Primary bone tumors span an exceptionally wide biological range, from benign incidentalomas such as enchondromas to highly lethal sarcomas including osteosarcoma and Ewing sarcoma. While radiographs, CT, and MRI define local anatomy and morphology, they cannot directly interrogate tumor metabolism, blood flow, or receptor expression. Nuclear medicine fills this gap by providing whole-body functional imaging in a single examination, making it uniquely suited to detecting multifocal disease, guiding biopsy to the most aggressive tumor region, and measuring the metabolic consequence of treatment before surgery reveals histologic response.
The core diagnostic challenge: Primary bone sarcomas are individually rare, histologically heterogeneous, and frequently large at presentation. Sampling error during biopsy is a recognized problem: a needle placed in a necrotic or low-grade region of a mixed-grade tumor will underestimate the disease. Moreover, conventional imaging cannot reliably distinguish a benign enchondroma from a low-grade chondrosarcoma, a distinction that governs whether a patient undergoes observation, curettage, or wide surgical resection. These are precisely the questions nuclear medicine is positioned to help answer.
Scope of this 2025 review: Published in Diagnostics (MDPI) and authored by Lee, Joo, and Carsote, this review surveys the full breadth of nuclear medicine modalities applicable to primary bone tumors. It covers bone scintigraphy and SPECT/CT, [18F]FDG PET/CT, and emerging tracers including [18F]sodium fluoride ([18F]NaF) and [68Ga]Ga-FAPI, as well as the integration of radiomics and artificial intelligence. Critically, the review addresses both malignant bone tumors (osteosarcoma, Ewing sarcoma, chondrosarcoma) and benign entities that can mimic cancer on functional imaging (osteoid osteoma, osteoblastoma, chondroblastoma, aneurysmal bone cyst), providing a practical framework for interpreting nuclear medicine findings across the full diagnostic spectrum.
The review situates nuclear medicine not as a replacement for anatomic imaging but as a synergistic complement: the combination of MRI for local delineation and PET/CT for whole-body metabolic staging addresses clinical questions that neither modality can resolve alone. This integrative philosophy underlies all the clinical evidence summarized in subsequent sections.
Bone scintigraphy uses intravenously administered Technetium-99m ([99mTc])-labeled diphosphonate (most commonly [99mTc]Tc-MDP) that adsorbs onto hydroxyapatite in bone, reflecting osteoblastic activity and regional blood flow. Its principal strengths are high sensitivity for skeletal lesions and the ability to survey the entire skeleton in a single examination. The technique can detect changes as small as 5% in bone formation, while plain radiographs require 40-50% mineral loss before abnormalities become visible. However, planar scintigraphy suffers from limited spatial resolution and poor specificity: fractures, degenerative changes, infection, and inflammation all produce increased uptake indistinguishable from neoplasia on two-dimensional images.
SPECT/CT: The integration of single-photon emission computed tomography (SPECT) with diagnostic CT substantially improves diagnostic confidence by adding three-dimensional imaging that eliminates overlapping activity and provides precise anatomic localization. Modern hybrid SPECT/CT scanners equipped with cadmium zinc telluride (CZT) solid-state detectors achieve higher sensitivity and spatial resolution than conventional Anger camera designs. Combined with iterative reconstruction and resolution recovery algorithms, quantitative SPECT now permits measurement of tracer uptake as absolute activity concentrations or standardized uptake values (SUV), moving the technique from qualitative to quantitative imaging and bringing it closer in capability to PET/CT.
[18F]FDG PET/CT: [18F]FDG is a glucose analog that accumulates in proportion to glycolytic activity. It is phosphorylated inside cells and becomes metabolically trapped, with uptake expressed as the maximum standardized uptake value (SUVmax), defined as the highest voxel value within a lesion normalized to injected dose and body weight. PET is now performed exclusively with hybrid PET/CT scanners, enabling precise co-registration of metabolic and anatomic information. Compared with bone scintigraphy, [18F]FDG PET/CT is superior for detecting lytic or marrow-based metastases and simultaneously identifies osseous and extra-osseous disease. Its primary limitation is nonspecificity: inflammatory and infectious processes, as well as some benign tumors, can generate high SUVmax values that overlap with malignancy.
A three-phase bone scan (perfusion, blood-pool, and delayed static imaging) adds incremental specificity by characterizing the vascularity pattern of a lesion. For suspected primary bone tumors where malignant transformation or skip lesions are clinical concerns, the combination of a three-phase scan with SPECT/CT at the primary site represents the minimum standard functional imaging workup before [18F]FDG PET/CT is applied.
