Management of Primary Osseous Spinal Tumors with PET

INTRODUCTION

Primary spine tumors account for 5% of all osseous tumors.¹ Given the imaging overlap of many benign and malignant lesions, tumors of the vertebrae can present a diagnostic challenge. Using imaging to accurately diagnose tumors can prevent unnecessary biopsy and determine the optimal medical or surgical therapy. Prompt and accurate diagnosis is necessary because both malignant and certain benign tumors can cause mass effect on the thecal sac and lead to irreversible neurologic compromise. Primary osseous tumors of the spine are for the most part similar to osseous tumors found within the rest of the body. Although they can often be diagnosed based on anatomic imaging alone, fluorodeoxyglucose (FDG) PET can help in more challenging cases. Additionally, knowledge of the PET findings of such tumors is important because they are often incidentally encountered when imaging patients for an unrelated primary neoplastic process.

The first means of differentiating tumor types is through consideration of patient demographics in combination with the location of the tumor. The greatest differentiator is simply age, with the vast majority of spine lesions presenting before 30 years of age benign, with the exception of Ewing sarcoma and osteosarcoma.²

Location of the lesion, vertebral body versus posterior elements, is another important differentiating factor. Lesions originating from the posterior elements are primarily benign and include osteoid osteoma, osteoblastoma, osteochondroma, and aneurysmal bone cyst (ABC). Malignant lesions of vertebral body, such as chondrosarcoma, osteosarcoma, and Ewing sarcoma, however, may extend into the posterior elements. Both benign and malignant lesions can be expansile and encroach on the thecal sac or nerve roots, causing neurologic symptoms. Many benign lesions, such as osteoblastomas or even vertebral hemangiomas, can be expansile, leading to spinal cord or nerve compression.³ Within the vertebral body, the majority of primary lesions are benign hemangiomas or enostoses. Other lesions centered primarily within the vertebral bodies include myeloma/plasmacytoma, lymphoma, chordoma, eosinophilic granuloma, and giant cell tumor (GCT).⁴

In addition to location, certain morphologic characteristics seen on radiography and CT can help differentiate benign from aggressive lesions. In general, benign lesions demonstrate well-defined borders, which are often sclerotic, whereas aggressive tumors have a wider zone of transition to normal bone. Aggressive lesions also often have more aggressive periosteal reaction, often showing a lamellated or sunburst appearance. On MR imaging, aggressive lesions are often found to demonstrate adjacent marrow edema as well as soft tissue component and edema. In addition to these general morphologic features, specific types of lesions can often be diagnosed with very specific findings. Fluid-fluid level seen on MR imaging or CT, for example, is most often associated with ABCs in the posterior elements, although it can also be seen with telangiectatic osteosarcoma and GCTs. Chondroblastomas, although rare in the spine, also may demonstrate fluid-fluid levels. A sclerotic lesion with a central nidus is often seen with osteoid osteoma, again usually centered in the posterior elements. Lesions that clearly contain fat are most often benign, with the majority representing hemangiomas.⁴ Vertebral hemangiomas demonstrate thickening of the primary vertical weight-bearing trabeculae on CT, resulting in corduroy sign on sagittal or vertical projections and polka-dot sign on axial images.

PET/COMPUTED TOMOGRAPHY IN THE DIAGNOSIS OF BENIGN OSSEOUS SPINAL LESIONS

FDG PET can be used in conjunction with CT or MR imaging to characterize lesions that appear similar or to grade lesions as more aggressive or less aggressive. As a whole, malignancies of the vertebrae, as elsewhere in the body, have greater FDG uptake than benign lesions.^5–8 FDG PET/CT is an excellent tool for differentiating metastatic disease from benign spinal lesions, proving more sensitive and specific than CT or MR imaging alone.⁹ Differentiation between lesions is most useful when comparing tumors of the similar histo-pathology, such as chondrosarcoma versus enchondroma.^5,6,10–12 Although cartilage lesions usually demonstrate characteristic ring and arc matrix on plain film and CT as well as high T2 signal on MR imaging, this does not necessarily indicate benignity, and differentiation between malignant and benign lesions can be difficult on the basis of anatomic imaging alone. Evaluation of glucose metabolism with FDG PET has a sensitivity of 91%, specificity of 100%, and accuracy of 97% in differentiating benign and malignant chondral lesions.^10,11 Specifically, the vast majority of malignant lesions studied demonstrate maximum standardized uptake value (SUVmax) of greater than 2, whereas all benign lesions show SUVmax of lower than 2. On the contrary, enchondromas demonstrate substantial uptake on 18F-sodium fluoride (NaF) PET due to their osteoblastic activity, potentially confusing them with chondrosarcoma.^13,14

