➢ Bone tumors can vary in aggressiveness. Therefore, appropriate characterization and diagnosis of primary bone tumors are critical for determining prognosis and therapeutic strategies.
➢ Conventional radiography is the first and most important imaging modality in the characterization of bone tumors.
➢ Conventional radiography can be used to determine the localization, size, margins, and matrix content of the lesion as well as to evaluate other findings such as the osteolytic pattern, cortical changes, and periosteal reaction associated with the lesion. Combined with patient history, this information can be used to evaluate bone tumor activity and to generate a preliminary differential diagnosis.
A primary bone tumor is a neoplasia that originates within the skeletal tissue. Primary musculoskeletal tumors are rare and account for <1% of all cancers. Approximately 2500 new cases of primary bone malignancy are diagnosed each year in the United States1,2. The vast majority of bone tumors are benign and asymptomatic and therefore are likely to remain undetected or are diagnosed incidentally1,2. In turn, the true incidence of benign bone tumors is difficult to estimate1,2. In one study, the prevalence of metastatic bone disease was estimated to be approximately 280,000 cases in the U.S. in 20083.
Primary bone tumors can be benign or malignant and can vary in terms of aggressiveness and activity. While some benign and indolent lesions may be asymptomatic, aggressive malignant lesions may present with pain, pathologic fractures, and metastasis. Appropriate characterization and diagnosis of bone tumors is necessary for determining therapeutic strategies1,2.
Conventional radiography is the first diagnostic tool that is used for the assessment of suspected bone tumors and is considered to be the most important initial imaging modality4-9. Conventional radiography can be used to determine the localization, size, margins, and matrix content of the lesion as well as to evaluate other findings such as the osteolytic pattern, cortical changes, and periosteal reaction associated with the lesion2,9-11. Combined with relevant clinical history, these radiographic features can be used to characterize the osseous lesion and can help to establish a preliminary differential diagnosis2,4,6-8,12.
Advanced imaging techniques may offer additional clinical information13-17. Computed tomography (CT) can be used to establish the transverse localization and is more sensitive for detecting periosteal and cortical changes, thereby allowing for the detection of osteolysis earlier than is possible on radiographs9,10. Magnetic resonance imaging (MRI) can be employed to determine the extent of the lesion and its relationships with neighboring structures, making it useful for staging of malignant lesions18. MRI can also can be used to determine the effect of the tumor on surrounding soft tissues and can characterize the fluid content of the tumor and surrounding areas14-16,18. While advanced imaging may be important for the characterization and diagnosis of bone tumors, a full discussion of MRI and CT findings of bone tumors is beyond the scope of this review.
We provide a narrative review of the role of radiographic imaging in the diagnosis and characterization of bone tumors. We review the radiographic features of bone tumors that can be utilized to estimate tumor activity, to evaluate the likelihood of malignancy, and to generate a differential diagnosis. We also provide an overview of the role of bone scintigraphy in the characterization of bone tumors.
Assessment of Tumor Activity and Likelihood of Malignancy
Estimation of tumor aggressiveness and the likelihood of malignancy is critical in determining prognosis and treatment options. Radiography can be employed to determine the size, shape, osteolytic pattern, margins, endosteal and periosteal changes as well as the presence of soft-tissue involvement.
Tumor Size and Shape
The size of the lesion is proportional to the likelihood of malignancy and can be a predictor of prognosis19. Kaste et al., in a retrospective study of forty-two patients, found that the absolute size of nonmetastatic osteosarcoma lesions was predictive of overall survival rates and event-free survival in a continuous manner (i.e., the greater the tumor size, the lower the survival rate)19. A categorical data analysis showed that osteosarcoma lesions measuring >150 cm3 in volume were associated with an overall five-year survival rate of 87.5%, whereas lesions measuring ≤150 cm3 were associated with a five-year survival rate of 61.1%. Certain lesions can also be differentiated from each other on the basis of size19. For example, the difference between a fibrous cortical defect and a nonossifying fibroma is that the former is <3 cm in length whereas the latter is ≥3 cm in length19.
