Guidelines for Acquisition, Interpretation, and Reporting of Whole-Body MRI in Myeloma: Myeloma Response Assessment and Diagnosis System (MY-RADS)

Citation for pulished version (APA): Messiou, C., Hillengass, J., Delorme, S., Lecouvet, F. E., Moulopoulos, L. A., Collins, D. J., Blackledge, M. D., Abildgaard, N., Østergaard, B., Schlemmer, H-P., Landgren, O., Asmussen, J. T., Kaiser, M. F., & Padhani, A. (2019). Guidelines for Acquisition, Interpretation, and Reporting of Whole-Body MRI in Myeloma: Myeloma Response Assessment and Diagnosis System (MY-RADS). Radiology, 291(1), 5-13. https://doi.org/10.1148/radiol.2019181949

Whole-body MRI as currently used includes DW MRI sequences that are sensitive to cellular density and viability and are important for disease detection, monitoring, and therapy response assessments. Inclusion of DW MRI allows highly sensitive and quantitative evaluation of soft tissue and bone marrow, which is widely available and quick to perform and interpret. The relationship of apparent diffusion coefficient (ADC) (in square micrometers per second) values with cell density permits response assessments ahead of changes in lesion size, in addition to assessments of response heterogeneity (8,(30)(31)(32).
Whole-body MRI including DW MRI has become established as the most sensitive technique for bone marrow imaging (23)(24)(25) with additional benefits of speed, coverage, and quantification in comparison with traditional MRI, obviating intravenous injections and radiation exposure. Wide anatomic coverage is essential because 50% of lesions would be missed by imaging the spine alone (33). Avoidance of ionizing radiation is likely to become increasingly relevant as surveillance imaging of high-risk patients with monoclonal gammopathy of undetermined significance and of those with smoldering disease gains momentum. Whole-body MRI is a generally well-tolerated technique (24) that offers the additional benefits of assessing skeletal complications, such as spinal canal and/or nerve root compression, and is the most accurate method for differentiating benign from malignant vertebral compression fractures (34).
Although the International Myeloma Working Group has taken a pragmatic approach in recommending a range of possible imaging investigations (including low-dose CT, FDG PET/CT, and MRI) for patients with a new potential diagnosis of myeloma, the high sensitivity of whole-body MRI is explicitly acknowledged (6,35) and it is now recommended as first-line imaging for all patients with a suspected diagnosis of asymptomatic myeloma or solitary bone plasmacytoma. In the United Kingdom, whole-body MRI is recommended as firstline imaging for all patients with a suspected new diagnosis of myeloma (26). Recently, the British Society for Haematology additionally recommended the use of whole-body MRI for monitoring response of nonsecretory myeloma, oligosecretory myeloma (which can occur at relapse in patients with previous secretory disease), and for those patients with extramedullary disease (36). The uses of whole-body MRI in myeloma are summarized in Figure 1.
Acknowledging the increasingly important role of whole-body MRI for directing patient care in myeloma, a multidisciplinary, international, and expert panel of radiologists, medical physicists, and hematologists with experience in whole-body MRI in myeloma convened to discuss the performance standards, merits, and limitations of currently available imaging methods. The recently published Metastasis Reporting and Data System for Prostate Cancer, or MET-RADS, guidelines on the use of whole-body MRI for metastasis evaluations (37) were used as a model to formulate the performance standards for whole-body MRI use in the assessment of involvement by myeloma. The Myeloma Response Assessment and Diagnosis System (MY-RADS) imaging recommendations are designed to promote standardization and diminish variations in the acquisition, interpretation, and reporting of whole-body MRI in myeloma. The system also provides a means Essentials n Whole-body MRI is now incorporated into international standards for imaging patients with myeloma.
n Core clinical and comprehensive protocols for whole-body MRI include anatomic and functional imaging and can be completed in 30 minutes and 50 minutes, respectively.
n Standardized acquisition protocols and structured reporting will support clinical deployment, training, and research in imaging of myeloma.
the copresence of degenerative changes, benign lesions, and osteoporosis in the population at risk. The excellent soft-tissue contrast of MRI allows direct imaging of the bone marrow, providing high sensitivity (8)(9)(10)(11). Perhaps the clearest benefit of using MRI is the early detection of focal myeloma disease. It is well established that patients with unequivocal focal lesions at MRI have poorer outcomes. If disease can be detected early, and patients are stratified and treated according to clinical risk, then survival advantages are conferred (12)(13)(14)(15)(16)(17)(18)(19). Furthermore, most patients with apparently solitary plasmacytoma at skeletal surveys are upstaged with MRI, which profoundly influences treatment strategies (11,12). The number and sizes of focal lesions at MRI have also been shown to predict outcome (20,21). Fluorodeoxyglucose (FDG) PET/CT can also be used for diagnosis of focal bone lesions; however, MRI is more sensitive for diagnosis and contemporary MRI techniques have increased the advantage (22-25) (sensitivity of diffusion-weighted [DW] imaging has been reported as 77% compared with 47% for FDG PET/ CT) (23). Gene expression profiling has revealed lower expression of hexokinase 2, which is involved in the glycolysis pathway, in lesions that are negative at FDG PET and positive at MRI. The prognostic significance of lesions that are discrepant between MRI and FDG PET/CT is not yet known (25). However, whole-body MRI does result in the largest rise in quality-adjusted life years compared with CT or FDG PET/CT (26).
Although whole-body DW MRI has emerged as one of the most sensitive tools for imaging bone marrow with increased lesion conspicuity compared with conventional MRI sequences (27)(28)(29), some debate remains as to its specificity. There are few studies relating to myeloma. Although Lecouvet et al (28) presented data to suggest high specificity for detection of metastatic bone disease (98%-100%), a recent meta-analysis showed a pooled specificity of 86.1% (29). The paucity of myeloma-specific prospective studies and marked heterogeneity in reference standards make current judgments on specificity challenging, and biopsy of all lesions is not feasible. The approach offered by the International Myeloma Working Group of 3-6-month follow-up of equivocal solitary small lesions is a pragmatic solution (6). for response assessment and standardized protocols to facilitate data sharing for future developments.

