Objective: To determine whether delayed-phase dynamic contrast-enhanced (DCE)-MRI improves MRI–histopathology size agreement in invasive lobular carcinoma (ILC) compared with early-phase imaging, and whether modifying the cortical thickness threshol...
Objective: To determine whether delayed-phase dynamic contrast-enhanced (DCE)-MRI improves MRI–histopathology size agreement in invasive lobular carcinoma (ILC) compared with early-phase imaging, and whether modifying the cortical thickness threshold on ultrasonography (US) enhances the diagnostic performance for axillary lymph node metastasis in ILC relative to invasive ductal carcinoma (IDC).
Materials and Methods: This retrospective single-center study included 1,473 consecutive women (mean age 54.2 years ± 11.1; range, 29 87 years) with 1,531 breast cancers (128 ILC, 1,403 IDC) who underwent preoperative US, DCE-MRI, and surgery from March 2020 to December 2021. Tumor size was measured on DCE-MRI using early (90 seconds post-contrast) and delayed (590 seconds)-phase images, respectively. Between ILC and IDC groups, MRI and histopathologic size discrepancy were compared using Wilcoxon signed-rank test, and Spearman’s correlation coefficients (ρ) were compared using bootstrapping. Diagnostic performance of lymph node evaluation using current 2.5 mm and modified 2.0 mm and 3.0 mm cortical thickness thresholds was compared by McNemar's test.
Results: Compared with histopathology, tumor size on early-phase MRI was more markedly underestimated in ILC than IDC (mean discrepancy −0.9±2.2 cm vs 0.1±1.4 cm, p<0.001). In ILC, the correlation with histopathologic size was higher on delayed-phase than on than early-phase images (ρ=0.714 vs 0.555, p<0.001). In IDC, the histopathologic tumor size showed a stronger correlation with early-phase MRI than with delayed-phase MRI (ρ=0.760 vs 0.726, p<0.001). For ILC, modifying the cortical thickness threshold did not improve diagnostic performance; compared to the current 2.5 mm (sensitivity 51.5% [17/33], specificity 90.5% [86/95], false-negative rate 48.5% [16/33]), at 2.0 mm, false-negative rate (39.4% [13/33], p=0.25) and sensitivity (60.6% [20/33], p=0.25) were comparable and specificity was reduced (75.8% [72/95], p<0.001), while at 3.0 mm, sensitivity decreased (33.3% [11/33], p=0.03) and the false-negative rate increased (66.7% [22/33], p=0.03) with comparable specificity (93.7% [89/95], p=0.25).
Conclusion: For ILC, delayed-phase DCE-MRI provides a more accurate assessment of tumor size, while the early-phase remains suitable for IDC. Lowering the cortical thickness threshold does not improve nodal diagnostic accuracy.