Evaluation of Matrix-Assisted Laser Desorption Ionization−Time of Flight Mass Spectrometry for Identification of Mycobacterium species, Nocardia species, and Other Aerobic Actinomycetes

Abstract

The value of matrix-assisted laser desorption ionization−time of flight mass spectrometry (MALDI-TOF MS) for the identification of bacteria and yeasts is well documented in the literature. Its utility for the identification of mycobacteria and Nocardia spp. has also been reported in a limited scope. In this work, we report the specificity of MALDI-TOF MS for the identification of 162 Mycobacterium species and subspecies, 53 Nocardia species, and 13 genera (totaling 43 species) of other aerobic actinomycetes using both the MALDI-TOF MS manufacturer's supplied database(s) and a custom database generated in our laboratory. The performance of a simplified processing and extraction procedure was also evaluated, and, similar to the results in an earlier literature report, our viability studies confirmed the ability of this process to inactivate Mycobacterium tuberculosis prior to analysis. Following library construction and the specificity study, the performance of MALDI-TOF MS was directly compared with that of 16S rRNA gene sequencing for the evaluation of 297 mycobacteria isolates, 148 Nocardia species isolates, and 61 other aerobic actinomycetes isolates under routine clinical laboratory working conditions over a 6-month period. MALDI-TOF MS is a valuable tool for the identification of these groups of organisms. Limitations in the databases and in the ability of MALDI-TOF MS to rapidly identify slowly growing mycobacteria are discussed.

INTRODUCTION

Currently, there are more than 170 recognized species and subspecies of mycobacteria, more than 100 Nocardia species, and several hundred other aerobic actinomycetes species distributed across approximately 16 genera (1). Some of these organisms are clinically relevant and cause a spectrum of disease presentations in humans, while others are environmental organisms that can be found as commensal organisms or in laboratory cultures as contaminants resulting from specimen collection or processing (2). Accurate identification of mycobacteria and the aerobic actinomycetes is important for patient care but can be difficult due to the low growth rates of some species, the large number of species which have small differences in genetic diversity, and the need for biosafety level (BSL) 3 facilities when unknown isolates that might be Mycobacterium tuberculosis or Mycobacterium bovis are processed. The current gold standard for the identification of mycobacteria and aerobic actinomycetes is DNA sequencing with several targets recognized as useful for the species identification of mycobacteria and aerobic actinomycetes, including the 16S rRNA gene, rpoB, secA, and hsp65 (3). However, many clinical laboratories do not have the resources to routinely perform sequencing because it is labor-intensive and technically complex.

In the last few years, matrix-assisted laser desorption ionization−time of flight mass spectrometry (MALDI-TOF MS) has proven to be a reliable method for the identification of a wide variety of bacteria and yeasts following growth on culture medium (4–8). Fewer studies have been performed on the identification of mycobacteria and aerobic actinomycetes by MALDI-TOF MS, and none have looked at the specificity of the method for the majority of recognized species within the Mycobacterium genus and for a wide range of aerobic actinomycete genera and species (9–11). Another challenge with the use of MALDI-TOF MS for the identification of mycobacteria is the need to render the organism nonviable before testing outside a BSL3 laboratory to prevent possible lab staff exposure to M. tuberculosis or M. bovis.

In this work, we describe the development of a custom MALDI-TOF MS database for Mycobacterium spp., Nocardia spp., and other aerobic actinomycete genera and report its specificity for the identification of 162 Mycobacterium species, 53 Nocardia species, and 43 species in 13 other aerobic actinomycete genera. After construction of the database, a head-to-head comparison between the manufacturer's database and the custom database was performed with partial (500-bp) 16S rRNA gene sequencing used as the comparator reference method in order to determine the utility of MALDI-TOF MS for the routine identification of mycobacteria, Nocardia spp., and other aerobic actinomycetes under standard clinical microbiology laboratory working conditions. In addition, we provide confirmatory data for an earlier report by Dunne et al. (12) describing a simplified processing and extraction method for mycobacteria to demonstrate that this method renders isolates of M. tuberculosis complex safe for MALDI-TOF MS analysis outside a BSL3 environment.