Osteolytic Patterns and Tumor Margins
Bone tumors can stimulate osteolytic activity along the advancing margins of the lesion, thereby generating an osteolytic pattern that can become visible on radiographs20. Reactive host-bone osteogenesis occurs in response to tumor activity, and the combination of these two processes generates the tumor margin7,8,11,20,21. Both the tumor-induced bone-destruction pattern and the lesion margins can be used to evaluate tumor aggressiveness and the likelihood of malignancy. With slow-growing benign tumors, reactive host-bone osteogenesis can lead to the creation of a sclerotic margin that sharply demarcates the tumor from the host bone; however, with fast-growing aggressive or malignant tumors, there is insufficient time for the host bone to make a reactive bone border around the tumor, resulting in a poorly defined or visualized margin between the tumor and the surrounding normal bone6,8,10,20.
A geographic osteolytic pattern is associated with a focal solitary radiolucent lesion with a narrow zone of transition between neoplastic and normal tissue7,8,10,20. The region inside the lesion is generally homogeneous. While some aggressive and metastatic lesions may have a geographic appearance, geographic bone lesions typically are slow-growing and tend to have low biologic activity22,23. The biological activities of bone tumors with geographic osteolytic patterns can be further estimated on the basis of the appearance of their margins (Fig. 1)20,24. Three types of margins were described by Lodwick et al.24. Type 1A is a thick, sclerotic margin that is indicative of an indolent tumor; Type 1B is a thick, faintly sclerotic margin that is indicative of an actively growing tumor; and Type 1C is a barely discernible margin that is indicative of a faster-growing, more aggressive tumor.
Geographic osteolytic lesions may present with a radiodense sclerotic rim. The sclerotic rim is formed when the host bone lays down reactive bone in response to tumor growth and activity7,8,10,21. These tumors generally are slower-growing benign processes or low-grade malignant lesions7,8,24. The thickness of the sclerotic rim is inversely proportional to tumor growth rate and activity12. Detailed examination of the sclerotic rim may have additional clinical value20. The sclerotic rims of less-active benign lesions tend to have sharper outer margins as compared with inner margins, whereas the outer margins of biologically active inflammatory lesions may be poorly defined4,20 (Fig. 1, A).
Other geographic lesions have well-defined sharp but non-sclerotic borders24. The absence of a sclerotic rim suggests that these tumors have greater growth rates than those with sclerotic rims20. However, the sharp margins imply that the tumor-induced osteolysis has not surpassed reactive host-bone osteogenesis, suggesting that the lesion is generally less aggressive and slow growing8,12,20. This type of lesion rarely extends beyond the radiographic margins of the tumor4 (Fig. 1, B).
A geographic lesion also may present without a sclerotic rim and with a poorly defined border4,20,24. These lesions also may have evidence of total tumor penetration of the cortex20. The poorly defined margins mean that tumor-induced osteolytic activity is dominant over reactive host-bone osteogenesis and that the tumor is infiltrating between adjacent trabeculae, suggesting that the lesion is likely aggressive and fast-growing12,20,24. The lesion is more extensive than its appearance on radiographs and should be further evaluated for malignancy7,12 (Fig. 1, C).
Moth-eaten and permeative patterns appear on radiographs as clusters of multiple punched-out radiolucent holes that are localized in both cancellous and cortical bone7 (Fig. 2). These holes have irregular and poorly defined boundaries, with a large zone of transition, giving the lesion a diffuse appearance20,24. These lesions tend to be more extensive and larger than their radiographic appearance and are likely to be aggressive, rapidly growing, and malignant12,20,24.
The borders between a lesion and the surrounding normal bone may be poorly visible and barely perceptible, making it difficult to identify the lesion on radiographs. These lesions represent a situation in which the host bone has had minimal time to respond or to wall off the tumor. In turn, this osteolytic pattern typically occurs in fast-growing and aggressive lesions20.
A single lesion may exhibit a combination of different osteolytic patterns or margin types. Combinations of different osteolytic patterns or margin types suggest that an antecedent benign lesion may be undergoing aggressive and possibly malignant transformation20.
