An international expert opinion statement on the utility of PET/MR for imaging of skeletal metastases

MR is an important imaging modality for evaluating musculoskeletal malignancies owing to its high soft tissue contrast and its ability to acquire multiparametric information. PET provides quantitative molecular and physiologic information and is a critical tool in the diagnosis and staging of several malignancies. PET/MR, which can take advantage of its constituent modalities, is uniquely suited for evaluating skeletal metastases. We reviewed the current evidence of PET/MR in assessing for skeletal metastases and provided recommendations for its use. We searched for the peer reviewed literature related to the usage of PET/MR in the settings of osseous metastases. In addition, expert opinions, practices, and protocols of major research institutions performing research on PET/MR of skeletal metastases were considered. Peer-reviewed published literature was included. Nuclear medicine and radiology experts, including those from 13 major PET/MR centers, shared the gained expertise on PET/MR use for evaluating skeletal metastases and contributed to a consensus expert opinion statement. [18F]-FDG and non [18F]-FDG PET/MR may provide key advantages over PET/CT in the evaluation for osseous metastases in several primary malignancies. PET/MR should be considered for staging of malignancies where there is a high likelihood of osseous metastatic disease based on the characteristics of the primary malignancy, hight clinical suspicious and in case, where the presence of osseous metastases will have an impact on patient management. Appropriate choice of tumor-specific radiopharmaceuticals, as well as stringent adherence to PET and MR protocols, should be employed.


Introduction
Positron emission tomography (PET) is a quantitative diagnostic imaging modality that investigates molecular processes in vivo. However, PET provides limited anatomic information. Magnetic resonance imaging (MRI) is a modality with high spatial resolution and soft tissue contrast that allows for the identification and characterization of bone and soft tissue abnormalities. Integrated PET/MR is a hybrid technology that allows the simultaneous acquisition of both metabolic and anatomic information. There are currently no clinical guidelines regarding the role of PET/MR in the evaluation of skeletal metastases. We will discuss technical and clinical considerations relevant to the usage of PET/MR in the evalua t i o n o f s k e l e t a l m e t a s t a s e s , b o t h w i t h 1 8 Ffluorodeoxyglucose ( 18 F-FDG) and with non 18 F-FDG radiopharmaceuticals. We will also review current evidence and provide recommendations for the use of PET/MR in patients with suspected skeletal metastases.

Technical considerations
Simultaneous acquisition of PET and MR data using a single device required overcoming several technical challenges not present in combined PET/CT imaging. One such challenge was adapting PET detectors to operate within a magnetic field. The development of solid-state photodetectors (i.e., avalanche photodiodes, silicone photomultipliers), which are unaffected by the magnetic field, allowed for simultaneous acquisition of PET data and, in the case of silicon photomultipliers, provides time-of-flight information [1]. Three equipment manufacturers are currently commercializing fully integrated PET/ MR scanners for clinical use.
On the methodological side, the major obstacle slowing the widespread clinical adoption of PET/MR has been the challenge of generating accurate attenuation correction (AC) maps [2,3]. A correction must be applied to PET data to account for attenuation of the emitted photons prior to reaching the PET detectors [4,5]. AC issues are particularly important for lesions in or adjacent to cortical bone for which incorrect AC might lead to erroneous PET quantification. Unlike PET/CT, where the CT images can be used to measure the linear photon attenuation coefficients of tissues, albeit at a lower energy, PET/MR generally relies on MR-derived image parameters to calculate an attenuation map. Bone heavily attenuates emitted photons and, if not appropriately accounted for, can result in considerable inaccuracy in the quantitative analysis of PET radiotracer distribution [6][7][8].
Early MR-based AC techniques used conventional T1weighted or Dixon sequences to segment classes of tissues and create attenuation maps [9][10][11]. Cortical bone, despite having a high linear attenuation coefficient, has a low signal intensity on T1-weighted images and was misclassified as air using standard MR-based AC methods. Although this resulted in substantial bias in anatomic areas surrounded by bone, such as the head and pelvis, the reduction in standardized uptake values (SUVs) compared to PET/CT was shown to have minimal impact on the clinical evaluation of malignant bone lesions [12][13][14]. The impact of remaining bone-tissue misclassification has also been shown to be reduced when incorporating the time-of-flight information [15]. Application of model-based approaches in which bony structures are incorporated into the Dixon-based AC maps was shown to substantially reduce the bias caused by bony structures [16]. Additionally, techniques using specialized MR sequences (i.e., ultrashort or zero echo time sequences) have been implemented to generate even more accurate AC maps, particularly in skull attenuation in brain PET/MR [17][18][19][20][21][22]. Recently, deep learning-based AC techniques have emerged to exploit information acquired either as part of diagnostic MR images or from separately acquired AC MR sequences to construct attenuation maps [23][24][25][26].