Despite its utility in differentiating malignant and benign chondral lesions, distinguishing between aggressive benign lesions of differing histologies and high-grade malignancies of spine by FDG PET is not always possible, as shown by a study of 202 patients by Schulte and colleagues.⁵ Additionally, many benign tumors with the exact same histology can have widely varying levels of FDG uptake. Of the benign spinal column tumors, several, including fibrous dysplasia, ABC, GCT, eosinophilic granuloma, osteoid osteoma, and osteoblastoma, often exhibit substantial FDG avidity.^7,8,10,15 Many of these lesions can have the same levels of uptake as highly aggressive neoplasms, such as osteosarcoma. The only benign vertebral lesions with consistently low FDG uptake are benign cartilaginous lesions, hemangiomas, and intraosseous lipomas. Some investigators, however, argue that benign cartilaginous tumors of the spine are rare. Similar to FDG, NaF can also not be definitively used to differentiate benign from malignant spine lesions as a whole.^16–18 When combined with morphologic details provided by MR imaging or CT, however, both FDG and NaF PET can prove diagnostically useful in certain instances. Langsteger¹⁹ studied 150 lesions in 20 patients who underwent both NaF PET/CT and FDG PET/CT to decipher whether one radiotracer was more sensitive than the other in detecting osseous metastases and found that approximately 50% of the metastatic lesions were found on both modalities, whereas approximately 20% were found on FDG PET/CT only and 20% were found on NaF PET/CT only. Specifically, FDG PET/CT was better at detecting osteolytic lesions overall and NaF PET/CT was better at detecting non–FDG-avid lesions, such as renal cell carcinoma and thyroid cancer.¹⁹ Additional studies have also shown the superiority of NaF in the diagnosis of sclerotic lesions.²⁰ This sensitivity, however, also leads to a high false-positive rate in NaF.²¹

GCT is a traditionally benign lesion that can prove difficult to differentiate from aggressive malignancies on both anatomic and physiologic imaging. This benign tumor is composed of ovoid mononuclear cells and osteoclastic giant cells, occurring in skeletally mature patients. Of the 7% of GCTs affecting the spine, 90% are found in the sacrum, especially within the sacral ala.^4,22,23 Due to the prevalence of endosteal scalloping, cortical destruction and even associated soft tissue mass, it is difficult to distinguish these lesions from malignancies on anatomic imaging alone. Unfortunately, the lesions are also highly FDG avid, with a mean SUVmax of approximately 4 to 5, limiting the role of FDG PET/CT in distinguishing them from malignancy.^5–7,11 Although not well studied using NaF PET, there are reports of increased uptake on this modality, even though the lesion is generally lytic.^13,24 Another lytic lesion that is often expansile and primarily exists within the posterior elements of the spine is an ABC. The spine can be involved in up to 20% of cases.^4,25 ABC is composed of blood-filled cavities within the bone, sometimes occurring within other primary lesions, most commonly GCT, followed by osteoblastoma and chondroblastoma.^4,25,26 They are expansile lytic lesions, often with associated cortical thinning. The characteristic finding on MR imaging or CT is multiple fluid-fluid levels within the hemorrhagic cavities. Unfortunately, telangiectatic osteosarcoma, an aggressive malignancy, can have a similar appearance on anatomic imaging. The 2 also cannot be reliably differentiated on FDG PET, because they are both often slightly hypermetabolic. ABCs typically demonstrate SUVmax ranging from 1 to 6, with a mean of approximately 3.^5,10