Bone lesions may erode or scallop the endosteum, resulting in concave depressions on the medullary aspects of the cortex that can be visualized on radiographs12. The finding of endosteal scalloping is suggestive of a more-active lesion12 (Fig. 3).
Underlying bone tumor activity may induce changes in the periosteum known as periosteal reactions25,26. Periosteal reactions tend to be more common in children because the periosteum is thicker and more metabolically active compared with that in adults27. While the periosteum normally is not visible on conventional radiographs, mineralization of the periosteum in patients with pathological lesions allows these structures to be detected radiographically.
Periosteal reactions are classified according to their appearance27. The type of periosteal reaction may help to assess tumor characteristics such as growth rate, aggressiveness, and malignancy. Some periosteal reactions are almost always correlated with aggressive or malignant lesions, making the understanding of periosteal reactions clinically valuable7,12,25,26.
Erosion of the endosteum as a result of tumor growth-induced osteolysis may stimulate appositional ossification on the periosteal surface, thereby generating an expanded cortical shell appearance7,12,25,26. These two concomitant processes generate an expanded cortical shell that is visible on radiographs25. A higher tumor growth rate induces greater endosteal erosion relative to the rate of appositional ossification, thereby generating a progressively thinner expanded cortical shell12,25. Therefore, the thickness of the expanded cortex is believed to be inversely proportional to the tumor growth rate and osteolytic activity25.
Expanded cortical shells may be further classified as smooth, lobulated, or ridged25. Smooth shells have a radiodense, expanded, and uniform outer contour on radiographs and are typically seen in association with tumors that are expanding and eroding the endosteum in a symmetrical manner25. Smooth-shell lesions tend to be benign25. Lobulated shells have the radiographic appearance of multiple compartments separated by radiodense septations25. Ridged shells are lobulated shells with denser septations25.
A single-layered periosteal reaction is characterized by a uniformly radiodense layer above the cortical surface that may join the cortex at the proximal and distal ends12,25,26. The single-layered pattern is classically associated with nonaggressive benign processes such as osteomyelitis or fracture-healing12,25,26. This pattern also can be seen in association with malignant lesions but rarely is seen in association with metastases12,25,26 (Fig. 4, A).
A multiple-lamellated or “onion skin” pattern has the radiographic appearance of alternating radiodense and radiolucent layers that lie above the cortex8,12,25-27. The alternating layers may connect to the cortex at the proximal and distal ends8,12,25,26. This appearance may be generated when tumor-associated hyperemia induces alternating sheets of fibroblasts to undergo sequential or simultaneous modulation into osteoblastic cells12,25,26. The multiple-lamellated appearance is associated with benign processes but also can be associated with malignant lesions12,25,26. Generally, more aggressive lesions tend to have thinner lamellations and more space between the ossified layers12,25,26 (Fig. 4, B).
A solid layered periosteal reaction has the radiographic appearance of a homogeneous radiodense layer of periosteal bone of variable thickness that joins the cortex at the proximal and distal ends8,12,25-27. This pattern roughly appears as a “bulge” protruding from the cortex12,25,26. The solid layered pattern arises from the sequential addition of osseous layers or from the fusion of the layers of a multiple-lamellated reaction12,26 (Fig. 4, C).
A solid layered pattern typically is associated with long-standing benign slow-growing processes and rarely is seen in association with aggressive lesions12,25-27. In a review of seventy-eight lesions with a solid layered appearance, the most commonly associated processes were osteoid osteoma and osteomyelitis26. Thicker and denser solid layered reactions and the existence of an undulating contour are associated with longer-existing processes7,12,25,26.
A spiculated periosteal pattern appears on radiographs as radiodense rays projecting from the cortical surface7,8,12,25,26. This pattern also can be described as a “hair on end” pattern (radiodense parallel rays projecting perpendicularly from the cortex) or a “sunburst” pattern (radiodense rays fanning out from a focal point on the cortex)26. A spiculated reaction occurs when reactive bone forms along the fibrous bands and blood vessels projecting from the cortex25,26. The orientation of the spicules often reflects the direction of tumor growth25 (Fig. 4, D).
