Conventional imaging modalities, such as plain radiography, ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI), have been used for many years to identify and characterize tumors based on anatomic differences in density, water content, shape, and size. More recently, functional imaging modalities like 18F-fluorodeoxyglucose-positron emission tomography (FDG-PET) have been developed that are capable of characterizing tumors based on biochemical changes at the molecular level. Since the year 2000, the number of FDG-PET scans performed in the United States has increased about 9-fold, and it is estimated that, in 2011, about 1.8 million studies were performed, and 94% of the studies were done for cancer. This dramatic increase in use is likely being driven by several factors, one of which is the higher sensitivity and specificity of FDG-PET for the detection and staging of many tumors compared with anatomic imaging modalities.[2, 3] In addition, PET is being used increasingly to assess therapeutic response and to characterize tumor biology. In this review, we highlight the various applications of PET imaging in cancer, including its role in personalized medicine. Although FDG continues to be the most widely used radiotracer for PET imaging (96% of PET studies in 2011 used FDG), some of the newer PET imaging agents that have been studied in humans will also be reviewed (Table 1).
|18F-FLT||Thymidine kinase||Phase 3|
|18F-FMAU||Thymidine kinase||Phase 1|
|18F-FMISO||Low oxygen||Phase 2|
|64Cu-ATSM||Low oxygen||Phase 2|
|18F-EF5||Low oxygen||Phase 2|
|18F-FAZA||Low oxygen||Phase 2|
|18F/99mTc-annexin V||Phosphytidylserine||Phase 2|
|18F-ML-10||Apoptotic changes||Phase 2|
|11C-MET||Amino acid transporters||Phase 2|
|18F-FET||Amino acid transporters||Phase 2|
|18F-FMT||Amino acid transporters||Phase 1|
|18F-FACBC||Amino acid transporters||Phase 2|
|11C-choline||Choline kinase||FDA approved|
|18F-choline||Choline kinase||Phase 2|
|18F-FES||Estrogen receptor||Phase 2|
Principles of PET Imaging
PET imaging uses radiopharmaceuticals labeled with positron emitting radioisotopes such as 11C, 13N, 15O, and 18F, which are produced in a cyclotron, and 68Ga and 82Rb, which are produced in a radioisotope generator. After a positron is emitted, it annihilates with a nearby electron and generates 2 annihilation photons (each with an energy of 511 keV), which travel in opposite directions. PET scanners are equipped with coincidence electronics to detect these pairs of photons as they hit opposing detectors nearly simultaneously. Single unpaired photons occur when one of the annihilation photons is absorbed in the body or is not detected. Although these single photons are excluded by the coincidence window, additional processes occur in PET imaging that degrade image contrast and need to be accounted for in quantitative PET imaging. Thus, various algorithms and techniques are required to correct for random coincidence events, scatter, dead-time, and various sensitivity among detectors. For accurate quantitative results, one of the most important corrections to the acquired PET image is attenuation correction. This correction is necessary because photons released from the center of the body, on the average, are attenuated to a greater extent than photons released from the periphery. One of the advantages of hybrid PET/CT systems is that the corresponding CT images can be used for anatomic localization as well as attenuation correction.
FDG has its origin in the work of Sokoloff et al, who developed a method to measure regional cerebral glucose metabolism in animals using 14C-deoxyglucose autoradiography. Although 14C-deoxyglucose is transported into cells in parallel with glucose and is phosphorylated by hexokinase to 14C-deoxyglucose-6-phosphate, because it lacks a hydroxyl group at the 2 position, it is prevented from being a substrate of enzymes farther down the glycolytic pathway. Thus, 14C-deoxyglucose-6-phosphate becomes trapped within cells, providing a measure of the metabolic rate. This concept was extended when FDG was developed, because FDG also lacks a hydroxyl group in the 2 position. In a similar fashion, FDG is transported into cells and phosphorylated, and it becomes trapped within cells as FDG-6-phosphate. When labeled with 18F, which is a positron-emitting radioisotope, FDG can be imaged with PET and used to quantify regional glucose metabolism in humans.
