advanced breast cancer undergoing neoadjuvant chemotherapy.

Abstract

Abstract Abstract #4016 Background: We studied the Magnetic Resonance Imaging (MRI) characteristics of breast tissue in patients with invasive cancers to better understand the relationship between tumors and their normal “host” tissue. There is a strong association between tumor morphologic patterns on MRI and both treatment outcome and survival. The objectives of this study were to assess the characteristics of normal host tissue to verify whether they were associated with treatment outcome.
 Materials and Methods: 42 women with locally-advanced breast cancer underwent MRI exams prior to the start of neoadjuvant adriamycin/cytoxan (AC) chemotherapy (MRI1), following one cycle of AC (MRI2) and at the completion of chemotherapy, prior to surgery (MRI3). Patients mean age was 48.56 years (range 29.7 - 71.5). Median follow-up was 178 weeks. Pathology and radiology reports were available for all patients. Total follow-up was 3 years. We quantified tumor and surrounding normal tissue enhancement. Signal Enhancement Ratio (SER) values have been histopathologically shown to positively correlate with microvessel density within tumor. We lowered the enhancement threshold necessary to detect MRI tumor-enhancing regions, to insure that slow or non-enhancing tissue regions could be analyzed. For all patients, tumor morphologic patterns were categorized according to the degree of tumor containment from 1, denoting well-circumscribed uni-centric masses to 5, corresponding to infiltrative tumors with ill-defined borders. MRI breast tissue composition quantitative measures included 1) volumetric breast density, 2) tumor volume, 3) normal fibroglandular tissue volume and 4) SER levels in tumor and normal surrounding tissue. In order to quantify the uninvolved breast tissue volume, we used a semi-automated technique that performs a segmentation of MRI breast tissue from fat regions without user interaction, therefore avoiding subjective threshold decision issues. Regions in normal tissue surrounding tumors were automatically selected and analyzed at all time points.
 Results: 10 patients recurred. We found that non-tumor mean SER did not change significantly with one cycle of chemotherapy, however the value of non-tumor mean SER after 1 cycle of chemotherapy was an independent predictor of disease free survival. We also observed a strong association between the initial (pre-treatment) tumor morphologic patterns on MRI with treatment outcome. We found that volume of normal breast tissue at baseline does relate to tumor phenotype (p<0.067), and change in non-tumor breast tissue volume after chemotherapy was also predictive of recurrence (p<0.075).
 Conclusion: We found that measures of uninvolved breast tissue volume were correlated with tumor phenotypes and were shown to be predictive of recurrence. These results suggest that the host tissue may provide treatment response information. Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 4016.