(This study was presented in part at the 113th General Meeting of the American Society of Microbiology, Denver, CO, May 18 to 21, 2013, and at the 114th General Meeting of the American Society of Microbiology, Boston, MA, May 17 to 20, 2014.)

MATERIALS AND METHODS

Culture isolates used for database construction.

A total of 173 type strains of mycobacteria and 96 type strains of aerobic actinomycetes were obtained from the Deutsche Sammlung von Mikroorganismen (DSMZ) (Braunschweig, Germany), the American Type Culture Collection (ATCC) (Manassas, VA), and the Culture Collection, University of Göteborg (CCUG) (Göteborg, Sweden) for development of the custom database. In addition, 298 archived clinical isolates were also used for database development following growth on Middlebrook 7H10 medium (Remel, Lenexa, KS) or a sheep's blood agar plate (Remel) with incubation at 37°C. Mycobacterium ulcerans was incubated at 30°C, and Mycobacterium haemophilum was grown on a Middlebrook 7H10 plate supplemented with X factor (Becton Dickinson, Sparks, MD). The identification of all type strains was confirmed using partial (500-bp) 16S rRNA gene sequencing (3).

Culture isolates used for comparison versus 16S sequencing.

In order to evaluate the performance of MALDI-TOF MS under routine clinical laboratory workflow conditions, a direct head-to-head comparison of MALDI-TOF MS identification and partial 16S rRNA gene sequencing was performed over a 6-month period using 285 clinical isolates of mycobacteria grown in culture on solid agar medium from clinical specimens. In addition, 197 aerobic actinomycete isolates (136 Nocardia species isolates and 61 isolates from other genera) were similarly tested in a head-to-head fashion using 16S rRNA gene sequencing and MALDI-TOF MS analysis. Clinical specimens were cultured to look for mycobacteria on Middlebrook 7H10 agar plates, and specimens for aerobic actinomycetes were cultured on Middlebrook 7H10 agar and/or sheep's blood agar plates. Isolates identified as Mycobacterium chelonae/Mycobacterium abscessus complex using 16S rRNA gene sequencing were further differentiated using a laboratory-developed, real-time LightCycler PCR assay (13). This PCR assay differentiates M. chelonae from the M. abscessus group (M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, and M. abscessus subsp. massiliense). For clinically significant or closely related taxa with low numbers of isolates in the prospective study, additional clinical isolates were obtained after the original 6-month study was completed to further evaluate the performance of MALDI-TOF MS. These organisms were M. haemophilum (n = 5), Mycobacterium marinum (n = 5), Mycobacterium gastri (n = 2), Nocardia otitidiscaviarum (n = 5) and Nocardia elegans/Nocardia veterana (n = 5) and are included in Tables 2 and 3.

TABLE 2.

Comparison of MALDI-TOF MS identification and partial (500-bp) 16S rRNA gene sequencing for 297 mycobacterial isolates using the manufacturer's libraries (MBL/BDAL) and the manufacturer's libraries combined with a custom library (MCL)

