The majority of cancer patients with solid primary tumors will ultimately develop metastatic disease to the bones, most of them via hematogenous spread. Only few of the circulating tumor cells survive the host’s defense mechanisms and manage to reach the bone marrow vasculature. An even smaller fraction of tumor cells will escape the marrow vasculature by destroying the vascular basement membrane. Once in the bone marrow, they interact with hematopoietic and stromal cells, initiating a vicious cycle of tumor growth and bone destruction (seed and soil theory) [16]. Osteoclast activation resulting in bone resorption is achieved through the release of various tumor-derived factors. In turn, bone resorption leads to the release of growth factors which promote tumor proliferation. It, therefore, seems that the interaction between specific tumor cells and the bone marrow microenvironment determines the development of bone metastases [17].

The skeleton is one of the most frequent sites of metastatic disease, and bone metastases are associated with increased morbidity [18]. The vast majority of patients with bone metastases have primary tumors of the breast, prostate, thyroid, lung, kidney, and pancreas. Lung, breast, and prostate cancers, which are the most common human cancers, display increased avidity for the skeletal system. Bone metastases are characterized as osteolytic or osteoblastic depending on whether there is bone resorption or bone formation, respectively. Actually, at any metastatic site there is a spectrum of bone tumor activity, ranging from pure lysis to pure sclerosis. It is the predominant feature which determines the characterization of a bone metastasis as osteolytic or osteoblastic.

Conventional radiographs, computed tomography (CT), skeletal scintigraphy, [(18)F]-fluoro-2-D-glucose positron emitting tomography (FDG-PET), and MRI may all be employed in the work-up of skeletal metastases, each with specific indications. Conventional radiographs, CT, and skeletal scintigraphy depend on changes in bone turnover for lesion detection, and FDG-PET assesses metabolic activity. MRI is the only imaging modality which directly visualizes the bone marrow. It does not require bone loss or a perilesional reaction to identify bone metastases but may depict lesions which are confined to the marrow, before they affect the bony substrate.

MRI may detect lytic bone metastases in patients with negative radiographs, CT or bone scans. This is explained by the fact that over 30 % of bone loss must occur before a lesion becomes apparent on CT images and a much higher percentage of bone loss (about 50 %) is required for osteolysis to become evident on plain radiographs. CT will adequately depict cortical destruction by a metastatic lesion, but it may miss lesions confined to the marrow of spongious bones (vertebrae, iliac bones) unless marked trabecular destruction has occurred [19]. Osteoporosis and degenerative osteoarthropathy may, at times, be difficult to distinguish from bone metastases [20]. Caution is required, particularly with the increasing use of whole-body CT (WBCT) for skeletal assessment, not to diagnose small intracortical hypodense foci as bone metastases when, in fact, they may represent normal structures such as vessels. As with MRI, we should refrain from characterizing lesions smaller than 5 mm.

Bone scans detect bone metastases with increased sensitivity and much earlier than plain radiographs. However, aggressive bone metastases with rapid bone destruction or bone metastases at an early stage lack the osteoblastic reaction which is responsible for the radiotracer uptake and may, therefore, not show on a bone scan [21]. Small intramedullary metastases may fail to incite sufficient bone remodeling, and a mild increase in uptake may be obscured by the more intense uptake of normal cortical bone. The vertebrae and the long bones are bones with large marrow cavities, and consequently involvement of the cortex by bone metastases that lodge in the bone marrow will occur later in the course of the disease [22]. This explains why in a study comparing bone scintigraphy to MRI in patients with disseminated metastatic disease, bone scans identified the majority of spinal metastatic lesions involving the vertebral cortex but only a third of those in a subcortical location and none of the purely intramedullary lesions [23]. Bone scans, on the other hand, may fare better than MRI in the assessment of bones with limited marrow cavities such as the ribs or in the depiction of small intracortical metastases. False-positive bone scans are common, for example, in cases of degenerative disc disease and fractures, quite frequent findings in the elderly population which is also the age group more likely to harbor bone metastases [24, 25].