A spiculated pattern typically is associated with rapidly growing malignant lesions7,25,26,28. In a retrospective review of forty cases of pathologically confirmed osteosarcomas, thirty-eight patients (95%) had radiographic evidence of a spiculated pattern29. Benign processes such as infection or healing fractures that display the spiculated pattern tend to have shorter and thicker spicules25.
Interrupted periosteal reactions are abruptly truncated25 (Fig. 5). These patterns are further classified according to their radiographic appearance. Interrupted patterns are generated when a lesion extends through or rapidly erodes a preexisting periosteal reaction25. In turn, processes that are associated with interrupted patterns are generally aggressive, fast growing, and, likely, malignant26.
The buttress pattern has the radiographic appearance of beak-like solid radiodense wedges that tend to be located at the extremes of the lesion7,8,12,25-27 (Fig. 5, A). The cortex directly below the radiodense wedges is usually intact, whereas the cortex between the wedges is usually absent8,25,26. This appearance also may be accompanied by a periosteal shell reaction25. The buttress appearance may be produced when an intramedullary lesion that previously generated a solid layered pattern becomes aggressive and disrupts the solid reaction12,25,26. Thus, a buttress pattern may indicate aggressive and possibly malignant transformation of a previously benign slow-growing lesion25. However, buttress reactions also may directly form at the margins of a shell periosteal reaction or in response to adjacent cortical saucerization8,12,25,26.
The Codman triangle appears on radiographs as a triangular-shaped reaction that forms an acute angle with the cortex at one end and borders the lesion on the other6-8,12,25-27. Unlike the buttress pattern, the Codman triangle contains an outer plane that is more radiodense than the rest of the reaction25. This appearance can be generated when a neoplasm elevates the periosteum and the regions of periosteum touching the extreme ends of the tumor ossify25,26. The Codman triangle pattern is classically associated with aggressive and malignant bone lesions7,12,25,26,29. As pus or hemorrhage also can elevate the periosteum, the Codman triangle also may be observed in association with osteomyelitis25,26. In a review of eighty-five lesions associated with the Codman triangle, sixty (71%) of these lesions were pathologically confirmed osteosarcomas or Ewing sarcomas26 (Fig. 5, B).
An interrupted spiculated reaction is similar to a continuous spiculated pattern except that it is only found at the lateral borders of the lesion25,26. Although the reaction typically has the appearance of rays projecting from the cortex, the spicules may be closely packed, thus generating a homogenous radiodense appearance25,26. While spiculated wedges can directly form at the lateral borders of a lesion, thereby generating a so-called interrupted appearance, interrupted spiculated patterns likely are the result of direct destruction of a continuous spiculated reaction by a rapidly growing lesion26. Therefore, an interrupted spiculated appearance suggests that the lesion is undergoing rapid extraosseous growth7,28,29 (Fig. 5, C).
Generating a Differential Diagnosis
The development of a differential diagnosis is critical for estimating prognosis and for determining appropriate treatment for bone neoplasia. While estimations of tumor activity and the likelihood of malignancy may contribute to the generation of a differential diagnosis, the localization and matrix patterns of the tumor and relevant patient history also can play an important role in the diagnosis of a bone lesion.
Patient age always should be considered in the diagnosis of a bone lesion. Bone tumors tend to arise in specific metabolic environments and cell populations21. As bone metabolic activity and cellular composition vary with age, most bone tumors tend to occur in specific age groups8 (Table I). Importantly, metastasis or myeloma should be a consideration in the evaluation of bone lesions in patients more than forty years of age4,8,12.
The sex of the patient may provide some additional clues to the diagnosis of the bone tumor. Most bone lesions occur in either equal proportions in males and females or have a slight male predilection16. Exceptions to this observation are giant-cell tumor, enchondroma, fibrous dysplasia, and parosteal osteosarcoma, which have a female predilection4.
