Introduction

Rapid and accurate identification of pathogens is essential for appropriate antimicrobial therapy to improve patient outcomes. However, conventional biochemical methods take days, automated systems require at least 4–12 hours and molecular methods may still be relatively expensive, labor intensive and not suitable for large-scale routine identification. Recently, new mass spectrometry (MS) based technology, the matrix assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF MS) has been introduced as a way to accurately identify bacteria and fungi in minutes. The technology is based on the analysis of highly abundant, mainly ribosomal, proteins in the mass range of 2,000 to 20,000 Daltons. A mass spectrum can be acquired from an unknown micro-organism, and its proteomic fingerprint is then compared to those in a reference spectral database to ascertain the likely genus and species. Compared to standard phenotypic identification, it requires reagents that are inexpensive and the technology as such seems as accurate as 16S rRNA sequencing [1].

The aim of the present study was to evaluate the performance of the MALDI-TOF Vitek MS for rapid identification of micro-organisms in real time in a routine-diagnostic microbiology laboratory.

Materials and methods

This study was performed at a 650-bed super-speciality hospital in New Rajinder Nagar, New Delhi. Non-repetitive aerobic microbial isolates recovered from various clinical samples including blood, body fluids, respiratory specimens, urine, pus and tissues received in the microbiology laboratory from May 2013 to April 2014 were identified by MALDI-TOF MS (Vitek MS system, bioMerieux, France). The isolates were recovered from clinical samples using standard methodology [2].

Briefly, the primary media used for isolation were 5% sheep blood agar (bioMerieux, France), MacConkey agar (HiMedia, India), Hichrome agar (HiMedia, India), chocolate agar (bioMerieux, France), Lowenstein Jensen media (HiMedia, India), and Sabrauds Dextrose agar (HiMedia, India), wherever applicable.

Sample preparation for Vitek MS identification from microbial growth

Bacteria and yeasts [3, 4]

Briefly, freshly grown bacterial and yeast isolates were smeared as a thin film onto disposable MALDI target slides using a 1 μL disposable loop. The dried microbial film was then overlaid with 1.0 μL α-cyano-4-hydroxycinnamic acid (CHCA) MALDI-TOF MS matrix and air dried for 1–2 min at room temperature. The slide was then inserted into the Vitek MS system for data acquisition. For yeasts, the dried film was overlaid with 0.5-μL formic acid before addition of matrix.

Filamentous fungi

Extractions were performed according to the Inactivation/Extraction Protocol for Moulds and Vitek MS Identification (27 September 2013) (bioMerieux, Hazelwood, MO) [5].

Mycobacteria/Nocardia [6]

Briefly, a 1-μL loop full of bacterial biomass cultured on Lowenstein Jensen media was collected in a 1.5-mL screw-top microcentrifuge tube containing 500 μL of 70% ethanol and 200 μL of 0.5-mm silica beads (MoBio Laboratories, Inc.). All tubes were vortexed for 15 min in a horizontal position using a Vortex-Genie 2 with a 24-tube adaptor at maximum speed. The tubes were then kept at room temperature for 10 min to complete inactivation of the mycobacteria. The liquid contents (excluding beads) were then transferred to a new 1.5-mL snap-top microcentrifuge tube and centrifuged at 10,000 g for 2 min. The supernatant was decanted and the pellet was resuspended and kept for 2–5 min in formic acid, after which 10 μL of 98% acetonitrile was added to each tube. Following mixing by vortexing for 20 s and subsequent centrifugation at 10,000 g for 2 min, the supernatant was deposited on the VITEK MS slide and overlaid with 1 μL CHCA matrix.

MALDI-TOF MS acquisition

Data were processed automatically by the instrument software, and the new spectra were compared with those in the reference IVD approved database. As recommended by the manufacturer, the Escherichia coli ATCC 8739 strain was used as a calibrator and internal ID control and inoculated on the calibration spots of each acquisition group. In case no identification was achieved, the whole procedure was repeated with a fresh colony. For mycobacterial isolates, the results were read additionally in the Saramis RUO database.

Other identification methods used in the study

The other identification system used was the Vitek 2 system (for bacteria and yeasts) using the GP, GN, YST card (bioMerieux, France) as applicable, serotyping (for Salmonella and Shigella spp.) or otherwise by nucleic acid-mediated methods. Such methods were applied when Vitek MS failed to generate identification results. All the moulds and mycobacterial isolates were also identified by Lactophenol blue mounts and AccuProbe (bioMerieux, France) and GenoType Mycobacterium CM (Hain Lifescience GmbH, Nehren, Germany) molecular identification systems, respectively. These methods are used widely and are considered standard of care in clinical practice. We used all methods according to their manufacturer’s instructions.

Results

MALDI-TOF MS identification performance

A total of 12,003 microbial isolates were tested using Vitek MS, which included 5,350 enterobacteriaceae, 2763 Gram-negative bacteria other than enterobacteriaceae, 1100 Enterococci, 179 Streptococci and related species, 1297 Staphylococci and related species, 20 gram positive aerobic bacilli, 1248 yeasts, 20 filamentous fungi and 26 mycobacteria. The Vitek MS could provide ID to the species level for 95.8% (n = 11509) isolates, and an ID only to the genus level was obtained for 1.11% (n = 134) of isolates (mostly Salmonella and Enterobacter cloacae). No ID could be provided in 2.93% (n = 352) of isolates and eight Shigella flexneri isolates were expectedly misidentified as E. coli. Such non-lactose fermenting organisms were subsequently identified using Vitek-2. Eight isolates belonging to four species were not yet included in the database of the Vitek MS system (2 Brucella melitensis, 3 Burkholderia pseudomallei, 1 Nocardia cyriacigeogrica- identified by 16S rRNA, 1 Mycobacterium abscessus and 1 Mycobacterium chelonae).

