PET imaging of the 18 kDa translocator protein (TSPO), a biomarker of microglial activity, receives growing interest in clinical and preclinical applications of neuroinflammatory and neurodegenerative brain diseases. In globally affected brains, intra-cerebral pseudo reference regions are not feasible. Consequently, many brain-independent approaches have been attempted, including SUV analysis and normalization to muscle- or heart uptake, aiming to stabilize quantitative analysis. In this study, we systematically compared different image normalization methods for static late phase TSPO-PET imaging of rodent brain.Methods:
We first obtained gamma counter measurements for gold standard quantitation of [18F]GE180 uptake in brain of C57Bl/6 mice (N = 10) after PET, aiming to identify factors contributing significantly to the quantitative results. Subsequently, data from a large cohort of C57Bl/6 mice (N = 79) were compiled to precisely determine the weighted influence and variance attributable these factors by regression analysis. Scan-rescan variability and agreement with histology were used to validate the tested normalization methods in an Alzheimer's disease (AD) mouse model with pathologically increased TSPO expression (PS2APP; N = 24). Longitudinal data from AD model mice (N = 10) scanned at four different ages were used to challenge and validate the different normalization methods in a practical application.Results:
Gamma counter results revealed that injected dose, body weight and PET-measured radioactivity concentration in the ventral myocardium all significantly accounted for [18F]GE180 activity in the brain. Skeletal muscle activity had high test-retest variance in this PET only application and was therefore pursued no further. Regression analysis of the large scale evaluation showed that scaling to injected dose or SUV analysis accounted for little variance in brain activity (R2 < 0.5), but inclusion of myocardial activity together with injected dose and body weight in the regression model accounted for most of the variance in brain uptake (R2 = 0.94). Scan-rescan stability, correlation with histology and applicability for longitudinal examination in the disease model were also significantly improved by inclusion of myocadial uptake in the quantitative model.Conclusion:
Cerebral and myocardial TSPO expression are highly coupled under physiological conditions. Myocardial uptake has great potential for stabilization of static late phase [18F]GE180 quantification in brain in the absence of a valid intra-cerebral pseudo-reference region.