16S rRNA gene sequencing identification (no. of isolates tested)	No. of isolates identified by MALDI-TOF MS at a cutoff score level of:
	≥2.0		≥1.7		<1.7 (no identification)
	MBL/BDAL	MBL/BDAL + MCL	MBL/BDAL	MBL/BDAL + MCL	MBL/BDAL	MBL/BDAL + MCL
M. chelonae/M. abscessus complex (n = 111)	39	108	73	108	38	3
M. chelonae (n = 29)a	0	28	4	28	25	1
M. abscessus group (n = 82)a	39	80b	69	80	13	2
M. arupense (n = 14)	13	14	14	14	0	0
M. asiaticum (n = 2)	1	2	1	2	1	0
M. avium/intracellulare complex (n = 18)
M. avium complex (n = 12)	5	6	8	11	4	1
M. intracellulare (n = 6)	2	2	2	2	4	4
M. canariasense (n = 1)	1	1	1	1	0	0
M. cosmeticum (n = 1)	1	1	1	1	0	0
M. conceptionense/M. houstonense/M. senegalense (n = 1)	1	1	1	1	0	0
M. fortuitum (n = 36)	32	35	34	35	2	1
M. gordonae (n = 9)	4	6	5	7	4	2
M. haemophilum (n = 5)	4	5	5	5	0	0
M. immunogenum (n = 4)	1	4	4	4	0	0
M. gastri/M. kansasii (n = 14)c	9	11	11	11	5	3
M. lentiflavum (n = 3)	3	3	3	3	0	0
M. mageritense (n = 3)	3	3	3	3	0	0
M. marinum/ulcerans (n = 6)d	6	6	6	6	0	0
M. mucogenicum/phocaicum (n = 18)	6	16	14	16	4	2
M. neoaurum (n = 1)	1	1	1	1	0	0
M. nebraskense (n = 1)	0	0	0	0	1	1
M. obuense (n = 1)	0	1	0	1	1	0
M. palustre (n = 1)	0	0	0	0	1	1
M. paraffinicum (n = 2)	0	1	0	1	2	1
M. parascrofulaceum (n = 1)	1	1	1	1	0	0
M. peregrinum/septicum (n = 6)e	3	4	4	4	0	0
M. porcinum (n = 10)	10	10	10	10	0	0
M. senuense (n = 1)	0	0	0	0	1	1
M. smegmatis (n = 2)	2	2	2	2	0	0
M. szulgai (n = 1)	1	1	1	1	0	0
M. triplex (n = 4)	2	4	3	4	1	0
M. tuberculosis complex (n = 6)	4	5	6	6	0	0
M. xenopi (n = 12)	7	8	8	8	4	4
Mycobacterium sp., not M. tuberculosis, not able to further identify (n = 2)	0	0	0	0	2	2
No. correct	162	261	222	269	73	26
% correct	54.6	87.9	74.8	90.6	25.6	9.1

^aThe M. chelonae/M. abscessus complex was further differentiated into the M. chelonae and M. abscessus group (M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, M. abscessus subsp. massiliense, M. franklinii) after 16S sequencing using a real-time PCR assay (13).

^b1 isolate was M. franklinii.

^c12 isolates were M. kansasii and 2 isolates were M. gastri.

^dThe 6 isolates were M. marinum.

^e4 isolates were M. peregrinum and 2 isolates were M. septicum.

TABLE 3.

Comparison of MALDI-TOF MS identification and partial (500-bp) 16S rRNA gene sequencing for 148 Nocardia spp. isolates using the manufacturer's library (BDAL) and the manufacturer's library combined with a custom library (MCL)

16S rRNA gene sequencing identification (no. of isolates tested)	No. of isolates identified by MALDI-TOF MS at a cutoff score level of:						No. of isolated misidentified
	≥2.0		≥1.7		<1.7 (no identification)
	BDAL	BDAL + MCL	BDAL	BDAL + MCL	BDAL	BDAL + MCL	BDAL	BDAL + MCL
N. abscessus/N. asiatica (n = 7)a	0	5	5	7	2	0	0	0
N. arthritidis (n = 5)	0	4	0	5	5	0	0	0
N. beijingensis (n = 4)	0	3	0	4	4	0	0	0
N. brasiliensis (n = 25)	2	25	4	25	21b	0	0	0
N. cyriacigeorgica (n = 29)	19	29	26	29	3	0	0	0
N. elegans/veterana (n = 6)	4	5	5	5	1	1	0	0
N. farcinica (n = 16)	13	16	15	16	1	0	0	0
N. gamkensis (n = 1)	0	0	0	0	1	0	0	1
N. kruczakiae (n = 1)	1	1	1	1	0	0	0	0
N. nova (n = 24)	19	23	22	24	2	0	0	0
N. otitidiscaviarum (n = 6)	4	6	6	6	0	0	0	0
N. pseudobrasiliensis (n = 2)	0	2	0	2	2	0	0	0
N. transvalensis (n = 4)	0	4	0	4	4	0	0	0
N. vinacea MCR (n = 1)	0	1	0	1	1	0	0	0
N. wallacei (n = 8)	0	8	0	8	8	0	0	0
Nocardia species, not able to further identify (n = 7)	0	2	2	3	6	4	0	0
No. correct	62	133	85	140	61	5	0	1
% correct	41.9	89.9	57.4	94.6	44.9	3.7	0	0.74

^a5/7 isolates were identified as N. abscessus, and 2/7 isolates were identified as N. asiatica.

