Differentiation of multiple myeloma and metastases: Use of axial diffusion-weighted MR imaging in addition to standard MR imaging at 3T

Background Metastasis and multiple myeloma are common malignant bone marrow lesions which may be difficult to distinguish because of similar imaging findings. The purpose of this study was to determine the value of adding diffusion-weighted imaging (DWI) to standard MR imaging to differentiate multiple myeloma from metastasis. Methods 25 patients with metastasis and 18 patients with multiple myeloma underwent 3T MR imaging with DWI (b = 0, 800 s/mm2) were enrolled. They all had pathologically confirmed bone lesions and were in a treatment naïve state. Two readers who were blind of final diagnosis measured the average ADC (ADCav) and minimum ADC (ADCmin) on the DWI. They then estimated the diagnosis, based on the standard MR imaging and measured ADC values. Another reader performed histogram analysis on the whole tumor volume and obtained mean ADC (ADCvol), standard deviation (SDvol), skewness, and kurtosis. Comparison of the obtained values from DWI was performed by the t-test or Mann-Whitney U test. The receiver operating characteristic (ROC) curve with areas under the curve (AUC) was used to obtain the cut off values and to evaluate the diagnostic performance of the two readers. Results ADCav, ADCmin, and ADCvol of multiple myeloma were significantly lower than those of metastasis: ADCav, 752 μm2/sec versus 1081 μm2/sec; ADCmin, 704 μm2/sec vs 835 μm2/sec; ADCvol 761 μm2/sec vs 1184 μm2/sec (p < .001). In histogram analysis, ADC values of multiple myeloma showed narrow distribution than metastasis: SDvol, 144 vs 257 (p < .001). Areas under the receiver operating characteristic curve was significantly higher with additive DWI than standard MR alone: 0.762 vs 0.953; 0.706 vs 0.950 (p < .05) for two readers. Conclusions This study suggested that the addition of axial DWI to standard MR imaging can be helpful to diagnose multiple myeloma from metastasis at 3T.


Introduction
Metastasis and multiple myeloma are common malignant disease involving bone marrow. Metastasis is most common, and appeared in various form of lytic or sclerotic bone lesions. The incidence of multiple myeloma is increasing in recent years [1]. It usually appears as lytic bone lesion in x-ray and CT images, and its MR imaging findings are classified into normal, focal, diffuse, and typical salt and pepper pattern [2][3][4][5]. They both present similar MR imaging manifestation and symptoms such as back pain, especially when involving the spine [6]. There were studies using standard MR imaging to differentiate these two diseases involving the spine, but there were some overlaps of MR imaging patterns between multiple myeloma and metastasis [7][8].
Diffusion-weighted imaging (DWI) has been a research topic in various field of musculoskeletal imaging. Previous studies proved that malignant marrow infiltration shows increased water content by destruction of trabecular bone and replacement of marrow fat, thus elevating apparent diffusion coefficient (ADC) value relative to normal background bone marrow [9][10][11]. In addition, DWI represented better signal to background ratio in detecting malignant bone lesions [12]. There have been various studies on malignant bone lesions including metastasis and multiple myeloma using DWI [10,11,[13][14][15][16][17]. However, so far comparative studies between the two groups are limited in our knowledge.
Thus, we hypothesized that adding axial DWI to standard MR imaging could help differentiate between multiple myeloma and metastasis. The purpose of our study was to retrospectively determine the value of adding axial DWI to standard MR imaging to differentiate multiple myeloma from metastasis at 3T.

Patient population
This retrospective study was approved by Seoul St. Mary's hospital institutional review board and informed consent was waived. From September 2010 to March 2014, the patients who underwent 3T musculoskeletal MR imaging including DWI were included (Fig 1). A total of 67 patients were diagnosed as multiple myeloma by bone marrow biopsy. Finally, 18 patients with treatment naive multiple myeloma were included in this study except 47 patients who underwent MR after the treatment, and 2 patients who had poor quality DWI for evaluation. Median age was 63 years old (range 41-77 years). The types of multiple myeloma were IgG κ (n = 10), light chain disease κ (n = 3), IgA κ (n = 2), IgA λ (n = 2) and light chain disease λ (n = 1).
There were a total of 35 patients who had confirmed pathology through biopsy or surgery for the bone lesion identified in musculoskeletal MR imaging. Six of these patients were excluded by performing MR after the treatment, and four were excluded due to poor quality of DWI. Finally, 25 patients with untreated metastatic bone lesions were included. Median age was 60 years old (range 29-81 years). The primary cancers were as follows; lung cancer (n = 7), hepatocellular carcinoma (n = 4), renal cell carcinoma (n = 3), stomach cancer (n = 2), prostate cancer (n = 2), ampulla of Vater cancer (n = 1), colon cancer (n = 1), adrenal gland cancer (n = 1), breast cancer (n = 1), ovary cancer (n = 1), melanoma (n = 1) and leiomyosarcoma of bone (n = 1). When these 25 metastatic lesions were classified according to radiographic nature, 14 were osteolytic, 1 was osteoblastic, 1 was mixed appearance, and 9 were not visible on x-ray or CT images.

