Dissected sentinel lymph nodes of breast cancer patients: characterization with high-spatial-resolution 7-T MR imaging.

ABSTRACT

PURPOSE: To investigate the association of 7-T magnetic resonance (MR) imaging characteristics with metastatic nodal invasion, determined with histopathologic assessment in dissected sentinel lymph nodes of breast cancer patients. MATERIALS AND METHODS: Institutional review board approval and informed consent were obtained. From November 2008 to July 2010, 114 dissected lymph nodes from 33 women (mean age, 57 years; range, 31-80 years) with breast cancer were included. For morphological analysis, three-dimensional (3D) T1-weighted fat-suppressed fast field- (gradient-) echo (isotropic resolution, 180 µm) MR was performed; 3D nodal dimensions, maximum cortical thickness, and presence of fatty hilum were noted. For quantitative parametric analysis, two-dimensional T1-weighted and 3D T2-, T2*-, and diffusion-weighted images were acquired. Statistical analysis included generalized estimating equations (GEEs), forward and backward stepwise regression analyses, and calculation of positive predictive value (PPV) and negative predictive value (NPV). RESULTS: Of 114 nodes, 26 (23%) were malignant. Morphological criteria showed weak discriminatory power A fatty center was absent in 35% of malignant nodes and 30% of benign nodes (P = .9). Nodal volume and length-width ratio were not significantly different (P = .11 and .75, respectively). Cortical thickness (threshold level, 3 mm; P = .02) showed 91% NPV for malignancy and 95% NPV for presence of macrometastases. Quantitative parametric analyses showed comparable mean T1, T2, and T2* relaxation time constants and apparent diffusion coefficient for metastatic and benign nodes 991 msec, 30 msec, and 18 msec and 0.17 mm²/sec versus 1035 msec (P = .14), 31 msec (P = .001; not significant after GEE), and 15 msec (P = .002) and 0.20 mm²/sec (P = .38), respectively. Mean T2* alone offered an additive discriminatory effect for identification of metastatic nodes. Consistent with the notion of pannodal changes accompanying tumor infiltration, mean T2* differed significantly even if only micrometastases were present. The interindividual differences were small, precluding easy clinical implementation. CONCLUSION: Morphological criteria showed poor discriminatory power, even with very-high-spatial-resolution imaging. T2* quantification allowed identification of metastatic nodal invasion.