The skeleton is a common site of cancer metastasis. Notably high incidences of bone lesions are found for breast, prostate, and renal carcinoma. Malignant bone tumors result in significant patient morbidity. Identification of these lesions is a critical step to accurately stratify patients, guide treatment course, monitor disease progression, and evaluate response to therapy. Diagnosis of cancer in the skeleton typically relies on indirect bone-targeted radiotracer uptake at sites of active bone remodeling. In this manuscript, we discuss established and emerging tools and techniques for detection of bone lesions, quantification of skeletal tumor burden, and current clinical challenges.
The treatment of a primary solid tumors may involve surgery, radiation therapy, chemotherapy, or a combination of these approaches. Treatment effectiveness and resulting survival rates are highest when the disease is still localized to the initial site. Metastatic disease, in which a malignancy has begun to spread to sites throughout the body, has more limited therapeutic options and poorer outcome. While it is difficult to generalize for the wide range of neoplastic diseases that are broadly labeled as cancer, there are several common patterns of metastatic progression. These can involve locoregional growth, spread through the lymph nodes, seeding of distant organs, and invasion of the skeleton.
The metastatic colonization of the skeleton represents the lethal form of several commonly diagnosed solid carcinomas including breast, prostate, thyroid, renal, and lung cancer. As such, it is a focus of basic biological and clinical research. Critical biological questions, such as why cancer cells have a tropism for bone, and how they are able to flourish in such a distinct microenvironment from their original soft-tissue sites, remain largely unanswered. From a clinical standpoint, detection and monitoring of bone metastasis is a key aspect of patient management in advanced disease, and will be the focus of this perspective.
Incidence and impact
The osteotropic nature of cell dissemination from many cancers results in a large cohort number of patients with skeletal tumor burden. Estimates of approximately one-third of renal cell carcinoma patients,1,2 as many as 70% of prostate and breast cancer patients,3 and a large number of patients with lung,4,5 skin,6 and thyroid cancer7 are affected each year. Recent assessments place the number of patients with metastatic bone disease is approximately 300 000∼400 000 in the United States alone.8,9
The presence and extent of bone metastases are associated with poor overall outcome in metastatic prostate, breast, kidney, and thyroid carcinoma.10,
It follows that risk stratification and patient management are intimately dependent on knowledge of the presence of bone metastasis. Clinical chemistry provides markers of skeletal metabolism that can be used for monitoring or to complement other diagnostic modalities. However, N-telopeptide, alkaline phosphatase, and urine deoxypyridinoline are hampered by renal factors and are less sensitive than direct radiological evaluation of the skeleton using X-ray computed tomography (CT), magnetic resonance (MR), or nuclear medicine (NM) approaches. Therefore, imaging of bone lesions has become a mainstay of clinical radiological practice.
Imaging skeletal activity
Tumor cell dissemination to the bone is a complex process that involves reciprocal interplay between cancer cells, cells in the surrounding microenvironment, and the stroma itself.17 Radiographic detection of cancer metastasis to the bone is most commonly performed by identifying sites of active bone remodeling. Thus, we commonly define the bone lesions that arise as a function of the cancer/stroma interaction rather than detection of the cancer cells themselves. Chemical and physical stimuli released by metastatic foci which lead to bone deposition (as is common in prostate cancer) or resorption (as in breast cancer) have gross effects which can be visualized using several clinical imaging modalities. These methods either visualize aberrant tissue morphology (as in CT and MR) or incorporation of radiotracers in the remodeling bone matrix.
The bone scan
The dominant approach to detect metastatic involvement of the skeleton is by a NM radionuclide bone scan (RNB) using 99mTechnetium-radiolabeled bisphosphonates. Patients are scanned using a planar gamma (γ)-camera, which acquires the whole-body spatial distribution of the 99mTc-tracer via the 140 keV γ-ray produced upon its radioactive decay. Scatter and absorption of these intermediate energy photons necessitates the collection of both anterior and posterior planar scans, which can subsequently be evaluated by a trained radiologist to identify lesions (Figure 1).
