PET probes beyond 18F-FDG

Abstract During the past several decades, positron emission tomography (PET) has been one of the rapidly growing areas of medical imaging; particularly, its applications in routine oncological practice have been widely recognized. At present, 18F-fluorodeoxyglucose (18F-FDG) is the most broadly used PET probe. However, 18F-FDG also suffers many limitations. Thus, scientists and clinicians are greatly interested in exploring and developing new PET imaging probes with high affinity and specificity. In this review, we briefly summarize the representative PET probes beyond 18F-FDG that are available for patients imaging in three major clinical areas (oncology, neurology and cardiology), and we also discuss the feasibility and trends in developing new PET probes for personalized medicine.


INTRODUCTION
During the past several decades, positron emission tomography (PET) has been a rapidly growing area of medical imaging, which is increasingly important in routine oncological practice worldwide [1,2] . This change is based on two primary factors: one is the combination of PET and computed tomography (CT), which take advantage of the traditional diagnostic imaging and functional imaging techniques, and the other is the broad utility of 18 F-fluorodeoxyglucose ( 18 F-FDG) for evaluating the staging and restaging of cancer, the response to treatment, the differentiation of post-therapy alterations from residual or recurrent tumor, and the assessment of prognoses [3][4][5] .
In 1978, Ido et al. [6] firstly prepared the 18 F-FDG using 18 F-F2 gas in the Brookhaven National Laboratory of the United States. Another breakthrough was then achieved in 1986 by Hamacher et al. [7] , who produced 18 F-FDG in high amounts based on 18 F-fluoride, and this method was further improved by Fuchtner et al. [8] in 1996. Currently, 18 F-FDG, as a biomarker of glucose metabolism, is the most widely used PET probe in clinics (,90%) [1] . After 18 F-FDG is injected into the human body, it is taken up by various tissues through glucose transporters and trapped intracellularly. The elevated glycolysis of the cancer cells and a corresponding increase in hexokinase activity lead to the high accumulation of 18 F-FDG by cancer cells, which is significantly distinguished from benign tissues. Meanwhile, the accumulation of 18 F-FDG is semi-quantified by PET based on the standardized uptake value (SUV; SUV 5 uptake / [injected dose / patient weight]), which is applied for predicting the likelihood of malignancy [9] . However, 18 F-FDG is not a highly specific radiotracer and cannot differentiate between cells that have a high metabolic rate associated with neoplasia and cells with an increased metabolism related to other etiologies, such as infection or inflammation [9] . In addition, many malignancies (such as renal carcinoma) do not display a high accumulation of glucose and are not readily diagnosed by 18 F-FDG PET [10][11][12] . Therefore, scientists and clinicians are greatly interested in exploring and developing other new PET probes, which are able to target some specific biomarkers, are easily transferred from preclinical to clinical settings, and may benefit as many patients as possible.
Many positron-emitting radionuclides may be applied in the development of successful PET radiotracers for research and clinical trials. These radionuclides include, but are not limited to, 11 11 C is an attractive and important positron-emitting radionuclide for labeling molecules targeted to biomarkers. However, the t 1/2 of 11 C is very short (only 20.4 minutes), and multisteps syntheses are not generally appropriate for the radiosynthesis of 11 C-containing molecules; in most cases, 11 C is applicable for labeling molecules in academic research and drug development projects [2] . Compared with short half-lived radionuclides, several nonconventional metallic radionuclides with longer half-lives can be prepared in high yields in small biomedical cyclotrons or generators and easily delivered. For example, the availability of 68 Ga and 82 Rb generators provides an opportunity to prepare PET radiotracers on site if necessary. However, some metallic radionuclides possess complex decay schemes and usually decay with the emission of low b + -percentage branching ( 86 Y, 33%; 68 Ga, 57%), high b + -energy ( 86 Y, 1.3 MeV; 68 Ga, 4.2 MeV), and co-emission of a substantial amount of c-radiation, which causes patients increased radiation dose. Due to the ideal chemical and nuclear properties of fluorine-18, there has been an increasing focus on establishing tracers labeled with 18 F for clinical utility. The half-life of 18 F is 109.8 minutes, which is long enough to allow for time-consuming multi-steps radiosyntheses and for imaging procedures to be extended over several hours. Moreover, 18 F has a low b + -energy of 0.64 MeV, which promises a short positron linear range in tissues, resulting in a particularly high spatial resolution of up to 1 mm in PET images and lower radiation doses to patients [2] . At present, the desired imaging targets include cell metabolism (glucose, amino acid, nucleoside, lipid, etc.), biological events (tumor hypoxia, apoptosis, proliferation, angiogenesis, etc.), or many other molecular biomarkers (growth factor receptors, specific enzymes, etc.) [9,13] . Several excellent reviews have summarized the progress of developing PET probes for these important targets [1][2]14] . In the following content, we review the representative PET probes beyond 18 F-FDG that are available for patients in three major clinical areas (oncology, neurology and cardiology), and we also discuss the feasibility and trends in developing future PET probes for personalized imaging.

