Whole body MRI (WBMRI) has an increasingly important role in the assessment and monitoring of multiple myeloma (MM) ^1,2. Dixon imaging is becoming more widely used in MM and can provide both functional ^3,4 and anatomical information. We observed that myeloma lesions seemed more conspicuous on fat only images than on conventional T1-weighted images, suggesting that Dixon imaging may offer additional information compared to T1-weighted images alone.

In this study, we evaluate myeloma lesion detection rates on the four Dixon image types: in phase (IP), out of phase (OP), water only (WO) and fat only (FO) (Figure 1). We hypothesised that lesion detection rates (sensitivity) would be highest on the fat only images.

Thirty patients (13 male and 17 female, median age 55, range 36-82) with suspected multiple myeloma underwent WBMRI imaging on a 3.0T wide-bore system (Ingenia; Phillips Healthcare, Best, Netherlands) using two anterior surface coils, a head coil and an integrated posterior coil. The WBMRI protocol included coronal pre- and post-contrast modified Dixon (mDixon) acquisitions from which fat and water images and calculated in and out of phase images were reconstructed (TR 3.0ms, TE 1.02-18, flip angle 15°, slice thickness 5mm, pixel bandwidth 1992Hz, acquisition matrix 196*238, SENSE factor 2, number of slices 120) and diffusion and post-contrast imaging, covering vertex to toe using ten anatomical stations.

Each patient had four image series (i.e. IP, OP, FO and WO). The individual series were randomised and read by two consultant radiologists with expertise in musculoskeletal and haematological imaging respectively. On each series, each observer was asked to count the number of myeloma lesions present in the bony pelvis (pubis, ischium, ilium and sacrum) and label these lesions on the images (up to a maximum of 20). If the disease was diffuse (or there were over 20 lesions), the patient was assigned a lesion count of 20. Additionally, the observers were asked to provide a confidence score based on their degree of certainty that there were myeloma lesions in the pelvis (1-no lesions, 2-indeterminate lesions, 3-likely myeloma lesions, 4-very likely myeloma lesions). After scoring, each labelled lesion was compared to a reference standard consisting of diffusion-weighted and post-contrast imaging, which had been evaluated by a further consultant radiologist with expertise in musculoskeletal imaging.

For each image series, we compared each lesion with the reference standard to determine the number of per-series true positive lesions (TP), false positive lesions (FP) (i.e. those that were incorrectly identified as lesions); and false negative lesions (FN) (these were the reference lesions which were not identified). For each Dixon image type, we recorded the mean per-series lesion count, sensitivity (TP/TP+FN), positive predictive value (TP/TP + FP) and mean confidence score.

Two observers read 120 image series each for 30 patients (four series per patient) and identified 609 and 789 lesions respectively. 1560 reference lesions were identified. The mean number of lesions for each image type was as follows: 8 for FO, 6.5 for WO, 6 for IP and 5.6 for OP (Figure 2). The sensitivity for each image type was: FO, 54%; IP, 38%; WO, 37% and OP 30%. The positive predictive values were: IP, 79%; FO 79%; WO, 73%; and OP, 64%. The mean confidence scores were: FO 2.8, WO 2.8, IP 2.6, and OP 2.6.

These results suggest that lesion detection rates are highest on FO images compared to other Dixon image types. Furthermore, the positive predictive values for FO images are comparable with IP images and higher than that of OP and WO, suggesting that the increase in sensitivity reflects a true increase in conspicuity rather than a lower reader ‘threshold’ for lesion identification. Readers also reported the highest overall confidence scores on the FO images. This may be because there is a proportionately greater change in bone marrow fat content than in water content when the bone marrow becomes infiltrated.

Importantly, many centres use solely T1-weighted (i.e. in phase) imaging in their WBMRI protocols for multiple myeloma; these results suggest that Dixon imaging should be used in preference to T1-weighted imaging alone.

In this study, we have not obtained confidence limits for sensitivity comparisons between modalities across the entire cohort. A potential solution would be to perform a meta-analysis (treating each patient in the cohort as an individual study) to definitively establish differences in sensitivity between modalities.

FO images offer improved lesion detection rates compared to IP imaging. Dixon imaging provides additional information with minimal increase in scan time and should be used in preference to T1-weighted imaging alone.