Radiotracer uptake in PET imaging can be quantified from dynamic acquisitions performed over a period of time or from static acquisitions performed at a fixed point in time. In the research setting, dynamic imaging is often performed to measure the radiotracer uptake over time, yielding plots of radioactivity concentration versus time known as time-activity curves. Kinetic modeling can then be used, sometimes in conjunction with blood sampling, to estimate the rate of uptake and trapping of a radiotracer or to estimate the density of a receptor of interest. Because dynamic imaging is time consuming and difficult to perform over the whole body, clinical studies usually acquire a single image at approximately 60 minutes after injection of FDG. This static image is then used to calculate a simplified measure of FDG uptake, known as the standardized uptake value (SUV). The SUV represents the concentration of radioactivity in the tumor tissue, normalized to the injected FDG dose and the body weight of the patient; the SUV is equal to 1 for an evenly distributed tracer. Because the SUV is typically calculated at approximately 60 minutes after injection of radiotracer, the radioactivity in metabolically active tissues is predominantly due to phosphorylated FDG rather than nonphosphorylated intracellular/intravascular FDG. Thus, the SUV is roughly proportional to the net rate of FDG phosphorylation. However, SUVs are also affected by a variety of factors other than tumor glucose use; these include the time interval between injection and imaging, the size of the lesion, image reconstruction parameters, and the spatial resolution of the PET scanner. In addition, differences in the affinity and kinetics of glucose transporters and hexokinase for FDG versus glucose lead to variation in the relative trapping of each substrate in different tissues and tumors. Thus, although SUVs do not provide an accurate absolute measure of tumor glucose use, they are useful for assessing interval change in metabolism, because many of the confounding issues with SUVs are negated by measuring relative change. With properly calibrated imaging devices and adherence to standard patient preparation and imaging protocols, the test-retest reproducibility of SUVs in tumors and normal tissues is high, and relative changes in SUVs as small as 20% can be used as a criterion for a metabolic response to therapy.[10, 11]
The use of FDG as an agent for cancer detection is based on the observation of Warburg in the 1920s that cancer cells have abnormally high rates of glycolysis. Even when tumor cells have sufficient oxygen supply, they preferentially generate energy using anaerobic glycolysis followed by metabolism of pyruvate into lactic acid. In addition to elevated glycolysis, tumors often have increased expression of glucose transporters (GLUTs). These transporters allow energy-independent transport of glucose across the cell membrane down a concentration gradient. In malignant tumors, GLUT-1 is frequently overexpressed, but expression levels of GLUT-2, GLUT-3, GLUT-4, GLUT-5, and GLUT-12 also reportedly are increased in several tumors types. The relative importance of GLUTs versus hexokinase activity continues to be debated, and it is likely that, in a given cell type, one may predominate.
FDG has gained widespread use in the clinic for several reasons. First, unlike glucose, FDG is excreted in the urine, which results in relatively rapid clearance of the radiotracer from the blood pool. Second, hexokinase, the target of FDG, is ubiquitous and very efficient. The net result of these 2 points is that very high target-to-background ratios are achieved by 60 minutes, with just enough background activity to enable localization of FDG-avid tumors relative to the physiologic pattern of FDG uptake. Third, there are few radiolabeled metabolites of FDG in the blood, which makes analysis of FDG uptake more straightforward. Finally, the 110-minute half-life of 18F allows FDG to be produced at a central facility and transported to nearby imaging centers; thus, it is not limited to large academic centers with cyclotron facilities.
FDG-PET for Cancer Detection
FDG was first used to study tumors in the 1980s by Di Chiro, who demonstrated that the degree of malignancy of brain tumors was correlated with their FDG uptake. Because FDG accumulates in most cancer cells to a greater degree than noncancer cells, FDG-PET is a very general approach to cancer imaging. Thus, FDG-PET has achieved widespread use for the detection of cancer, primarily in staging newly diagnosed or recurrent cancers. Results from the National Oncologic PET Registry, which included data from 85,658 patients with a wide variety of cancer types, concluded that FDG-PET imaging changed physicians' intended management in about 36% of patients. The dominant impact was a change from nontreatment to treatment, which occurred in 29% of patients; the reverse pattern, changing from treatment to nontreatment, occurred in about 7% of patients. The types of cancer that are most commonly imaged with FDG-PET include lymphoma, head and neck cancer, lung, colorectal cancer, breast cancer, esophageal cancer, melanoma, cervical cancer, thyroid cancer, and pancreatic cancer.[17-19] Some of these cancers, such as aggressive lymphoma, squamous cell carcinoma of the head and neck, and melanoma, are usually “hot” on FDG-PET. Other cancers, such as prostate cancer, neuroendocrine cancer, and well differentiated hepatocellular carcinoma, are often “cold,” which limits the utility of FDG-PET in those tumors.[21, 22] Still other cancers, such as breast and thyroid cancers, demonstrate a range of FDG uptake, which may be related to incompletely understood biologic factors.
Recent studies have demonstrated that the activation of a variety of oncogenes results in increased FDG uptake. Thus, in many patients, the degree of FDG uptake is a marker of tumor dedifferentiation. For example, when prostate cancer and thyroid cancer lose their ability to respond to androgens and trap iodine, respectively, they are much more likely to be observed with FDG-PET.[25, 26] Because many factors contribute to increased FDG uptake in cancer cells, some studies have suggested that FDG-PET provides a unique measure of tumor biology that is distinct from in vitro assays.