Imaging tumor amino acid metabolism
Similar to the mechanism of increased glucose transport and corresponding glucose accumulation by tumor cells, most tumor cells exhibit elevated protein synthesis and a high expression of cell membrane transporters of amino acids [15] . Thus, amino acid metabolism is also an important imaging target. Among the various radiolabeled amino acids, L-[methyl-11 C] methionine ( 11 C-Met) has been evaluated for a long time. In 1985, Kubota et al. [16] first reported the application of 11 C-Met for diagnosing lung tumors. In 1998, Sasaki M et al. [17] demonstrated that 11 C-Met PET is useful for distinguishing malignant from benign astrocytomas. More recently, Koizumi et al. [18] reported the usefulness of 11 C-Met PET for predicting the response to carbon ion radiotherapy (CIRT) for recurrent rectal cancer. Fig. 1 demonstrates that a patient with the glioblastoma could be diagnosed by 11 C-Met PET [19] .

Imaging tumor proliferation
As high proliferation is one of the most important characteristics of tumor cells, targeting tumor proliferation (that is, the nucleoside metabolism of tumor cells) is another important imaging goal for exploring new PET probes. Among the radiolabeled nucleoside analogs, 11 C-thymidine, a pyrimidine analog that is rapidly incorporated into the DNA of proliferating cells, was initially used [20] . However, the short t 1/2 of 11 C and the rapid degradation of the radiotracer in vivo limited its clinical application [1] . As an alternative to 11 C-thymidine, an 18 F-labeled pyrimidine analog, 18 F-deoxy-39-fluorothymidine ( 18 F-FLT), was developed and has become widely evaluated in some studies [21][22][23] .
The application of 18 F-FLT PET for evaluating proliferation activities has been reported in the diagnosis and evaluation of treatment responses in various types of malignancies, such as lung cancer, brain tumors, breast cancer and lymphoma [24][25][26][27][28][29] . Fig. 2 presents a case with rectal cancer, which is delineated by both 18 F-FLT and 18 F-FDG PET [30] . However, negative 18 F-FDG uptake F-FDG uptake in a metastatic regional lymph node (arrowhead). C and D: 18 F-FLT PET/CT images show positive 18 F-FLT uptake in the rectal cancer (SUV max 7.7, arrow) and in the metastatic regional lymph node (arrowhead). was observed in a metastatic regional lymph node on PET/CT images, but positive 18 F-FLT accumulation was seen in the lymph node.

Imaging tumor hypoxia
Tumor hypoxia results from an inadequate supply of oxygen from the vasculature to the growing tumors. Furthermore, the presence of tumor hypoxia promotes a more aggressive phenotype with resistance to treatment, which constitutes a major challenge to patient management, especially in cancer radiotherapy. Therefore, accurate judgment of tumor hypoxia is critical for oncologists to design treatment protocols and to predict treatment outcomes. For this purpose, various PET probes targeting tumor hypoxia have been developed.
At present, there are several PET probes showing high promise for imaging hypoxia. One is an 18 F-labeled 2-nitroimidazole derivative, that is, 18 F-fluoromisonidazole ( 18 F-FMISO). It has been used to predict the radiotherapy response of tumors (such as lung cancer and head-and-neck cancer) and indicates the prognostic value for patients prior to therapy in investigations [31][32][33][34] . However, with the slow blood clearance of 18 F-FMISO, the contrast between hypoxic tumors and normal tissues is relatively low, which limits its evaluation of changes in hypoxia during the therapeutic interventions. Another probe is 18 F-fluoroazomycin arabinoside ( 18 F-FAZA), a novel 2-nitroimidazole derivative, which has more favorable kinetics than 18 F-FMISO does and has displayed promise in imaging tumor hypoxia [ 3 5 -3 6 ] . Moreover, 18 F-EF5 is also a validated marker for PET imaging of tumor hypoxia [37] . 2-nitroimidazole derivatives, including a novel hypoxic tracer called Cu-diacetyl-bis (N4-methyl-thiosemicarbazone) (Cu-ATSM), which is retained in hypoxic cells, were developed by Fujibayashi et al. [38] . As Cu-ATSM has high membrane permeability and low redox potential, it rapidly enters into cells but is trapped only in hypoxic cells with a reducing environment by the reduction of Cu(II) to Cu(I) [39] . 62 Cu-ATSM-PET is currently ongoing in several institutions for studies. Fig. 3 shows the   comparison between 62 Cu-ATSM and 18 F-FDG PET in a patient with squamous cell carcinoma (SCC) [40] .