^b19/21 isolates were identified as “Nocardia species” using the BDAL library.

Sample preparation and extraction for MALDI-TOF MS.

An isolate was selected from a culture plate using a 1-μl sterile, disposable inoculating loop and placed into 500 μl of 70% ethanol (Sigma-Aldrich, St. Louis, MO) containing 0.05 ml of 0.1-mm silica beads (Biospec Products, Bartlesville, OK). When possible, single isolated colonies were selected from the primary culture plate for MALDI-TOF MS analysis. However, in instances where there was limited growth on a plate (i.e., some slowly growing mycobacteria), a 1-μl disposable inoculating loop was used to scrape the agar to collect organisms growing as a light film on the plate. After incubation of the specimen for 10 min at room temperature, the isolate was then mechanically lysed by vortexing for 2 min using a Disruptor Genie (Scientific Industries, Inc., Bohemia, NY). The suspension was pelleted by centrifuging for 5 min at 20,800 × g, the supernatant was decanted, and the pellet was then dried by placing the tube in a SpeedVac (Eppendorf, Hauppauge, NY) for 10 min. The pellet was reconstituted in 20 μl of 70% formic acid (Sigma-Aldrich) and vortexed for 10 s followed by the addition of 20 μl of acetonitrile (Sigma-Aldrich). The contents of the tube were vortexed for an additional 10 s, and then 1 μl of suspension was applied on a 96-spot MSP Big Anchor steel plate (Bruker Daltonics Inc.) and allowed to dry. Each dried spot was then overlaid with 2 μl of α-cyano-4-hydroxycinnamic acid (HCAA) matrix solution (Bruker Daltonics Inc.) and allowed to air dry. A bacterial test standard (BTS) (Bruker Daltonics Inc.) was used on each plate for instrument calibration. A positive control consisting of Mycobacterium fortuitum ATCC 27408 and a negative extraction control (a tube containing reagents only but no isolate material) were included on each plate.

Nonviability studies.

Since M. tuberculosis is a BSL3 pathogen, it was necessary to demonstrate that the sample preparation process discussed above rendered M. tuberculosis nonviable so that it was safe to perform subsequent MALDI-TOF MS analysis steps under standard BSL2 conditions. Viability testing was conducted to examine the effect of the inoculum size and incubation time to demonstrate that incubation of a small quantity of the organism in 70% (vol/vol) ethanol is sufficient to render nonviable strains of M. tuberculosis (n = 12), Mycobacterium avium (n = 2), Mycobacterium kansasii (n = 2), M. abscessus group (n = 2), and 4 aerobic actinomycetes isolates (n = 1 each for Nocardia nova, Nocardia brasiliensis, Streptomyces sp., and Tsukamurella sp.). Three of the 12 M. tuberculosis complex strains tested were multidrug-resistant strains. For each organism listed above, a 1-μl loop full of organism was placed into a 1.5-ml tube containing 500 μl of 70% (vol/vol) ethanol). The tube was vortexed for 10 s and incubated at room temperature for 10 min. Following the 10-min incubation period, tubes were centrifuged at 20,800 × g for 2 min, and the supernatant was removed. The pellet was washed with 500 μl of sterile water, and the tubes were vortexed for 10 s. Following the water wash step, the tube was centrifuged for 2 min, and the supernatant was removed. Then, 500 μl of sterile water was added, the tube was vortexed briefly to resuspend the cell pellet, and 500 μl of the suspension was added to a MGIT tube (mycobacteria growth indicator tube; Becton Dickinson, Franklin Lakes, NJ) and a Middlebrook 7H10 plate. The MGIT tubes were incubated for 42 days, and the Middlebrook 7H10 plates were incubated for 60 days and examined weekly to look for any growth of viable organisms.

