Imaging of monoamine oxidase-A in the human brain with [11C]befloxatone: quantification strategies and correlation with mRNA transcription maps

Introduction[11C]Befloxatone is a highly specific, reversible, and selective radioligand for brain PET imaging of monoamine oxidase-A and can be quantified by a two-tissue compartment model (2TCM) and an arterial input function. The aims of the present study were the following: (a) to assess whether in-vivo protein concentration, as measured by [11C]befloxatone total distribution volume (VT), is correlated with post-mortem mRNA expression; (b) to replicate in a population of tobacco smokers the results of a recent study on healthy nonsmokers, which showed that spectral analysis (SA) provides a highly accurate estimation of [11C]befloxatone-VT at the voxel level; and (c) to validate the use of an input function that would not require arterial sampling. Materials and methodsHealthy male nonsmokers (n=7) and smokers (n=8) were imaged with PET and [11C]befloxatone. Binding was quantified at the regional and voxel level with the Logan plot, multilinear analysis (MA1), and SA. VT values were compared with the reference values obtained by 2TCM at the regional level. [11C]Befloxatone binding was compared with mRNA transcription maps from the Allen Human Brain Atlas. A less-invasive input function was obtained with a population-based input function (PBIF) scaled with arterialized venous samples. ResultsmRNA expression was highly correlated with in-vivo 2TCM-VT values both for nonsmokers (R2=0.873; P<0.0001) and for smokers (R2=0.851; P<0.0001). At the regional level, both Logan and MA1 showed a moderate negative bias (−5 to −10%) compared with the reference VT values. With the exception of a single outlying individual, SA showed little bias and variability (+4.4±3.5%). Although variability was higher than at the regional level, SA provided the most accurate VT estimations at the voxel level: all but one participant had an error of less than 20%. Parametric Logan and MA1 analyses gave highly biased or unusable results. PBIF provided good results in all participants in whom the arterialization of venous blood was successful (all errors of about 10% or less). Conclusion[11C]Befloxatone binding is strongly correlated with the values of mRNA transcription measured in post-mortem brains. At the voxel level, SA is the best available choice for [11C]befloxatone quantification, although a higher variability must be expected. When an arterial input function is not technically feasible, a PBIF scaled with arterialized venous samples may provide an acceptable alternative, provided an optimal arterialization can be achieved.