Radiation dose
In current practice, radiation dose from PET/MR can be up to 80% lower than the vast majority of PET/CT [27]. This is primarily due to the elimination of radiation exposure from CT used for AC and, if required, from a separately acquired diagnostic CT. Radiation exposure with PET/MR occurs only from the injected radiopharmaceutical [28][29][30][31]. Radiation exposure can be further reduced by decreasing radiopharmaceutical activity given for PET/MR. This can be achieved by taking advantage of features of current PET/MR scanners, including the longer axial field of view, and reduced diameter compared to PET/CT devices, as well as the longer acquisition times needed for acquiring multiparametric MR information. The end result is a reduction of injected radioactivity by up to 50-65% [32,33].
The newest generation of PET/CT scanners, equipped with higher performance detectors (e.g., temporal resolution approaching 200 ps), allow for additional reductions in radiation dose. This class of detectors will likely be introduced in the next generations of PET/MR systems, with further expected decrease in radiation dose and acquisition time [34].

MR component of PET/MR and protocol considerations
A fundamental advantage of PET/MR over PET/CT is the superior diagnostic performance of MR compared to CT in the identification and, in many cases, the characterization of osseous lesions [35]. Detection of metastases by CT requires destruction of cortical or trabecular bone, adjacent sclerotic changes, or identification of soft tissue attenuation within the normal fat-attenuation marrow. In contrast, skeletal metastases on MR can be characterized by signal abnormality of the bone marrow on T1, STIR, and diffusion-weighted sequences [36][37][38][39][40]. Numerous studies have shown that MR is more sensitive than CT alone in the detection of focal marrow replacing lesions and that the higher soft tissue contrast provided by MR allows for better delineation of extra-osseous tumor spread and spinal cord compression (Fig. 1) [41][42][43][44][45][46][47]. Furthermore, MR can be a useful tool for distinguishing benign osteoporotic vertebral body fractures from pathologic fractures [48][49][50].
In addition to the superior performance of MR over CT in evaluating bone abnormalities, PET/MR benefits from the simultaneous acquisition of diagnostic anatomic images and PET data. This allows for more accurate co-registration and fusion of the MR and PET images. The combination of highquality anatomic imaging and improved co-registration may facilitate lesion detection and characterization. It may help guide percutaneous or surgical intervention. Although most PET/MR protocols involve long acquisition times, which is sub-optimal for clinical situations where rapid scanning is needed, they can be streamlined to ensure the whole body is acquired within 20-25 min, depending on body habitus [27].
Based on the clinical experience of the co-authors of this study and on the published literature, PET/MR acquisition protocols should be tailored to the selected radiopharmaceutical with a focus on its effective half-life. For example, while most radiotracers do not require rapid MR image acquisition and can permit scan times of up to 1 h, the short biologic halflife of 18 F-fluciclovine necessitates MR image acquisition in under 20 min from tracer injection. For most studies of adult patients, image acquisition should start from the mid-thighs and end at the vertex of the skull to ensure the pelvis is imaged before the bladder fills, in case of urinary excreted radiopharmaceuticals. Radiopharmaceuticals within the distended bladder can result in halo artifacts due to overestimation of scatter contribution in PET reconstruction, which can obscure abnormal uptake in regional structures. PET/MR protocols of children and teenagers typically require head-to-toe acquisition [27]. . Some authors, despite the longer acquisition time, prefer whole-body coronal and spine sagittal T1-weighted fast spin echo sequences instead of axial or/and coronal T1-weighted dual point Dixon gradient echo sequences. Gadolinium administration may be advantageous but not strictly necessary for evaluation of skeletal lesions on PET/MR [51]. 18F-FDG-PET images can be color-coded to reflect radiotracer uptake and superimposed on high-resolution gradient echo images. These fused images can be reconstructed in the coronal and sagittal planes. This protocol, depending on available PET/MR scanner hardware, takes approximately 20-25 min. If necessary, the primary tumor and any other areas of interest might be further interrogated with additional dedicated protocols as clinically appropriate [27].
For 18 F-fluciclovine-PET/MR, given its short relative halflife, we recommend acquisition of coronal T1-weighted highresolution dual point Dixon gradient echo sequences or T1weighted FSE simultaneously with PET from mid-thighs to vertex, and ensure completion within 20 min from injection. F-FDG-PET, c axial fused PET/ MR, and d same-day contrastenhanced CT images of the humerus. These images demonstrate a metastatic lesion (arrow) in the right humeral head which enhances after gadolinium administration and demonstrates associated increased 18 F-FDG uptake. No corresponding abnormality is seen on same day CT image, d Subsequently, the following sequences can be acquired: coronal STIR sequences and/or axial simultaneous multislice diffusion-weighted images with suggested b-values 50-400-800 s/mm 2 and axial and coronal contrast-enhanced T1weighted fat-suppressed gradient echo images. Coronal T1weighted high-resolution dual point Dixon gradient echo sequences can be secondarily reconstructed in the sagittal plane for evaluation of the spine. Additional imaging parameters may vary, given technical differences in currently available PET/MR scanner hardware.
Additional pre-contrast coronal STIR images can be obtained as deemed necessary.
In the authors' experience, single-shot fast spin echo T2weighted images are less useful in the evaluation of bony metastases. Although acquisition time is short and the sequence is relatively resistant to respiratory and other types of motion artifact, the contrast between tumor lesions and the bone marrow is less conspicuous, as compared with DWI, STIR, and Gd-enhanced T1-weighted scans.
Suggested MRI sequences with descriptions of their roles in assessment of osseous metastases are described in Table 1.