Another lytic lesion, eosinophilic granuloma, is found most often in children, usually occurring before age 30.²⁷ Lesions are often multiple and, when within the spine, can lead to partial or complete vertebral body collapse (vertebra plana). Due to high histiocyte content, lesions are usually FDG avid, with FDG PET having been found at least as sensitive as MR imaging in lesion detection, often used as the test of choice for screening and follow-up in multifocal disease.²⁸ Given their FDG avidity, FDG PET is of little utility in differentiating eosinophilic granuloma from malignancy. Similarly, despite being lytic, eosinophilic granuloma can also demonstrate substantial uptake of NaF.¹⁴

Osteoid osteoma is a benign mainly sclerotic lesion, which often presents as painful scoliosis in the spine. On CT, the lesion has a lucent nidus, often with a central sclerotic dot representing mineralized osteoid, with surrounding reactive sclerosis, usually within the posterior elements (Fig. 1A). Osteoblastoma, which is usually larger in size, by definition greater than 2 cm, also involves the posterior elements. Although osteoid osteomas usually regress, however, osteoblastomas grow and can be locally aggressive, often being confused with malignant lesions on imaging (Fig. 2). At times, osteoblastomas can also demonstrate malignant degeneration. Osteoblastomas are lytic and expansile with mixed sclerosis.^4,29 Both lesions demonstrate FDG uptake above background. Osteoid osteoma has a mean SUVmax between 2 and 3 whereas osteoblastoma has a minimally higher mean SUVmax slightly above 3.^5,30–32 Neither can be differentiated from a malignant lesion purely based on FDG uptake. Occasionally, osteoid osteomas are not active on FDG PET (Fig. 1B). In such situations, NaF PET may prove useful because these lesions have also been shown to be NaF avid (Fig. 1C).^18,33

Fig. 1.

Axial CT through the thoracic spine of a young male patient with symptomatic osteoid osteoma shows a sclerotic focus within the right pedicle of T5 (A, white arrow). The osteoid osteoma demonstrated no significant uptake on FDG PET (B). A coronal NaF PET, however, shows increased radiotracer uptake corresponding to the lesion (C, black arrow).

Fig. 2.

Axial CT (A) through the T4 vertebrae in a 58-year-old man shows an expansile lytic soft tissue mass within the body and posterior elements (white arrow). The mass was somewhat FDG avid on PET (B), similar to the aortic arch blood pool. Axial T2-weighted (C), sagittal precontrast T1-weighted (D), and axial postcontrast T1-weighted (E) images better define the mildly T2 hyperintense and T1 hypointense mass with diffuse enhancement, effacing the thecal sac and causing mass effect on the thoracic spinal cord. On surgical pathology, the lesion was found to represent an osteoblastoma.

Fibrous dysplasia is another lesion that has higher than background FDG avidity. Although fibrous dysplasia has a mean SUVmax of approximately 2, many lesions have a much higher standardized uptake value (SUV) and, therefore, cannot be differentiated from malignancies, such as metastatic disease, osteosarcoma, or chondrosarcoma based on FDG uptake.^5,7,32 Fortunately, its anatomic appearance is usually more characteristic of a benign entity than more aggressive appearing lesions, such as osteoblastoma and GCT. A more aggressive appearance, however, including secondary ABC formation and cortical breakthrough, can be seen on CT.³⁴ Although vertebral involvement of the disease is rare, it can be seen with polyostotic forms. On CT and radiography, there is often a classic ground-glass appearance to the osseous matrix, with many lesions being expansile and demonstrating lytic components and a sclerotic rim (Fig. 3).⁴ Given increased osteoblastic activity, fibrous dysplasia is also NaF avid, further potentially confusing it with more aggressive tumors on the basis of physiologic imaging alone.¹⁸

Fig. 3.