Whole-Body MRI Data Acquisition and Analysis
The core clinical protocol for whole-body MRI when used alone is designed for myeloma detection in bone marrow but can also image extramedullary disease and should be completed within 30 minutes of imaging time (Table 1). More comprehensive assessments can be performed within 45-50 minutes of imaging time ( Table 1). Details of machine setup, sequence specification, quality assurance procedures, and quality control and radiographic aspects can be found in Table E3 The core protocol for whole-body MRI is adequate for the detection of disease at diagnosis. Comprehensive assessments are recommended for assessment of soft tissue, extramedullary disease, or for those patients in whom serial tumor response assessments (including clinical trials) are planned.

Clinical Information
The following information should ideally be available to radiologists at the time of reporting (and increasing use of electronic patient records increases the feasibility of this): time of initial diagnosis or suspected diagnosis, serum paraprotein levels and light chain levels and trephine status (if performed, including site), symptomatic sites including any clinical indication of cord or nerve root compression, first line or relapse state, current treatment, history of autograft or allograft transplant, details of previous radiation therapy or surgical interventions including vertebroplasty, granulocyte colony-stimulating factor or steroid administration, and minimal residual disease status (if performed).
Assessing Images from Whole-Body MRI Multisequence evaluations should be performed by using all images from DW MRI (low, intermediate [if obtained], and high b values and ADC maps) in conjunction with the anatomic and fat fraction images by using image linking and scrolling workstation facilities and coregistration tools as diagnostic aids.
Maximum intensity projections of high b-value images displayed by using the inverted gray scale are useful for global tumor volume assessments and for localizing regional tumor distribution. These images are able to display disease (eg, for referring clinicians in multidisciplinary meetings or in discussion with patients), but maximum intensity projection images should not be used alone for reading because apparent false-positive and false-negative disease assessments can occur (eg, due to T2 shine-through, respiratory motion signal dephasing, sparse disease pattern) (37,38). Serial volume maximum intensity projection image comparisons can be facilitated by using windowing techniques; for example, by maintaining window width between studies but adjusting the window level to a normal tissue such as muscle or subcutaneous fat for each time point.
The evaluation of source images obtained with DW MRI sequences at b values of 800-900 sec/mm 2 is based on comparing high b-value image intensity to adjacent muscle signal intensity, but assessments of ADC maps are numeric (unit, 310 23 mm 2 / sec or 310 26 mm 2 /sec [µm 2 /sec]).
The definitions for hypointense and hyperintense signal on images at b values of 800-900 sec/mm 2 remain subjective but can be gauged by using adjacent muscle as the reference background tissue (39)(40)(41).
Not all hyperintense bone lesions on high b-value images are malignant in nature. Causes for apparent high b-value focal bone lesions that are false-positive for cancer include bone marrow edema caused by fractures, osteoarthritis, infection, bone infarcts, vertebral hemangiomas, chondromas, cysts, focal fat-poor bone marrow, and artifacts around metal implants (42,43). Furthermore, successfully treated focal lesions may undergo liquid transformation and can stand out on DW images due to the T2 shinethrough phenomenon.
The presence of focal bone lesions can also be obscured when background bone marrow hyperplasia occurs as the result of anemia, as a rebound after high-dose therapy, or due to use of bone marrow growth factors. This is commonly encountered after allogenic stem cell transplantation. The detection of skeletal  Images demonstrate focal lesion (white arrows) greater than 5 mm, which returns higher signal intensity at b value of 900 sec/mm 2 than does muscle and background marrow, and which has ADC resembling soft tissue that is higher than background marrow but lower than fluid, as seen in bowel anteriorly (black arrow). E, F, Images show marrow that has diffusely higher signal intensity than does muscle and returns ADC that is similar to soft tissues. G, H, Images demonstrate focal lesion (arrows) that returns higher signal intensity than does muscle and background marrow at b value of 900 sec/mm 2 . However, background marrow is also higher signal intensity than muscle, suggesting background diffuse marrow hypercellularity that is confirmed on ADC map. I, J, Images show widespread heterogeneity with tiny nodular areas of altered diffusion signal (5 mm) with preserved normal marrow between them. This variegated or micronodular pattern is confirmed on, K, water-only Dixon and, L, sagittal T1 images. M, Image shows paramedullary soft-tissue disease that is in direct continuity with marrow disease, as demonstrated around left femur in M (black arrow). In contrast, N, extramedullary disease is soft-tissue disease that is not in continuity with marrow disease, as demonstrated in M and N (white arrows), which show ischiorectal fossa and pancreatic disease, respectively.