There is still ongoing discussion regarding the value of whole-body MRI (WBMRI) versus FDG-PET/CT for the detection of skeletal metastases. Both modalities are more accurate than bone scans in detecting bone metastases, and both offer the advantage of complete TNM staging. Most authors conclude that both WBMRI and FDG-PET/CT are robust imaging modalities for screening for osseous metastases provided that, in the case of FDG-PET/CT, investigated primary tumors are FDG-avid [26–29]. Indeed, a relative drawback of FDG-PET/CT in this setting is the poor FDG avidity of certain tumors such as prostate cancer, myxoid gastrointestinal tumors, low-grade sarcomas, and renal cell carcinoma [30]. On the other hand, FDG-PET/CT may be used as a problem-solving tool for indeterminate MRI lesions. Although MRI is extremely sensitive in detecting marrow lesions its specificity is relatively poor. For example, an atypical hemangioma, which may simulate a bone metastasis on MRI, will show no metabolic activity on FDG-PET. Recently, whole-body FDG-PET/MRI combining the metabolic and morphologic information of PET and MRI has emerged as a promising new technique for the evaluation of malignant bone lesions, including metastases [31].

The spine is the most common site for bone metastases because of the abundance of red marrow. The thoracic spine is more often affected, followed by the lumbar and, lastly, the cervical spine. Spinal metastases initially involve the vertebral body. The past belief that bone metastases show a predilection for the pedicles was due to the fact that destruction of the pedicles was more easily detected on conventional radiographs in comparison to destruction of the vertebral body [22, 32]. The increase in the fatty component of the bone marrow which occurs with age, central skeleton included, facilitates the diagnosis of metastases even of very small size, with MRI (Fig. 5.8). In older adults, hypointense metastases are easily recognized against a background of hyperintense fatty marrow on T1-weighted MR images. In children and young adults, red marrow predominates in the central skeleton and the sensitivity of MRI for the detection of vertebral metastases may be somewhat lower. For example, it has been found that metastases from neuroblastoma are detected with greater accuracy after the first year of life, because at that age some amount of fatty marrow has appeared in the spine [33].

In the large majority of patients, osseous metastases manifest with a focal rather than a diffuse MR pattern. The most important sequence for determining the presence and pattern of bone metastases is the T1-weighted sequence. On T1-weighted images, osteolytic metastases appear as well-defined, round- or oval-shaped foci which are hypointense and, less often, isointense to muscle or intervertebral discs (Fig. 5.9). Hyperintense bone metastases are extremely rare and have been reported only in association with melanoma (Fig. 5.10) [34]. The signal intensity of metastatic lesions is homogeneous because they destroy bony trabeculae and replace normal bone marrow in contiguity, rather than intermingle with it. This may be a useful sign in differentiating bone metastases from bone marrow edema on T1-weighted images; bone marrow edema shows heterogeneous signal intensity due to coexistence of hyperintense foci of normal fatty marrow with hypointense foci of bone marrow edema (Figs. 5.11 and 5.12). Low-signal-intensity focal lesions on a T1-weighted sequence may also represent red marrow rests. The signal intensity of red marrow rests is usually higher than that of disc or muscle (Fig. 5.13). A centrally located hyperintense focus (bull’s eye sign), if present, is an indicator of hematopoietic marrow. However, additional sequences (STIR, chemical-shift imaging, post-contrast T1-weighted) may be required to confidently differentiate between the two. Osteoblastic metastases are very hypointense because of paucity of moving protons. Their signal intensity may approximate that of cortical bone (signal-void focal pattern) (Fig. 5.14). An enostosis (bone island) may appear quite similar to an osteoblastic metastasis on T1-weighted images, but it displays homogeneous, very low signal intensity on all pulse sequences. Radiating bony streaks (thorny border) when present are also characteristic of an enostosis (Fig. 5.15).