Axial CT (A) and coronal fused FDG PET/CT (B) images in a 38-year-old man with complaints of neck and shoulder pain as well as numbness of the right fourth and fifth digits demonstrate an expansile lytic lesion of the T1 vertebral body and adjacent right first rib. There is general preserved cortex with small foci of cortical breakthrough. There are some areas of ground-glass matrix within the rib lesion. The SUVmax, found within the rib, was 8. An additional lesion was seen in the left iliac bone (not shown). The rib lesion was biopsied, allowing for diagnosis of polyostotic fibrous dysplasia.

Osseous hemangiomas are common hamartomatous lesions of bone, which are composed of vessels interposed by normal marrow. They are the most common benign lesions of the spine, found frequently in middle-aged individuals.¹ Although most often clinically irrelevant, some can become expansile and compress on the spinal cord.³⁵ They are often simple to diagnose on both CT and MR imaging, demonstrating a classic corduroy appearance of vertical striations, with portions of fat equivalent signal on MR imaging. Expansile hemangiomas can appear more aggressive, sometimes being confused with Ewing sarcoma or other vascular lesions, such as hemangioblastoma. Fortunately, FDG PET can help distinguish these lesions because hemangiomas are classically not particularly FDG avid, with SUVmax less than 2 (Fig. 4).^5,7,10 Case reports of particularly FDG-avid lesions, however, have been described.³⁶ On NaF PET, hemangiomas can often demonstrate significant radiotracer uptake.¹⁸

Fig. 4.

Axial CT (A) through the level of T12 in a 37-year-old man demonstrates an expansile, osteolytic lesion involving the spinous process with mildly aggressive features and breakthrough into the posterior spinal canal. The lesion causes mass effect on the conus medullaris and demonstrates heterogeneous T2 hyperintensity (B) and avid T1 postcontrast enhancement (C). The mass demonstrates similar FDG avidity to the unaffected vertebral body seen on the same FDG PET axial image (D). On biopsy, the mass was found to represent an expansile hemangioma.

Osteochondroma, a cartilage-covered bony excrescence, rarely involves the posterior elements of the spine. Although these lesions are not FDG avid, PET has been used in the evaluation of suspected malignant/sarcomatous transformation of these lesions.

PET/CT AND PET/MR IMAGING IN THE DIAGNOSTIC MANAGEMENT OF MALIGNANT OSSEOUS SPINAL LESIONS

The most common malignant spinal lesions aside from metastatic disease are hematologic malignancies, such as multiple myeloma (MM) and lymphoma. MM is the most common primary bone malignancy in adults and solitary plasmacytoma is one of the most common primary osseous lesions of the vertebrae and pelvis.³⁷ As expected, FDG PET/CT has been shown superior to FDG PET alone in detecting diffuse spiny involvement of MM and in detecting lesions less than 10 mm (Fig. 5).³⁸ Furthermore, Fonti and colleagues³⁹ compared FDG PET/CT with technetium sestamibi (⁹⁹Tc-MIBI) and found that PET/CT was better at detecting focal lesions whereas ⁹⁹Tc-MIBI was superior in detecting multifocal disease. Sachpekidis and colleagues⁴⁰ similarly found that although NaF PET/CT can detect many myeloma lesions, FDG PET/CT is much more sensitive, detecting more than twice as many lesions. Although FDG PET/CT has good sensitivity for MM lesions, Breyer and colleagues⁴¹ have described that this sensitivity is greatly increased by MR imaging when small focal lesions are present. PET/MR imaging, therefore, is a promising modality that can help to assess the extent of lesions as well as better define neural compromise.

Fig. 5.

Sagittal FDG PET (A) and fused PET/CT (B) images of a 64-year-old woman with newly diagnosed MM demonstrates numerous FDG-avid spinal lesions (white arrows), most prominent in the lower thoracic and lumbar spine. The most FDG-avid lesion in L4 has an SUV of 6.2. Axial fused FDG PET/CT (C) and PET (D) images through the sacrum of a 61-year-old man with MM on maintenance therapy presenting for restaging show a large FDG-avid expansile plasmacytoma involving the left sacral ala as well as the left iliac bone (white arrows) with SUVmax of 3.6.

