Introduction Diagnosing infective endocarditis (IE) is challenging. The modified Duke’s criteria have shortcomings. European Society of Cardiology guidance (2015) suggests a potential role of¹⁸F-Fluorodeoxyglucose positron emission tomography (PET), based on class C evidence. There is a lack of data for native valve IE (NVE).

Methods Dual centre retrospective study of all patients with suspected IE, from 01/2010. Patients were classified as confirmed/probable/rejected IE pre- and post-PET, with incremental benefit assessed versus actual diagnosis. This was defined by surgical specimen or Endocarditis Team (MDT) consensus at least three months following index admission.

Results PET was undertaken in 71 patients from 2010 to date; 59 since the inception of the MDT in October 2015 (male=50; mean age 60.6 y (range 19–89)). At discharge, 27/39 (69%) had confirmed NVE and 21/32 (66%) confirmed prosthetic IE (PVE). 30/71 (42%) patients required surgical intervention with concomitant device extraction in 7. Whilst Staphylococcus was isolated in 30/71 (42%) patients, 22/71 (31%) were peripheral blood culture-negative. PET sensitivity, specificity, positive and negative predictive values were 72%, 100%, 67% and 100% respectively in NVE, and 84%, 54%, 70% and 73% in PVE. PET highlighted 12/71 (16.9%) patients as having an alternative non-cardiac source of infection. Receiver Operating Characteristic (ROC) curves showed incremental benefit of PET over Duke’s criteria alone (AUC 0.875 vs 0.750, p=0.003) in NVE, though no difference in PVE (AUC 0.682 vs 0.613, p=0.649) compared to discharge diagnosis.

Conclusion PET has incremental value above modified Duke’s criteria in diagnosing IE, especially in NVE. PET has reduced specificity in PVE, likely related to post-surgical uptake.