Positron emission tomography : development, evaluation and application of quantification methods

Abstract

Positron Emission Tomography (PET) is an imaging technique that allows for in vivo quantification of biochemical and physiological processes in the brain. Examples of targets in the brain that can be imaged using PET are dopamine receptors and the translocator protein 18kDa (TSPO). Following intravenous injection of a radio-labeled ligand and the ensuing PET examination of a subject, kinetic models are often used to estimate parameters of interest. Example of such parameters are binding potential or distribution volume, which are estimates of the availability of receptors in a specific region or volume-element of the brain. These parameters can then be inserted into statistical models to infer e.g. differences in target availability between patients and controls or relationships to behavioral traits. In order to detect effects of interest, it is important that the estimation of these parameters is precise, reliable and valid. The aim of this thesis was to evaluate different methods for estimating such parameters, and apply them on clinical data.

The thesis consists of two different themes. The focus of theme I was the quantification of dopamine receptor in striatum and the cortex, and their relationship to normal and dysfunctional social behavior. Study I and Study II examined the relationship between dopamine D1 receptor availability and self-rated pro and anti-social behavior in healthy subjects. Study I found a positive correlation between striatal D1 receptor availability and Social Desirability, and a negative correlation to Trait Aggression. Study II did however fail to replicate these results. In Study III, dopamine D2 receptor availability in limbic and cortical regions in patients with social anxiety disorder and healthy controls were compared. Exploratory analyses suggested that patients had higher D2 receptor availability in the lateral and orbitofrontal cortex, although the results warrant replication in a larger sample.

The focus of theme II was the quantification of TSPO in patients with psychosis and healthy subjects. The level of TSPO in the brain has been hypothesized to function as an index of microglial cell activity, which in turn is believed to be a proxy for immune activation in the central nervous system. In Study IV, [11C]PBR28 binding in the whole grey-matter in patients with first-episode psychosis and healthy controls were compared. Contrary to the hypothesis of elevated microglia activity, patients were found to have lower TSPO levels. Study V evaluated the test-retest reliability and convergent validity of different methods to measure TSPO levels using [11C]PBR28. Distribution volume ratios and standardized uptakes value ratios, derived using pseudo-reference regions, showed both poor reliability and convergent validity. Study VI carried out a meta-analysis of TSPO in patients with schizophrenia and psychotic disorders compared to healthy controls. Again, contrary to the hypothesis of higher microglia activity, strong evidence was found in favor of patients having lower TSPO levels in both cortical and subcortical regions. In Study VII, the test-retest reliability and convergent validity of different methods to estimate TSPO levels using (R)-[11C]PK11195 were evaluated. Outcomes derived using pseudo-reference region approaches were unreliable and showed no convergent validity to outcomes derived using arterial input function. Finally, Study VIII evaluated the reliability and accuracy of a new modeling method, applied to [11C]PBR28 data, in order to estimate specific binding without requiring a reference region. Simulations, a pharmacological challenge and test-retest analysis showed that non-displaceable distribution volume, and ensuing specific distribution volume values, derived using this method were accurate, precise and reliable.

Taken together, the results of the studies illustrate the importance of evaluating quantification methods prior to applying them on clinical data. The thesis also shows how robust kinetic and statistical modeling, and the use of direct replications or multi-center collaborations, can yield more trustworthy and reliable findings in PET.