Impact of nonsteady plasma glucose level on the calculated cerebral metabolic rates of glucose in mice

Objectives To investigate the accuracy of cerebral metabolic rate of glucose (CMRglc) calculation in mice when the plasma glucose concentration (Cglc) is not constant during FDG PET imaging.

Methods Dynamic PET scans were performed for 60 min on 11 isoflurane-anesthetized (1.5%) male C57BL/6 mice (26±2 g) after tail-vein injection of FDG (14.2±3.8 MBq). Six mice were fed ad libitum and 5 were fasted overnight (17±1 h). Serial blood samples were collected from the femoral artery to measure FDG and Cglc levels, and the rate of Cglc change was determined by linear regression. Time course of FDG uptake constant (Ki) was simulated using a previously determined inverse relationship between Ki and Cglc, according to the rate of Cglc change in each mouse. Brain time-activity curve (TAC) was simulated using the time-varying Ki and the plasma TAC in each mouse. Ki value estimated by Patlak analysis over 15-60 min of the brain TAC was then used to calculate CMRglc with Cglc at 60 min, mean Cglc over 60 min, or mean Cglc over 15-60 min and compared with the original CMRglc value by Friedman test followed by the Dunn’s multiple comparisons test at P<0.05.

Results Initial Cglc was higher in nonfasted than in fasted mice (9.7±2.7 vs 7.0±3.4 mM). A steady increase in Cglc was observed in nonfasted (0.051±0.069 mM/min) and fasted (0.069±0.043 mM/min) mice, probably due to isoflurane anesthesia. Mean bias in Patlak Ki due to variations in Cglc was -8%±9% and -17%±7% in nonfasted and fasted mice, respectively. For the combined group of animals, CMRglc values calculated using the mean Cglc over 60 min were in good agreement (bias: 5%±4%, range: -3% to 12%, P>0.05) with the original CMRglc value when compared to those using the mean Cglc over 15-60 min (bias: 10%±7%, range: -5% to 19%, P<0.05) or Cglc at 60 min (bias: 23%±16%, range: -11% to 42%, P<0.05).

Conclusions Mean plasma glucose level over the entire imaging period should be used in combination with Patlak analysis to calculate CMRglc when the plasma glucose level is not constant.

Research Support NIBIB grant R01-EB00194