Models and methods for derivation of in vivo neuroreceptor parameters with PET and SPECT reversible radiotracers

Abstract

The science of quantitative analysis of PET and SPECT neuroreceptor imaging studies has grown considerably over the past decade. A number of methods have been proposed in which receptor parameter estimation results from fitting data to a model of the underlying kinetics of ligand uptake in the brain. These approaches have come to be collectively known as model-based methods and several have received widespread use. Here, we briefly review the most frequently used methods and examine their strengths and weaknesses. Kinetic modeling is the most direct implementation of the compartment models, but with some tracers accurate input function measurement and good compartment configuration identification can be difficult to obtain. Other methods were designed to overcome some particular vulnerability to error of classical kinetic modeling, but introduced new vulnerabilities in the process. Reference region methods obviate the need for arterial plasma measurement, but are not as robust to violations of the underlying modeling assumptions as methods using the arterial input function. Graphical methods give estimates of V_T without the requirement of compartment model specification, but provide a biased estimator in the presence of statistical noise. True equilibrium methods are quite robust, but their use is limited to experiments with tracers that are suitable for constant infusion. In conclusion, there is no universally “best” method that is applicable to all neuroreceptor imaging studies, and carefully evaluation of model-based methods is required for each radiotracer.

Introduction

The objective of a PET or SPECT neuroreceptor imaging study is to obtain quantitative information about the distribution of the target receptor throughout the brain. Several factors must be taken into account in order to form accurate inferences about receptor parameters. The activity recorded by the scanner will be due to some combination of specifically bound, non-specifically bound, and unbound radiotracer. The proportions of total measured activity represented by each of these pools are time varying and interdependent quantities. In addition to the actual distribution of receptors, peripheral clearance, regional cerebral blood flow and transport across the blood brain barrier will all influence the profile of activity over time. These factors can vary greatly from subject to subject and should be accounted for if results are to be adequately interpretable. Therefore, analysis of neuroreceptor imaging studies should not be limited to empirical outcome measures, such as the ratio of activity in a region of interest (ROI) to that in a region of reference (R_REF) at one point in time. Model based methods relate the observed ROI time activity curves (the observed response function) to the arterial time activity curve (the observed input function) through a defined model (convolution with an unobservable but inferred impulse response function).

Model-based methods for in vivo quantification of neuroreceptors can be divided into kinetic, equilibrium and graphical methods. Kinetic methods yield quantitative information about the receptors from the estimation of the rate constants that characterize the tracer transfer between plasma, brain, and receptors [39]. Equilibrium methods derive this information from analysis of the activity distribution at equilibrium (i.e., when the receptor-ligand association and dissociation rates are equal) [13], [28]. In the graphical approach, data are transformed to new variables that are linearly related, and the parameters of interest are derived by linear regression [36], [44]. In this paper, we will look at several formulations in each of these categories. While our choices are not intended to comprise an exhaustive list, we do believe they are representative of the range of analysis techniques in current use.