Primary bone lymphoma, also known as reticulum cell sarcoma or osteolymphoma, is less common than secondary involvement of bone from systemic lymphoma. More commonly, they are from non-Hodgkin histologic subtypes, with only 6% of cases diagnosed as Hodgkin lymphoma. Primary bone lymphoma is distinguished from other PET-avid bone lesions as extremely FDG avid, with SUV values up to and beyond 10 compared with approximately 5 in the majority of other primary bone lesions.⁴¹ Primary lymphoma of the spine is a rare entity and makes for difficult radiographic diagnosis because it may mimic other primary or metastatic bony spine lesions.⁷ Epidural extension can occur in approximately 14% of primary lymphomatous lesions involving the vertebrae.^42,43 Because timely radiotherapy and chemotherapy can improve prognosis substantially, especially compared with other osseous malignancies, prompt and accurate diagnosis is imperative.^42–44 Lei and colleagues⁴⁵ described the utility of FDG PET/CT in the diagnosis of primary spinal lymphoplasmacytic lymphoma that had destroyed much of the thoracic and lumbar vertebrae and extended into the epidural space, causing cord compression.

The possibility of chordoma should also be raised when an FDG-avid sacrococcygeal lesion is noted on PET/CT.^46,47 Chordoma typically presents in the fifth and sixth decades, most often involving the clivus or sacrum, with other vertebral involvement less common. On MR imaging, they are classically extremely T2 hyperintense and T1 hypointense. Chordomas have heterogeneous FDG uptake and cause local bony destruction in the sacrum when present, often extending into the sacral epidural space and causing nerve root compromise.⁴⁷

The most common nonhematologic primary vertebral malignancies are Ewing sarcoma and osteosarcoma. FDG PET/CT has been successfully used to stage and restage osteosarcoma and Ewing sarcoma.^48,49 Both tumors are generally FDG avid, allowing detection of distant meta-static disease (Fig. 6). In addition to providing information regarding distant metastatic disease, FDG PET/CT is an excellent prognosticating tool for these malignancies. FDG PET–based imaging can provide information on tumor response to therapy before any change is seen on anatomic imaging alone.⁵⁰ In the case of osteosarcoma, pretherapeutic and post-therapeutic FDG uptake, as well as change in uptake between the 2 scans, have all been shown to correlate with histologic tumor response, progression-free survival, and overall survival.^51–54 FDG PET is particularly adept at determining histologic response, because tumor necrosis can be differentiated from residual metabolically active tumor in osteosarcomas. A post-therapeutic SUV of less than 2.5 and reduction in metabolic tumor volume by 50% have been found to independently correlate with tumor necrosis.⁵³ Evaluating response to treatment through FDG PET/CT can help alter the chemotherapeutic regimen in tumors that seem resistant to therapy. Although FDG PET/CT also allows prognostication in Ewing sarcoma, the evidence is somewhat less strong than for osteosarcoma. Favorable histologic response in Ewing sarcoma, for example, correlates with greater than 90% reduction in metabolic tumor volume, compared with 50% in osteosarcoma (Fig. 7). Additionally, only post-therapy FDG uptake has been found to correlate significantly with progression-free survival, with a post-therapy SUV of 2.5 or lower correlating with a 4-year progression-free survival of 72%.⁵¹ Pretherapeutic uptake and change in up-take between pretherapy and post-therapy scans have not shown similar significance.

Fig. 6.

Sagittal contrast-enhanced T1-weighted MR imaging through the thoracic spine (A) of a 21-year-old man with biopsy-proved Ewing sarcoma of the right anterior iliac bone shows multiple enhancing metastases throughout the visualized vertebrae (white arrows). Fused coronal PET/CT images in the same patient (B) demonstrate these vertebral lesions, as well as multiple sacral and bilateral iliac lesions, as having higher FDG avidity than normal marrow (white arrows).

Fig. 7.

