Introduction
Positron emission tomography (
PET) and single-photon emission tomography (
SPET) are powerful tools for investigating the pathophysiology of psychiatric illnesses and the action of psychotropic drugs. With these techniques monoaminergic, cholinergic, opioid and benzodiazepine receptors, regional cerebral blood flow, glucose and oxygen metabolism can be measured in the living brain (
Table 2.3.6.1). Thus, neural function of direct relevance to neurochemical and anatomical theories of psychiatric illnesses can be sampled.
Methodology of PET and SPET(1)
In brief, PET and SPET comprise the following:
The production and incorporation of a positron or gamma-emitting radio-isotope into a molecule of biological interest to form a radiotracer administered to humans (
Plate 3).
The use of PET or SPET cameras to detect the emitted gamma radiation from the decaying radio-isotope and hence the 3D distribution of the radiotracer, over minutes to hours, in living human brain (
Plate 4).
Quantification of a physiological parameter of interest, such as number of available receptors or regional cerebral blood flow, from the mathematical modeling of the measured radio-activity in the brain over time (
Plates 5 and
6).
Production of isotopes
Common PET radio-isotopes, produced by a cyclotron, are oxygen-15 (
(15)O), carbon-11 (
(11)C), and fluorine-18 (
(18)F) (with half-lives of 2.03 min, 20.4 min, and 109.8 min respectively) whilst SPET radio-isotopes include technetium (
(99m)Tc) and iodine-123 (
(123)I) (with half-lives of 6.02 h and 13.2 h respectively). With appropriate radiochemistry, isotopes can be incorporated into specific molecules to make radiotracers. Following quality control procedures, to estimate specific activity and radiochemical purity, the radiotracer is injected intravenously into subjects lying in the PET camera (
Plate 3). Importantly, the total mass of radiotracer injected is very small (typically less than 5 µg) and therefore the radiotracer has no pharmacological effect itself.
PET versus SPET
SPET radiotracers are less diverse than PET tracers. However, SPET is cheaper than PET and less technically demanding, making it more readily available in hospitals and research centres. PET radiochemical procedures require in-house automated rapid synthetic chemistry facilities in dedicated hot cells, whereas SPET chemistry is more straight-forward and does not require such extensive facilities. For research and quantitation purposes, PET is far superior to SPET, although any widespread commercial/clinical application is
likely to be SPET based because of the technology restraints and costs associated with PET scanning.
Imaging of radiotracer, data collection, and analysis
PET utilizes the disintegration of positrons emitted from unstable nuclei such as
(11)C (
Plate 4). Emitted positrons travel a short distance in tissue before annihilation by collision with an electron.
(1) On annihilation, two high-energy gamma rays are generated with a separation angle of 180° (
Plate 4). Radiation detectors (e.g. bismuth germanate), 180° apart and linked in electronic coincidence circuits, detect the resulting gamma radiation and therefore localize the source of radiation to a volume between any two detectors (
Plate 3). By arranging rings of detectors around the subject’s head and using computer-based back-projection techniques, the distribution of radiotracer within tomographic slices of the brain can be obtained.
(1) SPET radioisotopes, in contrast, decay by emitting a single gamma ray and therefore the radiation detectors are not linked in coincidence circuits. State-of-the-art PET and SPET cameras have transaxial spatial resolutions of the order of 4 to 5 mm and can detect subnanomolar concentrations of receptors.
(1)Positron-emitting isotopes can be incorporated into molecules associated with diverse biochemical processes in the brain. For example, the positron emitter
(11)C can be incorporated into a molecule WAY 100635, which selectively binds to 5-HT
1A receptors, and injected intravenously in tracer amounts. Brain regions will show different profiles of radio-activity accumulation over time as the radiotracer binds in areas with a high density of 5-HT
1A receptors (medial temporal cortex) whilst in regions with no or sparse receptors (cerebellum), it will be washed out (
Plate 5). By this means, specific and non-specific binding can be distinguished. With an appropriate model of the radiotracer’s history in tissue over time, a quantitative measurement of 5-HT
1A receptor number in tomographic slices of the human brain can be obtained.
(1) With some radiotracers (e.g. [11C]diprenorphine to label opiate receptors) it may be necessary to undertake radial artery cannulation to obtain an ‘input function’
(1) that describes the time course of presentation of radiotracer to the brain (
Plate 6), whereas others tracers can be modeled with a ‘pseudo’ input function from a reference region.
Technical and practical limitations of PET and SPET compared with other imaging modalities
PET and SPET excel in the measurement of neurochemical parameters
in vivo at very low (subnanomolar) concentration. Such sensitivity cannot be matched by other
in vivo methods such as proton magnetic resonance spectroscopy (millimolar range). However, radiation dosimetry limits the number of scans that subjects may receive. Full quantitation can often be achieved with PET, unlike
SPET. However, for imaging blood flow change, or its correlates such as BOLD arterial spin labeling (ASL) and functional magnetic resonance imaging (fMRI), now offer the possibility of repeated measures (without radiation exposure) that far exceed that possible with PET- and SPET-based methods of flow mapping. Full quantitation of blood flow is not yet readily achievable with functional MRI without injection of contrast agents. In contrast, MRI based ASL can achieve full quantification. One disadvantage of functional MRI over PET, for some subjects, is the noisy claustrophobic environment of the scanner, but generally subjects and paradigms studied with PET flow mapping can be readily investigated with functional MRI (see
Chapter 2.3.8), although all test materials in the vicinity of the scanner have to be non-magnetic.
Structural MRI scanning is often used in conjunction with PET activation and ligand binding techniques. The high-resolution anatomical information contained in MRI images can be used to precisely define areas of activation or radiotracer binding observed in PET studies from single subjects.
PET and SPET, and even functional MRI, have relatively poor temporal resolution (seconds) compared with electrophysiological methods such as EEG, event-related potentials, and magnetoencephalography (milliseconds), but these methods in turn suffer from poor spatial resolution. Attempts to integrate information from these different modalities are a major focus of methodological research in many imaging centres.
PET and SPET imaging strategies in psychiatry
These techniques (see
Table 2.3.6.2) are used to either measure brain receptors and neurochemistry, or map functional brain activity via the indices of regional blood flow and glucose utilization. Each approach attempts to define trait and state abnormalities of psychiatric illnesses or the effect of psychotropic drug action.
Because of the technical complexities, it is important to bear in mind the following questions when judging experimental results.
What assumptions are made about the behaviour of the radiotracer in vivo?
Has the radiotracer been well validated for the apparent physiological parameter measured ?
Does the mathematical model of the radiotracer’s behaviour give a good fit to the raw data?
Is the spatial resolution of the PET camera sufficient for the regions measured?
What is the test-test reliability for the PET radiotracer measure?
How have the raw PET images been modified/treated in the data analysis?
How have regions of interest been defined?
Is there a possibility of observer bias in the measurements made?
What statistical techniques have been used, and are the statistical thresholds appropriate?
Do the statistics reflect fixed or random effects and the multiple comparisons made?