The 99mTc radioisotope has been evaluated as a medical radiotracer for over half a century.18 It is the most widely used radioisotope for medical imaging because of its ease of on-site production, facile radiolabeling chemistry (vide infra), favorable half-life, and image quality. 99mTc can be produced in high quantity and purity from a 99Molybdenum (99Mo)/99mTechnetium (99mTc) generator. Here, the parent 99Mo is bound to a chromatography column and decays to 99mTc. This daughter radionuclide is mobilized with saline to produce 99mTc pertechnetate (99mTcO4−) while the 99Mo parent remains immobilized. This system obviates the need for expensive equipment such as a cyclotron (generally only found at academic hospitals) and the generator can be eluted every several hours. While concern has been mounting from the forecasted shortage of the parent 99Mo (as older research reactors which produce the isotope are decommissioned), novel cyclotron production for direct 99mTc are expected to fill any future requirements.19
Several phosphate compounds have been used to complex 99mTcO4− to form 99mTc-diphosphates (Figure 2). Coordination of the radiometal with the phosphates is facile and rapid in the presence of the reducing agent stannous chloride at neutral pH. The most commonly used phosphate in the United States is medronic acid to form 99mTc-methylene diphosphonate (99mTc-MDP).20 Selection of this tracer is a function of high bone uptake, relatively quick clearance from background tissues and in vivo stability (both in plasma and after binding bone).21
Three-dimensional radionuclide distribution can be realized using single-photon emission computed tomography (SPECT). This approach rotates a number of detectors around the patient, again collecting the gamma emissions, which are reconstructed to form a volumetric distribution of the tracer; as planar X-ray is to CT, γ-camera planar imaging is to SPECT. SPECT is not generally performed as a stand-alone diagnostic tool, but may be indicated following a planar RNB to clarify foci of uptake (Figure 3).
All diphosphonates have a high affinity for bone mineral. Unlabeled (non-radioactive) molecules are widely used therapeutically to combat low bone density leading to osteoporotic fracture or hypercalcemia. Radiolabeled diphosphonates are given at sub-therapeutic mass doses that are assumed to function through a similar mechanism. The accumulation of these radiolabeled complexes to sites of active bone remodeling are ascribed to the coordination of the phosphate groups to the calcium present in the hydroxyapatite of bone (Figure 2d).
Uptake of these compounds appears to be multifactorial. The proliferating cells at bone metastases encourage increased vascularization and therefore increased blood flow to site. This can alter the distribution of the bone-targeted radiotracer. In elegant experiments using blood flow tracers (including radioactive microspheres and blood-dissolved radioactive gas), it has been shown that some dependence on vascular delivery rates exists. However, extraction from the vessels is not the sole component to govern uptake in both normal and newly diseased bone as diffusion and chemical adsorption play a role.22,
Each 99mTc-diphosphonate possesses slightly different pharmacokinetics with respect to bone uptake and washout from background tissues26 but can generally be used interchangeably. A general protocol sees patient imaging at approximately 3 hours after intravenous administration of approximately 600 MBq of the tracer. It should also be noted that in addition to technical artifacts, renal function, medication, and metabolic status of the patient can strongly affect scan quality and outcome.27
Sodium fluoride positron emission tomography
An alternative radiotracer for imaging of the skeletal component of bone metastases is radioactive fluoride (18F) in the form of sodium fluoride (18F-NaF). First proposed as a RNB agent using γ-camera imaging in the 1960s, fluoride ion imaging was approved by the Food and Drug Administration in 1972. It was superseded by the widespread availability and more favorable emission profile of 99mTc-tracers for standard sodium iodide γ-camera technology. However, the development and availability of positron emission tomography (PET) led to the rediscovery of 18F imaging for bone scanning.28
18F decays primarily by positron emission. This positron undergoes an annihilation event with an electron to produce two 511 keV γ-rays which travel in opposite directions (180° from each other). Capture of dual γ-ray emissions which travel in opposite directions forms the basis for PET (Figure 4).29 Here, a ring of detectors is passed over a patient injected with a positron-emitting radionuclide. Lines of response are recorded when detector elements on opposite sides of the ring are struck at the same time, and the sum of these gated events can be used to reconstruct a detailed map of the three-dimensional location of decay events.
