Magnetic resonance imaging captures the biology of ductal carcinoma in situ

Abstract

Purpose: Magnetic resonance imaging (MRI) is an important tool for characterizing invasive breast cancer but has proven to be more challenging in the setting of ductal carcinoma in situ (DCIS). We investigated whether MRI features of DCIS reflect differences in biology and pathology.

Patients and methods: Forty five of 100 patients with biopsy-proven DCIS who underwent MRI and had sufficient tissue to be characterized by pathologic (nuclear grade, presence of comedo necrosis, size, and density of disease) and immunohistochemical (IHC) findings (proliferation, Ki67; angiogenesis, CD34; and inflammation, CD68). Pathology and MRI features (enhancement patterns, distribution, size, and density) were analyzed using pairwise and canonical correlations.

Results: Histopathologic and IHC variables correlated with MRI features (r = 0.73). The correlation was largely due to size, density (by either MRI or pathology), and inflammation (P < .05). Most small focal masses were estrogen receptor-positive. MRI enhancement patterns that were clumped were more likely than heterogeneous patterns to be high-grade lesions. Homogenous lesions were large, high grade, and rich in macrophages. Presence of comedo necrosis and size could be distinguished on MRI (P < .05). MRI was most likely to over-represent the size of less dense, diffuse DCIS lesions.

Conclusion: The heterogeneous presentation of DCIS on MRI reflects underlying histopathologic differences.

Figure 1

Geographic distribution of magnetic resonance imaging patterns: (A) focal = nonmass enhancement organized towards a center; (B) linear (left) ductal (right) = traces pattern of a duct or a branch; (C) segmental = triangular enhancement, apex points towards the nipple, suggesting ductal distribution; (D) regional = single quadrant enhancement not conforming to duct distribution; (E) multiregional (left, patchy; right, diffuse) = areas of enhancement in = two quadrants.

Figure 2

Patterns of gadolinium enhancement: (A) heterogeneous = nonspecific, nonuniform, between regions of nonenhancement; (B) clumped = cobblestone-like; (C) homogeneous = confluent uniform.

Figure 3

Magnetic resonance imaging (MRI) pixel density score (contrast-enhanced, three-dimensional, gradient-echo breast MRI, sagittal projection, with fat saturation). (A) Score 1 = subtle clumped enhancement in upper breast; (B) score 2 = more concentrated nodular, clumped enhancement in the upper breast; (C) score 3 = dense confluent homogeneous enhancement.

Figure 4

Pathology compactness score. Pathology specimens were scored by the space lesions occupied on ×40 field and the distance between individual lesions. (A) Score 1 = less than 25%, sparsely distributed, > 5 mm apart; (B) score 2 = 25% to 50%, 2 to 5 mm apart; (C) score 3 = more than 50%, densely packed, < 2 mm apart.

Figure 5

Biomarkers of inflammation (CD68) and proliferation (Ki67) in ductal carcinoma in situ at ×40. (A) Low-density and (B) high-density staining of tumor-associated macrophages, highlighted by macrophage marker CD68 (brown) immunostain. (C) Low-proliferation and (D) high-proliferation indices demonstrated by nuclear proliferation antigen Ki67 (brown) immunostain.

Figure 6

Example magnetic resonance images from each group: (A) focal = small, low-grade, estrogen receptor (ER) –positive tumors; (B) linear ductal = high proliferation, low inflammation, low necrosis; (C) heterogeneous = ER positive, low inflammatory component; (D) clumped regional = large, ER negative, highly proliferative; (E) segmental = high inflammatory and proliferative components, as did (F) homogeneous = large, ER negative.

Figure 7

Correlation of magnetic resonance imaging (MRI) size (mm) and pathology size (mm) of ductal carcinoma in situ, labeled by MRI group, shows moderately strong correlation (r = 0.55). Solid line represents 1:1 association. Outliers where MRI size was 200% that of pathology size were predominantly in groups 3 and 4.