Improved sensitivity for molecular detection of bacterial and Candida infections in blood

Abstract

The rapid identification of bacteria and fungi directly from the blood of patients with suspected bloodstream infections aids in diagnosis and guides treatment decisions. The development of an automated, rapid, and sensitive molecular technology capable of detecting the diverse agents of such infections at low titers has been challenging, due in part to the high background of genomic DNA in blood. PCR followed by electrospray ionization mass spectrometry (PCR/ESI-MS) allows for the rapid and accurate identification of microorganisms but with a sensitivity of about 50% compared to that of culture when using 1-ml whole-blood specimens. Here, we describe a new integrated specimen preparation technology that substantially improves the sensitivity of PCR/ESI-MS analysis. An efficient lysis method and automated DNA purification system were designed for processing 5 ml of whole blood. In addition, PCR amplification formulations were optimized to tolerate high levels of human DNA. An analysis of 331 specimens collected from patients with suspected bloodstream infections resulted in 35 PCR/ESI-MS-positive specimens (10.6%) compared to 18 positive by culture (5.4%). PCR/ESI-MS was 83% sensitive and 94% specific compared to culture. Replicate PCR/ESI-MS testing from a second aliquot of the PCR/ESI-MS-positive/culture-negative specimens corroborated the initial findings in most cases, resulting in increased sensitivity (91%) and specificity (99%) when confirmed detections were considered true positives. The integrated solution described here has the potential to provide rapid detection and identification of organisms responsible for bloodstream infections.

FIG 1

Workflow and timing of the steps in sample preparation and PCR/ESI-MS analysis of a single whole-blood specimen, with results reported in <6 h. A batch of up to 6 samples can be tested simultaneously and the results reported in 8 h.

FIG 2

Frequency distribution plot for white blood cell counts obtained from patients. Each bin has a width of 2 × 10⁶ white blood cells/ml. The y axis represents the frequency of each bin relative to the total number of samples, and the x axis shows the range of white blood cell counts. The bracket at the top indicates that 90% of samples fall into the range from 0 to 16 × 10⁶ cells/ml. The black regions of the bars correspond to the specimens that were positive by PCR/ESI-MS.

FIG 3

Microbial and human DNA loads define the functional limits of 16S sequence analysis of clinical specimens. Blood sample extracts were obtained using the 5-ml DNA isolation protocol from whole-blood specimens collected from patients suspected to have bacteremia. For tissues, 25-mg specimens were collected from patients suspected of sterile-site bacterial/candidal infection. All PCR/ESI-MS-positive samples were further analyzed by Sanger sequencing. The samples are plotted with respect to their total DNA load (y axis) and bacterial DNA detection level by PCR/ESI-MS (x axis). The horizontal dashed line indicates a 0.4-μg DNA load threshold above which only one tissue sample is present and below which only one blood sample is represented. The vertical solid line indicates the detection level of approximately 40 bacterial genomes per PCR, the threshold above which sequencing was successful.

FIG 4

Quantitative bacterial loads in whole blood determined by various methods. Q, interquartile range; I, range; S, ±1 standard deviation; |, cutoff; diamond, median. The PCR/ESI-MS values are those reported in Table 1 corrected for the dilution factors in sample preparation and converted to genome copies per ml of blood to be consistent with the reported values for quantitative PCR. The analytical LOD for PCR/ESI-MS is indicated by the vertical dashed line.

FIG 5

Spectra from representative PCRs. The spectra for primer pairs 348 (left) and 349 (right) are reported for sample 1,083 (Serratia marcescens detected in replicates 1 and 2, first and second rows) and for sample 933 (Acinetobacter baumannii detected in replicates 1 and 2, third and fourth rows). For each primer pair, the conserved horizontal scale emphasizes the reproducibility of the detections between replicates, as can be verified by the vertical alignment of the peaks corresponding to the internal positive control of the assay (calibrant, [gray]) and to the detected species (blue). The vertical scales are normalized to the highest peak present in the corresponding well. The spectra shown are for only two primer pairs; the reported species detections are further supported by similar detections of the corresponding amplicons using four additional primer pairs.