The role of MRI and clinicopathologic features in predicting the invasive component of biopsy-confirmed ductal carcinoma in situ

The institutional review board of our hospital approved this study and the requirement for informed consent was waived due to the retrospective nature of the analyses.

We conducted a retrospective review of the pathologic database of our institution for patients with a histologic diagnosis of DCIS on stereotactic vacuum-assisted biopsy (SVAB) or US-guided CNB (US-CNB) between January and December 2015. We enrolled patients who underwent preoperative breast MRI. The lesions detected on both mammography and US were confirmed by US-CNB. In general, SVAB was performed with either an 8- or 11-gauge vacuum-assisted biopsy needle (Mammotome; Devicor Medical Products, Cincinnati, OH, USA) on 12 samples harvested from each patient. US-CNB was performed with a 14-guage automated biopsy gun (Stericut; TSK Laboratory, Tochigi, Japan) with five samples obtained from each patient. We excluded patients who had undergone excision before the preoperative MRI (n = 14) or had not undergone surgery (n = 11). Finally, 201 patients (mean age 49.7 ± 10.7 years) with 206 breast lesions were included in the study cohort (Fig. 1). Five patients had a biopsy-confirmed DCIS in bilateral breasts. Ninety-five patients showed overlap with a previous study that compared surgical outcomes of DCIS with and without preoperative MRI [22]. All patients underwent both initial mammography and US prior to the MRI examination. Mammographic densities were recorded in accordance with the categories of breast composition [23], i.e., almost entirely fatty, scattered fibrograndular tissue (fatty), heterogeneously dense, or extremely dense (dense). The presence of calcification was also recorded from the mammographic images.

Bilateral MRI was performed using a 1.5 T or 3.0 T (Avanto; Siemens Medical Solutions, Erlangen, Germany, Skyra; Siemens Medical Solutions, Erlangen, Germany, Ingenia; Philips, Best, The Netherlands) MR scanner and a dedicated 18-channel phased-array breast coil (Siemens Medical Solutions) with the patient in a prone position. The imaging protocol included a T2-weighted short tau inversion recovery turbo spin-echo pulse sequence (repetition time [TR]/echo time [TE], 1300/131; matrix size, 384 × 384; field of view [FOV], 340 × 340 mm²; slice thickness, 1.5 mm for the 1.5 T scanner; TR/TE, 1100/131; matrix size, 256 × 416; FOV, 341 × 210 mm²; section thickness,1.5 mm for the 3.0 T scanner) and a dynamic contrast material-enhanced fat-saturated axial three-dimensional T1-weighted fast low-angle shot sequence (TR/TE, 5.0/2.4; matrix size, 384 × 384; FOV, 340 × 340 mm²; section thickness, 0.9 mm for the 1.5 T scanner; TR/TE, 5.6/2.5; matrix size, 384 × 384; FOV, 360 × 360 mm²; section thickness, 0.9 mm for the 3.0 T scanner), consisting of unenhanced and five contrast-enhanced acquisitions. Contrast material (0.2 mL/kg gadoterate meglumine; UNIRAY®; Dongkook Pharmaceutical Co., Ltd., Seoul, Korea) was power-injected (Spectris; Medrad, Pittsburgh, PA, USA) at a flow rate of 1 mL/s, followed by a 20 mL saline flush.

All MR images were reviewed in consensus by two dedicated breast radiologists (G.Y.Y. and W.J.C., with 4 and 10 years of experience in breast imaging, respectively) using the Breast Imaging Reporting and Data System (BI-RADs) [23]. Reviewers were blinded to the clinicopathologic findings. The level of background parenchymal enhancement (BPE; minimal, mild, moderate, or marked), as well as lesion multifocality, multicentricity, and morphologic features were determined. Lesions were classified into mass or non-mass enhancement (NME) types. In patients with multiple lesions, the index lesion, with the largest diameter, was evaluated. Masses were analyzed according to shape, margin, and internal enhancement; the presence of intratumoral high signal intensity (SI) and peritumoral edema was also evaluated on T2-weighted images (T2WI). Intratumoral high SI was visually determined when the SI of the tumor was higher than, or almost the same as, that of water or the vessels, or higher than that of the surrounding normal parenchymal tissue [24]. Peritumoral edema was defined by high SI around the tumor observed on T2WI [25]. NMEs were analyzed according to their distribution (focal, linear, segmental, regional, multiple regions, or diffuse) and internal enhancement. The kinetic features of each tumor were evaluated using a commercially available CAD system (CADstream, version 6.0.1; Confirma, Kirkland, WA, USA). To measure MRI parameters, pre- and post-contrast T1-weighted image series were transferred to a CAD system, which automatically segmented and calculated the tumor diameter (maximal size of the enhancing lesion), angio volume (volume of total enhancing lesion), peak enhancement (highest pixel SI in the first post-contrast series), and delayed enhancement profiles (proportions of persistent, plateau, and washout enhancing components within a tumor).

