Positron Emission Tomography (PET)
What is it? What does it measure?
Positron emission tomography is a non-invasive imaging method to obtain quantitative molecular and biochemical information of physiological processes in the body, which means that PET imaging shows the chemical functioning of organs and tissues in the living object. The concept of positron emission imaging was developed already in 1951 and the first human studies were published in 1953. Since then fundamental technical developments have taken place in such scientific fields as electronics, computer science, biology and biochemistry. These developments have also significantly advanced the development of imaging techniques based on positron emission. Even though positron imaging technology is 60 years old, its use in the clinical environment has been recognized in the past 20 years. The limited availability of suitable positron labeled ligands has been the biggest block to delay clinical applications.
How does it work? What equipment is needed?
Radioactive compounds (positron emitters) are needed to conduct PET imaging studies. To obtain PET image, radioactive compound has to be administered into an object. This radioactive compound can be gas to investigate hemodynamics or oxygen metabolism or it can be ligand, which has be labeled with radioactivity before administering into the object, like fluorine-18 labeled 2-fluoro-2-deoxy-D-glucose to investigate glucose metabolism or some radiolabeled receptor agonist or antagonist to investigate specific receptor function. PET is based on the detection of radioactivity (annihilation quanta) of the administered compounds (ligands, drugs).
Since the commonly used radioactive molecules have short half-life (oxygen-15 2 min, nitrogen-13 10 min, carbon-11 20 min and fluorine-18 110 min) they have to be produced in the proximity of the PET imaging device. An accelerator, what in the hospital environment is most often a cyclotron, produces these molecules. Since the cyclotron produces only radioactive molecules, an additional robotic system is needed to introduce radioactive atoms into the organic compounds to produce desired ligands (radiosynthesis) to be used in the imaging studies. Table 1 shows the most commonly used radioligands and their relation to the physiological functions and target areas.
When radioactive ligand is administered into the object, it will distribute with blood stream into the body and accumulate into the target tissue. Detection of this accumulation is based on nuclear physical process “annihilation”, where mass of particles is turned to energy. Briefly, in a positron emitter the nuclear origin is a proton. Under the influence of other nucleons the proton is converted into a neutron, a positron and a neutrino. The immediate result of the breakdown is that the positron and the neutrino are ejected from the nucleus while the neutron remains (Figure 1). The released positron carries kinetic energy depending on the binding energy in the nucleus and loses it while traveling in the surrounding media. The annihilation reaction is a result of a collision between the positron and a negatively charged electron. In annihilation, the masses of the two particles completely turn into energy. The energy is divided equally between the two photons (each of 511 keV), traveling to the opposite directions (180 +/-0.25o) of each other
This released photon pair can hit the scintillation crystals connected with coincidence logic and can be detected as a line between the crystals. This information is used to generate images of activity distribution (Figure 1).
Table 1. A summary of the most commonly used radioligands in PET imaging and their biological targets
Physiological function |
Radioligand |
Disease or Target |
|
|
Hemodynamics: |
Blood flow |
C15O2, H215O |
Stroke, Infarct, Activation, Cancer |
Blood volume |
C15O, 11CO |
Perfusion |
13NH3 |
Myocardium |
Metabolism: |
Glucose metabolism |
[18F]FDG |
Cancer, Stroke, Infarct, Alzheimer’s disease, Huntington’s disease |
Oxygen metabolism |
15O2 |
Amino acid metabolism |
[11C]methionine |
Cancer |
Fatty acid metabolism |
[11C]palmitic acid,
[18F]fluoro-6-thia-heptadecanoic acid |
Myocardium |
Receptor Function:
|
Dopamine transporter |
[11C]CFT, [18F]fluoro-L-dopa, [11C]altropane |
Parkinson’s disease,
Schizophrenia,
Drug addiction |
Dopamine D1-receptor |
[11C]-SCH 23390 |
Dopamine D2- receptor |
[11C]raclopride, [18F]fallypride |
5-HT (1A) -receptor |
[11C]WAY 100635 |
Schizophrenia |
Benzodiazepine-receptor |
[11C]flumazenil |
Epilepsy |
Opioid receptor |
[11C]diprenorphine,
[11C]carfentanil |
Pain |
Estrogen receptor |
16 alpha-[18F]fluoro-17beta-estradiol |
Cancer |
Molecular mechanism:
|
Proliferation potential |
3-deoxy-3-[18F]fluoro-thymidine, |
Cancer |
DNA synthesis |
[11C]thymidine |
Cancer |
Herpes simplex virus
type 1 thymidine
kinase (HSV 1-tk) |
[18F]5-fluorouracil,
8-[18F]fluoroganciclovir, [18F]acycloguanosine, [124I]FIAU |
Cancer |
 |
Figure 1. Schematic diagram of the process, where a positron is released from the nucleus of the radioactive marker. After that the released positron travels in the surrounding media and loses its energy, and finally collides with a negatively charged electron. In this process, annihilation, the masses of the two particles completely turn into energy, which is divided between the two photons (each 511 keV) traveling in opposite directions (180+/- 0.25 o) of each other. The released photon pair can be detected as a line between the detectors and the obtained information is used to generate images of activity distribution.
