Parametric PET imaging of 5HT2A receptor distribution with 18F-setoperone in the normal human neocortex

Because of 5HT2A receptor's (5HT2AR) putative role in several neuropsychiatric diseases, studying it in vivo is an important goal. 18F-setoperone is a well-validated and widely used PET radioligand for the study of neocortical 5HT2AR. We have previously developed and validated in baboons a method to generate parametric maps of the binding potential (i.e., the k3-to-k4 ratio) on a pixel-by-pixel basis, based on a single-dose tracer amount dynamic 18F-setoperone PET paradigm, and with the receptor-poor cerebellum as reference structure. However, previous semiquantitative PET human studies suggested that nonspecific (NS) binding in the neocortex might not be identical to that in the cerebellum.

As a first step in the development of k3:k4 parametric mapping in humans, we therefore estimated directly the NS binding of 18F-setoperone in the neocortex of four young healthy volunteers who were studied with PET both before and after 2 wk of daily therapeutic oral doses of sertindole, an atypical neuroleptic possessing strong 5HT2AR antagonistic activity.

Visual analysis of the dynamic PET data obtained over 120 min confirmed that virtually full receptor saturation had indeed been achieved; however, the late neocortical time-activity curves (TACs) progressively fell to lower uptake values than corresponding cerebellar TACs and could not be fitted according to a four-compartment (four-Cpt) nonlinear model, indicating lack of specific binding. The cerebellum TACs for both the control and the challenge conditions, as well as the challenge neocortical TACs, were fitted according to three-Cpt modeling, providing the k/k6 ratio and in turn the f2 fraction for both structures. Despite the small sample of only four subjects, the f2 fraction for the neocortex was significantly larger (i.e., NS binding was smaller) than that estimated for the cerebellum. This allowed us to determine the k3-to-k4 ratio for the control neocortex using the challenge neocortex as reference structure, that is, without using the cerebellum at all. This "assumption-free" approach was also successfully used to generate k3:k4 maps for these four subjects, which showed highest values for the temporal cortex.

This study shows that, for every new PET or SPECT radioligand and when estimation of specific binding is based on a reference structure, it is important to determine the uniformity of nonspecific binding before proceeding with human investigations.