lesions at whole-body DW imaging may also be impaired in areas of body movement such as the ribs and occasionally in the sternum, but review of the maximum intensity projections of high b-value images can be helpful. Evaluation of skull vault lesions is best performed by evaluating the axial source high b-value images and corresponding Dixon fat fraction images. The visibility of skull base disease is generally impaired because of susceptibility effects and because of the adjacent high signal intensity of the brain on high b-value images.
Strategies for mitigating false-positive results due to DW imaging hyperintensities include direct correlations with morphologic appearances, including T1-weighted spin-echo and Dixon images (44,45) and ADC values (ADC values of normal bone marrow are generally below 600-700 µm 2 /sec and viable tumor lies between 700-1400 µm 2 /sec (46,47,48). Tumor ADC values greater than or equal to 1400 µm 2 /sec are usually observed in treated or necrotic disease. However, myeloma is a disease where there is intermixing of myeloma cells, fat, and myeloid cells within the marrow space, the relative proportions of which can alter ADC values, which must also be considered (30,49). Every suspicious lesion on high b-value images should be evaluated with other sequences, particularly Dixon fat fraction images by using coregistration tools.
The authors also recognize there are limitations of the ADC cutoff values referred to previously, which are partly related to the fact that ADC values depend on the choice of b values of DW images used for calculations (hence, the constraints on the recommended choices of b values). The ADC values may also depend on the diffusion time achievable with diffusion sequences (which is dependent on sequence waveforms and imager specifications). Aside from imaging parameters, ADC quantification is influenced by additional factors, which are patient related caused by susceptibility effects (eg, implants, air-tissue interfaces) or motion and technique related caused by the specifications of 9 DW imaging. Otherwise, the ADC values can be erroneous, reflecting the distribution of noise on the images.
Preferably, water signal should be detectable for the assessed region on all b-value images. However, the absence of tissue signal intensity on very high (b value of 800-900 sec/ mm 2 ) b-value images does not in and of itself invalidate a tissue from ADC measurements. Where applicable, if signal intensity is detectable on low and intermediate b-value images, then ADC measurements should be performed. ADC is generally considered to be a parameter with good reproducibility, and the coefficient of variation in myeloma-involved bone marrow has been reported as low as 2.8% (31).
Diffuse disease.-For patients with a suspected new diagnosis of myeloma, diffuse disease can be suspected from diffuse decreased signal on T1 fast spin-echo or Dixon in-phase and fatonly images, and diffuse increased signal throughout the marrow relative to normal muscle on high b-value images (Fig 2) (48). Suspicion for diffuse infiltration can be documented but must be confirmed with trephine biopsy, and imaging must be performed to state whether posterior iliac crest sampling is likely to be representative. An ADC measurement using a region of interest greater than 1 cm 2 of representative involved bone marrow can be measured particularly where diffuse infiltration is suspected (50). the MRI unit, including magnetic field strength, gradients, and coils (39). Where there are deviations from the recommended b values due to machine, software, or technical factors, then institutions can determine their own musclenormalized high b-value signal intensity and ADC cutoff values for normal marrow and for untreated bone marrow lesions as described by Padhani et al (41).
Size measurements.-The International Myeloma Working Group stipulates 5 mm as the threshold for defining an unequivocal focal active lesion (Fig 2). Therefore, lesions less than or equal to 5 mm should be documented for surveillance but precise measurements not given due to limitations on image resolution. Lesions greater than 5 mm but less than 1.0 cm with appropriate signal characteristics should be documented as focal active disease, but once again, precise measurements not given.
In rare instances when there are unequivocal malignant lesions less than 1.0 cm and entry to clinical trials mandates the presence of measurable disease, a relaxation of the threshold of greater than or equal to 1.0 cm may be applied in the knowledge of the caveats on image resolution.
Obtaining ADC values.-ADC measurements should only be obtained from lesions or areas when water is detectable at 5, highly likely to be progressing (Table 3). More detailed case report forms for research are included in Appendix E1 (online).

