Can whole-tumor apparent diffusion coefficient histogram analysis be helpful to evaluate breast phyllode tumor grades?

Abstract

Purpose: To investigate whether whole-tumor apparent diffusion coefficient (ADC) histogram analysis could be helpful to evaluate breast phyllode tumor (PT) grades.

Materials and methods: This institutional review board-approved retrospective study included 56 PTs (23 benign lesions, 22 borderline lesions, and 11 malignant lesions) from August 2011 to November 2017. MRI was performed using a 1.5 T MR system equipped with a 4-channel SENSE breast coil. All cases were divided into two groups, benign PT (BPT) and borderline or malignant PT (BMPT). The conventional MR parameters included age, longest diameter, shape, margin, internal enhancement characteristics, cystic component of the tumor, wall of the cystic component, peritumoral edema on T2-weighted imaging (T2WI), T1-weighted imaging (T1WI) and T2WI signal intensity, time-signal intensity curve (TIC) patterns and early-stage enhancement ratio (EER). The ADC values were determined in three different types of regions of interest (ROIs), including a circular ROI (ROI-c), single-slice ROI (ROI-s), and whole-tumor ROI (ROI-w). All ADC values were measured twice by Observer A and B (with a 2-week interval). The Ki-67 index was determined, and cases were classified into a "negative group" (Ki-67<14%) and a "positive group" (Ki-67≥14%). SPSS Statistics V21.0 was used for the statistical analyses.

Results: Our study included 23 cases of BPT and 33 cases of BMPT (including 22 borderline PTs and 11 malignant PTs). Only 23 patients in BMPT group had Ki-67 results, and 17 of these were positive. Regarding conventional MR features, significant differences were observed in the margin (P = 0.011), cystic component (P<0.001), peritumoral edema on T2WI (P<0.001), and cystic wall (P = 0.011) of the PT between the BPT and BMPT groups. Regarding the ADC value, good intraobserver agreement for ROI-c, ROI-s and ROI-w measurements was obtained. For the three different ROIs, the intraclass correlation coefficient (ICC) values were 0.905 for ROI-c (P > 0.05), 0.965 (P > 0.05) for ROI-s and 0.994 (P > 0.05) for ROI-w. ADC parameter indicated that the figure of ROI-s tended to be higher than the ROI-c and ROI-w, while the ROI-c and ROI-w values were similar. However, no significant difference was found in ADC values between the BPT and BMPT groups for ROI-c, ROI-s and mean ROI-w values and the 10th, 25th, 50th and 75th ROI-w. The areas under the ROC curves for the mean ROI-w and the 10th, 25th, 50th and 75th ROI-w were 0.568, 0.613, 0.567, 0.544, and 0.540, respectively.

Conclusion: Based on the results obtained in our study, the whole-tumor ADC histogram could not improve differentiation of the breast PT grade, while conventional MR images could provide more meaningful information, so morphological characteristics may be valuable than ADC value, and ADC could be used as a supplemental method to differentiate PT grades.

Keywords: Apparent diffusion coefficient (ADC); Breast; Magnetic resonance imaging; Phyllode tumors.