False-positive findings are relatively common on FDG-PET, because elevated glycolysis is not limited to cancer cells. Typical causes of increased FDG uptake unrelated to malignancy include infectious and inflammatory etiologies, muscular activity, metabolism in brown fat, and changes in response to bone marrow-stimulating cytokines. Thus, the modest specificity and low sensitivity of FDG-PET for early stage disease limit its utility for cancer screening. However, in patients with an established diagnosis, or strong suspicion, of cancer, FDG-PET is a powerful tool that can help characterize the disease and determine its extent.
FDG-PET for Monitoring Tumor Response
It has been demonstrated that FDG-PET is useful for monitoring during treatment and assessing response after treatment has ended.[7, 27, 28] Many studies in breast cancer, lymphoma, lung cancer, and other tumors have demonstrated that FDG-PET can detect an early response to treatment (Fig. 1), which, in many patients, correlates well with clinical outcome, such as disease-free survival.[18, 29-31] In addition, FDG-PET has been used to improve management decisions in oncologic settings. For example, data from the National Oncologic PET Registry indicated that, of 10,497 treatment-monitoring PET scans performed for a wide variety of cancer types, physician-intended management changed in about 50% of patients; this included a switch to another therapy in 27% of patients, an adjustment in dose or duration of therapy in 17% of patients, and a switch from therapy to observation or supportive care in 6% of patients. The expectation is that there is a strong correlation between changes in physician-intended management and patient health outcomes; however, the relationship is imperfect, and more research is needed. The goal of early monitoring for treatment response is to evaluate the effectiveness of a therapy earlier than is feasible through symptoms or other clinical parameters; ultimately, this could shorten therapeutic regimens in patients who are responding, allow for earlier changes to a new therapeutic regimen in patients who are not responding, and expedite the development of new treatments for cancer, all of which have the potential to lead to better patient outcomes.
The timing and relative change in FDG uptake during treatment appears to depend on the type of treatment used as well as the type of cancer being treated. Studies of cytotoxic chemotherapy in breast cancer, lymphoma, gastrointestinal cancers, and others indicated that FDG-PET is capable of detecting a response after a single cycle of chemotherapy.[33-35] In addition, in some lymphomas, significant changes in FDG uptake have been observed as early as 1 day after starting chemotherapy. The reasons for the rapid decline in FDG uptake after cytotoxic chemotherapy are not known but are likely related to a decrease in the population of viable cells as well as a decrease in the rate of glycolysis per cell. Studies of targeted therapies like imatinib, which inhibits the c-kit growth factor pathway in gastrointestinal stromal tumors, have indicated that FDG uptake is reduced within hours of starting treatment and appears to be related to a rapid decline in glucose transporter expression.[37-39] Similarly, it has been demonstrated that the inhibition of epidermal growth factor receptor kinase by gefitinib in lung cancer cell lines results in translocation of glucose transporters from the plasma membrane to the cytosol within 4 hours. However, additional studies are needed to further elucidate the mechanisms behind the rapid decline in FDG uptake in response to these classes of drugs. It is noteworthy that, when treatment was temporarily held in patients with gastrointestinal stromal tumors who were taking sunitinib, a multitargeted tyrosine kinase inhibitor that also targets the c-kit pathway, many of the tumors demonstrated a rebound in FDG activity after 3 weeks, and this was followed by a decline in FDG activity when treatment resumed. Thus, the finding that FDG-PET is capable of demonstrating changes in tumor biology hours to days after starting therapy gives it tremendous potential as a biomarker for tumor response to therapy, in marked contrast to conventional response measures by CT, which take weeks to months to evolve and can be misleading at early time points. In most patients, successful treatment results in decreased FDG uptake; however, in certain patients, targeted therapy has resulted in increased FDG uptake at early time points during treatment. For example, a study of patients with breast cancer indicated that increased tumor FDG uptake at 7 to 10 days after the initiation of tamoxifen therapy was predictive of a subsequent clinical response.
Monitoring radiotherapy with FDG-PET can be challenging, because FDG uptake is often unchanged immediately after treatment and, in some patients, may be increased. The reason for this may be because, compared with chemotherapy, after radiotherapy, there is a different and longer period of viability for tumor cells or possibly a greater inflammatory response. Conversely, one study indicated that increased rather than decreased FDG uptake in high-grade gliomas after radiotherapy was predictive of a better outcome.