The advantages of PET include its inherently quantitative nature, enabling the absolute determination of activity per unit volume of tissue with and a higher resolution than SPECT reconstructions of 99mTc-diphosphonate uptake. The faster clearance of the fluoride allows for scanning earlier after injection than with 99mTc-MDP, increasing clinical workflow and minimizing inconvenience to the patient. Additionally, PET is approximately a log-order more sensitive for radioactive detection than planar imaging or SPECT. This feature permits lower activity amounts to be administered in order to detect pathological sites and faster scan times as compared to SPECT. Importantly, PET identifies more bone metastases than the planar RNB30 and 18F-PET/CT is able to detect more lesions than SPECT/CT.31
Fluoride chemistry and bone uptake
A cyclotron is used to produce (18F) by proton irradiation of 18O-water.32 The fluoride ion is physicochemically adsorbed on an anion exchange resin cartridge, and can be subsequently eluted in sterile saline to form 18F-NaF. The solution is sterile filtered and calibrated for patient administration. The relatively short half-life of the radionuclide (109.7 minutes) requires that processing be performed rapidly.33 While an on-site cyclotron is expedient, commercial services are able to produce the compound at off-site locations and deliver as needed for clinical scans.
Similar to 99mTc-bisphosphonate compounds, uptake of 18F in regions of upregulated bone activity is dependent on several parameters.34,35 After intravenous administration, 18F is rapidly cleared from the plasma to the bone or renally excreted.34 The majority of bone uptake occurs on the first pass of the ion through the circulation.36 As such, blood flow characteristics are a dominant factor in 18F-NaF uptake.37 At metabolically active bone surfaces, the radiofluoride is able to exchange with hydroxyl ions in the hydroxyapatite crystal to form fluorapatite and fluorohydroxyapatite.38,39 The increased mineralizing surface area at bone metastasis (as well as fracture and arthritis) result in increased 18F-fluoride uptake.40
Tracer uptake at sites of bone remodeling has critical value in detection and monitoring of bone lesions. However, structural anatomical imaging technologies such as CT and MR imaging often provide additional information to aid diagnosis. The calcified tissue contrast of CT can be used to rule out suspicious uptake of non-metastatic origin (such as increased tracer uptake at microfractures and benign degeneration).41 On the other hand, the excellent soft-tissue contrast afforded by MR imaging is also useful in that it may lead to identification of additional soft-tissue metastases outside of the skeleton.
Within the bone, MR is able to sensitively detect replacement of high-signal fat and marrow cellular material by incipient metastases.42 Thus, this modality has the potential to visualize metastasis in the marrow cavity and early in the bone growth stage, prior to gross bone remodeling detectable by RNB and CT. Nevertheless, MR imaging results may again capture secondary changes in bone and bone marrow, rather than directly visualize the cancer foci itself. Early small reports on the increased sensitivity and specificity of MR and specifically whole-body MR over RNB may point to greater use of this modality for metastatic bone involvement.43,44 These reports indicate anatomically dependent sensitivity for each modality (more rib lesions by RNB and more spinal lesions by MR) suggesting that multiple modalities might best be used in tandem. However, many centers are still in the feasibility stage for comparison of whole-body MR scanning to established techniques.45
Direct visualization of malignant cells rather than secondary bone events may be facilitated using NM techniques that involve the radiolabeling of targeted ligands, be they small molecules or full size antibodies. Such radiotracers facilitate detection of carcinoma through dysregulated cancer cell metabolism or aberrant cell surface receptor expression. The oncological PET tracer 18F-fluorodeoxyglucose (18F-FDG) is the most widely used agent by a significant margin.
