International investigation of breast MRI: results of a multicentre study (11 sites) concerning diagnostic parameters for contrast-enhanced MRI based on 519 histopathologically correlated lesions.

A multicentre study was undertaken to provide fundamentals for improved standardization and optimized interpretation guidelines of dynamic contrast-enhanced MRI. Only patients scheduled for biopsy of a clinical or imaging abnormality were included. They underwent standardized dynamic MRI on Siemens 1.0 (163 valid lesions > or = 5 mm) or 1.5 T (395 valid lesions > or = 5 mm) using 3D fast low-angle shot (FLASH; 87 s) before and five times after standardized bolus of 0.2 mmol Gd-DTPA/kg. One-Tesla and 1.5 T data were analysed separately using a discriminant analysis. Only histologically correlated lesions entered the statistical evaluation. Histopathology and imaging were correlated in retrospect and in open. The best results were achieved by combining up to five wash-in or wash-out parameters. Different weighting of false-negative vs false-positive calls allowed formulation of a statistically based interpretation scheme yielding optimized rules for the highest possible sensitivity (specificity 30%), for moderate (50%) or high (64-71%) specificity. The sensitivities obtained at the above specificity levels were better at 1.0 T (98, 97, or 96%) than at 1.5 T (96, 93, 86%). Using a widely available standardized MR technique definition of statistically founded interpretation rules is possible. Choice of an optimum interpretation rule may vary with the clinical question. Prospective testing remains necessary. Differences of 1.0 and 1.5 T are not statistically significant but may be due to pulse sequences.