Abstract P2-02-13: The clinical significance of breast MRI in the management of ductal carcinoma in situ diagnosed on needle biopsy

Abstract

Abstract Background: About 10-40% of ductal carcinoma in situ (DCIS) cases diagnosed on needle biopsy have invasive disease in the surgical specimens, which require surgical axillary staging. However, there is controversy regarding which subgroups should be indicated for sentinel lymph node biopsy (SLNB) because of the upstage on final diagnosis. Therefore, our aim is to identify the associated factors with invasive disease in DCIS diagnosed on needle biopsy by analyzing breast MRI findings with the histopathological factors of biopsy specimens. Methods: Seventy-five consecutive DCIS patients diagnosed on needle-biopsy in our institution who underwent preoperative MRI were retrospectively reviewed. The morphological features and kinetic curves were assessed using the ACR BI-RADS MRI lexicon. We measured the signal intensity (SI) of detected breast lesions and surrounding normal breast tissue on dynamic MRI and calculated the SI ratio as the SI of detected lesion divided by that of surrounding normal breast tissue. The MRI and histopathological variables were assessed between pure DCIS and invasive breast cancer (IBC) diagnosed on surgical specimens. Multivariable-analyses were performed to determine the independent factors for invasion using logistic-regression-model. Results: On dynamic MRI, 60 cases of 75 (80%) were classified as non-mass-like enhancement (NMLE) type and 15/75 (20%) were Mass type. In NMLE, 11/60 (18%) were ultimately diagnosed as IBC. Lesion size (P = 0.027), signal intensity (SI) ratios (calculated as the SI of detected lesions divided by the SI of surrounding normal breast tissue) (P = 0.032) on MRI and the number of biopsy-cores containing cancer nests (cancer-cores) (P = 0.012) were each independently associated with invasion. Table 1. Comparison of MRI findings and clinicopathological factors between pure DCIS and IBC groups in NMLE type.VariablesDCIS (n = 49)IDC (n = 11)P value (univariate)P value (multivariate)Age (years) 0.0480.067Median5544 Range34-7639-66 Lesion size on MRI (mm) 0.0180.027Median1634 Range5-606-59 Signal intensity ratio 0.0010.032Median1.352.01 Range1.04-2.891.56-2.55 The number of Cancer-cores 0.0050.012Median32 Range1-101-7 According to the ROC and statistical analyses performed for these independent factors, the area under the curve (AUC) was 0.72 for lesion size on MRI, 0.92 for the SI ratio and 0.80 for the number of cancer-cores (Table 2). Among these three factors, the SI ratio was the most accurately correlated with the presence of invasive disease. Table 2. The statistical results of the predictive factors in multivariate analysis for NMLE type.VariablesAUCCut-off valueSensitivitySpecificityPPVNPVLesion size on MRI (mm)0.723164%84%47%92%Signal intensity ratio0.921.7691%88%63%98%The number of Cancer-cores0.80282%80%47%95% In Mass type, there were no statistically significant differences between the DCIS and IBC cases. Conclusion: Our study demonstrated that needle biopsy-proven DCIS cases of NMLE type might be sufficiently managed using breast MRI features such as enhanced lesion size and signal intensity, incorporating the number of cancer-cores at needle biopsy specimen in the clinical setting. (Jpn J Clin Oncol. 2013 Epub ahead of print). Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-02-13.