Microarray-Based Nucleic Acid Assay and MALDI-TOF MS Analysis for the Detection of Gram-Negative Bacteria in Direct Blood Cultures

Seon Young Kim, MD, PhD; Jeong Su Park, MD, PhD; Yun Ji Hong, MD, PhD; Taek Soo Kim, MD, PhD; Kiho Hong, MD, PhD; Kyoung-Ho Song, MD, PhD; Hyunju Lee, MD, PhD; Eu Suk Kim, MD, PhD; Hong Bin Kim, MD, PhD; Kyoung Un Park, MD, PhD; Junghan Song, MD, PhD; Sun Hoe Koo, MD, PhD; Eui-Chong Kim, MD, PhD

Disclosures

Am J Clin Pathol. 2019;151(2):143-153. 

In This Article

Results

BC-GN Assay and MALDI-TOF MS for Gram-negative Bacterial Identification

Among 124 total blood cultures in which gram-negative bacteria were detected, 108 were monomicrobial, 10 had grown both gram-negative and gram-positive bacteria, and six were polymicrobial, having grown two different gram-negative bacteria. Therefore, a total of 130 gram-negative bacteria were detected Figure 1. Among the 130 gram-negative rods (GNRs), 117 were covered by the BC-GN panel and 13 were not (two Salmonella paratyphi, two Stenotrophomonas maltophilia, Vibrio alginolyticus, Morganella morganii, Aeromonas caviae, Bacteroides fragilis, Alcaligenes xylosoxidans, Ochrobactrum anthropi, Yersinia enterocolitica, Ralstonia mannitolilytica, and Raoultella planticola). Among 107 cultures in which single GNRs covered by the BC-GN panel had grown with/without gram-positive bacteria, 106 were correctly identified by BC-GN assay; therefore, the concordance rate was 99.1%. Among the six polymicrobial blood cultures with two different gram-negative bacteria (including 10 targeted bacteria among 12 total GNRs), 10 targeted gram-negative bacteria were correctly identified (100% concordance). Overall, the BC-GN test directly performed from BC bottles presented 116/117 (99.1%) concordant results for the identification of targeted gram-negative bacteria. Among 13 gram-negative bacteria not targeted, three were wrongly identified (76.9% concordance): Aeromonas caviae was identified as Enterobacter species, Raoultella planticola was identified as K oxytoca, and Stenotrophomonas maltophilia was identified as P aeruginosa.

Figure 1.

Identification of gram-negative rods (GNRs) via a Verigene gram-negative blood culture (BC-GN) assay in 124 positive blood culture specimens. aOne specimen was monomicrobial (Citrobacter koseri) by conventional culture but polymicrobial (Citrobacter species and Acinetobacter species) by BC-GN.

Meanwhile, among 124 blood cultures, 100 were tested by MALDI-TOF MS. Among them, 84 cultures were monomicrobial, 10 cultures were polymicrobial with both gram-negative and gram-positive bacteria, and six cultures were polymicrobial with two different GNRs Figure 2. Among the 84 monomicrobial cultures, 73 were correctly identified (86.9% concordance rates), and the remaining 11 samples were not identified with "not reliable id" results and a MALDI score from 1.32 to 1.66. Among the 73 correctly identified samples, four showed identification results in which the genus was correctly identified, but the species was misidentified (Enterobacter cloacae to E kobei, three Citrobacter freundii to two C braakii, and one C gillenii). Among the 10 polymicrobial samples with gram-negative and gram-positive bacteria, eight GNRs were correctly identified, and in two samples, only gram-positive bacteria were identified. Therefore, when single gram-negative bacteria were grown in blood culture, 81/94 (86.2%) were correctly identified by MALDI-TOF MS. Among the six polymicrobial samples in which two different GNRs had grown, one culture was identified to present two different GNRs via mixed-mode MALDI-TOF analysis, and in the remaining five cultures, only one species was correctly identified. Therefore, the overall concordance for these samples was seven of 12 (58.3%).

Figure 2.

Identification of gram-negative rods (GNR) via matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) in 100 positive blood culture specimens. aOne specimen was monomicrobial (Citrobacter koseri) by conventional culture and MALDI-TOF MA but polymicrobial (Citrobacter species and Acinetobacter species) by gram-negative blood culture (BC-GN). bIn four samples, the genus was correctly identified but the species was misidentified (Enterobacter cloacae to E kobei, three Citrobacter freundii to two C braakii, and one C gillenii by MALDI-TOF MS). cTwo GNRs were detected by mixed mode; one GNR was detected by standard mode. dIn one sample, the genus was correctly identified, but the species was misidentified (Citrobacter freundii by BC-GN and C braakii by conventional culture but C koseri by MALDI-TOF MS). eIn two samples, one GNR was detected by mixed mode, and no GNRs were detected by standard mode; in the remaining three samples, one GNR was detected by both mixed and standard modes.