All histopathologic information was obtained from the pathology reports of the surgical specimens. We divided the study patients into three groups according to their final surgical pathologic results: DCIS, mIDC, and IDC. When the histological analysis of the surgical specimen showed microinvasive or invasive foci within a tumor which had been preoperatively diagnosed as DCIS, the tumor was defined as a histologically upgraded lesion. Microinvasion was defined according to the 7th edition of the AJCC Cancer Staging Manual, as an extension of the cancer cells beyond the basement membrane into adjacent tissues but not exceeding 1 mm in the greatest dimension [3]. Invasion was also defined in tissues with basement membrane invasion [3]. Patients with multiple microinvasion foci were included in this classification if the size of the largest focus did not exceed 1 mm. Pathological data included nuclear grade, presence of necrosis, pathologic tumor size, the status of the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), and Ki-67 proliferation. Pathologic tumor size was recorded as the largest tumor diameter including both DCIS and invasive component. HER2 positivity was defined as a protein overexpression score of 3+ determined by immunohistochemistry or the presence of gene amplification (positive in situ hybridization) [26]. Ki-67 scores ≥20% were considered high.

Imaging findings and clinicopathologic features were compared between the DCIS, mIDC, and IDC groups using the Chi-square test or Fisher’s exact test for categorical variables, and the ANOVA or Kruskal-Wallis test for continuous variables. Independent predictors of microinvasive and invasive carcinomas were analyzed using cumulative logistic regression analysis. Variables with P values < 0.1 in univariate analysis were included in the multivariate analysis using a cumulative logistic regression method with backward elimination. The variable, tumor size on MRI, was excluded from the multivariate analysis due to lack of data (non-visible lesions on MRI). For the lesions detected on MRI, a subgroup analysis was performed in accordance with the MR lesion type. Kinetic features were analyzed using the Chi-square test or Fisher’s exact test for categorical variables, and the Kruskal-Wallis test for continuous variables. All statistical analyses were performed with SPSS software version 20.0 (SPSS, Chicago, IL, USA) or SAS version 9.3 (SAS Institute, Cary, NC, USA). A P value < 0.05 was considered statistically significant.

Among the 206 biopsy-confirmed DCIS lesions analyzed in this present study, 112 (54.4%) were found to be pure DCIS in the final surgical pathology, 50 (24.3%) were upgraded to mIDC, and 44 (21.4%) were upgraded to IDC. Thirty-eight and 168 patients were diagnosed by SVAB and US-CNB, respectively. The clinicopathologic features of these samples are summarized in Table 1. The SVAB method (P = 0.028) and breast-conserving surgery (P < 0.001) were significantly more frequent in the DCIS group than in the mIDC and IDC groups. However, the mean age at diagnosis did not significantly differ between the three groups. Histologically, a higher nuclear grade (P = 0.001), larger pathologic tumor size of the surgical specimen (P < 0.001), negative ER (P < 0.001), negative PR (P < 0.001), positive HER2 (P < 0.001), and high Ki-67 level (P < 0.001) were significantly more frequent in the groups with upgraded lesions.

Table 2 presents the imaging features of the DCIS, mIDC, and IDC groups. The presence of calcification on mammography was more frequent in the groups with upgraded lesions (P = 0.025). The mIDC and IDC groups tended to show more NME lesions than the DCIS group; however, this difference did not reach statistical significance (P = 0.053). There were no significant differences between the groups with respect to parenchymal density on mammography or BPE on MRI.

Among all variables, the followings that showed P values < 0.1 on univariate analysis were used as input variables in subsequent multivariate analysis: older age (P = 0.070), US-CNB (P = 0.009), nuclear grade 3 (P = 0.001), presence of necrosis (P = 0.044), negative ER (P < 0.001), negative PR (P < 0.001), positive HER2 (P < 0.001), high Ki-67 level (P < 0.001), presence of mammographic calcification (P = 0.072), minimal or mild BPE (P = 0.051), and mass (P = 0.041) or NME (P = 0.019) detection on MRI. The multivariate analysis revealed that a mass lesion detected on MRI (Odds ratio (OR) = 8.84, 95% confidence interval (CI) = 1.05–74.04, P = 0.045), NME lesion detected on MRI (OR = 11.17, 95% CI = 1.35–92.36, P = 0.025), negative PR (OR = 2.40, 95% CI = 1.29–4.44, P = 0.006), and higher Ki-67 level (OR = 2.42, 95% CI = 1.30–4.50, P = 0.005) remained significant independent factors associated with histologic upgrade (Table 3).

Table 4 lists the MRI features by lesion type in the DCIS, mIDC, and IDC groups. In the two invasive groups, the median tumor size for both mass and NME lesions was significantly greater than that in the DCIS group. The dominant imaging features of the mass lesions in the two invasive disease groups were irregular shape and not-circumscribed appearance with heterogeneous or rim enhancement (P = 0.001). Conversely, the DCIS lesions showed benign favored features more frequently (Fig. 2). Intratumoral high SI and peritumoral edema on T2WI were also more frequent in the invasive groups than in the DCIS group (P < 0.001). NME lesions with clumped or clustered ring enhancement were more frequent in the invasive disease groups than in the DCIS group (P = 0.001; Fig. 3). Segmental distribution of NME lesions was common in all three groups, but the invasive groups showed a higher percentage of segmental distribution than the pure DCIS group; however, this difference did not reach statistical significance (P = 0.075). Lesions with high peak enhancement percentages were more frequent in mIDC and IDC groups (P < 0.001). Other kinetic features of mass and NME lesions, including initial and delayed enhancement, persistent component, plateau component, and washout component did not significantly differ between the three groups.