Estimation of the binding potential BPND without a reference region or blood samples for brain PET studies

Abstract

Binding potential (BPND) is a commonly used PET outcome measure because it can be estimated without blood sampling if a brain reference region (RR) devoid of the target of interest exists. For many radioligands, however, no RR exists, and the total distribution volume (VT), whose estimation requires arterial blood sampling, is normally considered as the outcome measure. Here, we present a method that allows calculation of BPND without requiring either blood samples or a RR.The method extends our previous algorithm for estimating non-displaceable distribution volumes (VND) without using a RR. Here we show that if a template input function, with arbitrary amplitude but a shape similar to the actual arterial input function, is used in the algorithm, estimation of VT and VND are both proportionally biased, and thus this bias cancels out in the estimation of BPND. The method is evaluated using simulated data, human data acquired with the serotonin 1A receptor radioligand [11C]WAY-100635, and blocking data acquired in baboons using the serotonin 1A receptor radioligand [11C]CUMI-101.We evaluated two versions of template input functions: an arbitrarily downscaled version of the actual arterial input function, and an unscaled population-based input function. In addition, we evaluated how shape modifications of the template input function impact the estimates of BPND.With the downscaled input function, BPND values close to the gold standard were obtained. When the unscaled population-based based input function was used, greater variability was observed but no discernable bias was introduced. When the input function shape was modified, a systematic but small bias in BPND was introduced.We conclude that, provided the shape of the arterial input function is adequately described, determination of its amplitude is not necessary for estimation of BPND.