Acinetobacter is a genus that comprises a group of opportunistic pathogens responsible for a variety of nosocomial infections. The Acinetobacter calcoaceticus–Acinetobacter baumannii (Acb) complex includes some species of clinical importance, mainly A. baumannii, A. pittii and A. nosocomialis, which share phenotypic similarities that make it very difficult to distinguish between them using a phenotypic approach. The aim of this study was to evaluate two commercial matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) systems for the identification of different Acinetobacter species, with a special focus among those belonging to the Acb complex.
MethodsOne hundred and fifty-six Acinetobacter spp. clinical strains, identified by amplified ribosomal DNA restriction analysis (ARDRA) and rpoB gene sequencing, were analysed by two different MALDI-TOF systems.
ResultsConsidering only the 144 strains of the Acb complex evaluated in this study, the Vitek-MS™ and Microflex LT™ systems correctly identified 129 (89.6%) and 143 (99.3%) strains, respectively.
ConclusionAfter analysing 156 strains belonging to Acinetobacter spp., both Vitek-MS™ and Microflex LT™ proved to be rapid and accurate systems for the identification of Acb complex species showing a good correlation. However, both manufacturers should improve their databases to include new species in them.
Acinetobacter es un género que comprende un grupo de patógenos oportunistas responsables de varias infecciones nosocomiales. El complejo Acinetobacter calcoaceticus-Acinetobacter baumannii (Acb) reúne algunas especies de importancia clínica, principalmente A. baumannii, A. pittii y A. nosocomialis, que comparten similitudes fenotípicas que hacen muy difícil poder discriminar entre ellas utilizando un enfoque fenotípico.
El objetivo de este estudio fue evaluar 2 sistemas comerciales de espectrometría de masas de ionización por láser asistido con una matriz (MALDI-TOF MS) para la identificación de diferentes especies de Acinetobacter, con un enfoque especial entre los que pertenecen al complejo Acb.
MétodosAnalizamos 156 cepas clínicas de Acinetobacter spp., identificadas mediante análisis de restricción de ADN ribosomal amplificado (ARDRA) y secuenciación del gen rpoB, por 2 sistemas diferentes de MALDI-TOF.
ResultadosTeniendo en cuenta solo las 144 cepas del complejo Acb evaluadas en este estudio, los sistemas Vitek® MS y Microflex® LT identificaron correctamente 129 (89,6%) y 143 (99,3%) cepas, respectivamente.
ConclusiónDespués de analizar 156 cepas pertenecientes a Acinetobacter spp., Vitek® MS y Microflex® LT demostraron ser sistemas rápidos y precisos para la identificación de especies del complejo Acb mostrando una buena correlación. Sin embargo, ambos fabricantes deberían mejorar sus bases de datos incluyendo nuevas especies en ellas.
Acinetobacter baumannii belongs to the Acinetobacter calcoaceticus-Acinetobacter baumannii (Acb) complex that includes the most relevant nosocomial pathogens of genus Acinetobacter, which cause serious infections in critically ill and immunocompromised patients.1A. baumannii represents a major healthcare problem, being responsible for infections with high morbidity and mortality rates, particularly in intensive care units (ICU), because of its ability to persist in the environment and develop resistance to multiple antibiotics.2–4Acinetobacter nosocomialis and Acinetobacter pittii, two other important species of Acb complex, are also able to cause nosocomial infections, but they are more rarely associated with outbreak situations and usually display greater susceptibility to antimicrobial agents. An accurate identification of these microorganisms at the species level is, therefore, essential to minimize treatment failure or implement infection control measures.5,6
Since all Acinetobacter species included in the Acb complex are highly similar from a phenotypic point of view, as well as by DNA-DNA hybridization,7 simple phenotypic tests and other available automated biochemical identification systems routinely used in clinical laboratories cannot distinguish among them.8,9 In this context, Matrix-Assisted Laser Desorption Ionization Time of Flight Mass Spectrometry (MALDI-TOF MS) is a powerful tool that has been adapted for the identification of different microorganisms at the genus, species or subspecies level with high reliability.10,11 Nevertheless, some studies have also highlighted the struggles of MALDI-TOF MS for discrimination between members of the Acb complex as well as additional Acinetobacter species, particularly since many novel species have been recently included in this taxon.12–14
There are several previous studies concerning identification, detection of antimicrobial resistance, and virulence among Acinetobacter species by MALDI-TOF MS. However, only a few of them have focused on determining the accuracy of this technique at the species level identification.12,15–17 In addition, none of such studies have focused on comparing the performance of the two commercial MALDI-TOF MS systems most commonly used in clinical laboratories for the identification of isolates included in the Acb complex.