On STIR images, bone metastases of an osteolytic nature become hyperintense. Red marrow rests, either become imperceptible or appear mildly hyperintense to background marrow; occasionally a focus of red marrow may appear moderately hyperintense on STIR images. Osteoblastic metastases show either no or mild heterogeneous increase in signal intensity (Fig. 5.16). The T2 halo sign which consists of a hyperintense rim surrounding a very low-signal-intensity lesion on STIR images, when present, is a very specific additional feature of an osteoblastic metastasis (Fig. 5.17). The hyperintense rim is believed to be caused by fluid, filling the area of destroyed bony trabeculae around the metastatic focus [35].

With chemical-shift imaging, most bone metastases show no or less than 20 % signal loss on out-of-phase images [36]. In fact, osteolytic metastases may even show a small increase in signal intensity and appear hyperintense on the out-of-phase compared to the in-phase images (Fig. 5.18). This is in contradistinction to red marrow rests which usually show a more than 20 % signal dropout on opposed-phase imaging. Chemical-shift imaging is not helpful for the evaluation of osteoblastic metastases.

On contrast-enhanced T1-weighted images without fat saturation, untreated osteolytic lesions become iso- or hyperintense to normal marrow. A more than 40 % enhancement on post-contrast T1-weighted images is characteristic of malignancy [37]. Osteolytic metastases may not be discernible on the enhanced images unless careful comparison with the pre-contrast images is made or fat suppression is applied (Fig. 5.19). Unlike osteolytic metastases, osteoblastic deposits usually show mild or no obvious contrast uptake (Fig. 5.20).

On diffusion-weighted images with high b-values, lytic bone metastases are hyperintense and easily recognized. Their ADC values are high, often higher than 1.0 × 10⁻³ mm²/s, because they exhibit increased diffusivity relative to the very restricted diffusion of normal fat-containing marrow (Fig. 5.21) [38]. Osteoblastic metastases are generally characterized by more pronounced diffusion restriction (i.e., lower ADC values) than osteolytic lesions because of new bone formation; occasionally their ADC values overlap with those of normal marrow.

Bone metastases may break through the cortical bone and present with an extraosseous soft-tissue component. Cortical destruction is easily diagnosed when the markedly low signal of cortical bone is lost (Fig. 5.24). Occasionally, extraosseous extension of a metastatic bone lesion may occur without evidence of gross cortical or trabecular destruction. This is usually observed in metastases from small round cell tumors, such as neuroblastoma, small cell lung carcinoma, Ewing’s sarcoma, and rhabdomyosarcoma, possibly reflecting the permeative nature of these tumors. A similar pattern of extraosseous soft-tissue extension with apparent preservation of cortical bone integrity has also been classically described in bone marrow lymphoma [39]. T1-weighted images are best suited for the diagnosis of tumor extending to the epidural space or intervertebral foramina (Fig. 5.25). The draped curtain sign is caused by tumor spreading to the right and left of the meningovertebral ligament which attaches the anterior surface of the dural sac to the posterior longitudinal ligament (Fig. 5.26) [40]. Symptomatic cord compression occurs in 10–20 % of patients with vertebral metastases (Fig. 5.27) [41]. Long-standing compression by epidural tumor may cause hyperintense signal in the spinal cord on T2-weighted images; contrast enhancement of the cord may or may not be present.

