Simplifications in analyzing positron emission tomography data: effects on outcome measures

Initial validation studies of new radiotracers generally involve kinetic models that require a measured arterial input function. This allows for the separation of tissue binding from delivery and blood flow effects. However, when using a tracer in a clinical setting, it is necessary to eliminate arterial blood sampling due to its invasiveness and the extra burden of counting and analyzing the blood samples for metabolites. In some cases, it may also be necessary to replace dynamic scanning with a shortened scanning period some time after tracer injection, as is done with FDG (F-18 fluorodeoxyglucose). These approximations represent loss of information. In this work, we considered several questions related to this: (1) Do differences in experimental conditions (drug treatments) or populations affect the input function, and what effect, if any, does this have on the final outcome measure? (2) How do errors in metabolite measurements enter into results? (3) What errors are incurred if the uptake ratio is used in place of the distribution volume ratio? (4) Is one- or two-point blood sampling any better for FDG data than the standardized uptake value? and (5) If blood sampling is necessary, what alternatives are there to arterial blood sampling? The first three questions were considered in terms of data from human dynamic positron emission tomography (PET) studies under conditions of baseline and drug pretreatment. Data from [11C]raclopride studies and those from the norepinephrine transporter tracer (S,S)-[11C]O-methyl reboxetine were used. Calculation of a metabolic rate for FDG using the operational equation requires a measured input function. We tested a procedure based on two blood samples to estimate the plasma integral and convolution that occur in the operational equation. There are some tracers for which blood sampling is necessary. Strategies for brain studies involve using the internal carotids in estimating the radioactivity after correcting for partial volume and spillover in order to eliminate arterial sampling. Some venous blood samples are still required for metabolite measurements. The ultimate solution to the problem of arterial sampling may be a wrist scanner, which acts as a small PET camera for imaging the arteries in the wrist. This is currently under development.