Aims: A new diagnostic strategy to improve the detection of pathogens in heart valves (HVs) from patients with infective endocarditis (IE) was evaluated.
Methods and results: Three hundred and fifty seven HVs surgically removed from 326 patients with proven IE or suspicious intra-operative findings, examined by 16S rDNA polymerase chain reaction (PCR) and culture were retrospectively analysed according to the predictive value of various PCR methods. Patients were classified into four categories: active IE, IE with ambiguous infective status, healed IE, and valve diseases but no IE. Retained samples of 200 HVs were analysed by real-time PCR targeting bacterial 23S rDNA, fungal 28S rDNA, and mycoplasmal tuf gene. 16S rDNA PCR revealed 80.6% sensitivity, 100% specificity, 100% positive predictive value, and 71% negative predictive value (NPV), compared with cultivation with 33.4, 96.6, 95.5, and 40.9%, respectively. The use of real-time PCR increased diagnostic sensitivity to 96.4%, and NPV to 92.5%. Bacterial load, C-reactive protein, and white blood cell counts (WBCs) decreased during antibiotic treatment. Bacterial load showed no correlation to C-reactive protein or WBCs, whereas C-reactive protein and WBCs were significantly correlated.
Conclusion: 23S rDNA real-time PCR of surgically removed HVs improves the diagnosis of IE. Polymerase chain reaction analysis of explanted HVs allow the optimization of the antimicrobial therapy, especially in patients with culture-negative IE.