Correlation between MRI and biopsies under second look ultrasound

AbstractThe term “second look” lesions in MRI refers to lesions detected by MRI that were not initially seen on mammography or ultrasound. The objectives of our study were to analyse the displacement of targets between MRI and ultrasound; to define discriminating BIRADS morphological criteria to predict benign or malignant character and better establish the indications of second look ultrasound and biopsy; and to analyse the agreement between ultrasound and MRI in terms of morphological criteria. A retrospective and monocentric review was performed of the records of consecutive patients with breast abnormalities (mass or non-mass) initially detected by MRI that were not initially seen on mammography or ultrasound. All patients with abnormalities found during the performance of second look ultrasound and biopsied were included in the study. All lesions were documented using the BIRADS lexicon for MRI and ultrasound. Of 100 included patients, 108 lesions were detected by MRI, found via second look ultrasound and biopsied between January 2008 and 2010. All of the included patients were followed-up for a variable period, from 2 to 5 years. Eighty-two upon 108 biopsied lesions (76%) were benign and 26/108 lesions (24%) were malignant. This study confirmed the switch from procubitus to decubitus essentially displaces the tumour in the antero-posterior direction. It showed that the risk factors were not reliable criteria for establishing an indication for second look ultrasound. This study also showed that circumscribed contours and a progressive enhancement curve (type I) for masses on MRI had the strongest negative predictive value of greater than 0.85. In ultrasound, the round or oval shape, circumscribed contours and the parallel orientation to the skin favoured benignity with a NPV of greater than 0.85. For masses, the study showed that the agreement in interpretation of the benign versus suspicious morphological criteria between the MRI and the ultrasound was very weak for the shape (Kappa=0.09) and weak for the contours (Kappa=0.23). Finally, the MRI overestimated the size of the targets compared to ultrasound (Student t-test, p=0.0001). The performance of second look ultrasound has to be performed after the detection of an abdnormality on MRI even for lesion classified BIRADS 3. The biopsy indications must be wide with insertion of a clip and a control MRI. Only this control allows to stop the investigation if the biopsied lesion is benign