Association of Contrast-Enhanced Mammography Imaging Features with Molecular Subtypes of Breast Cancer: A Cross-Sectional Study

Abstract

Background: Breast cancer molecular subtypes differ biologically and may exhibit distinct imaging phenotypes. This study examined whether mammographic morphologic features corresponding to the low-energy structural component of contrast-enhanced mammography (CEM) were associated with molecular subtype. Methods: A retrospective cross-sectional secondary analysis was conducted using the molecular-subtype branch of the Chinese Mammography Database. The patient-level cohort included 749 malignant cases classified as luminal A, luminal B, HER2-enriched, or triple-negative. Mass-only, calcification-only, and combined mass-with-calcification presentations were compared using Pearson's chi-square test and Cramer's V. Multinomial logistic regression adjusted for age and breast laterality. Prespecified secondary contrasts used modified Poisson regression with robust variance. Results: The cohort included 152 luminal A, 376 luminal B, 135 HER2-enriched, and 86 triple-negative cancers. Morphologic feature distribution differed across subtypes (chi-square[6] = 52.72, P < .001; Cramer's V = .188, 95% bootstrap CI [.150, .242]). Calcification-containing presentations occurred in 68.9% of HER2-enriched cancers, whereas 75.6% of triple-negative cancers were mass-only. Relative to luminal A and mass-only presentation, HER2-enriched cancer was associated with mass plus calcification (adjusted relative risk ratio [RRR] = 3.72, 95% CI [2.16, 6.41]) and calcification only (RRR = 3.85, 95% CI [1.93, 7.68]). Secondary analyses confirmed higher HER2-enriched prevalence with calcification-containing presentation and higher triple-negative prevalence with mass-only presentation. Conclusion: Simple structural mammographic features were associated with breast cancer molecular subtype. These findings support a subtype-related low-energy morphologic signal relevant to CEM interpretation, but they do not address contrast enhancement and cannot replace pathological receptor testing.