Biopsy of mammographic microcalcifications identified in screening often yields benign histology, producing a substantial number of potentially avoidable interventions. While full diagnostic MRI is established for invasive breast cancer, its role for detecting precancerous lesions such as ductal carcinoma in situ (DCIS) in an average-risk screening cohort remains uncertain. This study prospectively assessed the diagnostic performance and interreader agreement of a standardized abbreviated breast MRI reading protocol as an additional evaluation tool for suspicious microcalcifications detected on full-field digital mammography (FFDM).
This was a prospective single-center trial. The study population comprised 58 women (mean age 58 ± 24 years) who presented with 59 suspicious microcalcification findings on FFDM. A total of 68 lesions underwent histopathological verification; the manuscript reports diagnostic results based on the microcalcification lesions identified.
Four radiologists independently reviewed the MRI datasets. Reader experience ranged from 5 to 15 years. Readers were blinded to mammographic findings and clinical data to simulate an independent imaging assessment.
The abbreviated MRI protocol used a limited but standardized set of sequences: maximum intensity projection (MIP), early post-contrast subtraction images, and native (non-subtracted) images. Readers evaluated the datasets using these components to reach a binary recall decision (presumably recall for biopsy versus no recall). The abbreviated approach was designed to reduce reading time while preserving diagnostic information from contrast enhancement patterns and morphologic detail.
Among the 59 microcalcification lesions included in the analysis, 21 were malignant and 38 benign on histopathology. Across the four readers, diagnostic metrics were as follows:
Importantly, no invasive carcinomas were missed by any reader using the abbreviated MRI protocol. All observed false-negative MRI readings corresponded to cases of ductal carcinoma in situ (DCIS).
These results indicate that the abbreviated protocol provided reasonably high NPV and acceptable sensitivity for detecting malignant lesions associated with microcalcifications in this cohort, with some variability in specificity between readers.
The study also examined performance when readers used only the MIP component. MIP-only evaluation showed substantially lower sensitivity (48% to 52%) despite higher specificity (83% to 94%).
This contrast suggests that while MIP alone can reduce false positives, it misses a considerable proportion of malignant lesions—particularly non–mass-like or subtle enhancement patterns—making MIP-only approaches insufficient as a standalone assessment for suspicious microcalcifications.
Interreader agreement for the final recall decision was reported as moderate. Fleiss’ kappa for agreement among all four readers was 0.47. Pairwise Cohen’s kappa values between individual readers ranged from 0.40 to 0.53. These values reflect moderate concordance in recall decisions when using the abbreviated MRI protocol.
Mean reading times with the abbreviated protocol were short, ranging from 1 minute 42 seconds to 3 minutes 26 seconds across readers, supporting the time-efficiency objective of the abbreviated approach.
In this screening-like cohort, an abbreviated breast MRI protocol combining MIP, early post-contrast subtraction, and native images produced consistent diagnostic performance across readers, detected all invasive cancers, and achieved a high negative predictive value. The abbreviated protocol reduced reading time to a few minutes per case, which is favorable for workflow and resource considerations.
However, the exclusively missed malignancies were DCIS lesions. This raises clinical considerations about the sensitivity of abbreviated MRI for detecting noninvasive disease and whether additional imaging or biopsy thresholds should be maintained for calcification-only presentations.
The poor sensitivity observed with MIP-only reading highlights that brief, projection-only approaches sacrifice diagnostic sensitivity and should not replace multi-component abbreviated protocols when evaluating suspicious microcalcifications.
As reported, the study was a single-center prospective investigation with a relatively small sample size (58 women with 59 microcalcification lesions analyzed). The source reports reader experience ranges but does not provide lesion-level breakdowns beyond aggregate malignant versus benign counts, nor does it detail histologic subtypes beyond noting DCIS among false-negatives. Further multicenter studies with larger cohorts are necessary to validate generalizability and to better characterize performance for different histologic subtypes and calcification patterns.
A standardized abbreviated breast MRI protocol combining MIP, early post-contrast subtraction, and native images is a feasible, time-efficient adjunct for evaluating suspicious screening-detected microcalcifications, with consistent diagnostic performance across readers and no missed invasive carcinomas in this cohort. MIP-only evaluation demonstrated inadequate sensitivity and is not recommended as a standalone approach. The authors state that further evaluation is required before broad adoption of abbreviated MRI as a screening tool or routine adjunct for microcalcifications.