Microflex LT™ distinguishes the different Acinetobacter genomic species of the Acb complex. Vitek-MS™ globally identifies the Acb complex by versions v1.0, v2.0, and v3.0, while v3.2 distinguishes some of the different A. genomic species of the Acb complex, according to the manufacture's specifications.
This study aimed to evaluate the reliability of two systems, Vitek-MS™ (bioMérieux, Marcy l’Étoile, France) and Microflex LT™ (Bruker Daltonik GmbH, Bremen, Germany) for the identification of Acinetobacter spp. isolates belonging to the Acb complex. We have also compared four Vitek-MS™ database versions (v1.0, v2.0, v3.0, and v3.2), with two versions of Microflex LT™ (v6.0.0.0 and v8.0.0.0), to evaluate the evolution of such databases over time. In the case of Vitek-MS™, the software was used in IVD (In Vitro Diagnostic) mode instead of RUO (Research Use Only) mode to resemble clinical practice.
Material and methodsBacterial strainsEighty-four Acinetobacter spp. clinical strains recovered between 2004 and 2008 in the Microbiology Service of the University Hospital Marqués de Valdecilla, Santander, Spain, were chosen. Clonal relatedness among those strains belonged to the same Acinetobacter species were evaluated by repetitive extragenic palindromic PCR (REP-PCR) and by pulsed-field gel electrophoresis (PFGE) as reported previously.18 For the global study, a collection of 72 strains, from the Hospital Clinic in Barcelona, Spain, of less frequent species of the Acb complex, previously identified by rpoB gene sequencing,19 was also added in the analysis.
Species identification was performed by amplified ribosomal DNA restriction analysis (ARDRA) and rpoB gene sequencingARDRA was used as the reference method to confirm the identification of Acinetobacter species. The 16S rRNA gene was amplified for all Acinetobacter spp. strains using universal primers, and further digested separately with the restriction enzymes CfoI, AluI, MboI, RsaI, MspI and BfaI (Promega Biotech Ibérica, Madrid, Spain), as previously described.18,20
Any disagreements in the species identification results between ARDRA and MALDI-TOF MS systems, were analyzed by partial rpoB gene sequencing.19 In these cases, DNA from those strains showing discordant results was extracted by InstaGene™ Matrix (Bio-Rad Laboratories, Hercules, CA, USA). The rpoB gene was amplified using specific primers and purified using NucleoSpin®Gel Kit (Macherey-Nagel, Duren, Germany) and sequenced with ABI PRISM™ 377 DNA sequencer (Applied Biosystems, Foster City, USA). The partial rpoB gene sequences obtained were compared to reference sequences available on GenBank database using BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi). To establish the correct identification, sequences of a given pairwise alignment with the lowest E-value and the highest number of identities (>98%) were selected as the most likely species.
The statistical analysis was done with the IBM SPSS Version 20.0 software including the analysis of ROC curves and the calculation of Pearson's kappa index of concordance.
MALDI-TOF MS Microflex LT™ systemAll strains were subjected to analysis with Microflex LT™ system (Bruker Daltonik GmbH, Bremen, Germany) following the manufacturer's recommendations. The bacterial biomass was smeared onto a metallic MALDI-TOF MSP 96 plate (Bruker Daltonik GmbH, Bremen, Germany) and covered with 1μL of saturated α-cyano-4-hydroxycinnamic acid–50% acetonitrile–2.5% trifluoroacetic acid. The MSP 96 plate was dried at room temperature for 5min. Measurements were performed with the Bruker Microflex LT MALDI-TOF MS (Bruker Daltonik GmbH, Bremen, Germany) using FlexControl software with Compass Flex Series version 1.3 software and a 60-Hz nitrogen laser (337nm wavelength). The protein spectra ranging from the mass-to-charge ratio (m/z) 2 to 20kDa were explored with the Bruker Biotyper 3.1 software package (Bruker Daltonik GmbH, Bremen, Germany) with default settings and compared with the reference spectra present in the database, showing the 10 most similar patterns for each isolate. The database for identification was the reference Biotyper library v6.0.0.0 and v8.0.0.0 MSP (Bruker Daltonik GmbH, Bremen, Germany), which included 6.903 and 7.854 species, respectively. The similarity of patterns was represented as a score (2.3 to 3, identification at the species level; 2 to 2.3, identification at the genus level; <1.700, no reliable identification). Escherichia coli ATCC 25922 was used as a standard for calibration and as a reference for quality control.17,21
MALDI-TOF MS Vitek-MS™ systemIdentification of strains with the Vitek-MS™ system (bioMérieux, France) was performed following manufacturer's recommendations. Cells from a single colony were directly applied onto the steel carrier overlaid with 1μL of VITEK™ MS-CHCA ready-to-use-matrix solution (bioMérieux, France), and allowed to air-dry. Two samples of the same strain were applied and read in two rounds, as recommended. Measurement was performed with a Vitek-MS™ instrument supported by SARAMIS MS-IVD v1.0, v2.0, v3.0, and v3.2 databases (Anagnos Tee GMBH, BioMérieux) in positive linear mode, with a mass range of 2–20kDa, using E. coli ATCC® 8739 as a molecular mass standard. The intensity of the 50-Hz nitrogen laser was under the control of the acquisition software. Only hits within the spectra with scores of 99.9% were accepted.
