PurposeTo evaluate the rate of concordance between pathology and preoperative breast MRI performed with an abbreviated protocol (AP) or a full diagnostic protocol (FDP) for estimation of the tumour extent in patients with pure ductal carcinoma in situ (DCIS). MethodsThis retrospective study included 164 patients with pathologically proven DCIS who underwent preoperative breast MRI. Two radiologists independently evaluated the tumour extent on MRI with (FDP) and without the delayed phase (AP) and compared the readings with the pathological tumour extent. The background parenchymal enhancement (BPE) and morphology were also evaluated. Furthermore, the influence of the degree of BPE, presence or absence of B2 and B3 lesions, and pathological DCIS grade on the accuracy of MRI findings was assessed. Concordance between MRI and pathology was evaluated using correlation analysis. ResultsSpearman’s rank correlation coefficients for the concordance between MRI and pathology were 0.63 (reader 1) and 0.69 (reader 2) with AP and 0.65 and 0.73 (readers 1 and 2, respectively) with FDP. For both readers, the difference in the measured value between FDP and pathology was significantly smaller than that between AP and pathology (p < 0.001). The inter-reader variation in the measured tumour extent was larger with FDP than with AP. The presence of B3 lesions, low-grade DCIS, and moderate/marked BPE lowered the rate of concordance between MRI and pathology. ConclusionsOur findings suggest that preoperative MRI with FDP is more accurate than that with AP alone for estimation of the tumour extent in patients with pure DCIS.
Read full abstract