Session 3: Ultrafast MRI in the Clinical Setting to Improve Breast MRI Performance

Purpose: Breast MRI affords high sensitivity with intermediate specificity for cancer detection. Delayed phase kinetics from dynamic contrast enhanced (DCE)-MRI increases specificity but requires longer scan times. Ultrafast imaging assesses early contrast inflow with potential to supplement or replace conventional kinetics. We sought to determine whether ultrafast imaging can increase specificity of a clinical breast MRI protocol.

Materials and Methods: After IRB approval, we identified all breast MRIs from 3/2019-8/2020 with a single BI-RADS 3, 4, or 5 lesion, excluding patients with known breast cancer. This resulted in 101 breast MRIs with both conventional DCE (post-contrast acquisitions: 2, 5, and 8min) and ultrafast imaging (post-contrast acquisitions: 8, 16, 24, 32, and 40s) where lesions met reference standard (biopsy, follow-up imaging downgraded BI-RADS, or no breast cancer >2 years follow-up). MRIs were performed at 3T; ultrafast imaging was acquired during an idle delay time in the protocol shortly after contrast injection, but prior to conventional DCE. A radiologist, blinded to outcomes, retrospectively determined time to lesion enhancement (TTE) on ultrafast-MRI and noted aortic enhancement timing. Differences in TTE were compared between benign and malignant lesions using Wilcox rank sum test (p < 0.05 considered significant).

Results: Of 101 total lesions, 19 were malignant (3 DCIS lesions, 15 invasive breast cancers, and one breast lymphoma) and 82 were benign (51 biopsy-proven and 31 downgraded or no cancer on imaging follow-up). Median age was 46 (range: 26-77) years. The aorta enhanced at 16s for all patients. Mean TTE for benign lesions was higher than mean TTE for malignant lesions (37.8s vs. 29.1s; p<0.01). All lesions demonstrating no enhancement by 40s were benign, while all but one malignancy demonstrated TTE at ≤32s. Applying a cutoff of visible lesion enhancement >40s to downgrade to BI-RADS 2 would eliminate 16/70 total (16/51 [31%] benign) biopsies and end follow-up of 8/26 (31%) BI-RADS 3 lesions without missing a malignancy. A cutoff of >32s would avoid 26/70 (25/51 [49%] benign) biopsies and end follow-up of 14/26 (54%) BI-RADS 3 lesions, but miss one DCIS (BI-RADS 4) lesion.

Conclusion: Ultrafast imaging can be added to a clinical breast MRI protocol to improve specificity without increasing acquisition time. Future studies are needed to determine optimal thresholds to improve breast MRI accuracy.

Clinical Relevance Statement: Ultrafast imaging can be feasibly added to conventional DCE-MRI, with potential to increase diagnostic accuracy and reduce scan times. Absence of early lesion enhancement on a short ultrafast-MRI series may help exclude malignancy and decrease unnecessary biopsies.