Imaging tumor apoptosis
Many effective treatments, such as radiation therapy and chemotherapy, may induce an early increase in cell death, often by apoptosis (programmed cell death). Therefore, in vivo apoptosis imaging offers another approach targeting tumor cells to monitor and evaluate anticancer therapy. The most investigation of these radiotracers is Annexin V and its derivatives. In 1999, Blankenberg FG et al. [41] labeled annexin V with 99m Tc for single photon emission computed tomography (SPECT) imaging. However, apoptosis is a relatively rapid progress, and only a fraction of tumor cells are expected to be in the apoptotic state [42] . The annexin V signal is frequently low and not easily detectable by SPECT. Many researches on the evaluation of annexin V radiolabeled with a wide variety of positron-emitting radionuclides (including 18 F and 11 C) have also been under investigation [43][44] . However, until now, few cases and clinical trials have been reported.

Imaging tumor angiogenesis
Angiogenesis, the formation of new blood vessels from pre-existing vasculature, plays a vital role in tumor growth and metastatic spread. Tumor growth beyond a few millimeters in diameter requires independent vasculatures for the cellular supply of oxygen and nutrients and the removal of waste products. Tumor angiogenesis is a complex, multi-steps process that follows a characteristic sequence of events mediated and controlled by growth factors, cellular receptors and adhesion molecules.
During this process, cell adhesion receptors of the integrin family are responsible for a wide range of cell-extracellular matrix and cell-cell interactions and have been mostly well-studied in many tumors, including glioblastoma, melanoma, ovarian cancer, breast cancer and prostate cancer types [45][46] . One of the most prominent members of this receptor class is avb3 integrin, which is highly expressed on activated endothelial cells. The induced expression of these integrins is crucial for mediating tumor angiogenesis, growth and metastasis. Moreover, avb3 integrin receptors are either not expressed or in very low level on mature vessels or non-neoplastic epithelium.

Imaging tumor receptors
Receptors for regulatory peptides are overexpressed in various cancer cells and are regarded as important targets for cancer molecular imaging. Many radiolabeled compounds have been developed for the imaging of tumor receptors, and some are under assessment. Receptor imaging can noninvasively evaluate the status of receptor expression in cancerous tissues, which is useful for characterizing the individual cancers, early diagnosis and selecting candidates for receptortargeted treatment.
For neuroendocrine tumors (NET), somatostatin (SS) receptors are overexpressed on cell surface and are imaged by radiolabeled SS analogs. There are five subtypes of SS receptors, and among these, SS receptor 2 is predominantly expressed in most NETs. In addition to 111 In-DTPA-octreotide ( 111 In-DTPAOC) for SPECT imaging, 68 Ga-DOTATOC has been proven to have a high affinity with SS receptor 2, which is widely evaluated in the practice of PET imaging (Fig. 5) [54][55][56] .
The growth factor receptors are another category of tumor receptors targeted by PET imaging probes. The most extensively studied growth factor receptors suitable for imaging and therapy are the epidermal growth factor receptor (EGFR) and human epidermal growth factor receptor 2 (HER2) [57] . EGFR is a membrane protein comprising an intracellular domain with tyrosine kinase activity, a hydrophobic transmembrane domain and an extracellular ligand-binding domain. Aberrant expression and activation of EGFR is commonly found in human tumors of epithelial origin, where this receptor promotes tumor growth and progression. Several types of PET probes have been used to target EGFR, specifically, monoclonal antibodies, epidermal growth factor, EGFR binding proteins and peptides and EGFR tyrosine kinase inhibitors [58] . The biodistribution and tumor localization of these probes have been evaluated by small animal PET in animal tumor models, and the gefitinib analog 11 C-PD153035 has been evaluated in humans by Liu et al. [59] and Fredriksson et al. [60] . The growth factor receptor HER2 is a transmembrane glycoprotein encoded by the HER2/neu protooncogene. The amplification and activation of HER2 PET Probes beyond 18 F-FDG promote tumor cell proliferation, growth, migration, adhesion and invasiveness [61] . The HER2 gene is overexpressed and/or amplified in 20-30% of all primary invasive breast cancers, and high levels of the receptor are closely associated with a poor prognosis of breast cancer [62][63] . At present, many targeting-HER2 chemotherapy drugs have been explored in pre-clinical research and trastuzumab has been widely used in clinic. HER2-PET, as a molecular imaging tool, thus can play important roles in screening patients for HER2 targeted therapy and monitoring treatment efficacy. Currently, the PET probes that target HER2 essentially include monoclonal antibodies (in various engineered formats) and affibodies, which are currently in preclinical evaluation [57] .