Vitek MS gave a good overall ID to the genus and or species level up to 97.7% for bacterial isolates (97.3% enterobacteriaceae, 98.1% of gram negative bacteria other than enterobactericeae, 98.5% Enterococci and 89.9% Streptococci and related species, 98.3% Staphylococci and related species and 95% gram positive aerobic bacilli) as shown in Table 1. Vitek MS results correlated completely with gram reaction of the isolates.

Table 1 MALDI-TOF MS identification of bacterial, fungal and mycobacterial isolates
Full size table

Out of the 1,248 fungal isolates tested, 1175 (92.6%) could be identified by Vitek MS, the identification rates being 92.8% and 80% for yeasts and moulds, respectively (Table 1). For 34 of the 35 yeast isolates identified as C. haemulonii by Vitek 2, no identification was obtained using VITEK MS. Subsequently, 17 of these isolates were identified as C. auris (a novel species, not present in the Vitek MS database) by 18S rRNA sequencing, and using these strains, a new superspectrum of C. auris was created in VITEK MS using the RUO module. Out of a total of 26 mycobacterial isolates tested, only 11 (42.3%) could be identified by Vitek MS, using the Saramis RUO database (Table 1).

Discussion

The introduction of MALDI-TOF MS technology in clinical microbiology laboratories can reduce the time required for identification while improving the accuracy of microbe identification. Our study prospectively analyzed the routine use of Vitek MS in a clinical microbiology laboratory over a period of one year. In our hands, overall 95.8% of isolates were correctly identified to the species level, with high identification rates (97.7%) for bacteria and yeasts (92.8%), consistent with that reported in literature [3, 7, 8]. Failure or wrong identification in routine workflow could be due to poor quality deposits (too many peaks, like in mucoid organisms, e.g. in Klebsiella spp.) or too little peaks in case of thick cell walls where prior extraction is necessary, especially in moulds and mycobacteria, lack of representative spectra in the instrument database, alteration of protein expression in response to variations in environmental stimuli (culture conditions), failure of internal calibration and controls (seen when much older colonies, i.e. around 10–13 day old of calibrator E. coli strain were used) and closely related species expressing similar proteins can lead to additional errors.

The misidentification of Shigella isolates as E. coli in our study because of close genetic relatedness has also been reported previously [3, 7]. The manufacturers have already accepted these anomalies of the technology. Salmonella typing using MALDI-TOF MS also poses a challenge for clinical laboratories as a large number of reproducible peaks are required for subspecies identification [9]. In our study, out of a total of 145 isolates of Salmonella, 76 were identified as Salmonella group by VITEK MS but serotyping needs to be done for final identification.

An in-house superspectrum for C. auris was created in the RUO module using genetically identified strains, thereby indicating the utility of Vitek MS technology to identify novel microbial species by creating separate databases. Girard et al. [10] recently also created a novel C. auris superspectrum in the Vitek MS database using 12 reference strains and could successfully identify 46 of the 50 C. auris isolates recovered from clinical specimens obtained from three different countries using this expanded Vitek MS database, proving that Vitek MS could be efficiently used for identifying this novel species.

Among the moulds, VITEK MS in our hands identified 80% of isolates which remains comparable to data in other reports [4]. There are relatively few studies involving the use of MALDI-TOF MS for identification of filamentous fungi, probably because of difficulties related to phenotypic heterogeneity and lack of standardization in protein extraction protocols. Panda et al. [11], in a recent study on 37 isolates of filamentous fungi, found that only 68% of the isolates could be identified by MALDI-TOF MS although after modification of extraction protocols the rates increased to 87%. Improvement of extraction protocols along with a continuous expansion of the reference database library is therefore necessary to further improve the reliability of identification.

We have tried a few protocol modifications to identify monomorphic organisms directly from positive blood culture broths using Vitek MS. This could achieve 70.7% correct identification within 2 hours of a blood culture bottle turning positive [12].

MALDI-TOF MS has recently been adapted for identification of mycobacteria, and the majority of the work has been done using the Bruker Daltonics Flex system. In our study, the ID rates of mycobacterial species using the VITEK MS Saramis database were found to be very low (42.8%). Higher ID rates ranging from 82 to 95% [6, 13] have been reported recently using the Saramis RUO database; whereas in contrast during a study using the IVD database, out of the 102 mycobacterial isolates, none could be identified [14]. In our study the IVD database resulted in similar results but the Saramis RUO database could identify at least 42.8% isolates as mentioned above. Many of these issues are expected to be sorted out by the new V3 version of IVD database. In the meanwhile, we are trying to modify the extraction procedure for mycobacteria and the trial runs are encouraging. The modified procedure improved identification rate to >80% (unpublished data) using the Saramis RUO database. Since the number of isolates tested are small the study is still being continued. Also it would be interesting to observe the sensitivity of the V3 version in Vitek MS IVD in identifying mycobacteria in due course of time.

The main limitation of our study was that all discrepancies between identification by Vitek MS and alternative methods could not be resolved since sequencing was limited due to economic constraints. However, isolates were tested in comparison with the Vitek 2 ID system and other methods, as listed, which are widely used and are considered standard of care in clinical practice.

Conclusions

We found that MALDI-TOF MS could accurately identify bacteria and yeasts to both the genus and species levels for gram positive and gram negative bacteria as well as commonly encountered yeasts which favors its potential application for almost all aerobic bacterial species and yeasts commonly encountered in clinical practice. The ability to modify or add entries to the MALDI-TOF MS database as demonstrated by creation of super-spectra for Candida auris can help to further improve identification. However, results for moulds and mycobacteria were not encouraging, hence improved sample extraction protocols and database upgrades with inclusion of locally representative strains is warranted.