Osteosarcoma is the most common primary malignant bone tumor, with a bimodal age distribution peaking in adolescents aged 10-14 years and in adults older than 60. It most frequently arises in the metaphysis of long bones in younger patients. Standard management involves neoadjuvant chemotherapy followed by surgical resection, and the single most powerful prognostic factor is the degree of tumor necrosis in the resected specimen: good response (defined as at least 90% necrosis) is associated with substantially better outcomes than poor response (less than 90% necrosis). Nuclear medicine informs every phase of this management pathway.
Staging: Osteosarcomas are intensely [18F]FDG-avid tumors, and PET/CT is now a cornerstone of the initial staging workup. Multiple studies have demonstrated its superiority over conventional bone scintigraphy for detecting skeletal metastases, an advantage particularly evident for lytic metastases near active growth plates where physiologic bone scan uptake can mask adjacent lesions. The whole-body survey capability of a single PET/CT acquisition detects osseous and extra-osseous metastases simultaneously, informing surgical planning and systemic therapy decisions.
Biopsy guidance: Osteosarcomas are often large and heterogeneous, containing areas of variable histologic grade, necrosis, and hemorrhage. Sampling error during biopsy can lead to underestimation of the tumor's true grade. [18F]FDG PET/CT generates a metabolic map of the entire tumor mass, identifying the most glycolytically active and typically highest-grade components. Directing biopsy toward the SUVmax region improves diagnostic yield and avoids sampling of necrotic or low-grade areas that would not be representative of the lesion's overall biology.
Treatment response assessment: A significant reduction in SUVmax following neoadjuvant chemotherapy is a robust indicator of a good pathologic response (at least 90% tumor necrosis in the surgical specimen). This non-invasive metric can predict prognosis and guide decisions about chemotherapy regimen modification or surgical timing. For detecting recurrence post-treatment, a meta-analysis pooling data from seven studies reported composite sensitivity of 91% (95% CI: 81-96%) and specificity of 93% (95% CI: 87-97%) for [18F]FDG PET/CT, establishing it as the surveillance standard.
Ewing sarcoma is an aggressive small round blue cell tumor of bone and soft tissue, primarily affecting children and young adults, with common sites including the pelvis, femur, humerus, ribs, and spine. Its biology creates a specific imaging challenge: Ewing sarcoma metastases are often purely lytic and reside within the bone marrow without provoking significant reactive osteoblastic activity. Bone scintigraphy, which depends on detecting osteoblastic reactions to identify lesions, therefore performs poorly. Published studies document a stark sensitivity gap: one study reported PET/CT sensitivity of 88% for detecting bone metastases from Ewing sarcoma compared to only 37% for conventional imaging including bone scan, and another found 88% for PET/CT versus 50% for bone scan. Based on this evidence, [18F]FDG PET/CT is now the preferred nuclear medicine modality for systemic staging of Ewing sarcoma.
Chondrosarcoma and the grading problem: Chondrosarcoma presents a fundamentally different diagnostic challenge. Rather than staging metastases, the critical question is distinguishing enchondromas and atypical cartilaginous tumors (ACT, also called grade 1 chondrosarcoma of the appendicular skeleton) from higher-grade chondrosarcomas. This distinction determines management: benign enchondromas and many ACTs can be observed or treated with curettage given their limited metastatic potential, whereas grade 2 and 3 chondrosarcomas require wide surgical resection.
[18F]FDG as a grade surrogate: There is a well-established correlation between SUVmax on [18F]FDG PET and histologic grade in cartilaginous tumors. Benign enchondromas and ACTs are metabolically indolent, demonstrating low uptake. Grade 2 and 3 chondrosarcomas are more glycolytically active and show significantly higher uptake. However, no universally accepted SUVmax threshold has been established, and overlap between groups limits precision. Biopsy itself is problematic in cartilaginous tumors because histologic heterogeneity means a needle placed in a low-grade area can miss a high-grade focus elsewhere, making [18F]FDG PET/CT a valuable adjunct for identifying where to biopsy.
Bone SPECT/CT as an alternative: Where PET is unavailable, bone SPECT/CT offers a complementary functional approach. In a 2020 study of long-bone cartilage tumors, ACTs had significantly higher SPECT-derived SUVmean and SUVmax than enchondromas, yielding approximately 86% sensitivity and 75% specificity for differentiation. Focal regions of markedly elevated uptake on bone SPECT/CT can also guide biopsy toward the highest-grade focus of a heterogeneous cartilaginous lesion.
A critical practical problem in musculoskeletal oncology is that several benign bone tumors are intensely metabolically active and produce high SUVmax values on [18F]FDG PET/CT that fall within or exceed the range of malignant sarcomas. Understanding the specific nuclear medicine signature of each benign entity is therefore essential for avoiding misdiagnosis.