NEUROLOGICAL IMAGING Ab imaging
Alzheimer9 s disease (AD) is defined histologically by the presence of extracellular b-amyloid (Ab) plaques and intraneuronal neurofibrillary tangles (NFTs) in the cerebral cortex. The senile or neuritic plaque or Ab is one of the signature lesions of the AD brain. In 2003, Mathis et al. [64] and Klunk et al. [65] from the University of Pittsburgh Medical Center reported the development of a 11 C-labeled PET tracer, 6-OH-BTA ([Nmethyl-] 2-(45-methylaminophenyl)-6-hydroxybenzothiazole), also known as Pittsburgh compound B or PIB. 11 C-PIB both displays a high affinity for aggregated Ab and provides the high level of brain entry necessary for PET imaging. Based on the PIB molecule, the 18 F-labeled analog 39-18 F-PIB was also developed by the Pittsburgh group and is now under evaluation. Moreover, in 2004, Kung et al. [66] from the University of Pennsylvania at Philadelphia developed a small, neutral stilbene derivative, 11 C-SB-13, which also demonstrates an excellent binding affinity for Ab. Additionally, two structurally similar fluoropegylated agents being developed commercially are 18 F-AV-45 (shown in Fig. 6) and 18 F-AV-1 [1] . Taking 18 F-AV-1 as an example, it has been demonstrated to have the potential to distinguish AD from frontotemporal lobe dementia (FTLD). Rowe et al. reported that in a preliminary clinical study, 18 F-AV-1 PET imaging displayed extensive cortical accumulation in AD patients but negative in those patients with FTLD. There was no 18 F-AV-1 uptake shown in pure vascular dementia or Parkinson9 s disease (PD), but intermediate grade accumulation of 18 F-AV-1 in mild cognitive impairment and dementia with Lewy bodies (DLB) [67] .

Imaging vesicular monoamine transporter type II
In the brain, vesicular monoamine transporter type II (VMAT2) is responsible for the vesicular packaging and storage of monoamine neurotransmitters. In contrast to the situation on the presynaptic membrane, where there are specific and distinct transporters for active reuptake of three monoamines (dopamine, serotonin and norepinephrine), a common adenosine triphosphatedependent transporter, VMAT2, is responsible for the movement of all three monoamines from the cytosol in the presynaptic neuron into the vesicular lumen. Therefore, imaging VMAT2 in the brain can provide a measurement of the integrity (total number) of all three types of monoaminergic neurons.
During the past decade, a plethora of dopamine transporter imaging probes have been developed, most of which are tropane (or cocaine) derivatives that have varying degrees of affinity to serotonin and norepinephrine transporters. Imaging of the VMAT2 has been proposed as an alternative for monitoring the degeneration of monoaminergic (primarily dopaminergic) neurons in PD. In the 1990s, 11 C-(+)dihydrotetrabenzamine ( 11 C-DTBZ), a radioligand specific for VMA2, was first developed [ 6 8 -6 9 ] .