There is no generally accepted threshold for a reduction in SUV that represents a response to treatment. Reproducibility studies suggest that changes in SUV >20% are likely to be significant, and current guidelines recommend using a threshold of 25% or 30% for tumors with significant baseline activity.[47, 48] However, the SUV thresholds used in clinical studies range from 20% to 70%, with smaller thresholds used for studies performed after a single cycle of chemotherapy and larger thresholds used for studies performed later during the course of treatment. To confound matters, inflammatory tumor changes potentially can blunt the apparent reduction in SUV. For example, one study revealed that 29% of the tumor glucose use could be attributed to macrophages and granulation tissue. Because different tumors and treatments result in various changes in FDG uptake, more data are needed to validate specific response criteria for FDG-PET.
FDG-PET for Prognosis
The ability of FDG to serve as a prognostic indicator in patients with cancer is incompletely understood. Intuitively, FDG uptake would be expected to reflect tumor grading, because less differentiated and more rapidly proliferating tumors should need more glucose for energy production. However, the correlation between FDG uptake and cellular proliferation, although positive, is not very strong. Compared with FDG, the thymidine analog 3′-deoxy-3′-18F-fluorothymidine (FLT) demonstrates a much closer correlation between uptake and labeling with Ki-67, a cellular marker of proliferation. In addition, the correlation between FDG uptake and tumor grading is fairly weak for the majority of tumors; only gliomas, sarcomas, and thyroid cancers demonstrate strong correlations.[15, 52, 53] To make matters more complicated, some benign tumors, such as giant cell tumors, juvenile pilocytic astrocytomas, and Warthin tumors, frequently demonstrate intense FDG uptake.[54-56] Thus, although FDG uptake is modulated by several factors that often serve as poor prognostic markers, such as hypoxia, increased cellular proliferation, and the activation of a variety of oncogenes, the multifactorial etiology of increased FDG uptake limits its ability to assess for parameters other than glucose metabolism and may explain the variable success of FDG as a prognostic marker.
Aberrantly increased cellular proliferation is a hallmark of cancer, and a decline in proliferation is one of the earliest events in response to effective cancer therapy. It has been demonstrated that, in responding tumors, both chemotherapy and radiotherapy reduce proliferation rates, which precedes a reduction in tumor size.[58, 59] Thus, imaging the rate of cell proliferation could permit the differentiation of benign from malignant tumors and could provide an earlier measure of therapeutic response than anatomic imaging. In addition, cell proliferation imaging has a marked advantage over anatomic imaging with regard to assessing treatment response to novel targeted agents, such as protein kinase inhibitors or antiangiogenesis agents, because some of these agents have a predominantly cytostatic effect and may not lead to rapid tumor shrinkage.
Several different radiotracers have been developed to image cell proliferation. Initial efforts focused on synthesizing analogs of thymidine, because thymidine is used by proliferating cells for DNA synthesis during the S-phase of the cell cycle but, unlike other nucleosides, is not incorporated into RNA. The gold standard for assessing proliferation in vitro has long been 3H-thymidine incorporation. More recently, studies using 11C-thymidine and PET demonstrated promise in assessing response to therapy; however, the short half-life of 11C and its rapid metabolism to thymine and other metabolites made 11C-thymidine impractical for routine clinical use outside of academic centers. This problem spurred the development of FLT, an 18F-labeled, nonmetabolized thymidine analogue that has become the most extensively studied radiotracer for imaging tumor proliferation.[62, 63]
After intravenous injection, FLT enters the cell by facilitated diffusion through nucleoside transporters and is trapped in the cytosol through phosphorylation. Because FLT lacks a hydroxyl group at the 3′ position, it is not incorporated into DNA, and its accumulation serves as a measure of cellular proliferation. Correlation of FLT uptake with the rate of cell proliferation in various tumor types has been demonstrated using histopathologic markers of cell proliferation, such as the Ki-67 labeling index, as the gold standard. Unfortunately, most tumors demonstrate relatively low FLT uptake compared with FDG, which reduces the sensitivity of FLT-PET for detecting cancer. In addition, FLT demonstrates significant interfering physiologic activity in bone marrow and liver, and proliferating lymphocytes in reactive lymph nodes still have the potential to yield false-positive findings.[60, 66] Thus, it is unlikely that FLT will have a dominant role in the initial diagnosis and staging of most cancers. Conversely, there is a general consensus that FLT-PET can be a valuable method for monitoring tumor response to treatment, especially as an early measure of response. This may be particularly important for cytostatic therapy, ie, agents that stop tumor growth but may not kill tumor cells, for which proliferation imaging may play a key role in judging therapeutic effectiveness.
