This retrospective study was granted institutional approval by our Institutional Service Evaluation committee and the requirement for informed consent for data use was waived.
Consecutive patients referred to our centre with a suspected or confirmed new diagnosis of myeloma that underwent 18F-FDG PET/CT and WBMRI between March 2014 and May 2018, prior to treatment, were included. Patients were excluded if they did not fulfil inclusion criteria.
Following a 6-h fast, up to 400 MBq 18F-FDG was administered intravenously if blood glucose was < 10 mmol/L. At 60 +/− 5 min post-injection, imaging was acquired from skull vertex to feet at 3.5 min per bed position with an axial field of view of 15.7 cm and an 11-slice overlap between bed positions, using a GE Discovery 710 PET-CT scanner (GE Healthcare, Chicago, US). A low-dose CT scan (140 kV, mA 15-100, noise index 40, 0.5 s rotation time, and 40 mm collimation) was performed at the start of imaging to provide attenuation correction and an anatomical reference. PET image reconstruction included scanner-based corrections for radiotracer decay, scatter, randoms, and dead-time. Emission sinograms were reconstructed with an ordered subset expectation maximisation algorithm (2 iterations, 24 subsets).
WBMRI consisting of T2-weighted fast-spin echo, pre-and post-contrast agent T1-weighted Dixon spoiled 3D gradient-recalled echo and diffusion-weighted echo-planar imaging (with b value = 50 and 900 s/mm2) sequences from the skull vertex to below knees was performed at 1.5 T (Magnetom Aera, Siemens Healthcare, Erlangen, Germany) (Supplemental Table 1). WBMRI study duration was approximately 45 min, depending on patient height.
18F-FDG PET/CT and WBMRI imaging were reviewed independently by an experienced nuclear medicine physician and radiologist, respectively specialising in whole body and myeloma imaging. The presence/absence of focal and/or diffuse disease and number of focal lesions were recorded. Where focal lesions were present, their number was recorded as follows: <5; 5 to 10; or >10. Focal bone lesions were categorised as 18F-FDG-positive, i.e. standardised uptake value (SUV) greater than background bone marrow, or 18-FDG-negative. The presence or absence of an osteolytic lesion typical of myeloma on the CT component using bone window settings was also recorded, whether 18F-FDG-avid or not. For the purposes of qualitative assessment for diffuse marrow infiltration on 18F-FDG PET/CT, the background bone marrow was considered positive if bone marrow 18F-FDG uptake exceeded hepatic activity (bone marrow SUVmax: hepatic SUVmax ratio >1.0) . Bone marrow involvement at WBMRI was recorded as follows: normal, focal, diffuse, salt and pepper or a combination of abnormal patterns .
Further evaluation was performed in a subset of patients (25%) by 2 additional readers (a nuclear medicine physician and radiologist, respectively) to assess inter-observer agreement. For WBMRI, a total lesion score was assigned to facilitate assessment of inter-reader agreement of number of lesions across the different sites. The number of bone lesions were scored as follows: score 0=0 lesions; 1=1-4 lesions; 2=5-10 lesions; 3 =>10 lesions; for following 7 skeletal regions: skull, cervical spine, thoracic spine, lumbar spine, pelvis, long bones, ribs/other.
Clinical data collection
Electronic patient records were reviewed and the following patient data were collected and anonymised; demographic data, serum haemoglobin, creatinine, free light chain ratio, paraprotein level and bone marrow trephine plasma cell percentage at diagnosis; final recorded diagnosis following multidisciplinary consensus clinical discussion and international myeloma stage (ISS) served as the reference standard for presence of disease status.
Anonymised clinical and imaging data were reviewed by two consultant haematologists in consensus. Imaging data included; study positive or negative for disease by IMWG criteria, number of focal bone lesions (and for 18F-FDG PET/CT, whether bone lesions were FDG positive or negative, and with/without associated lytic component on CT), WBMRI disease pattern and likely presence/absence of diffuse infiltration on 18F-FDG PET/CT based on objective assessment of background bone marrow avidity compared to the liver. Clinical data alone, clinical and 18F-FDG PET/CT data and clinical and WBMRI data were reviewed separately and virtual management decisions made (either treat as myeloma or active surveillance) as per institutional treatment guidelines . Data was randomised and reviewed on separate occasions, with an interval of at least 1 week, to minimise recall bias.
Concordance and diagnostic performance of WBMRI and 18F-FDG PET/CT were assessed against the IMWG definition of myeloma bone disease which states that there should be one or more 5 mm osteolytic bone lesions at CT or 18F-FDG PET/CT and more than one unequivocal bone marrow lesion (measuring at least 5 mm) at WBMRI  and clinical reference standard for disease status, including bone marrow biopsy and final recorded diagnosis following multidisciplinary consensus clinical discussion.
Sensitivity of bone marrow SUVmax: hepatic SUVmax ratio >1.0 for detection of bone marrow infiltration (where bone marrow trephine plasma cell percentage of >10% was taken as the reference standard) was assessed. Spearman rank correlation was performed for assessment of correlation between background bone marrow avidity, SUVmax at the iliac crest, and bone marrow plasma cell percentage; and WMBRI bone marrow infiltration pattern and bone marrow plasma cell percentage. Inter-reader agreement for disease scoring by 18F-FDG PET/CT or WB-MRI was assessed using Cohen’s Kappa for categorical and Intraclass Correlation Coefficient for continuous data. The McNemar test was used to assess the significance of the difference between the proportion of different management decisions (either active treatment or surveillance) based on (i) clinical information alone compared with decision based on (ii) clinical information + 18F-FDG PET/CT result, or (iii) clinical information + WBMRI result, and finally (iv) comparing decision based on clinical information + 18F-FDG PET/CT and clinical information + WBMRI. P<0.05 denoted statistical significance.