Race may play an important role in differentiating bone lesions. Blacks have a twofold increased risk of multiple myeloma compared with whites. An analysis using the Surveillance, Epidemiology, and End Results database suggests that whites are more likely to develop Ewing sarcoma compared with other racial groups and have a ninefold increased risk compared with blacks30. Patient symptoms also can be important in the diagnosis and characterization of bone tumors. Osteoid osteoma is classically characterized as bone pain at night that is alleviated with nonsteroidal anti-inflammatory drugs4,12.
The cellular composition and activity as well as the biological behavior of bone is dependent on anatomical location4,21. As tumor cells tend to arise in sites where the corresponding precancerous cells have the greatest metabolic and mitotic activity, lesions have predilections for specific bones and locations within a bone4,21. Conventional radiographs can be used to determine tumor localization, thereby helping to establish a differential diagnosis (Table I).
A mineralized matrix that forms in the void created by tumor-induced osteolysis may become visible on radiographs22. The radiographic appearance of this matrix is dependent on its histological composition4,7,10,12,22. As chondroid, osteoid, and fibrous matrices are produced by chondrocytes, osteoblasts, and fibroblasts, respectively, knowledge of the matrix composition may help to determine the lesion’s cellular composition, thereby aiding in its diagnosis and characterization7,22. Furthermore, certain types of lesions, such as unicameral and aneurysmal bone cysts, giant-cell tumors, and lymphoma, do not present with a tumor matrix and therefore the detection of a mineralized matrix on radiographs excludes these entities from the differential diagnosis4. Also, the soft-tissue component of Ewing sarcoma typically does not mineralize, and therefore a finding of soft-tissue mineralization on radiographs reduces the likelihood of a diagnosis of Ewing sarcoma31. Serial radiographs also can be used to detect loss of mineralization, which can be a sign of evolution toward malignancy7,12,22. It is important to note that a lesion can produce more than one type of matrix, and therefore the radiographic appearance of the matrix mineralization is a reflection of the activity of the predominant cellular makeup of the tumor4.
Cartilaginous matrix mineralization tends to occur in the more mature central portions of the lesion7,22. Mineralized cartilaginous matrix appears on radiographs as radiodense stipples, floccules, and “rings and arcs”7,12,22,32. Stipples are the result of punctated calcification, and the coalescence of these stipples produces the flocculent pattern, whereas the rings-and-arcs pattern is generated from enchondral bone formation that predominately occurs around the periphery of the cartilaginous lobules7,12,23,32. Thus, the prevalence of rings and arcs or stipples and floccules suggests the predominance of enchondral bone formation or punctated calcification, respectively22,23,32. Knowledge of the type of cartilaginous calcification can contribute to a differential diagnosis7,22,23 (Fig. 6, A).
The radiographic appearance of mineralized osseous matrix is dependent on tumor maturity22. Immature bone-producing tumors such as high-grade osteosarcomas appear cloud-like and hazy on radiographs and are more likely to be aggressive and malignant4,22. Mature osseous tumors such as low-grade parosteal osteosarcoma may have the radiographic appearance of large, fairly homogenous, radiodense areas with well-defined edges22. Well-organized bone trabeculae appearing as radiodense struts also may be observed in these mature bone-forming tumors22 (Fig. 6, B).
Mineralized fibrous matrix is characterized by an opaque and hazy appearance on radiographs22,33. Lesions with this appearance typically are tumors in which fibroblasts have converted into osteoblasts and typically are fibrous dysplasia or nonossifying fibroma4,7,22 (Fig. 6, C).
Limitations of Radiography
Radiography is limited by its inability to visualize insufficiently mineralized structures as a result of its low contrast resolution4,14. A substantial proportion of trabecular bone (30% to 50%) needs to be lost before the lesions become detectable on radiographs4,7,8. In turn, early changes to the bone structure that may be reflective of tumor presence or activity may be undetectable with use of conventional radiography. Furthermore, as the visualization of an osteolytic pattern is dependent on the contrast of the surrounding bone, lesions in osteopenic or osteoporotic bone may be more difficult to detect with radiography4. In addition, tumors that are localized to the diaphysis of long bones may be difficult to detect because of the decreased contrast resulting from the lower proportion of cancellous trabeculae in that region4. Radiography also is limited by its lack of cross-sectional imaging capability and is therefore difficult to use to visualize overlapping structures14.
