ABSTRACT
PURPOSE: We tested a novel multi-parametric (mp) whole body (WB)-MRI evaluation algorithm for medullary lesions in comparison to positron emission tomography (PET) radiotracers 18F-fluorodeoxyglucose (18F-FDG) and 11C-methionine (11C-MET). METHODS AND MATERIALS: This retrospective single-center study included 44 MM patients, who received both 18F-FDG-PET and WB-MRI within ten days. MRI classified focal lesions as vital when showing 1) significant diffusion-restriction, 2) a fat fraction (FF) less than 20 % and 3) homogenous hypointensity on T2-weighted images. On a lesion-by-lesion level the findings were compared to 18F-FDG PET by using a 5-point scoring system (analogous to the Deauville score [DS]). In 24/44 (55 %) patients additional comparison to 11C-MET PET was available. RESULTS: Among two radiologists, an excellent inter-observer reliability for mpWB-MRI in a total of 84 medullary lesions was observed (ICC = 1, k = 1, p <.01). 16/17 (94.1 %) MRI-classified vital lesions had a DS of 4 or 5 on either 18F-FDG-PET or 11C-MET-PET. MRI-rated non-vital lesions correlated with PET-based DS ≤ 3. When results of mpWB-MRI were compared to 18F-FDG, a fair inter-observer agreement was recorded (ICC = 0.52, k = 0.53, p <.01), while for 11C-MET, an excellent concordance rate was achieved (ICC = 0.81, k = 0.79, p <.01). CONCLUSION: The proposed mpWB-MRI interpretation algorithm allowed to assess tumor activity of myeloma lesions with high inter-observer reproducibility. We observed a substantial concordance between the mpWB-MRI classification of lesions and PET assessment based on a semi-automatically calculated 5-point scoring system analogous to the Deauville scores.
