PDF
Citation
Print
INTRODUCTION
There has been increasing global concern regarding the clinical importance of nosocomial infections caused by Acinetobacter species, including ventilator-associated pneumonia, catheter-related sepsis, meningitis, and urinary tract infections [1, 2]. Using conventional phenotypic assays alone, it is difficult to accurately identify Acinetobacter at the species level [3]. Four Acinetobacter species, namely, Acinetobacter baumannii, A. calcoaceticus, A. nosocomialis, and A. pittii, are collectively called the Acinetobacter baumannii group (Ab group) because they share biochemical and genetic similarities [2]. However, differences in clinical progression and pathogenicity have been observed among Acinetobacter species. A. baumannii is the most frequently isolated [4] and most clinically significant Acinetobacter species owing to its association with severe nosocomial infections, high incidence of multidrug resistance, and high mortality [5]. Advances in molecular techniques have allowed the identification of various Acinetobacter species belonging to the non-Ab group from clinical samples, some of which are associated with human infections [6]. Acinetobacter junii is a rare pathogen that causes bacteremia, pneumonia, and neonatal sepsis [7-9]. Acinetobacter soli also causes bloodstream infections in newborns, and carbapenem-resistant A. soli reportedly causes bloodstream infections [10-12]. Bloodstream infections caused by A. lwoffii, A. ursingii, and A. radioresistens have also been reported [13]. Accordingly, the need for reliable identification techniques at the species level is rapidly increasing to ensure rapid and accurate diagnosis.
In the past, to classify Acinetobacter at the species level, it was essential to use molecular genetic methods, such as sequencing housekeeping genes, including rpoB and gyrB, which are both labor-intensive and time-consuming. These methods are difficult to perform routinely in many clinical microbiology laboratories [2]. However, as microbial identification using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) can be performed more rapidly and inexpensively, this method is preferably used to identify Acinetobacter species.
The Bruker Biotyper (Bruker Daltonik GmbH, Bremen, Germany) and VITEK MS (bioMérieux S.A., Marcy l’Etoile, France) are two MALDI-TOF MS systems used worldwide for microbial identification [14]. The ASTA MicroIDSys (ASTA Inc., Suwon, South Korea) is a relatively new MALDI-TOF MS instrument that was introduced in 2016. Some studies have compared the performance of ASTA MicroIDSys with existing MALDI-TOF MS instruments for identifying commonly isolated microorganisms [14, 15]. While previous studies analyzed most species requested for culture and strain identification in clinical microbiology laboratories, our study included clinically isolated strains and type or reference strains of Acinetobacter consisting of various species previously identified using molecular methods [16, 17]. The aim of our study was to compare the performance of ASTA MicroIDSys with the Bruker Biotyper for identifying various Acinetobacter species.
MATERIALS AND METHODS
1. Bacterial isolates
A total of 207 Acinetobacter strains, consisting of 22 type or reference strains and 185 clinical isolates, were analyzed in this study. The 22 Acinetobacter type or reference strains used in this study were obtained from the Korean Collection for Type Cultures (KCTC), the National Culture Collection for Pathogens (NCCP), and the German Collection of Microorganisms and Cell Cultures (DSM). The 185 Acinetobacter clinical isolates were randomly chosen from previously identified and stored Acinetobacter strains at Chonnam National University Hospital (collected between January 2010 and December 2012) and from Chosun University Hospital (collected between September 2005 and May 2012). These clinical Acinetobacter strains were first identified by routine biochemical methods using the VITEK 2 system and then by molecular methods using rpoB gene sequencing. Based on the studies performed at Chonnam National University Hospital [16] and at Chosun University Hospital [17], a 450-base pair (bp) sequence (variable zone 2) and a 350 bp sequence (variable zone 1) were selected as the target regions of the rpoB gene. Based on molecular identification using rpoB gene sequencing, reference identification of the strain was established.
Of the 207 strains used in this study, 175 strains (84.5%) belonged to the Ab group, which consisted of 79 strains of A. baumannii, 70 strains of A. nosocomialis, 25 strains of A. pittii, and one strain of A. calcoaceticus. The other 32 strains belonged to the non-Ab group and consisted of eight strains of A. soli, six strains of A. berezinae, two strains of A. ursingii, and one strain each of the following: A. baylyi, A. bouvetii, A. gerneri, A. guillouiae, A. haemolyticus, A. indicus, A. johnsonii, A. junii, A. lwoffii, A. marinus, A. oleivorans, A. parvus, A. qingfengensis, A. radioresistens, A. schindleri, and A. tandoii.