F-FDG PET/CT vs F-FDG-PET/MR
18 F-FDG PET/CT is currently used to image a variety of cancers, including lymphoma, small-cell and non-small-cell lung cancers, head and neck squamous cell cancer, melanoma, colorectal cancer, breast cancer, esophageal cancer, gastric cancer, pancreatic adenocarcinoma, cervical cancer as well as bone and soft tissue sarcomas. In general, these tumors show avid 18 F-FDG uptake and can therefore be easily detected. Malignancies such as prostate cancer, hepatocellular carcinoma, renal cell carcinoma, and well-differentiated gastroenteropancreatic neuroendocrine tumors have generally low 18 F-FDG uptake, underscoring the limited utility of 18 F-FDG PET/CT in assessment of these diseases.
Although PET/MR has been approved for clinical use for the last 10 years, rigorous comparison with PET/CT is still underway. Studies that have compared 18 F-FDG PET images from PET/CT to PET/MR in a variety of malignancies have generally shown similar diagnostic performance despite differences in quantitative and semi-quantitative assessment of 18 F-FDG uptake [58][59][60][61][62]. A single study of two sequential PET/MR exams following a PET/CT exam showed comparable diagnostic performance of PET from PET/MR versus PET/CT and acceptable reproducibility between sequential PET/MR exams [63].
Currently, few dedicated studies have explored the performance of PET/MR compared to PET/CT for assessment of malignancies involving the musculoskeletal system. Available studies focus primarily on multiple myeloma, extra-nodal osseous involvement of lymphoma, and osseous metastases from breast and lung cancer. PET/MR offers higher diagnostic confidence and improved conspicuity than PET/CT for detection of bone lesions, especially in the case of early osseous metastatic disease (Fig. 2) [64]. • Identification of replacement of T1 hyperintense fatty marrow signal with T1 hypointense metastases [40]. • Differentiation of metastases from benign processes.
• Signal drop out on out of phase images can indicate intralesional fat, distinguishing focal red marrow from marrow replacing lesions [52].
• Distinguishing benign osteoporotic compression fractures from pathologic fractures of the spine [54]. • Evaluation of ADC maps can be helpful in distinguishing red marrow from metastases and may have a role in assessment of treatment response in some malignancies [38,[55][56][57].
High-resolution post-contrast T1-weighted fat-suppressed GRE Axial (optional in sagittal or coronal planes) • Identification of enhancing marrow replacing lesions as well as associated extra-osseous soft tissue extension.
Short tau inversion recovery (STIR) (optional, since time intensive)

