Comparison of combined whole-body 18F-NAF and 18F-FDG PET/CT versus MRI for the detection of myeloma lesions

Objectives Imaging requirements for the diagnosis of multiple myeloma (MM) recently changed and 蠅 1 osteolytic bone destruction (蠅 5 mm in size) seen on whole-body (WB) low-dose computed tomography (ldCT) or positron emission tomography combined with CT (PET/CT) does fulfill the criteria for bone disease. The present work assessed the lesion detection rate of WB combined [18F]NaF and [18F]FDG PET/CT versus ldCT alone and MRI in patients with newly diagnosed MM.

Methods Patients with newly diagnosed MM, prospectively included, underwent WB (from vertex to toes) XR, MRI and combined [18F]NaF and [18F]FDG PET/CT (median delay between scans: 6 days). PET/CT scans were acquired after injection of 134 ± 13 MBq [18F]NaF and 249 ± 18 MBq [18F]FDG (median uptake time: 64 min). The ldCT (3 mm slice thickness; 120 kV; 50-80 mAs) followed by PET emission scan (90 seconds per bed position) were performed. The MR images were acquired in coronal planes in T1-weighted and T2-weighted short-tau inversion recovery. Diffusion-weighted with background suppression images were acquired in the axial plane and reconstructed on coronal planes. PET and LdCT images were reviewed by 2 experienced nuclear medicine physicians and 1 radiologist to detect focal lesions (FLs) and/or diffuse bone marrow involvement. The focal areas of visually detectable increased tracers’ uptake were considered as PET FLs. The MR, ldCT alone and XR images were analyzed by 3 radiologists blinded to each other and to PET/CT results. The FLs were classified according to their location: pelvis, skull, limbs, spine, ribs and one location including the sternum, scapula and clavicles. The McNemar’s test was used to compare the detection rate of each technique and the Kruskal-Wallis test was used to estimate a relationship between the detection rate and the size of FLs measured with ldCT.

Results Out of 14 patients initially included, two were excluded (one for a delay > 40 days with XR and one who experienced claustrophobia during MRI acquisition). Twelve myeloma patients (median age 64y) with stage 1 (n = 4), 2 (n = 5) or 3 (n = 3) were included in the analyses. The pattern of bone marrow involvement was focal (n = 7) or combined diffuse and focal (n = 5). Per patient, 1-3 FL (n = 4), 4-10 FLs (n = 2) or > 10 FLs (n = 6) were detected. The total number of FLs detected was 281; no extramedullary disease was detected. The detection rate of MM lesions between techniques was significantly different (p < 0.05): XR (89; 32%) < PET (158; 56%) < MRI (183, 65%) < LdCT alone (219; 78%) < PET/CT (277; 99%). Out of 158 FLs detected with PET, 125 (79%) were also detected with MRI. Out of 183 MM lesions detected with MRI, 125 (68%) were detected with PET; PET positivity was significantly associated with lesion size (p = 0.002). Out of 145 FLs (蠅 5 mm) detected with ldCT, the detection rate of MRI (n = 87; 60%) and PET (n = 96; 66%) was similar (p = 0.17) and significantly associated with lesion size only for MRI (p = 0.014). Whatever FLs location, the detection rate of PET and MRI was similar except for rib MM lesions for which PET was superior to MRI (p = 0.0005). Seventeen osteolytic rib lesions detected with PET were not detected with MRI, of which only one corresponded to a pathologic fracture. At the patient’s level, the diagnosis of MM was based on biological data in 7/12 patients. For the 5 remaining patients, MM diagnosis required imaging. PET/CT and ldCT alone correctly identified bone involvement in all 5 patients; MRI would have missed the correct MM diagnosis in 3 patients (2 with diffuse pattern only and one with 1 FL); WBXR would have missed MM diagnosis in 1/5 patient (one pelvic FL missed).

Conclusions The MM lesion detection rate of PET/CT was superior to ldCT alone and MRI, respectively. At the patient’s level, the accuracy of PET/CT and CT alone was superior to MRI and WB-XR for the diagnosis of MM.