Malignantly transformed cells have altered metabolic requirements to furnish their high proliferation rate and altered glycolysis pathways, referred to as the Warburg effect.46,
Metabolic substrates beyond glucose have also been evaluated, most notably Carbon-11 (11C-) and 18F-radiolabeled acetate and choline. These agents identify cancer based upon upregulated lipid synthesis and choline kinase activity, respectively. Synthesis of these compounds is more challenging than commercially available 18F-FDG, further complicated by the 20-minute half-life of the carbon isotope. Due to its reduced glycolytic phenotype, prostate cancer has been the primary application for these agents,58 but adoption of these techniques is not currently widespread.
In addition, there are a plethora of targeting moieties that can be functionalized with single-photon and/or positron-emitting radionuclides to identify and monitor cancer. Enumerating the different agents is well beyond the scope of this article, however only a small minority are in common clinical use for detection of bone metastatic disease (Figure 6). Radioiodine (typically Iodine-131 and/or Iodine-123) is widely used for identification of differentiated thyroid metastases (which continue to accumulate iodine similar to the healthy thyroid). Peptide-based somatostatin analogs are often relied on to target neuroendocrine tumors, which overexpress somatostatin receptor type 2. These peptides are labeled with radiometals, most commonly Indium-111 (111In) for planar imaging and SPECT,59 and Gallium-68 for PET.60 Finally, radioiodine-labeled MIBG has been used to identify metastatic neuroblastoma in children and malignant pheochromocytoma (Figure 7).
Many agents are in the validation stage for lesion detection in early stage clinical trials. For prostate cancer, radiolabeled urea-based small molecules and antibodies targeting prostate-specific membrane antigen (PSMA) have been widely explored.61,
Analysis and quantitation
Qualitative evaluation of whole-body RNB or structural imaging is a relatively straightforward task that requires radiographic identification of abnormal focal uptake. Quantification of disease burden using these methods is considerably more difficult. Despite the challenge, quantitative assessment opens the opportunity to stratify patients’ disease burden, monitor progression and response to therapy, and evaluate novel cancer therapies.67
Ideally, rigorous and universally accepted analysis techniques are applied across patient scans to establish imaging as a biomarker to guide all relevant stages of patient assessment and management.68 A biomarker is a factor that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.69 Such an approach has been achieved with the response evaluation criteria in solid tumors (RECIST) criteria for solid tumors established in 2000.70 In this approach, the longest axis of soft-tissue tumor burden is assessed by CT or MR and changes in this value respective to the baseline measurement define categories of outcomes. This criterion defines overall response and progression-free survival in clinical trials as an accepted imaging biomarker for the Food and Drug Administration.
However, a single one-dimensional measurement of burden is not applicable in the context of metastatic disease, and less so for bone disseminated disease. Several of the imaging methods discussed above have been evaluated for their use as imaging biomarkers. The focus for tomographic RNB as a biomarker has been has been on evaluating intensity of tracer uptake. PET is inherently quantitative, as one directly measures activity per unit volume. 18F-NaF PET uptake at a lesion is most commonly quantified as standardized uptake value (SUV), defined as the tissue concentration of tracer as measured by a PET scanner divided by the activity injected divided by body weight. The uptake value is represented by pixel or voxel intensity value in the defined volume of uptake, which is then converted into the activity concentration.
Quantification of 99mTc-diphosphonate bone scans is more intricate. True quantification of the intensity of uptake on a planar RNB is not possible as images are formed as a projection through the patient for a given acquisition (anterior or posterior). Enumeration of the number of lesions has been historically used to evaluate planar γ-camera scans. A sophistication over this approach is the Bone Scan Index (BSI), which is an algorithm for quantitative analysis of skeletal metastasis that represents the tumor burden in a bone scan as a percentage of the total skeletal mass.71 Although the method has shown to be of important clinical significance, popularity has been hampered by the manual labor required. Automated image analysis software has been shown to significantly decrease the time required.72 These methods are in development and still require a physician to supervise the scan evaluation. As well, automated methods tend to underestimate BSI scores in patients with extensive bone disease.73
Although both BSI and quantification (SUV) of 18F PET show robust reproducibility when analyzing the same image, the intra-patient variability is currently unknown (i.e. the variability of a repeated measurement after reinjection of the compound within the same week). Future prospective clinical trials are needed to fully validate these methods as clinical applicable biomarkers.
