To evaluate the diagnostic performance of liver 18F-FDG PET/MRI in addition to whole-body PET/CT and to compare it with MRI in the detection and clinical management of liver metastasis in patients with colorectal cancer (CRC).

Seventy-eight patients with CRC who underwent whole-body 18F-FDG PET/CT followed by liver PET/MRI were prospectively included. Histopathological confirmation and/or at least 3 months of clinical follow-up after PET/MRI were accepted as gold standard. Lesion and patient-based analyses were performed to evaluate the diagnostics performances of PET/CT, PET/MRI and MRI. In addition, changes of clinical management were evaluated.

On lesion-based analysis, for PET/CT, PET/MRI and MRI; sensitivity (Se): 55.6%, 97.2% and 100%; specificity (Sp): 98.5%, 100% and 80.5%; and accuracy (Acc): 70.7%, 98.2% and 93.1% were calculated, respectively. Se and Acc of PET/MRI and MRI were significantly superior than PET/CT (P < 0.001). Se and Acc of PET/MRI and MRI were comparable; however, Sp of PET/MRI was significantly better than MRI (P < 0.001). On patient-based analysis, Se: 75.6%, 100% and 100%; Sp: 97.3%, 100% and 86.5%; and Acc: 85.9%, 100% and 93.5% were calculated, respectively. Se and Acc of PET/MRI were significantly superior than PET/CT (P < 0.001). Also, Se of MRI was significantly superior than PET/CT (P < 0.001). Se of PET/MRI and MRI were comparable, but Sp and Acc of PET/MRI were significantly better than MRI. The additional information obtained from liver PET/MRI changed treatment strategy in 14/78 (18%) patients compared to PET/CT or alone liver MRI.

Diagnostic performances of PET/MRI and MRI for detection of CRC liver metastasis is superior to PET/CT. PET/MRI especially helps in the accurate detection of liver metastases that are suspicious on MRI and has the potential to change the clinical management of especially oligometastatic patients by identifying uncertain liver lesions.