FLT has several attractive properties as an agent for measuring cancer response. Measures of FLT uptake are precise and repeatable, with a test-retest reproducibility <10% for SUV measurements. This is an important feature in using FLT-PET for serial imaging to assess response. Also, studies have indicated that changes in FLT uptake reflect changes in tumor proliferation with treatment. Preclinical work has demonstrated that a decline in FLT uptake occurs early in the course of treatment with a variety of approaches, including chemotherapy, radiotherapy, and biologically targeted agents, and several studies have noted that changes in FLT activity reflected the effects of therapy better than changes in FDG uptake.[68, 69] Also, preliminary clinical studies confirmed the potential of FLT-PET for monitoring tumor response to therapy (Fig. 2). In patients with recurrent malignant brain neoplasms, tumor response assessed by kinetic measures of FLT uptake after 1 or 2 weeks of treatment with irinotecan and bevacizumab correlated well with overall survival. Early studies on breast cancer have also demonstrated that changes in FLT uptake during treatment predict clinical response, and multicenter trials for breast cancer are ongoing.[71, 72] In patients with nonsmall cell lung cancer, a reduction in FLT uptake after 7 days of therapy with gefitinib was highly predictive of tumor response on CT and of progression-free survival. This is a notable result, in that proliferation imaging was effective for predicting response to a biologically targeted agent that often has a cytostatic response when used as monotherapy.
However, some studies have reported less striking results. In lymphoma, treatment with rituximab did not result in an early change in FLT uptake, although it has been a highly effective agent in the clinic, and some data support its impact on cellular proliferation. Some more recent studies suggest that FLT may be taken up in proliferating immune cells, and it is possible that the immune effect of rituximab, a monoclonal antibody, could confound assessment of response by FLT-PET. Conversely, in some patients, tumor FLT uptake changed significantly in patients who did not go on to have a subsequent clinical response. For example, a study of patients with rectal cancer who received neoadjuvant chemoradiation revealed that FLT uptake was reduced significantly at day 14 in all patients, but there was no difference in uptake reduction for histopathologically responding versus nonresponding lesions, suggesting that inhibition of proliferation, although necessary, may not be sufficient for a favorable response in some tumor types or with some types of treatment. In such a case, the decrease in FLT uptake in histopathologic nonresponders may be because of treatment-induced growth arrest rather than cell death. An alternative hypothesis relates to the finding that FLT traces only the salvage pathway for incorporation of thymidine into DNA, whereas many gastrointestinal cancers rely heavily on the de novo pathway, which is not tracked by FLT.[60, 61] These findings indicate some potential limitations of FLT for monitoring the response to treatment, and it is likely that the choice of approach for early assessment of therapeutic response will need to be matched to the cancer and treatment types under study. Also, there is one report of a temporary rise in FLT uptake in patients with lung cancer who were receiving chemoradiation. Thus, FLT uptake may be influenced by factors other than proliferation, such as altered vascular permeability, perfusion, or uptake in proliferating inflammatory cells, and additional study is needed.
Other PET proliferation imaging probes have also been tested. The thymidine analog (1-[2′-deoxy-2′-18F-fluoro-beta-D-arabinofuranosyl]thymine) (FMAU) can be phosphorylated and subsequently incorporated into DNA. Because FMAU has minimal bone marrow uptake and urinary excretion, it is a promising agent for studying bone metastases and genitourinary malignancies. Preliminary in vitro studies have indicated that FMAU uptake is proportional to the rate of tumor proliferation. However, additional studies are needed to evaluate the ability of FMAU to serve as a PET imaging biomarker of cell proliferation. There have also been efforts to image proliferation using methods distinct from probes like FLT and FMAU, which rely on the thymidine salvage pathway for DNA synthesis. A promising example is the use of sigma-2 receptor binding agents to measure tumor proliferation. Like FLT, sigma-2 ligand binding levels increase with increasing cellular proliferation; however, sigma-2 agents have the potential advantage that, unlike FLT, they also produce some uptake in quiescent (nonproliferating) but viable cells. Promising early studies in humans motivate further work in this area of research.
Tumor hypoxia is a pathologic state in which tumor tissues lack enough oxygen for normal metabolism. In solid tumors, oxygen delivery to the tumor cells is frequently reduced or even abolished by a deteriorating diffusion geometry, severe structural abnormalities of tumor microvessels, and disturbed microcirculation. These microregions of hypoxia are distributed heterogeneously within the tumor mass and may be located adjacent to regions with normal oxygen partial pressures. It is noteworthy that >50% of locally advanced solid tumors may contain hypoxic tissue areas, and the presence or absence of hypoxia cannot be predicted by clinical size, stage, grade, histology, or site. In solid tumors, hypoxia is associated with restrained proliferation, differentiation, apoptosis, and necrosis. At the same time, hypoxia may promote tumor progression through mechanisms that enable tumor cells to overcome nutritive deficiency, escape from a hostile environment, and favor unrestricted growth. Sustained hypoxia also may lead to cellular changes, resulting in a more clinically aggressive phenotype. In addition, tumor hypoxia has been associated with treatment failure after radiotherapy and chemotherapy.[86, 87] Thus, knowledge of the degree and extent of tumor hypoxia has the potential to play a significant role in staging and treatment planning for a wide variety of tumor types and, in certain patients, may help to direct patients to regional therapy and/or agents that are selective for hypoxic tissue.
