Normoglycemic plasma glucose levels affect F-18 FDG uptake in the brain

The aim of this study was to investigate whether normoglycemic glucose concentrations interfere with cerebral F-18 FDG uptake.

The analysis was based on 2 sets of paired PET scans in 94 patients who were in complete metabolic remission after the successful completion of treatment for lymphoma. For these 188 PET scans, 2 subgroups were defined according to the plasma glucose level at the time of scanning. Group 1 contained the PET images that were associated with the lower of both normoglycemic plasma glucose levels, whereas group 2 contained the PET images that were associated with the higher of both plasma glucose levels. SUVs (standard uptake values) in the cerebellum between both groups were compared using paired sample T test. Subsequently, SUVs were normalized to a standard glucose concentration and normalized SUVs were again compared. Further, we calculated the coefficient of variation of SUVs in group 1 and 2 both before and after the normalization step.

Mean plasma glucose level was 86 mg/dL (SD of 9 mg/dL) in group 1 and 97 mg/dL (SD of 10 mg/dL) in group 2. Mean SUV was 3.8 (SD of 1.1) for group 1 and 3.5 (SD of 1.1) for group 2. Mean SUV in group 1 was slightly but statistically significantly higher than the mean SUV in group 2 (p < 0.01). Mean normalized SUV was 3.6 (SD of 1.1) in group 1 and 3.7 (SD of 1.3) in group 2. A paired comparison between normalized SUVs in both groups indicated that there was no statistically significant difference (p < 0.31). The coefficient of variation for the SUVs in group 1 and 2 before normalization was 29 and 30%, respectively. The coefficient of variation for the normalized SUVs in group 1 and 2 was 30 and 34%, respectively.

Our results indicated that plasma glucose levels that are within the normoglycemic range have a small but systematic effect on F-18 FDG uptake in the brain (following an inverse relationship). Normalizing plasma glucose levels to a standard glucose concentration successfully reduced the intra-subject variability of SUV measures. Inter-subject variability, however, remained high suggesting that other factors have an influence as well.