Coronal
• Identification of hyperintense marrow replacing lesions in the background of low marrow signal due to fat suppression [40]. • Identification of extra-osseous tumor extension.
• Improved identification of pathologic fracture and assessment of fracture acuity based on the presence of surrounding bone marrow edema. • Identification of degenerative disc disease and osteoarthritis. 18

Breast cancer
A recent meta-analysis that explored the overall performance of PET/MR for whole-body staging of breast cancer reported high whole-body patient-based pooled sensitivity (0.98) and specificity (0.87) and high lesion-based sensitivity (0.91) and specificity (0.95) [65]. In a study focused specifically on the performance of PET/MR in evaluating osseous metastases in breast cancer, 109 patients underwent same-day contrast-enhanced PET/CT and PET/MR, demonstrating a sensitivity of PET/CT and PET/MR of 0.85 and 0.96, respectively. More importantly, PET/MR was positive for osseous metastases in 12% of the patients who did not have osseous metastases detected by PET/CT [66].
A study of whole-body staging by PET/MR and PET/CT showed similar accuracy when evaluated using a patientbased analysis [67]. However, PET/MR was superior to PET/CT in identifying bone metastases (sensitivity of 0.92 vs 0.69, respectively). Another study of 50 patients showed improved sensitivity of 18 F-FDG PET/MR in the detection of bone and liver metastases versus PET/CT [30]. The superior performance of PET/MR in detecting osseous metastatic disease is in large part due to the conspicuity of marrow replacing lesions on the MR portion of the study. Such a finding, in conjunction with even faintly increased 18 F-FDG update on the PET images, allows for identification of metastases. On PET/CT, if an osseous metastasis is CT occult, faint 18 F-FDG uptake without a corresponding CT abnormality might be misinterpreted as normal marrow heterogeneity. The addition of MR can help improve detection of focal osseous lesions ( Fig. 3) and may also help in distinguishing metastases from 18 F-FDG-avid benign bone lesions [68].
Lung cancer 18 F-FDG PET/CT is an important tool in the initial staging and restaging of both small-cell (SCLC) and non-small-cell lung cancer (NSCLC). This is reflected in NCCN guidelines [69,70]. There are very few studies assessing the performance 18 F-FDG PET/MR in the evaluation of NSCLC, none of which have specifically investigated differences in PET/MR versus PET/CT in the detection of bone metastases [71]. Current literature, however, points to a comparable diagnostic accuracy for identification of M1 disease overall, regardless of the site of origin, with potential incremental value of MR over CT for detection of brain, bone, and liver metastases. [72][73][74][75].

18
F-FDG PET/CT plays a critical role in staging and restaging of Hodgkin lymphoma and of certain subtypes of non-Hodgkin lymphoma [76,77]. 18 F-FDG PET/MR has similar accuracy to PET/CT in the evaluation of lymphoma, particularly in identification of nodal disease [29,78]. In selected cases, current National Comprehensive Cancer Network (NCCN) guidelines suggest the usage of MR and PET/MR for the initial workup of Hodgkin lymphoma [79].
Despite the paucity of literature specifically addressing the evaluation of lymphomatous involvement of bone marrow by PET/MR, there is substantial literature showing the superior sensitivity and specificity of MR compared to CT for the detection of marrow replacing lesions [43,47]. CT may miss Heterogeneous enhancement in the T7 vertebral body (arrow), in b, could reflect marrow replacement or normal erythropoietic bone marrow. However, focally increased 18 F-FDG-PET accumulation in the T7 vertebral body, consistent with osseous metastasis, is shown in c and d. Avid 18 F-FDG uptake helped increase the conspicuity of the metastasis in a background of hyperplastic red marrow sites of marrow involvement that can be detected by MR and considered suspicious if there is corresponding abnormal 18 F-FDG uptake. This was demonstrated in a study evaluating 28 consecutive lymphoma patients, where, although PET/MR and PET/CT were concordant in 96.4% of patients and demonstrated a similar sensitivity, lymphomatous bony involvement of bone was detected in one patient by PET/MR where it was missed by PET/CT. This patient was therefore correctly upstaged to stage IV by PET/MR, with potentially important treatment implications [80]. Based on our unpublished personal experience, integrated PET/MR can further improve the sensitivity of stand-alone 18 F-FDG PET and of standalone MR, detecting lesions which might be apparent on the MR or PET alone. The superior performance of the MR component of PET/MR over PET/CT may also help distinguish extra-nodal involvement of cortical bone from bone marrow involvement by lymphoma, a distinction that has important staging and treatment implications [81]. PET/MR may provide additional prognostic implications, given that chemotherapy first decreases glucose metabolism and then increases hydrogen proton diffusion [82]. Future investigation is needed to determine if tumors with concordant response on 18 F-FDG PET and DWI on interim scans show better outcomes compared to tumors with early 18 F-FDG PET response and delayed DWI response.