Two major classes of PET radiotracers have been developed to image hypoxia: 2-nitroimidazoles, such as 18F-fluoromisonidazole (FMISO), 18F-EF5, and 18F-fluoroetanidazole (FETA), and nucleoside conjugates, such as 18F-fluoroazomycin arabinoside (FAZA) and 64Cu-diacetyl-bis(N4-methylthiosemicarbazone) (Cu-ATSM). All of these agents undergo intracellular trapping at a rate inversely proportional to intracellular oxygen concentration. In normal cells, the imaging agents are rapidly reoxidized and cleared, thus providing good demarcation between hypoxic and normoxic regions. Of the hypoxia imaging agents, FMISO has the largest body of preclinical validation studies and clinical experience, although the other PET hypoxia tracers also have been studied in patients. Recent studies have demonstrated that tumor hypoxia measured by PET is predictive of patient outcome; patients with hypoxia by PET had considerably earlier relapse or progression.[89, 90] However, not all studies demonstrate this correlation. For example, a study in patients with head and neck cancer demonstrated no correlation between hypoxia on PET imaging and patient outcome. Thus, although PET hypoxia imaging has tremendous potential in treatment planning, more research is needed to better delineate the role of hypoxia imaging in cancer care.
Programmed cell death, or apoptosis, is an essential component of normal human growth and development, immunoregulation and tissue homeostasis. It is generally accepted that effective therapy of tumors by radiation, chemotherapy, or both, leads to induction of apoptosis. Preclinical and clinical studies have shown that the detection of apoptosis can potentially be used to provide an early indication of the success of therapy. Thus, several different classes of apoptotic imaging agents have been developed.
The first class of imaging agents targets phosphytidylserine residues that normally reside on the intracellular membrane surface but that are translocated to the extracellular surface during apoptosis. The single-photon emission CT (SPECT) agent technetium-99m (99mTc)-annexin V, a radiolabeled protein with nanomolar affinity to phosphytidylserine, is the most widely used imaging probe in this class. Although 99mTc-annexin V has a suboptimal biodistribution profile characterized by high background activity, including confounding uptake in the liver, multiple studies have demonstrated that 99mTc-annexin V is able to image apoptosis in vivo and to predict patient outcome after chemotherapy or radiation therapy.[94, 95] However, more data are needed to support the ability of 99mTc-annexin V to predict long-term treatment outcomes. More recently, 18F-annexin V has been synthesized, and it has an improved biodistribution profile and can be imaged using PET; preliminary studies of 18F-annexin V have been done in animals but not in humans.
A second class of imaging agents targets caspases, which are cysteine-aspartate proteases that play a pivotal role in the regulation of apoptosis. Two of the main imaging agents in this class are 18F-ICMT-11 (a caspase-3-specific small molecule PET tracer based on the caspase inhibitor isatin) and 18F-CP18 (a pentapeptide-based PET tracer that is a substrate of caspase-3). Both of these agents have demonstrated promise for the early detection of drug-induced tumor apoptosis in several animal models, and the biodistribution of both agents has been studied in humans.[97, 98] However, to date, no clinical trials have evaluated the ability of these imaging agents to predict patient outcome.
A third class of imaging agents is thought to detect plasma membrane depolarization, although the exact mechanism has not yet been fully elucidated. The most notable of these imaging agents is the PET tracer 18F-ML-10 (an 18F-labeled small molecule malonic acid-based probe). A recent study of 10 patients with brain metastases who received whole-brain radiation therapy demonstrated a significant correlation between early changes in 18F-ML-10 PET scans and size changes on MRI at 6 to 8 weeks after treatment. Multicenter clinical trials are currently taking place to assess 18F-ML-10 in multiple tumor types.
One of the limitations of imaging apoptosis is related to its rapid evolution, which renders imaging of the transient apoptotic biomarkers particularly challenging. In fact, in some patients, the ideal temporal window for imaging occurs between 6 and 24 hours after treatment. Thus, concern has been expressed that the relatively small number of cells undergoing apoptosis at any one time and the small time window to have access to the biomarkers during the apoptotic process may limit the widespread use of apoptosis imaging. However, these same considerations may provide an advantage for mechanistic studies of early response that are important in determining optimal timing in multiagent therapy.
