Evaluation of Apparent Diffusion Coefficient to Predict Grade, Microinvasion, and Invasion in Ductal Carcinoma In Situ of the Breast

Abstract

To evaluate the role of apparent diffusion coefficient (ADC) in distinguishing ductal carcinoma in situ (DCIS) grades and identifying microinvasive and/or invasive disease in the preoperative evaluation of patients with core biopsy-proven DCIS. Research Ethics Board-approved study with informed consent from 81 women (age, 36-84 years) scheduled for core-biopsy with results of 82 noninvasive breast carcinomas. All patients were assessed preoperatively by diffusion sequence in addition to contrast magnetic resonance imaging (MRI). Lesion morphology and ADC values were recorded. The Kruskal-Wallis or one-way analysis of variance test and Pearson correlation coefficient were used to study the association between ADC and MRI lesion characteristics. Logistic regression analysis was used to evaluate the ability of ADC to predict the presence of invasion. Surgical pathology demonstrated associated invasive cancer in 26.8%, microinvasion in 14.6%, and pure DCIS in 58.5%. The minimum regions of interest (ROI)-based ADC was significantly different among the following three groups (P < .001, Kruskal-Wallis test): 0.98 × 10(-3) mm(2)/s ± 0.25 for pure DCIS, 0.82 × 10(-3) mm(2)/s ± 0.20 for DCIS with microinvasion, and 0.71 × 10(-3) mm(2)/s ± 0.27 for DCIS with invasive disease. Based on logistic regression analysis, the minimum ROI-based ADC of 0.56 × 10(-3) mm(2)/s was a significant predictor for invasive disease (odds ratio = 0.02, 95% confidence interval [0.002, 0.207], P = .001). Regardless of the field strength (1.5 vs. 3.0 T) ADC values of high-grade and non-high-grade DCIS were not significantly different. Pure DCIS had the highest "ROI-based" ADC measured using 1.5 T or 3.0 T. The ADC was able to identify microinvasion or invasive cancer in biopsy-proven DCIS lesions but not to distinguish the DCIS grades.