Background & aims: The liver is inaccessible to organ balance measurements in humans. To validate [(18)F]fluorodeoxyglucose ([(18)F]FDG) positron emission tomography (PET) in the quantification of hepatic glucose uptake (HGU), we determined [(18)F]FDG modeling parameters, lumped constant (LC), and input functions (single arterial versus dual).
Methods: Anesthetized pigs were studied during fasting (n = 6), physiologic (n = 4), and supraphysiologic (n = 4) hyperinsulinemia. PET was performed with C(15)O (blood pool) and [(18)F]FDG (glucose uptake). 6,6-Deuterated glucose ([(2)H]G) was coinjected with [(18)F]FDG and blood collected from the carotid artery and portal and hepatic veins to compute LC as ratio between tracers fractional extraction. HGU was estimated from PET images and ex vivo from high-performance liquid chromatography measurements of liver [(18)F]FDG versus [(18)F]FDG-6-phosphate and [(18)F]-glycogen. Endogenous glucose production was measured with [(2)H]G and hepatic blood flow by flowmeters.
Results: HGU was increased in hyperinsulinemia versus fasting (P < .05). Fractional extraction of [(18)F]FDG and [(2)H]G was similar (not significant), intercorrelated (r = 0.98, P < .0001), and equally higher during hyperinsulinemia than fasting (P <or= .05), with an LC of 0.98 +/- 0.10 and 1.18 +/- 0.26, respectively. [(18)F]FDG-PET modeling provided HGU values that did not differ from, and were correlated with, those from ex vivo measurements (r = 0.61, P <or= .02); proportional estimates of liver perfusion and endogenous glucose production were also obtained. Single and dual input functions produced strongly intercorrelated results (r > 0.95, P < .0001), with a modest underestimation of HGU by the former.
Conclusions: [(18)F]FDG-PET-derived parameters provide accurate quantification of HGU and estimates of liver perfusion and glucose production. In the liver, LC of [(18)F]FDG is nearly unitary. Using a single arterial input introduces only a small error in estimation of HGU.