|
Data acquisition is a compromise of the characteristics of the tomograph and the requirement of the biological application. Injected activity (radiation dose) and acquisition time are often limited by the biological system. The acquisition parameters in the system are adjusted with these guidelines. The old classical data acquisition method is 2-dimensional, which means that the data is acquired in a plane. The recent technical development in detector design has enabled 3-dimensional data acquisition, which means that data can be acquired both in the cross plane and in the axial direction. Three dimensional data acquisition increases sensitivity, but also internal scatter and random coincidences will be higher than in 2-dimensional acquisitions. Additional technical factors affecting data acquisition and final image quality are sampling, statistical noise, scatter and attenuation.
 |
Figure 2. (left) The acquired data includes information of location of the originated photoemission (X, Y, Z), time and intensity. This information is converted to polar coordinates and expressed as sinograms. (right) A PET image of the resolution phantom demonstrates the fact that PET imaging can visualize objects, which are far smaller than the resolution element. The Resolution of this PET device was 4.5 mm (FWHM) and the smallest block of radioactive rods had diameter of 2.0 mm and were separated by 10 mm (between the midpoints of the rods). The rods are clearly visible as separate dots. Convolution backprojection was used for image reconstruction. Profiles of the activity distribution are presented at the midlevel of different dots. |
Photons created through annihilation loose energy when passing through the tissue depending on the thickness and density of the tissue. Especially if surrounding tissues have asymmetrical bone structures, attenuation correction is necessary to obtain activity distribution. Measurement for attenuation correction can be done in two ways. The most accurate method is to perform a separate imaging for transmission and correct the sinograms with the transmission data. This is done in most commercial scanners by using rotating line or point sources. The other method is to use a mathematical approach based on determining the length of the photon range inside the surface contours. In combined PET-CT imaging systems, transmission image is conducted by CT.
Independently of the data acquisition mode (2- or 3-dimensional), reconstruction is often done in a 2-dimensional mode to obtain transverse, coronal and/or sagital slices with a variety of slice thickness depending on the statistical count distribution and biological questions. To reconstruct a data set in 2-dimensional mode after 3-dimensional acquisition requires rebinning of the sinograms. The sinograms are corrected for uniformity, scatter and attenuation before reconstruction of images. Also corrections for acquisition time, decay and injected radioactivity can be included to the corrections before reconstruction.
Reconstruction algorithms have been developed both for 2- and 3-dimensional approaches. The most commonly used algorithm is filtered backprojection. Also convolution backprojection can be used to reconstruct 2-dimensional data sets (Figure 2). Resolution in filtered backprojection as well as in convolution backprojection is affected mostly by the selection of the filter function and its cut-off value, which depends on the noise level of the data. Also several arithmetic iterative reconstruction techniques are available for 2 and 3-dimensional data analyses including the basic ordered-subsets expectation-maximization (OSEM) algorithm and attenuation weighed OSEM (WA-OSEM) as well as Fourier binning OSEM (FORE). Selection of the reconstruction algorithm depends mainly on the biological application.
Since quantitative parameters of metabolism or receptor function are normally required for organ or tissue, a good contrast is required to outline the borders of the different tissues. With the present multimodality imaging techniques it is possible to fuse PET images with anatomical magnetic resonance images and obtain the needed anatomical borderlines. A number of different biological models have been developed for quantification of PET imaging studies so that the final results can express biological function in metric units. There are generally approved models to quantify hemodynamics, glucose metabolism and several models for receptor binding.