ResultsA total of 144 Acinetobacter spp. strains belonging to Acb complex were evaluated, as follows: A. pittii (n=54), A. baumannii (n=47), A. nosocomialis (n=16), Acinetobacter dijkshoorniae (formerly A. phenon 5 by ARDRA; n=14), Acinetobacter seifertii (n=12), and Acinetobacter calcoaceticus (n=1). In addition, other 12 Acinetobacter spp. strains identified as belonging to other Acinetobacter species or genomic species (gen. sp.) not belonging to the Acb complex were also evaluated: Acinetobacter bereziniae (n=5), Acinetobacter gen. sp. 16 (n=1), Acinetobacter gyllenbergii (formerly Acinetobacter phenon 3 by ARDRA; n=2), Acinetobacter haemolyticus (n=2), and Acinetobacter vivianii (n=2). In the case of the strains identified as Acinetobacter phenon 5 by ARDRA, they were reclassified as A. dijkshoorniae by rpoB gene sequencing.
Considering the 144 strains of the Acb complex evaluated in this study [A. pittii (n=54), A. baumannii (n=47), A. nosocomialis (n=16), A. dijkshoorniae (n=14), A. seifertii (n=12) and A. calcoaceticus (n=1)], Vitek-MS™ and Microflex LT™ systems correctly identified 129 (89.6%) and 143 (99.3%) strains, respectively. Two A. baumannii strains were misidentified as A. nosocomialis and A. pittii by Microflex LT™ v6.0 and v8.0, respectively. Since there is no reference spectrum for A. dijkshoorniae in the Vitek-MS™ v3.2 library, one of the strains of this species was misidentified as Corynebacterium aurimucosum, as well as one A. nosocomialis strain as Burkholderia cenocepacia.
In the non-Acb complex group, Vitek-MS™ v3.2 misidentified one A. bereziniae and one Acinetobacter gen. sp. 16 strains as A. pittii and A. gyllenbergii, respectively. In addition, for six strains of the control group (non-Acb complex), the obtained spectra were not included in the databases. Meanwhile, Microflex LT™ v8.0.0 misidentified one A. bereziniae as A. pittii (n=1), two A. vivianii as A. haemolyticus and A. haemolyticus/parvus (n=2), two A. gyllenbergii as Acinetobacter proteolyticus and A. proteolyticus/haemolyticus (n=2), and one Acinetobacter gen. sp. 16 as A. proteolyticus/haemolyticus (Table 1). Vitek-MS™ v3.2 and Microflex LT™ showed a high sensitivity (>80%) and an area under the ROC curve (AUC)>0.8, demonstrating that both systems are effective in discriminating the Acb complex from the other Acinetobacter species not belonging to the complex. On the other hand, according to the interpretation criteria proposed by Landis and Koch, the high positive Pearson Kappa index (>0.75) showed an excellent correlation with the identification data obtained by ARDRA plus rpoB. It is worth noting that Microflex LT™ v 8.0.0.0 showed the best results as a whole, comparing only the eighty-four Acinetobacter spp. strains from University Hospital Marqués de Valdecilla (Table 2).22
Identification of Acinetobacter species with two MALDI-TOF MS systems.