Imaging Protein and Cell Membrane Metabolism
The process of protein synthesis in tumors is increased because of the uncontrolled and accelerated growth of cancer. Consequently, the demand for amino acids—the building blocks of proteins—is increased. In this regard, studies have demonstrated that many types of tumor cells have significantly increased L-type amino acid transporter 1 (LAT1) expression, which is the major transporter of large neutral amino acids. Thus, increased protein synthesis/LAT1 expression is an excellent target for tumor imaging. Although almost all of the amino acids have been radiolabeled, only a few have been used clinically as PET imaging agents. These include 11C-methionine (MET), 18F-fluoroethyl-tyrosine (FET), 18F-fluoromethyl-tyrosine (FMT), 1-amino-3-18F-fluorocyclobutane-1-carboxylic acid (FACBC), and 3,4-dihydroxy-6-18F-fluoro-L-phenylalanine (FDOPA).
MET-PET is the most popular amino acid imaging modality in oncology, with over 300 basic scientific and clinical publications. MET-PET has been used primarily to image cerebral gliomas, and it has been used for initial diagnosis, differentiation of tumor recurrence from radiation necrosis, grading, prognostication, delineation of tumor extent, biopsy planning, surgical resection and radiotherapy planning, and assessment of response to therapy. Because large neutral amino acids like MET are able to cross the blood-brain barrier and enter normal brain tissue, disruption of the blood-brain barrier (ie, contrast enhancement on CT or MRI) is not a prerequisite for intratumoral accumulation of these amino acids. Uptake of FLT, conversely, depends on blood-brain barrier damage, because transport across the normal blood-brain barrier is slow for thymidine. Thus, nonenhancing, low-grade gliomas can be observed with MET, but not with FLT. MET-PET also has been used clinically to distinguish benign from malignant tissue in head and neck cancer, melanoma, ovarian cancer, and other tumors. One of the advantages of MET and other radiolabeled amino acids over FDG is their relatively low uptake in normal brain, which makes them more specific for tumor detection and better at tumor delineation and staging. In addition, amino acid radiotracers generally are affected less by inflammation compared with FDG, although tumor specificity is still not perfect.
The short half-life of 11C prevents the widespread use of MET-PET for tumor imaging, and awareness of this limitation has stimulated the development and evaluation of 18F-labeled amino acids like FET, FMT, FACBC, and FDOPA. Several clinical studies with small numbers of patients have suggested that FMT is superior to FDG for the detection of musculoskeletal tumors, similar to FDG for the detection of head and neck tumors, and is superior to FDG for determining treatment response in lung cancer.[109, 110] Conversely, small clinical studies that used FET indicated that it was generally worse than FDG for the detection of many tumor types, including head and neck cancer.[111, 112] FET-PET did exhibit better specificity in several of the studies, raising the possibility that it may be better at differentiating tumor from inflammatory tissue. Also, when FET-PET was compared directly with MET-PET in 42 patients with gliomas or brain metastases, the 2 imaging agents provided comparable diagnostic information. FDOPA-PET also was compared with MET-PET in patients with primary brain tumors or brain metastases and, similarly, yielded comparable diagnostic information. FACBC has been studied primarily in prostate cancer, in which it had higher sensitivity for tumor detection compared with 111In-capromab pendetide (Prostascint) and 11C-choline.[115, 116] In addition, FACBC has demonstrated increased uptake in papillary renal cell carcinoma, but not in clear cell carcinoma. FACBC is especially notable for the lack of associated radioactivity in the urine, which allows for better evaluation of the genitourinary tract and pelvis.
Choline is a precursor of phosphatidylcholine, which is a major constituent of membrane lipids. Because both membrane lipid synthesis and protein synthesis are activated during cell proliferation, choline is consumed in large quantities by tumor cells. Thus, 11C-choline accumulation is significantly greater in malignant tumors than in benign lesions. Most reports of 11C-choline PET imaging involve prostate cancer, although 11C-choline uptake also is increased in a wide variety of other tumors.[119, 120] Given its lack of radiotracer activity in the urine, 11C-choline is well suited to evaluate pelvic malignancies. In recent studies of patients with prostate cancer, 11C-choline PET identified recurrent disease in 66% of patients with a mean prostate-specific antigen (PSA) level of 8.3 ng/mL, 28% of patients with a PSA level <1.5 ng/mL, and 21% of patients with a PSA level <0.5 ng/mL.[121, 122] In addition, 11C-choline PET demonstrated better sensitivity than MRI for lymph node metastases in prostate cancer. More recently, 18F-choline (FCH), which has the advantage of a much longer half-life compared with that of 11C, has been developed. To date, studies of FCH have focused largely on prostate cancer, for which FCH has performed similarly to 11C-choline.