FDG PET/CT and separate MR imaging were obtained in a 13-year-old girl with history of Ewing sarcoma, status post treatment with chemotherapy and local radiation. Residual epidural soft tissue at the L5-S1 level is seen on axial unenhanced T1-weighted MR imaging (A, white arrow). Because of its stability on 1-year follow-up (previous images not shown) and lack of uptake on FDG PET/CT (B, white arrow), the findings likely represent post-treatment changes rather than recurrent/residual disease.

More recently, FDG PET/MR imaging has been used in an attempt to improve evaluation of therapy response in osteosarcoma. Byun and colleagues⁵⁵ obtained fused, separate acquisition PET/MR imaging images in 30 patients who had undergone 2 cycles of neoadjuvant chemotherapy for osteosarcoma. Although not focused on spine lesions, they found that this modality was also able to predict histologic response using both metabolic tumor volume and total lesion glycolysis. Specifically, in a study of 27 patients undergoing pretreatment and post-treatment sequential PET/MR imaging, both change in mean apparent diffusion coefficient and change in SUVmax was found to significantly correlate with histologic response. The combination of the 2 variables improved on the sensitivity, specificity, and accuracy of either value alone.⁵⁵ Again, using fused separate acquisition PET/MR imaging, Kong and colleagues³⁰ found that the percentage of tumor necrosis at the location of SUVmax correlated significantly with overall histologic response of the entire resected tumor to chemotherapy. Similar analyses attempted on PET/CT have not been as successful, presumably due to the inferior contrast differentiation between necrotic and enhancing tissue. Using hybrid PET/MR images, the same group has also found an additional benefit of the physiologic information provided by MR imaging in predicting histologic response.

As opposed to FDG PET, NaF PET evaluation of osteosarcoma has not been studied as extensively. The few reports available, however, show that NaF is potentially useful in this setting. Hoh and colleagues⁵³ first described osteosarcoma as an intensely NaF-avid osseous lesion. Both the primary lesion and pulmonary metastases have been shown NaF avid, with substantial reduction in NaF uptake after successful chemo-therapy.^20,56,57 The level of NaF avidity of osteosarcoma is such that NaF PET is more sensitive than FDG PET in the detection of pulmonary and soft tissue metastases from osteosarcoma.^58,59 Combined NaF and FDG PET has been shown in at least 1 documented case to alter clinical management. Specifically, Brunkhorst and colleagues⁶⁰ described the case of a 15-year-old girl whose treatment changed from curative intent to palliative intent after combined NaF and FDG PET revealed greater extent of disease than initially thought. Even less has been studied regarding the uptake of NaF in Ewing sarcoma, with only a few scattered cases showing substantial radiotracer avidity.⁶¹

SUMMARY

Familiarity with the PET findings of common and rare primary spine tumors is paramount for anyone who regularly reads PET studies for cancer staging. Unfortunately, there is much overlap between the FDG avidity of many benign and malignant primary osseous spinal lesions, making differentiation between the 2 entities based on physiologic imaging alone difficult. When combined with morphologic characteristics seen on CT or MR imaging, however, PET can help point the radiologist toward the correct diagnosis. PET/CT and PET/MR imaging are also of utmost importance in the follow-up of primary malignant processes of the spine, proving better than MR imaging or CT alone in assessing response to treatment. Although research on the use of alternative radiotracers, such as NaF, and alternative modalities, such as PET/MR imaging, in the diagnosis of primary spine tumors is sparse, the available data point to their utility beyond standard diagnostic techniques. As such, further research is necessary to explore and define the utility and limitations of PET/MR imaging and non-FDG PET radiotracers in the realm of spinal lesions (Boxes 1 and 2).

Box 1. Diagnostic criteria of primary osseous lesions on PET/CT and PET/MR imaging.