Osteoid osteoma: This small, benign bone-forming tumor classically presents in young males with nocturnal pain relieved by NSAIDs. On three-phase bone scintigraphy, it shows the pathognomonic "double-density sign": a central hot spot of intense uptake corresponding to the vascular nidus, surrounded by a less intense zone of reactive sclerosis. Bone scintigraphy has sensitivity approaching 100% for osteoid osteoma detection. SPECT/CT adds precise localization particularly valuable for small or intra-articular lesions. [18F]FDG PET/CT typically shows high metabolic activity in the nidus but some lesions demonstrate no uptake, creating variability.
Osteoblastoma: A larger and more aggressive counterpart of osteoid osteoma, osteoblastoma demonstrates markedly elevated [18F]FDG uptake with wide reported ranges: average SUVmax around 3.2 in some series, individual cases reaching 6.2, and some spinal lesions exhibiting SUVmax as high as 16.0. This level of avidity exceeds that of many malignant sarcomas. Differentiating osteoblastoma from malignancy cannot be accomplished based on SUVmax alone; correlation with the characteristic radiographic appearance and location is essential.
Chondroblastoma and aneurysmal bone cyst (ABC): Chondroblastoma is a rare epiphyseal tumor in adolescents that demonstrates intense uptake on both bone scintigraphy and [18F]FDG PET/CT. Its diagnosis rests on context: epiphyseal location in a skeletally immature patient combined with a lytic lesion with thin sclerotic rim on radiograph should suggest chondroblastoma despite the high metabolic activity. ABC produces the characteristic "doughnut sign" on bone scintigraphy: a ring of intense peripheral uptake in the hypervascular fibrous septa surrounding a central photopenic core corresponding to the blood-filled cavity. Telangiectatic osteosarcoma can produce a similar pattern, requiring careful correlation with anatomic imaging.
Radiomics applies computational analysis to extract high-dimensional quantitative features from medical images, capturing textural, morphological, and heterogeneity characteristics that are invisible to the human eye. In the context of primary bone tumors, early [18F]FDG PET/CT-based radiomic studies have demonstrated that texture-derived features can complement conventional SUV metrics, enhancing response prediction and prognostic assessment in sarcomas beyond what SUVmax alone provides. The underlying rationale is that intratumoral metabolic heterogeneity, expressed through radiomic texture features, reflects biological heterogeneity in ways that single scalar measurements do not capture.
SPECT/CT radiomics: Radiomics applied specifically to SPECT/CT imaging has shown promise for differentiating malignant from benign cartilaginous tumors. In a study of long-bone cartilage tumors, Yoon et al. identified zone-length non-uniformity (ZLNUGLZLM), a gray-level zone matrix-derived textural feature, as an independent predictor of ACT diagnosis. This radiomic feature achieved reported sensitivities of 83-85% and specificities of 58-91% across validation cohorts, suggesting that radiomic modeling of quantitative SPECT/CT data provides incremental discriminatory value over SUVmax alone for the benign-versus-low-grade-malignant distinction in cartilage tumors.
AI for bone lesion characterization: A systematic review by Ong et al. evaluated AI techniques for differentiating benign from malignant bone lesions. PET/CT-based AI studies achieved accuracies of 0.74-0.88, sensitivities of 0.84-0.90, and specificities of 0.74-0.85, with AUC values ranging from 0.76 to 0.95. A large multicenter study by von Schacky et al. involving 880 patients developed machine learning models based on radiomic features extracted from radiographs combined with demographic information. Their best-performing artificial neural network achieved an AUC of 0.90 on an external validation set, with 75% accuracy, 90% sensitivity, and 68% specificity. Importantly, this performance was lower than expert musculoskeletal radiologists but higher than radiology residents, positioning AI as a decision-support tool for less experienced clinicians rather than a replacement for subspecialty expertise.
Current limitations: The review explicitly notes that no dedicated AI studies have specifically addressed nuclear medicine imaging of primary bone tumors as of 2025, representing a significant and actionable gap. Existing studies in adjacent areas are limited by small sample sizes, retrospective designs, and lack of multicenter validation. Standardization of acquisition protocols and harmonization of radiomic feature extraction pipelines across institutions will be prerequisites for any clinically deployable AI tool in this domain.
