18F-FDG PET/CT in the diagnosis of infective endocarditis

Abstract

Abstract Background Increased uptake of 18F-fluorodeoxyglucose (FDG) by positron-emission tomography/computed tomography (PET/CT) is one of the diagnostic criteria for prosthetic valve endocarditis (PVE). However, there is limited data on the diagnostic value of FDG PET/CT in diagnosing native valve endocarditis (NVE) and semi-quantitative parameters of FDG uptake. Purpose The objective of this retrospective study was to evaluate the accuracy of FDG PET/CT and additional value of semi-quantitative measures of FDG uptake in the detection of NVE and PVE. Methods All patients who had undergone FDG PET/CT due to suspected infective endocarditis (IE) at a single tertiary referral center between 2010 and 2020 were identified. Additional patients who underwent FDG PET/CT for detecting an infection focus were included as controls. Clinical reports were evaluated and images were analyzed for maximal standardized uptake value (SUVmax) and target to background ratio (TBR = SUVmax normalized to mean blood pool SUV) around valves. The scan was defined positive based on either clinical reading or previously proposed cut-off values for SUVmax and TBR (>4.2 and >2.1, respectively). The final diagnosis of IE was defined based on all available information in electronic medical records. Results The study cohort included 16 patients with suspected NVE (median age 58 years, 88% male), 24 patients with suspected PVE (median age 72 years, 92% male) and 37 patients who underwent FDG PET/CT for detecting an infectious focus (8 with a prosthetic valve). The final diagnosis was definite IE in 25 (14 PVE), possible IE in 7 (6 PVE), and no IE in 45 (12 with a prosthetic valve). Overall, clinical reading of FDG PET/CT identified 17 (sensitivity 68%) of definite IE with a specificity of 91% and accuracy of 83%. Compared with NVE, the sensitivity of FDG PET/CT was higher in the detection of PVE (46% vs. 86%) whereas specificity was lower (97% vs. 75%) resulting in small difference in accuracy (84% vs. 81%). In PVE, SUVmax had higher sensitivity (93%) than clinical reading, resulting in the same accuracy (81%) despite lower specificity (67%). A combination of either positive clinical reading or positive SUVmax identified all patients with PVE (sensitivity 100%). In NVE, TBR had higher sensitivity (64%) than clinical reading, but accuracy (66%) and specificity (67%) were significantly lower. Conclusions FDG PET/CT shows high sensitivity in the detection of PVE and high specificity in the detection of NVE in a real-world clinical cohort. Semi-quantitative parameters of FDG uptake can improve sensitivity of detecting PVE maintaining diagnostic accuracy. However, in NVE semi-quantitative parameters were associated with a prominent reduction in specificity.