Temporal resolution
Resolution in positron imaging is limited by positron range, small angle deviation, sensitivity, count rate capability and sampling. Especially in quantification of radioactivity concentration, the spatial resolution is an important factor. The size of the object should be at least two times the spatial resolution to avoid partial volume effects. If the object is smaller there will be an underestimation in activity concentration. However, on the visualization point, partial volume effect may not be a problem if there is a high activity level or high contrast between the lesion and the background activity (Figure 2).
The spatial resolution of the positron imaging devices is 4–5 mm in the whole body scanners and 2 mm in the head scanners. In the animal scanners the resolution is 1-2 mm.
The biggest limitation of PET studies is limited availability of radioligands. Being developed more than 60 years ago, and immersed to the clinical environment during the past 20 years based only by studies of glucose metabolism, it will take a lot of efforts to simplify and standardize radiopharmaceutical production methods to introduce new radioligands with competitive prices for the needs of PET imaging. In addition, since PET isotopes are short-lived, they have to be processed near the imaging facility.
The variety of biological applications of PET imaging is based on the superior sensitivity of detection of radioactivity. The obvious applications are studies of functional mechanisms, where a tracer amount of radioactivity is introduced into the system without changing its physiological environment. Biological applications have developed together with radiopharmaceutical development, especially with the development of chemical synthesis and labeling techniques. Carbon-11 and fluorine-18 are the most commonly used positron emitters in these labeling procedures. Recently, there is also a trend to develop labeling techniques for small peptides using copper-64 (t1/2= 12.8 hours) and iodine-124 (t1/2= 4.2 days). The commonly used radiopharmaceuticals with their targeted application are summarized in Table 1.
Imaging of Hemodynamics
Stroke: Detection of hemodynamic changes in brain blood flow has been one of earliest clinical applications of PET techniques. These techniques were used intensively in stroke studies to localize a site and extend as well as follow up the recovery. Recently, MRI techniques have overcome many of these applications. When separate information is needed of blood flow or blood volume changes and the possible connection to oxygen consumption, only PET studies can provide information of all these hemodynamic aspects.
Activation studies: PET studies have been extensively used to evaluate hemodynamic changes induced by stimulations to evaluate the activation and connectivity in different brain circuitries. These stimulations can be introduced through all the senses, but visual stimulation is the most commonly used technique combined with different tasks.
Infarct: PET studies of the hemodynamic changes in the heart are often combined with studies of glucose utilization to obtain more clinically relevant information, especially differentiation of ischemic and infarcted tissue areas (Figure 3). If only blood flow studies are done, it is necessary to repeat the studies during rest and exercise or combine blood flow studies with evaluation of wall motion in order to differentiate viable versus non-viable tissue areas.
|
|
Figure 3. PET studies of blood flow using 13NH3 as tracer and glucose metabolism using [18F]FDG as tracer. The upper row shows mismatch of blood flow and glucose utilization indicating an ischemic heart condition. Lower row shows similarly decreased blood flow and glucose utilization indicating an infarct. |
Imaging of Metabolism
The availability of fluorine-18 labeled 2-fluoro-2-deoxy-D-glucose ([18F]FDG) has significantly enhanced PET studies based on glucose metabolism as well as instrumentation development to the level of the concept of FDG-PET. The most important application of FDG-PET is in cancer studies, especially in whole-body imaging in order to localize primary tumors and their metastases.
Glucose is the main source of energy in the brain. Therefore all the changes in functionality create some changes in regional glucose metabolism. Glucose metabolic rate can be used as an index of malignancy of brain tumors as well as to prognosis of the therapy. Using [18F]FDG, it is possible to obtain invaluable information of all other biological processes except those that are based on specific receptor recognition mechanisms.
Imaging of receptor function
Imaging of the receptor function has significantly enhanced understanding of the underlying pathophysiological mechanisms especially in many neurological disorders. There are several specific ligands to investigate dopaminergic (Figure 4) and serotonergic receptor functions (Table 1). PET studies of dopaminergic function have enabled an early diagnosis of Parkinson’s disease as well as to develop and test therapeutic regimens for repairing the function. Dopaminergic function is affected in neurodegenerative disorders, schizophrenia, drug addiction (Figure 5) and in many psychological conditions such as anger, anxiety, and hyperactivity.
PET imaging is used also to assess the kinetics of drug-receptor interaction to determine the presence and level of in vivo expression of receptors in tumors without the necessity of biopsy and to aid in selection of the appropriate therapy.
