Purpose: To compare tumor size and the relationship to breast landmarks between preoperative supine (psMRI) and prone breast MRI (ppMRI). To assess performances of intraoperative supine MRI (isMRI) in predicting resection margins after breast-conserving therapy (BCT).
Materials and Methods: An IRB-approved, HIPAA-compliant prospective study from 4/1/2012-12/31/2022 included 43 patients with stage I-II breast cancer. All patients with measurable lesions on ppMRI eligible for BCT were included and three patients with multicentric disease were excluded. The psMRI was performed after completion of standard ppMRI for staging with injection of a half dose of gadolinium using a body/torso coil. Tumor sizes; closest distances to nipple, chest wall, skin; and tumor long axis-to-chest wall angles were independently measured by three breast radiologists on ppMRI and psMRI and compared using a paired T-test. After lumpectomy, the surgical cavity was filled with saline and temporarily sutured. Resection margins were assessed on 110-second postcontrast T1-weighted images of isMRI with and without CAD. Standard shave margins were then taken to correlate pathologic margin status with isMRI findings. Intraclass correlation (ICC) was determined for inter-reader reliability of measurements. Performance metrics of isMRI and re-excision rate were analyzed.
Results: Out of 40 included patients, 29 had psMRI and all had ppMRI. 27/29 cancers were visualized on both psMRI and ppMRI. Tumor sizes of psMRI vs ppMRI were similar (long axis mean±SD, 1.57±0.71 cm vs 1.71±1.29 cm, p=0.40; perpendicular short axis mean±SD, 1.11±0.47 cm vs 1.14±0.60 cm, p=0.61). Anatomical distances were significantly shorter on psMRI vs ppMRI for tumor-to-nipple (mean difference, -1.96; p< 0.001) and tumor-to-chest wall (mean difference, -1.91; p< 0.001). Tumor-to-skin distance (mean difference, 0.06) and tumor long axis-to-chest wall angle (mean difference, -8.48) were similar (p=0.51). Good to excellent reliability of measurements among three readers was found (ICC 0.86-0.98). Average isMRI scan time was 31.67 minutes. Resection margins were pathologically positive for carcinoma in 8/40 (20.0%) patients. Margin assessment on isMRI yielded accuracy, sensitivity, specificity, PPV, and NPV of 80.0%, 50.0%, 92.9%, 75.0%, and 81.3%, respectively. Re-excision rate was 12.5%.
Conclusion: Tumor-to-nipple and tumor-to-chest wall distances were significantly shorter on psMRI than ppMRI. Tumor size, tumor-to-skin distance, tumor long axis-to-chest wall angles did not change significantly. Assessment of resection margin on isMRI yielded favorable accuracy.
Clinical Relevance Statement: In this pilot use of supine breast MRI, psMRI assessment may alter tumor localization relative to nipple and chest wall compared to ppMRI, imitating patient positioning during BCT. Resection margins can be evaluated with isMRI with favorable accuracy.