This experiment was repeated three additional times using a 1-μl inoculating loop full of organism with a shorter 5-min exposure to 70% ethanol, a 10-μl loop full of organism with a 5-min ethanol exposure, and a 10-μl loop full of organism with a 10-min exposure to ethanol. M. tuberculosis (ATCC 27294) was used as a positive control for each experiment with the isolate placed in sterile water instead of 70% ethanol and processed as described above.

MALDI-TOF MS analysis.

Mass spectrometry was performed on a MALDI Biotyper 3.0 Microflex LT system using the manufacturer's settings. Captured spectra were analyzed using flexControl 3.0 software and library version 3.3.1.0 (5,627 entries), supplemented with mass spectra profiles (MSPs) from the Bruker mycobacteria-specific library (Mycobacteria Library [MBL] v. 2.0). Manufacturer-recommended cutoff scores were used for identification with scores of ≥2.000 indicating identification to the species level. Isolates with scores of ≤2.000 or those that failed to produce peaks were tested again by spotting the original extracted isolate in triplicate.

Custom library creation.

The spectral library supplied with the Biotyper instrument (Bruker BDAL 5627 library) contains 77 entries for mycobacteria covering 43 species and subspecies, 72 entries for Nocardia covering 32 species, and 166 entries for other aerobic actinomycetes covering 9 genera (see Tables S1 to S3 in the supplemental material). In addition, the Bruker MBL also supplied with the MALDI-TOF MS instrument contains 313 library entries for mycobacteria covering 131 species and subspecies. In order to enhance the mycobacteria and aerobic actinomycete species coverage and strain diversity, a custom library (Mayo Custom Library [MCL]) containing 479 entries from 162 species and subspecies of mycobacteria, 232 entries from 53 Nocardia species, and 185 entries from 13 genera and 43 species of aerobic actinomycetes was constructed. All isolates used for library construction were type strains or well-characterized clinical isolates identified by 16S rRNA gene sequencing, nucleic acid hybridization probes (AccuProbe; Hologic GenProbe, San Diego, CA), and, where appropriate, phenotypic properties (e.g., colonial morphology and photoreactivity). Multiple library entries were created for species that are more frequently encountered in the laboratory (see Table S1 in the supplemental material). Each entry added to the library was created using 24 to 36 spectra collected from 9 to 12 distinct spots on the target plate. Before creation of the entry, all spectra were analyzed using the manufacturer's recommended criteria, and the new entry was run against all other entries (MSPs) in all of the libraries to ensure that the identification score was acceptable and specific for that species of mycobacteria. Only a few species produced nondistinct entries (MSPs), and these are discussed in the Results. Isolates used for library-building purposes were distinct from those used in the head-to-head analysis that compared the performance of the manufacturer's libraries and the custom library to 16S rRNA gene sequencing.

Statistical analysis.

Results for paired proportions for both libraries and for adjusted score cutoff values were compared using the McNemar test. P values of <0.05 were considered statistically significant.

RESULTS

Nonviability studies.

Twenty-one of 22 tested isolates, including all 12 strains of M. tuberculosis were rendered nonviable by a 5-min incubation in 70% ethanol following selection with a 1-μl inoculation loop. A single strain of Streptomyces sp. remained viable after incubation in 70% ethanol for 5 and 10 min using either a 1- or 10-μl loop. Neither the loop size nor the incubation time made a difference in viability for any of the isolates tested. As expected, the M. tuberculosis positive control that was processed with water instead of ethanol was positive in all experiments.

Custom library verification.

Each entry in the custom library was checked for specificity by in silico comparison of the new MSP against all other MSP entries in both the manufacturer's library and the custom library. Notable specificity issues or improvements for the identification of Mycobacterium and Nocardia species by MALDI-TOF MS compared with partial (500-bp) 16S rRNA gene sequencing are presented in Table 1. Mycobacterium and Nocardia species in the MALDI-TOF MS libraries that are not listed in Table 1 have specific patterns by MALDI-TOF MS, but it should be noted that the number of strains deposited in the libraries for less common species is often low, and therefore additional specificity issues may be recognized in the future as larger numbers of strains are added to the libraries for the uncommon species. As noted in Table 1, MALDI-TOF MS can differentiate some species of mycobacteria and Nocardia that are indistinguishable by partial gene 16S sequencing so the two techniques can provide complementary tools for some closely related species. In other instances, most notably M. tuberculosis complex, neither method is able to identify an isolate to the species level, and alternative methods are required if species identification is necessary. An alternative sequencing target such as hsp65, rpoB, or secA may be useful for identification of some mycobacterial species that are not able to be identified by MALDI-TOF MS or partial 16S rRNA gene sequencing.