Figure 4. Distribution of [11C]2ß-carbomethoxy-3ß- (4-fluorophenyl)tropane, [11C]CFT in a primate brain. [11C]CFT is a sensitive ligand for dopamine transporters. High accumulation (red color) is an indication of high receptor density in caudate and putamen.
|
 Figure 5. Distribution of 11C-raclopride in a rat brain before and 24 hours after cessation of self-administration of cocaine. 11C-raclopride is a sensitive ligand for dopamine D2 receptors. The decreased accumulation of [11C]raclopride in the striatum after cocaine administration indicates deficit in dopamine D2 receptor function. |
Imaging of Molecular Mechanism
In molecular imaging, 18F-labeled compounds with high specific activity have a central role, since biological processes, incorporation of ligands into the cells, genes or DNA may take hours. For example 18F is used in the synthesis of 18F-labeled acycloguanosine analogues and other compounds used to image Herpes simplex virus-1 thymidine kinase to follow gene therapy and viral oncolysis (Figure 6) or production of 18F- labeled 3’-deoxy-3’-fluorothymidine (FLT) to image proliferation potential in the cell. DNA synthesis can be imaged by 11C-labeled thymidine. Thus PET studies can be used to evaluate cell-based responses on chemotherapy.
 |
Figure 6. A mouse carries identical MC26 cell line tumors on the left and right flank. The tumor on the right was administered with virus expressing HSV-1 tk.
(Upper row): PET imaging with [18F]FHBG shows enhanced accumulation in the right tumor indicating active viral function.
(Middle row): PET imaging with [18F]FLT shows enhanced proliferation potential in the untreated left tumor and significantly decreased accumulation in the right tumor as an indication of cell loss.
(Lower row): PET imaging of glucose metabolism shows enhanced metabolism in the left untreated tumor and decreased metabolism in the right viral treated tumor. |
Drug Development
PET has the potential to be a powerful tool in drug discovery and development. PET imaging is able to non-invasively assess drug distribution and action at the molecular level. Positron emitters can be introduced to the drug candidates and used as radiolabeled ligands. It has been well established that substituting a stable atom with a radioisotope of the same element does not affect the physicochemical, pharmacokinetic or biological properties of a drug. Preliminary studies indicate that dynamic PET imaging, using repeated images over time, can be a valuable technique for defining the time course of uptake and retention of radiolabeled anticancer drugs in tumors and in the surrounding normal tissue in patients. These drugs are designed to inhibit key
processes in cancer initiation and progression: angiogenesis, proliferation, avoidance of cell death or apoptosis, invasion and metastasis, and transduction of signals that modulate these processes. Many of these drugs are inhibitors of cell surface receptors or inhibitors of intracellular enzymes that transmit signals from receptors to the nucleus. PET can evaluate effects of drugs in small targets, and only trace amounts (nano or picomoles) of the labeled ligand (drug) are necessary to be administrated into the object.
|
| Figure 7. Coronal PET images of the accumulation of metabotropic glutamate 5 receptor (mGluR5) ligand, 2-(2-(5-[11C]methoxypyridin-3-yl)ethynyl)pyridine, [11C]MPEPy in a normal primate brain show location and receptor density of the mGluR5 receptors in caudate (Cd), putamen (P), nucleus accumbens (Acc), amygdala (Am), thalamus (Th), hippocampus (Hp) and visual area (Va). Accumulation of [11C]MPEPy in the cerebellum (Ce) is low as an indication of missing receptor sites. |
In the clinical environment, PET has established its power in cancer studies. However, these studies utilize only a fraction of the possibilities that can be obtained by PET imaging. In the future, the main challenge of PET imaging is in drug development based on different biological and functional mechanisms. The new ligands and drugs can be used in humans to enhance early diagnosis and predict and follow the progression of the therapy. The biggest challenge is to develop new radioactive ligands to investigate in vivo different receptor and enzyme functions. For example, the amino acid, glutamate, mediates most of the excitatory synaptic transmissions in the brain and the glutamatergic pathways are involved in diverse processes like epilepsy, ischemic brain disease, pain, learning etc (Figure 7). The potential applications of PET imaging in drug development have just begun to be studied and the preliminary successes of these approaches are evidence of the future power of PET imaging.
by Anna-Liisa Brownell, updated 05/2011