MR imaging protocols
All 43 patients in this study underwent MR imaging with a 3T MR scanner (Magnetom Verio, Siemens Healthcare, Erlangen, Germany). The standard MR protocols included sagittal turbo spin echo (TSE) T1-and T2-weighted images axial TSE T1-and T2-weighted sequences. Other sequences were added according to the anatomical location. Chemical shift selective pulse sequences was used for fat suppression. In 40 patients, sagittal or axial fat-suppressed contrast enhanced T1-weighted sequences were also performed. A single-shot echo planar image (SSEPI) prototype pulse sequence was used for DWI in the axial plane before contrast injection. SSEPI MR parameters were as follows: field of view (FOV) 140-370 mm, matrix size 64 x 45-120 x 128, TR 4032-8300 msec, TE 52-85 msec, section thickness 3-9 mm (no gap), number of excitation 1-5, parallel factor 2. ADC map was obtained with two b values of 0 and 800 sec/mm 2 .

MR imaging analysis
The MR imaging analysis was performed by two radiologists (W.H.J., S.Y.L., with 17 and 5 years of experience in musculoskeletal radiology), who were blind to final diagnosis. They first estimated the diagnosis by assessing standard MR imaging alone. They decided the diagnostic confidence grade on estimated diagnosis with scale of 0 to 4, where 0 = definite metastasis, 1 = probable metastasis, 2 = possible multiple myeloma, 3 = probable multiple myeloma, and 4 = definite multiple myeloma. And then, two readers measured average value of ADC (ADCav) and minimum value of ADC (ADCmin) by manual drawn of regions of interest (ROIs) on the ADC maps in areas corresponding to hyperintense areas on diffusion weighted images with b = 800 sec/mm 2 on picture archiving and communication system (PACS). They excluded the regions of hemorrhage, necrosis or severe collapse, comparing standard MR imaging side by side. ADCav was obtained by as large as possible drawn ROIs, and ADCmin was chosen among the lowest ADC value of several small ROIs. They measured the ADC values of up to three lesions in each patients in a consensus. Total 72 lesions were included: multiple myeloma (n = 36) and metastasis (n = 36). Finally, they determined the final estimated diagnosis based on the combined information from standard MR imaging and from obtained ADC values, using the diagnostic confidence grade. This process was conducted six weeks later to avoid recall bias. All the images were analyzed in a random order.
Another musculoskeletal radiologist (J.Y.J. with 10 years of experience in musculoskeletal radiology) performed a histogram analysis of the whole tumor volume, using prototype software (OncoTreat; Simens Healthcare, Erlangen, Germany). He knew the location of the lesions which were evaluated by two other radiologists and drew the multiple volumetric ROIs in three dimensional reformatted images of the b = 800 s/mm 2 using semi-automated technique. ADC values for each voxel within the segmented volume were displayed as histogram. We obtained ADCvol, SDvol, skewness and kurtosis. ADCvol means the mean value of total ADC values from each voxel, and SDvol means the standard deviation of the obtained histogram of whole tumor volume. A total of 46 lesions (24 for multiple myeloma, 22 for metastasis) were evaluated in histogram analysis. The main contributors of measurement failure was small lesions, as the lesions were poorly delineated in reformatted images.

Statistical analysis
In quantitative analysis, comparison of the obtained values from DWI between multiple myeloma and metastasis groups was determined by the t test or Mann-Whitney U-test. Interobserver mesurement reliability of two readers for ADC measurement was determined using intraclass correlation coefficient (ICC). An ICC value of less than 0.40 was indicative of poor agreement; 0.40-0.75, fair to good agreement; and more than 0.75, excellent agreement. The receiver operating characteristic (ROC) curve with areas under the curve (AUC) was used to obtain the optimal cut off values for the values that showed significant differences between two groups. Then, the sensitivity, specificity, and accuracy of each cut off values were analyzed.
In qualitative analysis, the sensitivity, specificity, and accuracy of estimated diagnosis by two readers were evaluated, using clinical diagnosis as a gold standard. Diagnostic confidence grade determined by two readers from standard MR image alone (step 1) and from combined DWI with standard MR image (step 2) were evaluated by the ROC curve with AUC. Diagnostic confidence grade 0-1 was considered metastasis, and 2-4 was considered as multiple myeloma. Interobserver variability was evaluated using the kappa value to measure the degree of agreement in diagnosis at each steps. A kappa value of 0.2 or less was regarded as poor agreement, 0.21 to 0.40 as fair, 0.41 to 0.60 as moderate, 0.61 to 0.80 as good, and 0.81 or more as very good. The statistical analysis was performed using commercial software (SPSS, version 19, Chicago, III and MedCalc Software, version 11.3.0.0, Belgium), with p < .05 being considered as significant.
There was one misinterpreted case of metastatic prostate cancer for both reader 1 and reader 2 (Fig 5). With standard MR imaging, reader 1 made a wrong interpretation and reader 2 made a correct interpretation. However, since the ADC value obtained from DWI was quite low, this case was assessed as multiple myeloma in both reader 1 and 2.