ARDRA/rpoB seq | Vitek-MS™ v3.2 | Microflex LT™ v8.0.0.0 | |||
---|---|---|---|---|---|
N° strains Species | N° strains Species | N° strains Species | |||
54 | A. pittii | 55 | +(A. bereziniae) | 57 | A. bereziniae |
47 | A. baumannii | 47 | A. baumannii | 46 | −1(A. pittii) |
16 | A. nosocomialis | 15 | −1(B. cenocepacia) | 16 | A. nosocomialis |
14 | A. dijkshoorniae | 0 | −13(P150a)−1(C. aurimucosum) | 14 | A. dijkshoorniae |
12 | A. seifertii | 12 | A. seifertii | 12 | A. seifertii |
1 | A. calcoaceticus | 1 | A. calcoaceticus | 1 | A. calcoaceticus |
5 | A. bereziniae | 0 | −4(P150a)−1(A. pittii) | 4 | −1(A. pittii) |
1 | A. gen. sp. 16 | 0 | −1(A. gyllenbergii) | 0 | −1(A. proteolyticus/haemolyticus) |
2 | A. gyllenbergii | 3 | +1(A. gen. sp. 16) | 0 | −2(A. proteolyticus, A. proteolyticus/haemolyticus) |
2 | A. haemolyticus | 2 | A. haemolyticus | 2 | A. haemolyticus |
2 | A. vivianii | 0 | −2(P150a) | 0 | −2(A. haemolyticus, A. haemolyticus/parvus) |
Kappa | 0.81 | 0.93 | |||
Sensitivity | 98.6% | 100% | |||
AUC | 0.94 (0.83–1) | 0.99 (0.98–1) |
Comparison of Acinetobacter species belonging to the Acb complex, from University Hospital Marqués de Valdecilla, identified by different database versions of Vitek-MS™ and Microflex LT™ system.
Identification | Vitek-MS™ | Microflex LT™ | ||||
---|---|---|---|---|---|---|
v.1.0 | v.2.0 | v.3.0 | v.3.2 | v6.0.0.0 | v8.0.0.0 | |
Acb complex | 76 | 76 | 75 | ni | ni | ni |
A. baumannii | ni | ni | ni | 33 | 32 | 32 |
A. calcoaceticus | ni | ni | ni | 1 | 1 | 1 |
A. nosocomialis | – | – | – | – | 1 | 0 |
A. pittii | ni | ni | ni | 37 | 39 | 39 |
A. dijkshoorniae | ni | ni | ni | ni | ni | 2 |
Total | 76 | 76 | 75 | 71 | 73 | 74 |
Kappa index | – | – | – | 0.78 | 0.77 | 0.81 |
Sensitivity | 97.3% | 96.1% | 92.4% | 98.6% | 98.6% | 100% |
AUC | 0.89 (0.74–1) | 0.84 (0.67–1) | 0.69 (0.48–0.9) | 0.94 (0.83–1) | 0.94 (0.83–1) | 0.99 (0.97–1) |
Abbreviations: ni, not identified; AUC, area under the ROC curve.
MALDI-TOF MS is a good methodology for the routine activity in clinical microbiology laboratories, since it rapidly identifies a wide range of bacterial and fungal species of clinical importance. In this study, both Vitek-MS™ and Microflex LT™ systems allowed acceptable identification of Acb complex. Our results demonstrated that ARDRA is still a powerful method that remains valid in those laboratories lacking a MALDI-TOF MS system; in addition, it is also becoming more cost-effective than conventional phenotypic based automated systems.23
Previous publications have already underlined some of the limitations we have found in this study, including the need to add newly recognized species in databases,12 which has been considered in the latest available versions, or the difficulty of differentiating species from the haemolytic clade, a situation improved by using an alternative matrix solution (strongly acidified ferulic acid).13 Although both Vitek-MS™ and Microflex LT™ systems recognize strains from other less common taxa as belonging to the genus Acinetobacter, identification at the species level using MALDI-TOF MS is recommended but with due caution for this purpose. Microflex LT™ discrepancies have minimal to no clinical impact on patient's antibiotic treatment, but Vitek-MS™ discrepancies, on the other hand, would have an enormous clinical impact. The current database of Vitek-MS™ v3.2, should be revised, as already done with previous versions, to include the new Acinetobacter species to force the system to simply identify at genus level, which is also of interest in routine practice. More studies, testing in IVD mode with higher number of strains, are still necessary in order to improve the databases of both MALDI-TOF MS systems, since non-A. baumannii species are important emerging nosocomial pathogens able to acquire antimicrobial resistance genes from A.baumannii24 and having an actual clinical importance.
Transparency declarationsAll authors have nothing to declare. This study has not been financially supported by any Diagnostic/Pharmaceutical company.
Ethical approvalNot applicable.
FundingThis study was supported by Plan Nacional de I+D+i 2013–2016, Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Economía y Competitividad, Spanish Network for Research in Infectious Diseases (REIPI RD16/0016/0007; REIPI RD16/0016/0010) and the 2017 call for Strategic Action on Health (PI17/01932), co-financed by European Development Regional Fund “A way to achieve Europe” and operative program Intelligent Growth 2014–2020. I.R. was supported by the Department of Health, Generalitat de Catalunya, grant SLT002/16/00349.
Conflict of interestThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.