Imaging Tumor-Specific Agents
Although tracers for metabolism, proliferation, and hypoxia provide useful imaging of neoplasms, they are relatively nonspecific and are usually less useful for imaging tumors that have very low growth rates. Therefore, the development of tracers targeting intracellular and cell-surface receptors that are uniquely expressed or overexpressed in cancer cells is essential for the development and usefulness of clinical PET. Examples of radiotracers that target these tumor-specific biomarkers are outlined below.
Prostate-specific membrane antigen (PSMA) is a cell surface protein that is highly expressed in prostate carcinoma cells. Several radiotracers that target PSMA have been developed and studied preclinically. Recently, both 18F-DCFBC (N-[N-([S]-1,3-dicarboxypropyl)carbamoyl]-4-18F-fluorobenzyl-L-cysteine) and 68Ga-PSMA have been studied in humans, and both reportedly were capable of detecting sites of metastatic prostate cancer.[125, 126] In a subsequent, small clinical trial, 68Ga-PSMA detected more lesions than 18F-choline.
Somatostatin receptors are overexpressed on neuroendocrine tumor cells. Imaging of this biomarker is currently accomplished using SPECT with 111In-diethylenetriaminepentaacetic acid (DTPA)-octreotide; however, several new radiotracers have been developed that can be imaged using PET. Examples include 68Ga-1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA)-Tyr3-octreotide (68Ga-DOTA-TOC), 68Ga-DOTA-1-Nal3-octreotide (68Ga-DOTA-NOC), and 68Ga-DOTA-Tyr3-octreotate (DOTA-TATE), all of which demonstrate better sensitivity than 111In-DTPA-octreotide for neuroendocrine tumors, and all have the additional benefit of better resolution because they are imaged using PET. Although all 3 radiotracers demonstrate variable affinity to somatostatin receptor subtypes, there is currently no evidence of a clinical impact from these differences; thus, there is no indication suggesting the preferential use of one compound over the others.
In addition, there are several cancer therapies that target a specific receptor or enzyme in tumor cells. Current examples of specific targets (and examples of treatments directed at them) include the estrogen receptor (tamoxifen and letrozole), HER2 (trastuzumab), epidermal growth factor receptor (gefitinib), and angiogenesis (bevacizumab). Measuring the target expression at each site of disease is a task for which PET is well suited. PET imaging can determine whether the target is expressed at all disease sites and can quantify the level of target expression at each site. Current examples of PET imaging to measure target expression include estrogen receptor imaging and androgen receptor imaging; HER2 imaging; imaging angiogenesis nonspecifically by measuring blood flow and by imaging specific components expressed in neovasculature; and measuring novel targets, such as matrix metalloproteinases.[130-135] In all of these cases, PET imaging has the potential to provide prognostic information and can help predict the likelihood that a patient will respond to a targeted chemotherapeutic agent. One of the major imaging agents in this class is 18F-fluoroestradiol (FES), which has exhibited the most promise in quantifying the functional estrogen receptor status of primary or metastatic breast cancer. Studies by Mortimer et al have demonstrated that a high level of FES uptake in advanced breast cancer predicts a greater likelihood of response to tamoxifen. More recent studies report similar results for patients with recurrent or metastatic breast cancer who received a variety of hormonal agents. These preliminary results demonstrate the exciting potential of PET imaging to help guide appropriate, individualized treatment for cancer, and they point the way for future clinical use.
PET is an established technique for cancer detection, and FDG continues to be the most widely used radiotracer for the staging and restaging of most tumors. However, many of the newer radiotracers described in this article will likely play a larger role in cancer care in the future and, along with FDG, will be used increasingly to determine a patient's prognosis, make decisions regarding treatment planning, and evaluate treatment response. In addition, new radiotracers will continue to be developed as diagnostic agents for tumors with minimal FDG uptake, such as prostate cancer and neuroendocrine tumors. Unfortunately, several barriers to the development and use of new radiotracers exist; these include a costly FDA approval process, relatively small profit potential (and, thus, little interest by the pharmaceutical industry in sponsoring clinical trials), as well as uncertain reimbursement by the US Centers for Medicare and Medicaid Services and by private insurers. Thus, future applications of PET for cancer imaging will need to rely on robust and reproducible quantitative data to validate the ability of PET to serve as an effective biomarker, and the passage of new radiotracers through the FDA and Centers for Medicare and Medicaid Services approval processes must be supported.