Direct Screening of Blood by PCR and Pyrosequencing for a 16S rRNA Gene Target from Emergency Department and Intensive Care Unit Patients Being Evaluated for Bloodstream Infection

Here we compared the results of PCR/pyrosequencing to those of culture for detecting bacteria directly from blood. DNA was extracted from 1,130 blood samples from 913 patients suspected of bacteremia (enrollment criteria were physician-ordered blood culture and complete blood count [CBC]), and 102 controls (healthy blood donors). Real-time PCR assays for beta-globin and Universal 16S rRNA gene targets were performed on all 1,232 extracts. Specimens identified by Universal 16S rRNA gene PCR/pyrosequencing as containing staphylococci, streptococci, or enteric Gram-negative rods had target-specific PCR/pyrosequencing performed. Amplifiable beta-globin (melting temperature [Tm], 87.2°C ± 0.2°C) occurred in 99.1% (1,120/1,130) of patient extracts and 100% (102/102) of controls. Concordance between PCR/pyrosequencing and culture was 96.9% (1,085/1,120) for Universal 16S rRNA gene targets, with positivity rates of 9.4% (105/1,120) and 11.3% (126/1,120), respectively. Bacteria cultured included staphylococci (59/126, 46.8%), Gram-negative rods (34/126, 27%), streptococci (32/126, 25.4%), and a Gram-positive rod (1/126, 0.8%). All controls screened negative by PCR/pyrosequencing. Clinical performance characteristics (95% confidence interval [CI]) for Universal 16S rRNA gene PCR/pyrosequencing included sensitivity of 77.8% (69.5 to 84.7), specificity of 99.3% (98.6 to 99.7), positive predictive value (PPV) of 93.3% (86.8 to 97.3), and negative predictive value (NPV) of 97.2% (96.0 to 98.2). Bacteria were accurately identified in 77.8% (98/126) of culture-confirmed sepsis samples with Universal 16S PCR/pyrosequencing and in 76.4% (96/126) with follow-up target-specific PCR/pyrosequencing. The initial PCR/pyrosequencing took ∼5.5 h to complete or ∼7.5 h when including target-specific PCR/pyrosequencing compared to 27.9 ± 13.6 h for Gram stain or 81.6 ± 24.0 h for phenotypic identification. In summary, this molecular approach detected the causative bacteria in over three-quarters of all culture-confirmed cases of bacteremia directly from blood in significantly less time than standard culture but cannot be used to rule out infection.