Role of Bone Scintigraphy
Radionuclide bone scintigraphy utilizing technetium-99m-labeled diphosphonates is a cost-effective and commonly available diagnostic tool. Bone scintigraphy is typically performed in three phases. The first phase is performed within the first five seconds following technetium-99m injection and demonstrates perfusion to a bone lesion. The second phase, performed five minutes after the injection, produces the “blood pool image” and demonstrates inflammation. The third phase, or “delay phase,” is performed three hours following the injection and demonstrates bone turnover17.
Several bone tumors have scintigraphic patterns that may aid in diagnosis. Aneurysmal bone cysts have an appearance of moderate to intense tracer accumulation at the periphery, with little activity in the center of the lesion, thereby producing a “donut pattern.” It is important to note that this scintigraphic pattern is not specific and can also be seen in association with chondrosarcomas, giant-cell tumors, and telangiectatic osteosarcomas17.
Monostotic fibrous dysplasia has the scintigraphic appearance of an area with markedly increased uptake on all three phases of the bone scan. The appearance of a donut pattern may indicate the presence of secondary aneurysmal bone cyst. Bone scintigraphy is also important for determining whether the bone lesion is polyostotic17,34.
Giant-cell tumors demonstrate increased tracer uptake in all three phases of bone scintigraphy. Similar to fibrous dysplasia and chrondrosarcoma, these lesions may demonstrate a donut pattern. As bone scans may inaccurately estimate giant-cell tumor extent or fail to determine extraosseous involvement, the clinical utility of bone scintigraphy in evaluating giant-cell tumors is limited17,35.
Enchondroma typically demonstrates mild-to-moderate uptake during the third phase of the bone scan. The clinical utility of bone scintigraphy for enchondromas is the detection of multiple lesions (multiple enchondromatosis), which have a greater risk of sarcomatous transformation into chrondrosarcoma. Thus, bone scintigraphy is important for evaluating the malignant potential of this lesion36,37.
In contrast to enchondroma, chrondrosarcoma demonstrates an intense and heterogeneous uptake on bone scintigraphy. Bone scintigraphy with use of technetium-99m dimercaptosuccinic acid can be used to evaluate the malignant potential of this lesion15,17,36.
Osteosarcoma appears on bone scanning as an area of intense homogeneous uptake17. Neighboring bones in the extremity of the tumor may demonstrate mild to moderate diffuse uptake that tends to be more intense around the joint38. Importantly, bone scintigraphy is a clinically effective tool for evaluating metastasis of this tumor, differentiating between nonspecific calcifications, and detecting a multifocal variant of osteosarcoma17. While Ewing sarcoma typically demonstrates intense uptake, in a very few cases Ewing sarcoma appears as an area without tracer accumulation on a technetium-99m bone scan39. Bone scintigraphy is not recommended for evaluating the response of Ewing sarcoma to chemotherapy17.
The ability of bone scintigraphy to detect active lesions throughout the entire skeletal system makes it important for the evaluation of metastatic disease. Bone scans may reveal metastatic lesions in patients without overt malignant disease. The ability of bone scans to visualize the entire skeletal system also allows this modality to provide imaging data in areas that may be difficult to visualize on radiographs (i.e., ribs, sternum, or pelvis)17 (Fig. 7).
Medical history, physical examination, and radiographic findings are important tools in the assessment and diagnosis of bone lesions4-8,15. Radiography can be used to evaluate the tumor location, lytic patterns, margins, matrix, periosteal reaction, and effects on the cortex and endosteum4,6-8. Information derived from radiography can be used to estimate tumor aggression, growth rate, histological content, and likelihood of malignancy4,6-8. Other imaging modalities may provide additional information.
Source of Funding: No external funding was received for the present study.
Investigation performed at Hackensack University Medical Center, Hackensack, New Jersey, and Mount Sinai Medical Center, New York, NY
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. None of the authors, or their institution(s), have had any financial relationship, in the thirty-six months prior to submission of this work, with any entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, no author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2014 by The Journal of Bone and Joint Surgery, Incorporated