Non 18 F-FDG PET/MR
18 F-Sodium fluoride 18 F-Sodium fluoride ( 18 F-NaF) PET has proven clinically useful in the identification of both benign and malignant conditions. 18 F-NaF localizes to areas of elevated osseous blood flow and bone formation [83]. NCCN guidelines recommend bone scintigraphy or 18 F-NaF-PET/CT for evaluation of specific bone lesions in locally advanced or metastatic breast cancer [84]. Radiotracer uptake by osteoblasts is nonspecific and may be elevated in benign conditions such as degenerative changes, osteonecrosis, fractures, inflammatory conditions, and benign bone neoplasms [85,86]. This underscores the importance of careful interpretation of the radiotracer uptake pattern and correlation with anatomic imaging, where available. Although 18 F-NaF-PET/MR has been used in rat models, there are to date no published studies using 18 F-NaF-PET/MR to assess for bony metastases in humans [87]. In unpublished work, the authors have performed 18 F-NaF-PET/MR on humans showing radiotracer uptake corresponding to marrow signal abnormalities on MR at sites of bone metastases (Fig. 4). 18 F-Fluciclovine 18 F-Fluorocyclobutane-1-carboxylic acid ( 18 F-FACBC or 18 F-fluciclovine) is an amino acid analogue that localizes to areas of increased amino acid transport. Normal distribution of 18 F-fluciclovine uptake includes the pancreas, liver, salivary glands, pituitary glands, gastrointestinal tract, bone marrow, and muscle [88,89]. Prostate cancer and glial brain lesions show uptake of 18 F-fluciclovine [90,91].
Given that prostate cancer has a predilection for osseous metastases, it is conceivable that 18 F-fluciclovine PET/MR would be helpful for evaluation of metastatic prostate cancer. However, most research has focused on identification of lymph node or soft tissue disease involvement [18,92]. Evaluation of bone metastases using 18 F-fluciclovine has largely been performed with PET/CT [93]. Densely sclerotic bony lesions might not take up enough 18 F-fluclicovine to be easily detected by PET. Moreover, the background marrow tracer uptake may further obscure small subtle areas of increased tracer accumulation. Only one study to date has evaluated the diagnostic performance of 18 F-fluciclovine PET/MR for the evaluation of bone metastases from prostate cancer, demonstrating detection rates of 0.68 for stand-alone PET and of 1.00 for PET/MR. In this study, the lesions most commonly missed on stand-alone PET belonged to the densely sclerotic category. These were identified readily on T1weighted MR sequences [94]. 68 Ga and 18 F -PSMA Prostate-specific membrane antigen (PSMA) radiopharmaceuticals labeled with 68 Ga and 18 F target cell surface receptors and have shown clinical utility in investigating prostate cancer. Since 68 Ga-PSMA-11 was first developed, most of the available studies have employed this radiopharmaceutical. The normal distribution of 68 Ga-PSMA-11 in the body is in the lacrimal and salivary glands, liver, spleen, kidneys, gastrointestinal tract, and neural ganglia [88]. Other PSMA based imaging agents labeled with 68 Ga have a very similar biodistribution. A critical limitation of PSMA imaging is that hypointense lesion in the right peripheral zone of the prostate at the level of the mid-gland (arrow) consistent with a primary prostate adenocarcinoma. A T2w hypointense, enhancing lesion in the left medial acetabulum (arrowhead) with associated PSMA uptake is consistent with a prostate adenocarcinoma metastasis renal excretion results in accumulation of radiotracer in the bladder and renal collecting system. The resultant halo phenomenon can obscure abnormal uptake in the prostate or prostatectomy bed as well as regional metastatic disease in the pelvis [95]. Strategies for reducing the halo phenomenon include late imaging with or without administration of diuretics or early imaging prior to substantial accumulation of radiotracer in the urinary tract [96][97][98][99]. Development of PSMA imaging agents with hepatobiliary excretion, such as 18 F-PSMA-1007, has shown progress in mitigating this problem [100].