Discussion
In this study, multiple myeloma showed significantly lower ADC values of ADCav and ADCmin for axial DWI and ADCvol of whole tumor. Moreover, multiple myeloma showed significantly lower SDvol than metastasis on ADC histogram analysis of whole tumor volume. Padhani et al [15] reported about ADC measurement of 34 myeloma lesions and 69 breast cancer lesions in 1.5 T MR imaging with two b values (b = 50, 900 sec/mm 2 ). In this study, ADC values of myeloma (875 ± 187 μm 2 /sec) were lower than those of the breast cancer (942 ± 154 μm 2 /sec), although the overlap of ADC values was considerable. Horger et al [16] reported DWI of 12 patients of multiple myeloma with 1.5 T MR imaging, using two b values (b = 50, 800 sec/mm 2 ) with mean ADC of 660 ± 150 μm 2 /sec at baseline. Giles et al [17] performed volume based ADC analysis in 18 patients with multiple myeloma with 1.5 T MR imaging, using two b values (b = 50, 800 sec/mm 2 ). Mean ADC for whole volume in each patients ranged from 659-971 μm 2 /sec (mean ± SD, 802 ± 89 μm 2 /sec).
In this study, multiple myeloma showed lower ADC values in DWI and narrow distribution in histogram analysis. We speculate that these results may be related to histopathological feature of multiple myeloma. Multiple myeloma is classified as a small cell round tumor of bone. This type of tumor usually represented as uniform, round or oval shaped cells with highly packed arrangement and high nuclear cytoplasmic ratio [18,19]. This compact arrangement of myeloma cell and associated different molecular pathway limit the free water movement in both extracellular and intracellular space. Therefore, even though the fatty bone marrow is replaced with myeloma cells, there is a certain level of limit on free water movement, so multiple myeloma showed lower ADC value and narrow distribution in histogram.   Lang et el. used DCE-MR to differentiate between multiple myeloma and metastasis [20], and multiple myeloma showed a more aggressive DCE kinetics of wash-out pattern. They interpreted this kinetics as rapid fill up and diffuse back of contrast material, because of the limited cellular space of multiple myeloma.
There was a misinterpreted case of metastatic prostate cancer for both readers. Since the ADC value obtained from DWI was quite low, this case was assessed as multiple myeloma in final decision. Previous studies [5,9] reported that prostate cancer with diffuse sclerosis showed false negative hypointensity on DWI and lower ADC value. Increased bone trabeculae in sclerotic lesions acts as barriers for free water movements. We used only the MR images for the analysis in this study, and because of this, we could not consider the sclerotic lesion and made a wrong interpretation. Thus, correlation with other imaging modalities such as radiographs or computed tomography scan would be mandatory for correct interpretation.
In this study, we showed that multiple myeloma had lower ADC values than that of metastasis in quantitative analysis. Also, we represented that these quantitative information were helpful to differentiate these two disease in qualitative analysis. Given the different diagnostic and therapeutic approaches of these two diseases, the differential diagnosis using DWI in addition to standard MR imaging may have a clinical significance.
This study has several limitations. First, it is a retrospective study with small study population. Second, metastasis group was composed of heterogeneous histopathology. Third, there may be a selection bias since only patients with pathologically proven bone lesions were included. Because of this, the proportion of multiple myeloma patients in the study population was higher than the general incidence, known as approximately 2.1 per 100,000 persons [1]. Fourth, the ROIs could have contained sclerosis and calcifications that affect ADCs. Finally, only two b values (b = 0 and 800 s/mm 2 ) were used for ADC calculation, because these are the most common set of values among the variable MR sequences in our institution. Inclusion of the b value of 0 s/mm 2 and its perfusion effect of ADC values cannot be ignored, and obtained ADC values can vary between the different DWI protocols.
In conclusion, multiple myeloma showed lower ADC values and standard deviation than metastases on DWI. The addition of axial DWI to a standard MR imaging may helpful to differentiate multiple myeloma from metastases at 3T.