Current challenges in identification of bone metastasis
Inherent biological and physical factors limit the effectiveness of bone metastasis imaging technologies. For the NM approaches, the complexities in analysis of patient scans arise predominantly from the difficulty in distinguishing uptake due to bone activity decoupled from cancer cell activity. Complications include benign pathologies that mimic the signal of metastatic disease, bone flare, and superscans. Anatomical imaging techniques which may detect smaller deposits of disease prior to macroscopic bone activity are also confounded by asymptomatic benign findings which often require biopsy to confirm. Together, these issues revolve around the lack of specificity of these techniques despite high sensitivity for detection of bone metabolism, which may not be associated with oncological phenomenon.
Cancer patients suspected of having bone disease are often elderly and present with age-associated benign skeletal degeneration. Uptake of bone-targeted radionuclides is also increased at sites of benign bone degeneration and joint inflammation.74,
The flare phenomenon refers to the increase in radiotracer uptake at previously diagnosed,80 or new previously undetected lesions,81 soon after initiating therapy. Increased uptake is transient, but variable, lasting between three and six months. The observation of the flare phenomenon occurs often but has been historically under-investigated.82 It is thought that the increase is a response to bone deposition and repair at treated sites. This increase in signal makes the distinction between progression of bone lesions (or increased metastatic spread) and successful treatment considerably difficult.
Theoretically, if all cancer burden in the bone was ablated, bone healing would continue for some time before returning to a baseline state. In such a hypothetical case, one might detect an increase in the number and intensity of bone lesions despite the lack of remaining cancer cells. Patient management in such situation is problematic as medical or radiation oncologists may be unsure if the current treatment is effective or not. Flare has been observed following systemic endocrine therapy83,
Lesions are most often identified as individual and distinct areas of intense uptake. When metastases are closely spaced, or as the area of skeletal involvement overlaps over time, it becomes difficult to identify individual sites. When the entire skeleton presents with elevated uptake, it becomes impossible to identify, enumerate or quantify the intensity of individual lesions. This so-called superscan pattern is found in advanced metastatic bone disease with diffuse activity throughout much or all of the skeleton (Figure 8). The likelihood of a superscan is increased in late stage and elderly patients with compromised kidney function, as this population is predisposed to bone metabolic disorders such as hyperthyroidism and fibrous dysplasia.90,91
Pitfalls in monitoring
RNB and structural imaging are important tools to follow disease progression and response to therapy.92 These techniques are also valuable in order to assess the potential of novel drugs – particularly as imaging response is an accepted outcome for early stage clinical trials.93 A pitfall with monitoring of bone activity is that pleotropic drug effects on bone remodeling rather than cancer cells may be deceptive. An illustrative example is cabozantinib, a tyrosine kinase inhibitor that primarily affects hepatocyte growth factor receptor and vascular endothelial growth factor receptor. The drug is approved for treatment of medullary thyroid cancer and has shown impressive early phase results in the metastatic setting in several solid cancers. Interestingly, rapid resolution of bone lesion uptake by planar RNB was seen in advanced prostate cancer patients, with significant palliative effect.94 Yet, the drug did not demonstrate a statistically significant benefit for overall survival.95 This dichotomy may be explained by effects on bone turnover rather than significant anti-cancer impact, as suggested by resolution of 18F-PET foci at sites of tumor-free fracture in a preclinical imaging study.96
Additional issues include scan duration, resolution, and artifactual uptake. From a pharmacokinetic standpoint, 99mTc-bisphosphonates must be imaged several hours after administration to accommodate soft-tissue, serum and renal clearance of non-bone bound agent. This inconveniences the patient and requires administration of greater activity such that sufficient signal is present at the later imaging time point (thereby increasing radioactive exposure to patient and staff). Planar scintigraphy has limited resolution. Longer on-camera acquisition times are required for SPECT scans. The ability to resolve smaller sites is achieved by PET, with the trade-off of greater cost. The 18F-NaF clearance is faster and imaging can be accomplished as early as tens of minutes after administration. However the dosimetry of the positron emitting 18F does not have an advantage over 99mTc compounds.