Much like 18 F-fluciclovine, most research on PSMA labeled with 68 Ga has focused on PET/CT rather than PET/ MR [101]. In limited studies of 68 Ga-PSMA PET/MR, bone metastases were detected following definitive treatment with early biochemical recurrence even at low serum prostatespecific antigen levels (PSA < 0.2 ng/mL) [102]. One recent study demonstrated similar performance of 68 Ga-PSMA PET/ MR and PET/CT for the evaluation of bone metastases from prostate cancer. However, the MR of the PET/MR was able to detect two bone metastases which were not visible on CT and may have been otherwise missed [103]. Furthermore, in another study, four lesions, including one bone lesion, were indeterminate on 68 Ga-PSMA PET/CT but were definitively characterized as metastases on 68 Ga-PSMA PET/MR [104]. Given that benign osseous lesions such as fibro-osseous lesions can demonstrate radiotracer uptake and therefore be misinterpreted as metastases, it is important to assess the characteristics of osseous lesions on anatomic imaging [105]. This is a clear point of strength of PET/MR versus PET/CT. Another recent study comparing the performance of 68 Ga-PSMA PET/CT and PET/MR showed agreement on sites of pelvic and distant metastatic disease, including osseous metastatic disease. However, 68 Ga-PSMA PET/MR was superior in detection of localized disease such as extracapsular tumor extension and seminal vesicle involvement, primarily owing to the high soft tissue contrast and multiparametric nature of MR [106]. Quantitative metrics incorporating 68 Ga-PSMA uptake and multiple MR imaging parameters have also been shown to distinguish normal prostatic tissue from clinically significant prostate cancer [107]. Therefore, given the benefits of MR in detection and characterization of bone lesions as well as the superior performance in evaluation of prostatic and local extra-prostatic disease, it is expected that PET/MR might be advantageous over PET/CT (Figs. 5, 6, 7). 68 Ga-DOTATATE, 68 Ga-DOTATOC, and 68 Ga-DOTANOC 68 Ga-DOTATATE is one of several radiopharmaceuticals targeting the somatostatin receptor (SSTRA) for PET imaging. 68 Ga-DOTATATE, unlike other somatostatin analogs 68 Ga-DOTANOC and 68 Ga-DOTATOC, binds with highest affinity to the surface somatostatin receptor subtype 2, which tends to be overexpressed in well-differentiated neuroendocrine tumors (Fig. 8) [108]. 68 Ga-DOTATATE PET/CT has proven to have a significant impact on the management of neuroendocrine tumor patients when compared to conventional anatomic and nuclear   Ga-DOTATATE PET/CT compared to conventional nuclear medicine studies may result in a change in management in up to 75% of patients [109]. A meta-analysis showed a sensitivity and specificity of 68 Ga-DOTATATE PET/CT for neuroendocrine tumors of 0.93 and 0.96, respectively [110]. However, poorly differentiated neuroendocrine tumors (WHO G3) typically show decreased 68 Ga-DOTATATE uptake and greater 18 F-FDG uptake, likely in part owing to decreased expression of somatostatin surface receptors [111]. As a result, greater 18 F-FDG uptake in neuroendocrine tumors is associated with a poorer prognosis mostly due to a greater tumor heterogeneity and presence of hepatic metastatic disease at presentation [112]. Beyond gastrointestinal neuroendocrine tumors, somatostatin analogs have shown early promise with pheochromocytomas, paragangliomas, neuroblastomas, and meningiomas [113][114][115][116][117][118].
Osteoblasts express subtype 2 somatostatin surface receptor [119]. Although osteoblastic activity is associated with sclerotic osseous lesions including metastases, it can be seen in response to a wide variety of nonmalignant processes in bone, including osteoarthritis and fractures. 68 Ga-DOTATATE uptake can also be seen with benign osseous lesions such as fibrous dysplasia and hemangiomas [120,121]. MR is more accurate in distinguishing these benign processes from malignant-appearing marrow replacing lesions than CT. Although there are currently no studies comparing the performance of 68 Ga-DOTATATE PET/CT and PET/ MR for skeletal or other metastatic disease, we believe that the benefits of superior anatomic imaging gained by simultaneous MR will better help assess the burden of skeletal metastatic disease.
PET/MR radiopharmaceuticals with indications, interpretation criteria, and limitations/pitfalls for assessment for skeletal metastases are described in Table 2.

