Breast MRI With Abbreviated Reading Protocol as Additional Evaluation Tool for Suspicious Microcalcifications in a Screening Cohort: Diagnostic Performance and Interreader Agreement

Le résumé fourni par la source

Background/Aim: Biopsy of suspicious microcalcifications (MC) detected in breast cancer screening yields benign results in the majority of cases, resulting in a substantial number of avoidable interventions. While magnetic resonance imaging (MRI) demonstrates high accuracy for invasive breast cancer, its diagnostic value for precancerous lesions remains uncertain - particularly in average-risk screening populations. This study aimed to assess the diagnostic performance of an abbreviated breast MRI reading protocol in a screening-like cohort. Patients and Methods: In this prospective single-center study, 58 women (mean age 58±24 years) with 59 suspicious MC on full-field digital mammography (FFDM) were included. A total of 68 lesions underwent histopathological verification. MRI datasets were independently assessed by four radiologists (5-15 years of experience), blinded to mammographic findings, using a standardized abbreviated protocol including maximum intensity projection (MIP), early post-contrast subtraction, and native images. Results: Among 59 microcalcification lesions, 21 were malignant and 38 benign. Sensitivity ranged from 74% to 83%, specificity from 58% to 83%, positive predictive value from 56% to 75%, negative predictive value from 83% to 86%, and accuracy from 68% to 81%. No invasive carcinomas were missed by any reader. False-negative findings were exclusively ductal carcinoma in situ (DCIS). Mean reading time ranged from 1 min 42 sec to 3 min 26 s. MIP-only evaluation demonstrated markedly lower sensitivity (48-52%) despite higher specificity (83-94%). Interreader agreement for the final recall decision was moderate (Fleiss’ κ=0.47), with pairwise Cohen’s κ values ranging from 0.40 to 0.53. Conclusion: Abbreviated breast MRI is a feasible and time-efficient approach with consistent diagnostic performance across readers for evaluating suspicious MC, further evaluation as screening tool is needed. MIP-only evaluation, does not appear to provide sufficient diagnostic reliability.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.