Reading of RNB can be complicated by soft-tissue uptake. While limited to a fraction of scans (1%–2%), uptake artifacts in the muscle and tissues often indicates the presence of pathologies such as blood flow defects, renal insufficiency, splenic uptake due to sickle cell disease or calcified soft-tissue masses.97,
The number of targeted agents that enable precise delineation of lesions in the bone continues to increase. This progress will improve the specificity of detection and monitoring of bone tropic metastases and may help to motivate more personalized patient management. In addition, incremental improvement of existing and well-validated targeted compounds are expected to have a high impact. For example, a higher resolution and more sensitive fluorine derivate of 123I/131I-MIBG, 18F-MFBG, has recently demonstrated impressive preclinical results of improved sensitivity to detect neuroblastoma with lower background.100 The intense interest in targeted tracers continues to increase as new methods emerge which enable image guided-targeted focal therapy (guided external beam or surgical intervention). Many of these tracers also have theranostic potential, and can be used to identify patients, which may benefit from therapeutic derivatives of imaging compounds (for example, high-dose 131I-MIBG).
Tracer developments have been matched with advances in instrumentation and analysis methods to better quantitate and monitor disease in both patients and preclinical models. From the small animal imaging field, pinhole SPECT, and refined reconstruction algorithms now routinely enable sub-millimeter.101,102 This technology is now being translated to the clinical arena and may enable resolution of single-photon emitters to 3 mm. Higher field-strength MR and multimodal PET/MR are also emerging with improved resolution, decreased scan time, and novel applications for fused imaging.103 Monitoring of cancer associated bone events by RNB with the fine soft-tissue detail afforded by MR is expected to decrease false-positive lesion enumeration and improve disease monitoring. However, challenges still remain with implementation of this nascent technology with questions regarding absolute quantification, economic concerns of device cost and upkeep, and issues regarding procedure reimbursement.104,105
Skeletal involvement during cancer progression is directly associated with poor outcome. Effective and long-lasting treatment of disseminated bone disease remains elusive, compounding the health-care impact of this common site of metastasis. The remodeling and destruction of bone at these sites produce pain and nerve compression, fracture, and hypercalcemia which together severely impact quality of life. Methods to image the presence and track the progression of bone metastasis generally rely on indirect visualization of bone activity, primarily through radiolabeled bisphosphonates for planar scintigraphic or tomographic RNBs.
Planar RNB reveal sites of active bone remodeling associated with lesions, which can be identified and the extent of the involved skeleton quantified. Sodium fluoride PET, MR, and CT provide companion and alternative methods to detect bone involvement in advanced disease. These techniques benefit from high sensitivity for bone events but lower specificity for directly cancer-related bone involvement. This lack of specificity is a problem exacerbated by the skeletal phenotype of the elderly population that makes up the bulk of cancer patients, possessing numerous benign but degenerative bone and joint sites. Advanced methods are emerging which include disease-specific targeted imaging agents, semi-automated image processing and whole-body imaging techniques.
We thank Dr. Diane Abou for the chemical structures. For funding support, we thank the Prostate Cancer Foundation Young Investigator Awards (DU and DLJT), National Institutes of Health National Cancer Institute (DU: P50-CA92629, P50-CA86438; DLJT: P50-CA058236), the Wallenberg Foundation Diagnostic Consortium Grant (DU) and the Patrick C. Walsh Prostate Cancer Research Fund (DLJT).
About this article
Current Osteoporosis Reports (2017)