Diagnostic performance of abbreviated non-contrast MRI for liver metastases in patients with newly diagnosed breast cancer

Purpose: To compare the diagnostic performance of non-contrast abbreviated liver MRI (abMRI) and standard MRI (sMRI) with gadoxetic acid enhancement in the detection of liver metastasis during the initial workup for patients with breast cancer. Methods: Of 7621 patients diagnosed with breast cancer who underwent abdominopelvic CT for their initial staging, 222 underwent sMRI between January 2016 and June 2019 to evaluate and/or characterize CT-indeterminate liver lesions. The abMRI protocol included diffusion-weighted images, apparent diffusion coefficient maps, and T2-weighted fat-suppression images, while the reference standard was histopathology or composite imaging follow-up. Two radiologists utilized a five-point scale to determine the probability of malignancy for each lesion. The per-patient diagnostic parameters were compared using generalized estimating equation and chi-square test. Results: A total of 222 female patients (age, 49.8 ± 10.4 years) including 17 with metastases (7.7 %) were included in the present analysis. When defining scores ≥4 as metastasis, there were no significant differences in the per-patient sensitivities (82.4 % vs. 82.4 %; p > 0.99), specificities (97.6 % vs. 98.1 %; p = 0.61), positive predictive values (73.7 % vs. 77.8 %; p = 0.63), negative predictive values (98.5 % vs. 98.5 %; p = 0.99), or accuracies (96.4 % vs. 96.9 %; p = 0.99) between the abMRI and sMRI groups, respectively. Additionally, there were no significant differences in the subgroups of patients with subcentimetre and stage II or higher disease. Conclusion: During the patients' initial workup, the diagnostic performance of non-contrast abMRI was comparable to that of sMRI with gadoxetic acid for CT-indeterminate liver lesions.