The occasional occurrence of peripheral skeletal metastases led to the evaluation of whole-body MRI (WBMRI) for bone metastases screening. Initially, WBMRI was performed utilizing T1-weighted and STIR sequences; more recently Diffusion-Weighted Imaging (DWI) was added to oncology protocols. Meta-analysis of data from 11 WBMRI studies showed a pooled sensitivity of 89.9 % and a pooled specificity of 91.8 % for the detection of bone metastases [42]. The implementation of DWI to WBMRI seems to increase its sensitivity and positive predictive value for the detection of bone metastases [43]. However, even though the sensitivity increases, the specificity of WBMRI may be compromised by the addition of DWI [42]. In the case of bone metastases from prostate cancer, DWI with high b-values shows increased lesion conspicuity compared to STIR, because the pronounced osteoblastic component and reparative bone formation of these lesions decreases T2 relaxation time; on the other hand, for breast cancer bone metastases, DWI and STIR seem to have similar sensitivities [44]. DWI is therefore recommended for WBMRI protocols of prostate cancer. WBMRI detects more lesions than MRI limited to the spine; the clinical impact of the surplus data, however, is not clear. In a study of prostate carcinoma patients, comparing WBMRI to MRI of the axial skeleton (AS-MRI covering the cervicothoracic and lumbosacral spine as well as the pelvis and proximal femora), peripherally located metastases visible at WBMRI were only present in patients who also harbored metastases in the axial skeleton [45]. In our experience, isolated peripheral metastases confined to the bone marrow are extremely unusual; rarely, intracortical bone metastases may be observed in the periphery without involvement of the axial skeleton.

Although MRI is very sensitive in detecting skeletal metastases even of small size, its ability to differentiate them from other malignant bone marrow lesions or to diagnose the neoplasm of origin is more limited. Bone metastases are quite similar in appearance with foci of multiple myeloma, both of which affect mainly the elderly population. When bone metastases occur close to the endplates, the differential diagnosis may also include Modic I degenerative endplate changes and spondylodiscitis. Modic I changes display signal intensity similar to metastases (low on T1-weighted and high on T2-weighted images), but they are typically located along the vertebral endplates on both sides of a degenerated disc. However, occasionally, they may be more focal and unilateral and difficult to distinguish from a metastatic lesion. On high b-value DWI, Modic I changes show either no high signal or linear high signal at the interface of normal with abnormal marrow (referred to as the claw sign), while osseous changes related to metastases and spondylodiscitis both show increased signal intensity (Figs. 5.28, 5.29, and 5.30) [46, 47]. Spondylodiscitis is additionally characterized by T2 hyperintensity and contrast enhancement of the intervertebral disc; osseous, intradiscal, or soft-tissue abscesses, when present, are pathognomonic of infection.

Multiple myeloma is a hematologic malignancy caused by proliferation of a single clone of plasma cells in the bone marrow. It affects 5–6/100,000 individuals/year and accounts for 13 % of hematologic malignancies and 1 % of all cancers. Median age at presentation is 70 years, and men are affected slightly more often than women [48–50].

Multiple myeloma is characterized by the presence of ≥10 % abnormal plasma cells in the bone marrow and ≥3 g/100 ml serum monoclonal protein (with or without the presence of monoclonal protein in the urine). The presence or absence of end-organ dysfunction manifesting with hypercalcemia, renal insufficiency, anemia, or bone lesions (referred collectively by the acronym CRAB) classifies myeloma as symptomatic or asymptomatic (smoldering) disease.

A monoclonal gammopathy of uncertain significance (MGUS) is an asymptomatic, premalignant proliferation of monoclonal plasma cells (<10 % plasma cells in the bone marrow and <3 g/100 ml monoclonal protein in the serum) which is believed to precede the development of multiple myeloma with an annual risk of about 1 % [51]. A monoclonal gammopathy of borderline significance (MGBS) differs from MGUS in the percentage of bone marrow plasmacytosis (10–30 %). Plasma cell disorders are believed to form a continuum, progressing from MGUS to asymptomatic myeloma, and eventually to symptomatic multiple myeloma [52].

The biology of multiple myeloma is complex because of the various genetic abnormalities of plasma cells [50]. These abnormal cells interact with the microenvironment of the bone marrow and cause both neoplastic cell proliferation and bone resorption. Myeloma cells initially attach to bone marrow cells and extracellular matrix proteins. Through a series of events, mediated by the release of various cytokines and growth factors, tumor growth and angiogenesis is induced and the balance between osteoblast and osteoclast formation is disrupted. The bone marrow microenvironment itself actually plays an active role, assisting neoplastic proliferation [53].