Determinants of [18F]-FDG Uptake in Adipose Tissue: Insights from a Generalized Additive Mixed Model Approach from a large retrospective PET cohort

Background Beyond its oncological role, [18F]-FDG PET is increasingly used to quantify adipose tissue (AT) metabolism. Technical and clinical parameters may impact [18F]-FDG accumulation and should be accounted for. We aimed to model how these parameters impact [18F]-FDG activity in AT using a large retrospective cohort. Methods 3230 [18F]-FDG PET-CT scans (2011-2014) from 2694 patients were semi-automatically segmented for subcutaneous AT (SAT) and visceral AT (VAT). A generalized additive mixed model (GAMM) was fitted on log-transformed [18F]-FDG activity per volume (kBq/cm3), decoupled from weight and dose. Fixed effects included AT type, sex, and cancer stage; non-linear smooth terms modeled blood glucose, blood volume, and dose; and a tensor product smooth of age and BMI was allowed to differ by sex and AT type. Results SAT had significantly lower [18F]-FDG activity than VAT (β = -0.48, p < .001). A sex effect (female β = 0.11, p < .001) was reversed in SAT by a significant interaction (β = -0.15, p < .001). Higher cancer stages reduced activity in SAT but not VAT. Blood glucose showed a U-shaped relationship with activity, minimum near 6 mmol/l. In females, SAT activity was U-shaped with age, inflecting at 50-60 years, while VAT showed a linear positive association. Conclusion VAT exhibited higher [18F]-FDG uptake than SAT during fasting, significantly modulated by sex. SAT activity increased with age in women after the age of about 50 years, possibly related to the menopausal transition; menopausal status itself was not recorded. A novel U-shaped blood glucose relationship suggests competitive uptake at euglycemic states, while increased glucose carrier levels at dysglycemic states may increase activity.