CARDIAC IMAGING
The application of PET in ischemic heart disease falls within two main categories. First, PET is a wellestablished imaging modality for the evaluation of myocardial blood flow (MBF). Second, PET has been used to assess myocardial metabolism and viability in patients with ischemic left ventricular (LV) dysfunction. The combined study of MBF or myocardial perfusion imaging (MPI) and metabolism by PET has led to a better understanding of the pathophysiology of ischemic heart disease.
On the one hand, several PET radiopharmaceuticals are available for evaluating the relative distribution of Unifocal, weak uptake is present in a lymph node (arrow) of the right abdomen. C-F: 68 Ga-DOTATOC PET. There is evidence of bifocal, intense uptake in two abdominal lymph nodes (arrows). C: maximum intensity projection. D: coronal slice. E: transaxial slice. F: sagittal slice.
MBF, including 82 Rb, 15 O-water, 13 N-ammonia (NH 3 ) and other agents [71][72] . 15 O-water closely meets the requirements for an ideal radiotracer for MBF measurement but with very short half-life. As a K + analog, 82 Rb is actively transported into myocardial cells via Na + /K + pump. For 13 N-NH 3 , it is actively transported into myocardial cells via Na + /K + pump or by passive diffusion of neutral lipid soluble ammonia. On the other hand, a PET radiopharmaceutical, 18 F-Flurpiridaz, has been explored as a PET probe for MPI studies. 18 F-Flurpiridaz, which was also called RP1012 [73] or BMS747158 [74] , is a member of the class of potentiometric lipophilic phosphonium cations originally developed to measure the mitochondrial membrane potential [73][74][75][76] . Based on the preclinical research, it has been observed that high and flow-independent first-pass extraction fraction promises linearity between tracer uptake and MBF. The first human results of BMS747158 were presented at the 2010 SNM annual meeting and then published on the Journal of Nuclear Medicine, and the images demonstrate excellent quality (Fig. 7) [77] .

MEDICINE
The development of new PET probes begins with target identification and functional analysis, followed by probe development, evaluation in animal models, and finally, the probes are evaluated for their usefulness in clinical settings. To achieve and foster the development of PET probes from preclinical to clinical applications, cooperative efforts are required from biologists to identify and validate novel imaging targets; from chemists and radiochemists to synthesize and optimize PET probes; and from engineers and physicists to develop and explore advanced imaging devices and instruments, such as PET/MRI or PET/ optical imaging (OI) and FDA, to accept and approve the clinical trials in human studies.
As shown in Fig. 8, the time intensive and expensive preclinical steps are involved in drug development [78] . Molecular contrast agent design and clinical use involve  a series of steps similar to those used in drug development, including molecular target identification, validation, chemical synthesis and characterization (in vitro and in vivo testing for activity, specificity, biodistribution, pharmacokinetics, off-target effects, toxicity, etc.) for new molecular imaging probes. In fact, the majority of current molecular imaging agents used in the clinic were initially discovered through these exhaustive preclinical experiments at academic institutions [79] . However, as high investment required for imaging probe development and potential limited profit comparing to therapeutics, developing an FDA approved PET probe for clinical use is highly changeling. The creative solutions for translation and approval the use of the personalized probes should be explored and figured out by researchers and clinicians.
Personalized imaging probes and personalized medicine are necessary and important. First, although it is estimated that a molecular imaging agent costs approximately $150 million over 10 years to discover, test and move to the clinic, the ultimate yield would be a $200-400 million per year revenue for successful contrast agents [79] . Second but more importantly, at present, healthcare spending accounts for almost 18% of the United States gross domestic product (GDP) and is growing rapidly, with projections by the Centers for Medicare & Medicaid Services (CMS) that it will reach 20% of the GDP by 2021 if allowed to grow at the current rate [80][81] . The vast majority of current healthcare dollars goes to treatment, with only a small fraction of total expenditures devoted to prediction, diagnosis and treatment monitoring [81] . The personalized imaging probes are expected to guide the treatment for patients, especially the targeted treatment, and significantly improve healthcare delivery with the potential to reduce costs simultaneously. Finally, molecular imaging and therapy can be integrated, whereby a biologically targeted agent provides both efficient contrast for molecular imaging and the delivery vehicle for molecular therapeutics [81] .

CONCLUSION
Currently, PET probes beyond 18 F-FDG are widely explored and developed, and several PET probes have been applied in clinical practice or clinical trials. Thus, in the foreseeable future, although 18 F-FDG will continue to be used in clinics, an increasing number of target specific PET probes will be widely and routinely used in clinical practice, which will greatly contribute to realize personalized medicine. Preclinical steps (green-shaded boxes) involve target identification, validation, chemical labeling, in vitro cellular characterization, and in vivo animal testing. After many optimized steps (arrows between boxes 1, 2 and 3), the imaging probes can be tested in humans after FDA applications for investigatory new drugs (INDs) or exploratory INDs (eINDs). Clinical testing (orange boxes) involves rigorous testing for agent properties (effectiveness, specificity, toxicity, off-target effects, kinetics, etc.) before potential FDA approval for routine clinical practice, which is also heavily monitored for the response to outcomes.