Differentiation between physiological and infective fluorodeoxyglucose uptake (FDG) in patients suspected of prosthetic valve endocarditis (PVE) is challenging. The aim of the study was to provide a quantitative analysis of FDG uptake patterns using the dispersion index.

The study enrolled 142consecutive patients referred to Fluorodeoxyglucose Positron Emission Tomography (FDG-PET/CT). Half had a PVE suspicion (n=70) and the other half was referred for 3-month follow-up after cardiac surgery (n=72) and served as controls. Using the free software LifeX developed by researchers from the Université of Paris Saclay, Paris Sud, valvular FDG uptake was segmented and isolated from the PET/CT images. FDG uptake was characterized by its intensity, volume and number of foci. The dispersion index was defined as the product of the number of foci and the relative volume of positive uptake (volume of the FDG uptake divided by the volume of the valvular region). Normal value (<90th) in control was used to rule out endocarditis diagnosis in the PVE group.

PET/CT examination was qualitatively graded as positive in 49% (n=35) of controls, 46% (18/39) of rejected and 89% (25/31) of definite PVE patients (n=31). Uptake intensity (SUV_max and SUV_ratio) was similar in rejected and definite PVE, while relative uptake volume was smaller in definite PVE (12±8%) than in control (39±19%) and rejected PVE (36±11%, P<0.001). The dispersion index averaged 0.16±0.31 (<0.5 for <90th) in control and was greater in definite than in rejected PVE patients (0.79±0.20 vs. 0.18±22, P<0.0001). Sensitivity and specificity of the dispersion index with a cut-off <0.5 for PVE were 97% and 95%, respectively. The inter- and intra-observer reproducibility of the index was 7±5% and 2±2%, respectively.

Dispersion index is accurate and reproducible for differentiating physiological and infective uptake in patients with left-sided prosthesis.