Can magnetic-resonance imaging help elucidate natural history of breast cancer multicentricity?

Abstract

We evaluated prospectivelywhether small enhancing foci, seen separately from the maintumour on contrast-enhanced MRI, were indeed cancer fociand whether MRI could detect all cancer foci identified byradiological-histological correlation.We studied ten patients and did preoperative contrast-enhanced breast MRI. High resolution transverse T1-weighted 3D FLASH images (TR=18 ms, TE=7 ms,FA=40°, TA=4 m 56 s, FOV=410 mm) before and after anintravenous bolus hand injection of dimegluminegadopentetate (Magnevist, 0·2 mL/kg) were acquired at 1·0T (Siemens Magnetom Scanner 42 SP with dedicated breastcoil). The 3D volume was 64 mm thick with 32 partitionsgiving an effective slice thickness of 2 mm and this wassufficient to cover the entire breast in all cases. After surgicalexcision (four mastectomies, six wide local excisions), thespecimens were fixed and sliced at 5 mm intervals in thesame plane as the MRI. Routine histopathologicalexamination was performed by an experienced breastpathologist and the remaining material was radiographed.Two observers identified radiological abnormalities(calcifications, densities, or spiculations) and all lesions thatwere deemed suspicious by either observer were sampledand examined histologically. MRI images were reviewedindependently and findings were compared withhistopathology results.On MRI, 19 enhancing foci separate from the maintumour were identified in seven out of ten patients (figure).On radiography of specimen slices, 71 suspicious areas weresampled and histological examination revealed 15 areas ofin-situ cancer (nine) or invasive cancer (six) in five patients.All five patients with cancer foci were amongst the sevenpatients who had enhancing foci on MRI. In two of these fivepatients, the tumour was surrounded by widespreadenhancement on MRI and all 11 (four+seven) areas sampledshowed cancer foci. In all wide local excision specimens, theenhancing foci on MRI were within 11 mm of the tumouredge and therefore within the resected specimen.Our data suggest that enhancing foci on MRI representcancer foci and that MRI detected 14 out of 15 cancer foci(sensitivity 93%). Of course, this is based on the assumptionthat the radiological-histological correlational method isindeed the gold standard. If that is so, the specificity of MRIfor tumour detection would be 79% (15/19). However,bearing in mind that the spatial resolution of MRI is of theorder of 1–2 mm, it may yet transpire that MRI could haveeven greater sensitivity and specificity which may becomeapparent with an even more obsessional sampling of thespecimen.