Meta-analysis assessing the sensitivity and specificity of ¹⁸F-FDG PET/CT for the diagnosis of prosthetic valve endocarditis (PVE) using individual patient data (IPD)

Importance: The use of ¹⁸F-FDG PET/CT in diagnostic algorithms for PVE has increased since publication of studies and guidelines advocating its use. The assessment of test accuracy has been limited by small study sizes. We undertook a systematic review using individual patient data (IPD) meta-analysis techniques.

Objective: To estimate the summary sensitivity and specificity of ¹⁸F-FDG PET/CT in diagnosing PVE. We also assessed the effect of patient factors on test accuracy as defined by changes in the odds ratios associated with each factor. The effect of the PET/CT study on the final diagnosis was also assessed when compared to the preliminary Duke classification to determine in which patient group ¹⁸F-FDG PET/CT had the greatest utility.

Study selection: Studies were included if PET/CT was performed for suspicion of PVE and IPD of both the PET/CT result and final diagnosis defined by a gold-standard assessment was available. There were 3 possible final diagnoses ("definite PVE," "possible PVE," and "rejected PVE").

Results: Seventeen studies were included with IPD available for 537 patients (from 538 scans). The summary sensitivity and specificity were 85% (95% CI 74.2%-91.8%) and 86.5% (95% CI 75.8%-92.9%) respectively when patients with final diagnosis of "possible PVE" were classified as positive for PVE. When this group was classified as negative for PVE, sensitivity was 87.4% (95% CI 80.4%-92.1%) and specificity was 84.9% (95% CI 71.5%-92.6%). Patients with a known pathogen (especially coagulase negative staphylococcal species), elevated CRP, a biological or aortic valve infection appeared more likely to have an accurate PET/CT diagnosis. Those with a mechanical valve, prior antibiotic treatment or a transcatheter aortic valve replacement valve were less likely to have an accurate test. Time since valve implantation and the presence of surgical adhesive did not appear to affect test accuracy. Of the patients with a preliminary Duke classification of "possible PVE," 84% received a more conclusive final diagnosis of "definite" or "rejected" PVE after the PET/CT study.

Conclusions and relevance: ¹⁸F-FDG PET/CT has high sensitivity and specificity in diagnosing PVE and the diagnostic utility is greatest in patients with a preliminary Duke classification of "possible PVE." Some patient factors appear to affect test accuracy, though these results should be interpreted with caution given low patient numbers for subgroup analyses.