Cartilaginous lesions	Characteristic ring and arc matrix on CTand T2 hyperintense on MR imaging. SUVmax greater than 2 is a highly sensitive and specific cutoff for malignancy on FDG PET. NaF PET is not as specific.
Giant cell tumors	Lytic lesions, often with endosteal scalloping or even at times cortical destruction on CT. Highly FDG-avid and, therefore, not able to be distinguished from malignancy based on FDG PET.
ABC	Expansile lucent lesion on CT, often with hematocrit levels on MR imaging. Often FDG-avid and cannot be differentiated from malignancy on the basis of FDG PET.
Eosinophilic granuloma	Lucent lesion on CT found in children and can lead to vertebra plana. Both FDG and NaF avid.
Osteoid osteoma	Characteristic sclerotic lesion with lucent nidus and a central sclerotic dot on CT. Usually involves the posterior elements. Demonstrates surrounding T2 hyperintense marrow edema on MR imaging. Highly NaF avid with usually intermediate uptake on FDG PET, having an SUVmax of 2–3.
Osteoblastoma	Lytic expansile lesion of the posterior elements with mean SUVmax slightly >3. Highly NaF avid.
Fibrous dysplasia	Classic ground-glass matrix with usually benign morphologic characteristics on CT, including intact cortex. Variable FDG uptake, with mean SUVmax of 2.
Osseous hemangioma	Classic corduroy appearance on CT with fat intensity signal on MR imaging. Typically not FDG avid, with SUVmax similar or below background marrow. Often highly NaF avid.
Myeloma	Lucent lesion within the Marrow on CT. T2 hyperintensity, T1 hypointensity, and enhancement on MR imaging. Usually no cortical involvement unless expansile plasmacytoma. FDG has greater sensitivity than NaF PET.
Lymphoma	Lytic, sclerotic, or mixed lesion on CT. T2 hyperintensity, T1 hypointensity, and enhancement on MR imaging. Can be distinguished from other primary bone pathologies by its intense FDG avidity, often with SUVmax of 10 or greater.
Chordoma	Lytic lesion with cortical breakthrough on CT predominantly within the sacrum or clivus. Light-bulb bright signal on T2-weighted MR imaging. FDG uptake is heterogeneous and inconsistent.
Ewing sarcoma	Generally lytic lesion on CT with intense enhancement on MR imaging. Moderate to intense FDG avidity. Treatment response can be assessed using FDG PET.
Osteosarcoma	Mixed sclerotic and lytic lesion on CTwith aggressive periosteal reaction and bone formation. Avidly enhancing on MR imaging. Intensely NaF avid and moderately FDG avid. Treatment response can be assessed using both FDG PET and NaF PET.

Box 2. What the referring physician needs to know about PET-based imaging for primary spine tumors.

As a general rule, PET cannot definitively differentiate benign from malignant primary spine tumors of unknown histology. In general, more aggressive lesions have higher FDG uptake. There are only 3 primary spine tumors that nearly always have low FDG uptake and, therefore, can be confidently diagnosed as benign: hemangioma, enostosis, and benign cartilaginous lesions. Given the greater sensitivity and specificity of MR imaging, PET/MR imaging is superior to PET/CT in characterizing many primary osseous lesions. FDG PET/CT and PET/MR imaging are of greater utility than anatomic imaging alone in assessing treatment response in osteosarcoma and Ewing sarcoma. NaF PET and FDG PET can play complementary roles in the detection and diagnosis of primary spine tumors, with the former having greater overall data and especially useful for lytic lesions and the latter useful for sclerotic lesions.

KEY POINTS.

• One of the principal ways of distinguishing spine tumors is their location within the vertebrae, with lesions originating in the posterior elements primarily benign. With the exception of hemangiomas and enostoses, lesions originating in the vertebral bodies are primarily malignant.

• Although PET/CT and PET/MR imaging can be used to distinguish benign from malignant chondral lesions, their utility in distinguishing aggressive benign lesions from malignancies of other histologic categories is not as consistent.

• 18^F-sodium fluoride PET and fluorodeoxyglucose (FDG) PET can play complementary roles in the detection and diagnosis of primary spine tumors, with the latter having greater overall data and especially useful for lytic lesions and the former useful for sclerotic lesions.

• FDG PET/CT and PET/MR imaging are superior to anatomic imaging alone in predicting histologic response of osteosarcoma and Ewing sarcoma to treatment, helping to differentiate viable tumor from necrosis and sclerosis.