This study was approved by the Institutional Review Board of Chosun University Hospital (CHOSUN NON2018-003). Informed consent was waived.
2. MALDI-TOF MS analysis
Identification of the bacterial isolates was conducted simultaneously at both institutions using the MALDI-TOF MS system per the manufacturer’s instructions. For the Bruker Biotyper, one colony was selected per strain, smeared on a target plate, and overlaid with 1 μL of saturated MALDI matrix solution containing α-cyano-4-hydroxycinnamic acid (CHCA) (Bruker Daltonik GmbH) in 50% acetonitrile and 2.5% trifluoroacetic acid. Peptide mass spectra were acquired using a Bruker Microflex LT system, Biotyper software 3.1, and the MALDI Biotyper reference library version 4.0.0, which included 945 species of gram-negative bacteria and 20 Acinetobacter species. For the ASTA MicroIDSys, one colony was isolated per strain and smeared on the spot. Subsequently, 1.5 μL of 70% formic acid was added and dried for 2 minutes, 1.5 μL of CHCA matrix solution was added and dried for 2 minutes, and the MALDI-TOF MS analysis was performed. The ASTA MicroIDSys CoreDB 1.27-build001, which included 715 species of gram-negative bacteria and 26 Acinetobacter species, was used for the spectrum analysis.
According to the manufacturer’s instructions, an identification score of <1.7 on the Bruker Biotyper system or a value of <140 on the ASTA MicroIDSys system was considered an unreliable identification score. The samples reported as unreliable were retested up to two times with the most favorable identification results used for comparative analysis.
3. Results interpretation
The identification results obtained using each MALDI-TOF MS system were classified into three categories based on their consistency with the reference identification determined by molecular identification results of previous studies [16, 17]: (a) correct identification to the species level when the MALDI-TOF result was consistent with that of the reference identification, (b) misidentification when the MALDI-TOF result was inconsistent with that of the reference identification, and (c) invalid identification when the instrument failed to identify the isolate.
4. Statistical analysis
The McNemar test was used to compare the correct identification rates of the two systems using IBM SPSS Statistics version 23 (SPSS Inc., Chicago, IL, USA). Statistical significance was set at P<0.05. For Acinetobacter species identified from more than five isolates, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated with 95% confidence intervals (CIs) and compared between the systems.
RESULTS
The identification results for the 22 type or reference strains, consisting of 20 type strains and two reference strains, are shown in Table 1. The Bruker Biotyper correctly identified 18 (81.8%) strains, but misidentified A. oleivorans KCTC 23045 and failed to identify A. marinus KCTC 12259, A. qingfengensis KCTC 32225, and A. indicus KCTC 42000. These four species were absent from the Bruker database. In contrast, the ASTA MicroIDSys correctly identified 21 (95.5%) strains but misidentified A. oleivorans KCTC 23045, despite this species being registered in the ASTA database.
The identification results of the 185 clinical isolates are presented in Table 2. Incorrectly identified strains were as follows. The most misidentified species in the Ab group were A. nosocomialis (six strains for the Bruker Biotyper, three for the ASTA MicroIDSys). Among the non-Ab group, the Bruker Biotyper misidentified eight strains of A. soli and one strain of A. ursingii, whereas the ASTA MicroIDSys misidentified one strain of A. berezinae.
The identification results of all 207 strains using the two systems are compared in Table 3. The correct identification rates of MALDI-TOF MS for all Acinetobacter were 89.4% (185/207) and 96.6% (200/207) using the Bruker Biotyper and ASTA MicroIDSys, respectively, showing a statistically significant difference (P<0.001). For the Ab group, the Bruker Biotyper and ASTA MicroIDSys demonstrated similar performances of 94.9% (166/175) and 97.1% (170/175), respectively (P=0.219). For the non-Ab group, however, the correct identification rates at the species level were significantly different (P<0.01) at 59.4% (19/32) and 93.8% (30/32), respectively. We reanalyzed the data after excluding A. indicus, A. marinus, A. oleivorans, A. qingfengensis, and A. soli, which were absent from the Bruker database. Consequently, the correct identification rates of the Bruker Biotyper and ASTA MicroIDSys were 94.9% (185/195) and 96.9% (189/195), respectively (P=0.289, data not shown).
