Methodologic Considerations for Quantitative 18F-FDG PET/CT Studies of Hepatic Glucose Metabolism in Healthy Subjects

PET with the glucose analog (18)F-FDG is used to measure regional tissue metabolism of glucose. However, (18)F-FDG may have affinities different from those of glucose for plasma membrane transporters and intracellular enzymes; the lumped constant (LC) can be used to correct these differences kinetically. The aims of this study were to investigate the feasibility of measuring human hepatic glucose metabolism with dynamic (18)F-FDG PET/CT and to determine an operational LC for (18)F-FDG by comparison with (3)H-glucose measurements.

Eight healthy human subjects were included. In all studies, (18)F-FDG and (3)H-glucose were mixed in saline and coadministered. A 60-min dynamic PET recording of the liver was performed for 180 min with blood sampling from catheters in a hepatic vein and a radial artery (concentrations of (18)F-FDG and (3)H-glucose in blood). Hepatic blood flow was determined by indocyanine green infusion. First, 3 subjects underwent studies comparing bolus administration and constant-infusion administration of tracers during hyperinsulinemic-euglycemic clamping. Next, 5 subjects underwent studies comparing fasting and hyperinsulinemic-euglycemic clamping with tracer infusions. Splanchnic extraction fractions of (18)F-FDG (E*) and (3)H-glucose (E) were calculated from concentrations in blood, and the LC was calculated as ln(1 - E*)/ln(1 - E). Volumes of interest were drawn in the liver tissue, and hepatic metabolic clearance of (18)F-FDG (mL of blood/100 mL of liver tissue/min) was estimated.

For bolus versus infusion, E* values were always negative when (18)F-FDG was administered as a bolus and were always positive when it was administered as an infusion. For fasting versus clamping, E* values were positive in 4 of 5 studies during fasting and were always positive during clamping. Negative extraction fractions were ascribed to the tracer distribution in the large volume of distribution in the prehepatic splanchnic bed. The LC ranged from 0.43 to 2.53, with no significant difference between fasting and clamping.

The large volume of distribution of (18)F-FDG in the prehepatic splanchnic bed may complicate the analysis of dynamic PET data because it represents the mixed tracer input to the liver via the portal vein. Therefore, dynamic (18)F-FDG data for human hepatic glucose metabolism should be interpreted with caution, but constant tracer infusion seems to yield more robust results than bolus injection.