Clinical Trial Assessments
Most clinical reports will be performed as outlined in the previous section. However, academic centers wishing to collect assessments of disease for clinical trials may require more sophisticated scoring systems. Detailed instructions for a proposed clinical trials scoring system and an exemplar case study can be found in Table E2 and Figure E1 (Appendix E1 [online]).
Knowing that quantitative assessments such as ADC and fat fraction measurements can change over time (for example, responding myeloma may show an early increase in ADC values at 4-6 weeks due to cell death, followed by a decrease in ADC with return of normal fatty marrow [51]), it is difficult to be prescriptive regarding the frequency of imaging follow-up. Therefore, for correlation, we recommend that imaging should coincide with clinical routines where serum and marrow assessments are performed.

Conclusion
The Myeloma Response Assessment and Diagnosis System (MY-RADS) system provides comprehensive characterization of the myeloma state, both at diagnosis and the start of treatment, as well as over time as the disease evolves during Although useful data are emerging to differentiate normal age-matched bone marrow from diffuse disease by using ADC (47,48,49), whole-marrow quantitative measures of ADC are not yet practical measures and therefore not part of the MY-RADS standard. However, marrow ADC values above 600-700 µm 2 / sec in a nontreated and newly diagnosed patient with multiple myeloma could be used to increase confidence for the diagnosis of diffuse marrow involvement (8). For patients who have been treated, there is increased potential for false-positive appearances of diffuse infiltration due to rebound hypercellularity related to treatment effects or granulocyte colony-stimulating factor (8). Therefore, when diffuse disease is suspected, administration of granulocyte colony-stimulating factor must be excluded and if progression is suspected based on suspicion for diffuse infiltration alone, then this must be confirmed with serum biochemistry and/or marrow sampling.

Structured Reporting
Structured clinical reporting (Table 2) should be performed for each examination. For response assessments, we recommend that response assessment category, or RAC, be assigned according to anatomic regions. For each region, RAC can be assigned by using the criteria given in Table 3. The RAC should use a five-point scale as follows: 1, highly likely to be responding; 2, likely to be responding; 3, stable; 4, likely to be progressing; State level of concern regarding incidental findings Note.-RAC = response assessment category. * See also Figure 2. † See also Table 3. therapy and follow-up. MY-RADS recommendations are likely to fulfill the need to promote standardization and diminish variations in the acquisition, interpretation, and reporting of whole-body MRI and allow better response assessments. This system is designed for guiding patient care but has potential for incorporation into clinical trials when lesion measurements become more important. MY-RADS allows the categorization of patients with specific patterns of disease for clinical trial stratification. MY-RADS requires validation within clinical trials, including assessments of reproducibility. We suggest that MY-RADS be evaluated in studies that assess the effects of life-prolonging myeloma treatments, including novel treatment strategies such as immunotherapy. In these studies, MY-RADS assessments of the depth and heterogeneity of response should be compared with established myeloma response criteria. Given the high rates of complete response seen in patients with multiple myeloma with new treatment approaches, new response categories need to be defined that can identify responses that are deeper than those conventionally defined as complete response.
Correlations with quality-of-life measures, rates of skeletal events, and rates of progression-free survival are also needed. The latter are prerequisites for the introduction of whole-body MRI into longer term follow-up registry studies that prospectively collect appropriate meta-data, which would allow objective assessments of whether whole-body MRI is effective in supporting patient care and drug development. It is anticipated that, as evidence accrues from clinical trials, more specific recommendations and/or algorithms incorporating MY-RADS will emerge. Thus, we recommend that Myeloma Response Assessment and Diagnosis System is now evaluated in clinical trials to assess its impact on the clinical practice of myeloma.
Disclosures of Conflicts of Interest: C.M. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: received payment for lectures including service on speakers bureaus from Janssen and Takeda; received payment for development of educational presentations from Amgen. Other relationships: disclosed no relevant relationships. J.H. Activities related to the present article: institution received grant from Celgene. Activities not related to the present article: is a consultant for Amgen and Celgene; has grants/ grants pending with Celgene and Sanofi; received payment for lectures including service on speakers bureaus and payment for travel/accommodations/meeting expenses unrelated to activities listed from Amgen, BMS, Celgene, Janssen, Novartis, and Takeda Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: is a consultant for and received payment for lectures including service on speakers bureaus from Amgen, Celgene, Janssen, and Takeda; has grants/grants pending with Celgene; received payment for travel/accommodations/meeting expenses unrelated to activities listed from Takeda. Other relationships: disclosed no relevant relationships. A.P. disclosed no relevant relationships.