Protocol requirements and diagnostic value of PET/MR imaging for liver metastasis detection

To compare the accuracy of PET/MR imaging with that of FDG PET/CT and to determine the MR sequences necessary for the detection of liver metastasis using a trimodality PET/CT/MR set-up.

Included in this single-centre IRB-approved study were 55 patients (22 women, age 61 ± 11 years) with suspected liver metastases from gastrointestinal cancer. Imaging using a trimodality PET/CT/MR set-up (time-of-flight PET/CT and 3-T whole-body MR imager) comprised PET, low-dose CT, contrast-enhanced (CE) CT of the abdomen, and MR with T1-W/T2-W, diffusion-weighted (DWI), and dynamic CE imaging. Two readers evaluated the following image sets for liver metastasis: PET/CT (set A), PET/CECT (B), PET/MR including T1-W/T2-W (C), T1-W/T2-W with either DWI (D) or CE imaging (E), and a combination (F). The accuracy of each image set was determined by receiver-operating characteristic analysis using image set B as the standard of reference.

Of 120 liver lesions in 21/55 patients (38%), 79 (66%) were considered malignant, and 63/79 (80%) showed abnormal FDG uptake. Accuracies were 0.937 (95% CI 89.5 - 97.9%) for image set A, 1.00 (95% CI 99.9 - 100.0%) for set C, 0.998 (95% CI 99.4 - 100.0%) for set D, 0.997 (95% CI 99.3 - 100.0%) for set E, and 0.995 (95% CI 99.0 - 100.0%) for set F. Differences were significant for image sets D - F (P < 0.05) when including lesions without abnormal FDG uptake. As shown by follow-up imaging after 50 - 177 days, the use of image sets D and both sets E and F led to the detection of metastases in one and three patients, respectively, and further metastases in the contralateral lobe in two patients negative on PET/CECT (P = 0.06).

PET/MR imaging with T1-W/T2-W sequences results in similar diagnostic accuracy for the detection of liver metastases to PET/CECT. To significantly improve the characterization of liver lesions, we recommend the use of dynamic CE imaging sequences. PET/MR imaging has a diagnostic impact on clinical decision making.