F-FDG avid malignancies
For staging of 18 F-FDG avid malignancies including lymphoma, small-cell and non-small-cell lung cancers, head and neck squamous cell cancer, melanoma, colorectal cancer, breast cancer, esophageal cancer, gastric cancer, pancreatic adenocarcinoma, and sarcoma, consider 18   MR may permit for more definitive characterization of osseous lesions.

Well-differentiated neuroendocrine tumors
For staging of well-differentiated neuroendocrine tumors, consider 68 Ga-DOTATATE or 68 Ga-DOTANOC or 68 Ga-DOTATOC PET/MR. PET/MR may be particularly helpful for evaluating skeletal metastatic disease, both in the identification of neoplastic focal marrow replacing lesions and in distinguishing them from benign osteoblastic activity. In poorly differentiated neuroendocrine tumors, 18 F-FDG PET/MR should be considered.

Conclusions
Technical advances in PET/MR have made it well suited for the evaluation of skeletal involvement of malignancies. To date, relatively few studies specifically explore the performance of PET/MR for identifying bone metastases. Despite their heterogeneity and small numbers, these studies suggest that PET/MR detects more bone metastases than PET/CT both on a per-lesion analysis and, more importantly, on a perpatient analysis. Moreover, PET/MR may also improve specificity owing to superior lesion characterization by the MR component [124]. In our opinion, based on experience imaging thousands of patients at 13 major PET/MR centers on 3 continents, when a hybrid PET study is indicated, PET/MR should be considered for staging of malignancies where there is a high likelihood of osseous metastatic disease based on the characteristics of the primary malignancy, high clinical suspicion, or where the presence of osseous metastases will have an impact on patient management. Tumor-specific tracers continue to emerge and should be considered, when available. However, to do so, both the MR and PET components must be used optimally, which means dedicated and accurate MR protocols, choice of the proper radiopharmaceuticals, and stringent up-to-date PET acquisition protocols.
Author contribution All authors contributed to the study conception and design. The first draft of the manuscript was written by Jad S Husseini and Onofrio A Catalano and all authors commented on versions of the manuscript. All authors read and approved the final manuscript.

Declarations
Conflict of interest A Torrado-Carvajal declares no conflicts of interest. The work by Dr. Torrado-Carvajal was partially supported by Young Reserchers R&D Project Ref M2166 (MIMC3-PET/MR) financed by Community of Madrid and Rey Juan Carlos University. Dr. Umutlu receives speaker and consultant payments from Bayer Healthcare and Siemens Healthcare, and research funds from Siemens. Dr. Herrmann has received personal fees from Bayer, Sofie Biosciences, SIRTEX, Adacap, Curium, Endocyte, IPSEN, Siemens Healthineers, GE Healthcare, and Amgen; non-financial support from ABX; personal fees from grants; and personal fees from BTG, outside the submitted work. Dr. Queiroz received research support from GE Healthcare. Dr. Herold is a member of scientific advisory board of Siemens Healthineers. He is also a recipient of grants/research support from Siemens Healthineers, Bayer Healthcare, Bracco (through contract with his University). Dr. Laghi has received honoraria for invited lectures from Bracco, GE Healthcare, Bayer, Guerbet and Merck. Dr. Mayerhoefer has received speaker honoraria from Bristol Myers Squibb and Siemens (paid to him), and research support from Siemens Healthineers (paid to his institution). Dr. Mahmood is a cofounder/shareholder, consultant, and grant recipient of