A new framework for metabolic connectivity mapping using bolus [¹⁸F]FDG PET and kinetic modelling

AbstractPurposeMetabolic connectivity (MC) has been previously proposed as the covariation of static [18F]FDG PET images across participants, which we callacross-individualMC (ai-MC). In few cases, MC has also been inferred from dynamic [18F]FDG signals, similarly to fMRI functional connectivity (FC), which we termwithin-individualMC (wi-MC). The validity and interpretability of both MC approaches is an important open issue.Here we reassess this topic, aiming to 1) develop a novel methodology for wi-MC estimation; 2) compare ai-MC maps obtained using different [18F]FDG parameters (K1, i.e. tracer transport rate,k3, i.e. phosphorylation rate,Ki, i.e. tracer uptake rate, and the standardized uptake value ratio,SUVR); 3) assess the interpretability of ai-MC and wi-MC in comparison to structural and functional connectivity (FC) measures.MethodsWe analyzed dynamic [18F]FDG data from 54 healthy adults using kinetic modelling to quantify the macro- and microparameters describing the tracer behavior (i.e.Ki,K1, k3). We also calculatedSUVR. From the across-individual correlation ofSUVR, Ki, K1, k3, we obtained four different ai-MC matrices. A new approach based on Euclidean distance was developed to calculate wi-MC from PET time-activity curves.ResultsWe identified Euclidean similarity as the most appropriate metric to calculate wi-MC. ai-MC networks changed with different [18F]FDG parameters (k3MC vs.SUVRMC, r = 0.44). We found that wi-MC and ai-MC matrices are dissimilar (maximum r = 0.37), and that the match with FC is higher for wi-MC (Dice similarity: 0.47-0.63) than for ai-MC (0.24-0.39).ConclusionOur data demonstrate that individual-level MC from dynamic [18F]FDG data using Euclidean similarity is feasible and yields interpretable matrices that bear similarity to resting-state fMRI FC measures.