We compared the agreement of the BC-GN and MALDI-TOF MS assay results with the conventional culture results according to the different species Table 1 . Overall, 89.2% (116/130) of gram-negative bacteria were correctly identified by BC-GN, and 83.0% (88/106) of gram-negative bacteria were correctly identified by MALDI-TOF MS (Table 1). The identification rates of Acinetobacter baumannii, Enterobacterspecies, K oxytoca, and P aeruginosa were lower for MALDI-TOF MS than for BC-GN assay. Among 13 nontargeted gram-negative bacteria by BC-GN, eight (61.5%) were correctly identified by MALDI-TOF MS. The results of polymicrobial cultures identified by BC-GN and MALDI-TOF are presented in Table 2 . The BC-GN assay can detect two gram-negative bacteria simultaneously, including resistance determinants. Using MALDI-TOF, standard mode analysis could not identify two or more bacteria. Using mixed-mode analysis, polymicrobial bacteria could be detected in some samples.

Signal Intensity of Microarray Spots and MALDI-TOF MS Results

We investigated the signal intensity of BC-GN target spots and MALDI-TOF MS identification scores in each sample Figure 3. Some samples were misidentified by MALDI-TOF MS; however, the signal intensity of BC-GN target spots was as high as that of other samples. In Acinetobacter species and Enterobacter species samples, monomicrobial samples presented high target spot intensity with BC-GN; however, MALDI-TOF MS could not correctly identify the species. Among samples with P aeruginosa and K oxytoca, some polymicrobial samples presented decreased target spot intensity with BC-GN assay, while MALDI-TOF MS did not correctly identify the species. Therefore, in these samples, the burden of the specific microorganism might be lower.

Figure 3.

Intensity of BC-GN target spots (squares) and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) scores (circles) in each sample with Acinetobacter species (A), Enterobacter species (B), Escherichia coli (C), Pseudomonas aeruginosa (D), Klebsiella pneumoniae (E), and Klebsiella oxytoca (F). Polymicrobial samples with gram-negative and gram-positive bacteria are indicated as blue squares, and two different gram-negative bacteria are shown as red squares. Samples that were misidentified in the MALDI-TOF MS results are indicated with an X. The broken lines indicate the MALDI-TOF score 2.0.

Antibiotic Susceptibility

The results of detection of antimicrobial resistance determinants are presented in Table 3 . bla CTX-M was found in more than half of K pneumoniae samples (58.1%) and more than 20% of Proteus species and E coli (30.0% and 27.3%, respectively). In total, gram-negative bacteria with bla CTX-M presented third-generation cephalosporin resistance in 79.2% of samples. When carbapenem resistance was considered, most of the Acinetobacter species had bla OXA positivity (91.7%); bla NDMwas found in K pneumoniae and Citrobacter species (38.7% and 16.7%, respectively). In addition, bla KPC was found in K pneumoniae (12.9%). Most CRE by BC-GN exhibited imipenem resistance in antimicrobial susceptibility tests (26/28, 92.9%). One discrepant specimen was monomicrobial (Citrobacter koseri) by conventional culture and MALDI-TOF MS but polymicrobial (Citrobacter species and Acinetobacterspecies) by BC-GN, and this specimen was imipenem susceptible (Table 2, M023 sample). The mean signal intensity of Citrobacter species was 65,397; however, the spot intensity for Acinetobacter species was 30,161, and that of species with bla OXAwas 21,199, which is significantly lower than the usual target spot intensity of greater than 60,000. Therefore, the bla OXA positivity of the sample may be from low-dose Acinetobacter species that were bla OXA positive.

Reducing Reporting Time With BC-GN Assay in a Clinical Laboratory

The time from specimen sampling to reporting of the final identification results was compared between the BC-GN and conventional culture methods as simulated under clinical laboratory conditions, where BC-GN assays are performed in the daytime on weekdays. The median time from sample admission to blood culture positivity was 13.4 hours (range, 2.8–68.8 hours), and that from blood culture positivity to reporting time was 14.7 hours (2.4–56.1 hours) for BC-GN, which is significantly lower than the 57.2 hours (range, 23.6–256.2 hours) for conventional culture (P<.001). For polymicrobial samples, the median time from blood culture positivity to reporting was 11.6 hours (3.9–45.9 hours) for BC-GN, whereas it was 86.5 hours (55.5–126.5 hours) for conventional culture. For monomicrobial samples, shorter times of 14.0 hours (2.4–56.1 hours) for BC-GN assay and 53.1 hours (23.6–195.7 hours) for conventional culture were observed.

processing....