The sensitivity, specificity, PPV, and NPV of the systems are listed in Table 4. Both instruments provided similar sensitivity, specificity, PPV, and NPV for the Ab group. With respect to the non-Ab group, the sensitivities for identification of A. berezinae of the Bru-ker Biotyper and ASTA MicroIDSys were 100.0% (95% CI, 51.7–100.0) and 83.3% (95% CI, 36.5–99.1), respectively.
DISCUSSION
The results of our study indicate that the ASTA MicroIDSys identified Acinetobacter species more accurately than the Bruker Biotyper. For the Ab group, the systems showed similar identification performance. However, the ASTA MicroIDSys demonstrated superior identification ability for the non-Ab group. This result may be attributable to the ASTA database, which included more Acinetobacter species within the non-Ab group.
Our study showed that the ASTA MicroIDSys identified the Ab group with high accuracy comparable to that of the Bruker Biotyper. This finding is consistent with earlier studies showing that the ASTA MicroIDSys, Bruker Biotyper, and VITEK MS have equivalent identification performances [14, 15, 18]. Lee et al. [14] reported agreement rates of 99.8% and 100% for A. baumannii (N=436) and A. nosocomialis (N=18) using the Bruker Biotyper and the ASTA MicroIDSys systems, respectively. In a study by Jung et al. [18], the correct identification rates using the VITEK MS and the ASTA MicroIDSys systems were 99.3% and 100.0% for A. baumannii (N=138), respectively. Liu et al. [15] showed that within the Ab group, A. baumannii (N=47), A. nosocomialis (N=28), and A. pittii (N=13) had correct identification rates at the species level of 100.0%, 100.0%, and 100.0% for the Bruker Biotyper system and 100.0%, 78.6%, and 100.0% for the ASTA MicroIDSys system, respectively. Regarding the non-Ab group, 31 strains consisting of 14 species were evaluated in the same study. ASTA MicroIDSys correctly identified A. bereziniae (N=8), A. haemolyticus (N=4), A. ursingii (N=4), A. radioresistens (N=2), A. baylyi (N=1), and A. johnsonii (N=1) at the species level. One strain of A. oleivorans was not identified correctly at the species level. These findings were generally consistent with our results, differing only for A. soli. Two strains of A. soli could not be correctly identified by ASTA MicroIDSys with CoreDB 1.26.02 [15]. Conversely, in our study eight strains of A. soli were correctly identified at the species level using the ASTA MicroIDSys with CoreDB 1.27-build001. Hence, the discrepancy between our results and those of Liu et al. [15] can be explained by the updated ASTA database.
Our study is another example demonstrating that database expansion leads to improved identification performance of MALDI-TOF MS systems. The statistically significant difference in identification rates in our study was primarily caused by Acinetobacter species whose protein spectra were only included in the ASTA database, such as A. indicus, A. marinus, A. oleivorans, A. qingfengensis, and A. soli. A study by Jeong et al. [19] showed that when representative mass spectra of 63 Acinetobacter strains were added to the default Bruker database, the agreement rate between the Bruker Biotyper system and rpoB sequencing increased from 69.8% to 100.0%. This strengthened the notion that constant updating of the database will improve MALDI-TOF MS-based microbial identification, leading to timely management of infection and better clinical outcomes. A. oleivorans and A. soli, which are becoming increasingly common etiologic agents of nosocomial infections, can carry antimicrobial resistance genes to various antibiotics such as β-lactams, aminoglycosides, tetracyclines, and macrolides [20]. This greatly impacts the selection of antibiotics and effective treatment strategies. MALDI-TOF MS would allow rapid identification and, consequently, antibiotic susceptibility testing. Although A. indicus, A. marinus, and A. qinfengensis are currently rare bacteria isolated from the natural environment [6], MALDI-TOF MS systems could be utilized to identify these bacteria rapidly and accurately, allowing the evaluation of their clinical significance.
In conclusion, we expect the identification capability of MALDI-TOF MS systems will continue to improve with continuous database enhancements. Such improvements may enhance our understanding of the clinical implications of rarely isolated species. Additional comparative studies following database upgrades are needed to evaluate performance improvements.
XML Download