[18F]sodium fluoride ([18F]NaF) is a PET radiotracer targeting bone metabolism through a mechanism analogous to [99mTc]Tc-MDP: the [18F]fluoride ion exchanges with hydroxyl groups in hydroxyapatite to form fluorapatite. Unlike bone scintigraphy, however, [18F]NaF benefits from the superior pharmacokinetics of PET tracers, higher spatial resolution, and quantitative capabilities. Multiple studies and meta-analyses have documented significantly higher sensitivity and diagnostic accuracy for [18F]NaF PET/CT compared to conventional planar bone scintigraphy and SPECT for detecting osseous metastases, making it a more powerful bone turnover imaging tool.
[18F]NaF in sarcoma: In osteosarcoma specifically, [18F]NaF PET/CT has revealed rare metastatic sites and has been applied for therapy monitoring. The NAFCIST (NaF PET Response Criteria in Solid Tumors) criteria were developed for assessing response in patients receiving radium-223 therapy ([223Ra]RaCl2). In early studies, NAFCIST correlated with biomarkers and overall survival, whereas [18F]FDG-based PERCIST criteria did not in the same cohorts, suggesting [18F]NaF may provide complementary prognostic information that [18F]FDG misses for bone-forming tumors.
[68Ga]Ga-FAPI: Fibroblast activation protein (FAP) is a type II serine protease overexpressed on cancer-associated fibroblasts and mesenchymal tumor cells in the stroma of sarcomas. [68Ga]Ga-FAPI tracers target this FAP-rich tumor microenvironment, yielding exceptionally high tumor-to-background contrast with minimal physiologic uptake, providing superior lesion conspicuity within the musculoskeletal system. Lanzafame et al. reported higher SUVmax with [68Ga]Ga-FAPI-46 than [18F]FDG (10.4 plus or minus 8.5 versus 7.0 plus or minus 4.5, p = 0.01). Sakir et al. found improved detection of bone and hepatic metastases with [68Ga]Ga-FAPI PET despite [18F]FDG identifying more total lesions, suggesting complementary rather than competing performance profiles.
FAPI as a theranostic platform: Beyond diagnostics, [68Ga]Ga-FAPI serves as the imaging component of a theranostic pair, with early therapeutic experience using [90Y]Y-FAPI-46 and [177Lu]Lu-FAPI-2286 demonstrating favorable safety profiles and preliminary efficacy in advanced sarcomas. However, variable FAP expression across sarcoma histotypes and uptake in inflammatory tissue remain limitations. Larger multicenter trials are required to validate diagnostic, prognostic, and therapeutic roles and to standardize quantitative parameters.
Radiomics and AI limitations: Despite promising results, current radiomics and AI studies in bone tumor nuclear medicine are universally constrained by small sample sizes, single-center retrospective designs, and absence of multicenter validation. The review explicitly identifies the complete lack of dedicated AI studies applied to nuclear medicine imaging of primary bone tumors as a critical gap. For quantitative SPECT specifically, evidence on SUV measurement utility in distinguishing benign from malignant primary bone tumors is limited to small single-center cohorts, and no multicenter validation has been performed. Standardization of acquisition protocols, scanner calibration, and radiomic feature extraction pipelines must precede any attempt at multicenter model training or deployment.
FAPI and [18F]NaF validation needs: Both novel tracers demonstrate clear biological rationale and promising early data but require larger multicenter trials before their diagnostic and prognostic roles can be sufficiently characterized. Variable FAP expression across sarcoma histotypes means [68Ga]Ga-FAPI will not perform uniformly across all bone sarcoma subtypes. [18F]NaF for osteosarcoma response assessment requires validation of NAFCIST criteria in larger, prospectively designed cohorts before the criteria can be adopted as a standard response assessment tool.
Quantitative SPECT as a growth area: The maturation of quantitative SPECT technology, including CZT detectors, resolution recovery reconstruction, and system-specific calibration enabling absolute activity quantification, positions SPECT/CT to provide PET-like quantitative capability at lower cost and broader availability. Early studies show quantitative SPECT/CT improves lesion characterization and reader agreement for bone metastases, with diagnostic accuracy often exceeding conventional qualitative reads. Expanding this quantitative framework to primary bone tumor differentiation, particularly for the enchondroma-versus-chondrosarcoma problem, is a priority future application.
Multimodal integration: The most clinically meaningful future direction is systematic multimodal integration: combining [18F]FDG PET metabolic data with [18F]NaF bone turnover measurements, [68Ga]Ga-FAPI stromal activity, quantitative SPECT radiomic features, MRI morphological and diffusion data, and clinical variables into unified predictive frameworks. Machine learning models that fuse these data streams have the potential to provide non-invasive tumor grading, response assessment, and recurrence detection with a precision that no single modality can achieve alone. Reaching this potential requires prospective multicenter data collection and collaborative consortium efforts analogous to those that advanced AI in other solid tumor types.