TABLE 1.

Notable specificity issues or improvements for the identification of Mycobacterium and Nocardia species by MALDI-TOF MS compared with partial (500-bp) 16S rRNA gene sequencing^a

MALDI-TOF MS result	Partial (500-bp) 16S rRNA gene sequencing result
M. avium complex	M. avium complex
•M. avium/M. avium subsp. hominissuis/M. avium subsp. paratuberculosis/M. avium subsp. silvaticum	•M. avium/M. avium subsp. hominissuis/M. avium subsp. paratuberculosis/M. avium subsp. silvaticum)
•M. intracellulare/M. chimaera/M. colombiense/M. marseillense/M. yongonense	•M. chimaera
•M. vulneris	•M. colombiense
•Not tested by MALDI-TOF: M. bouchedurhonense, M. timonense	•M. intracellulare
	•M. marseillense/M. yongonense
	•M. bouchedurhonense
	•M. timonense
	•M. vulneris
•M. chelonae	•M. chelonae/M. abscessus complex (includes M. abscessus subsp. abscessus, M. chelonae, M. abscessus subsp. massiliense, M. abscessus subsp. bolletii, M. franklinii)
•M. abscessus group (includes M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, M. abscessus subsp. massiliense)
•M. franklinii
•M. gastri	•M. gastri/M. kansasii
•M. kansasii
•M. marinum	•M. marinum/M. ulcerans
•M. ulcerans
•M. mucogenicum/M. phocaicum	•M. mucogenicum/M. phocaicum
•M. murale	•M. murale/M. tokaiense
•M. tokaiense
M. fortuitum complex	M. fortuitum complex
•M. alvei	•M. alvei/M. setense
•M. boenickei	•M. boenickei
•M. conceptionense/M. houstonense/M. senegalense	•M. conceptionense/M. houstonense/M. senegalense
•M. farcinogenes	•M. farcinogenesb
•M. fortuitum	•M. fortuitum
•M. neworleansense/M. porcinum	•M. neworleansense/M. porcinum
•M. peregrinum	•M. peregrinum/M. septicum
•M. septicum
•M. setense
•M. paraseoulense/M. seoulense	•M. paraseoulense/M. seoulense
M. terrae complex	M. terrae complex
•M. algericum	•M. algericum/M. terrae
•M. arupense	•M. arupense
•M. engbaekii	•M. engbaekii
•M. heraklionense	•M. heraklionense
•M. hiberniae	•M. hiberniae
•M. longobardum	•M. kumamotonense
•M. minnesotense	•M. longobardum
•M. nonchromogenicum	•M. minnesotense
•M. senuense	•M. nonchromogenicum
•M. terrae/M. kumamotonense	•M. senuense
M. tuberculosis complex (includes M. africanum, M. bovis, M. bovis subsp. caprae, M. bovis BCG, M. canettii, M. microti, M. mungi, M. pinnipedii, M. tuberculosis)	M. tuberculosis complex (includes M. africanum, M. bovis, M. bovis subsp. caprae, M. bovis BCG, M. canettii, M. microti, M. mungi, M. pinnipedii, M. tuberculosis)
N. abscessus complex	N. abscessus complex
•N. abscessus/N. asiatica	•N. abscessus/N. asiatica/N. arthritidis
•N. arthritidis
N. nova complex	N. nova complex (includes N. africana, N. cerradoensis, N. elegans, N. kruczakiae, N. mikamii, N. nova, N. veterana)
•N. africana
•N. elegans
•N. kruczakiae
•N. mikamii/N. cerradoensis
•N. nova
•N. veterana
N. higoensis complex	N. higoensis complex
•N. higoensis	•N. higoensis/N. shimofusensis
•N. shimofusensis

^aBulleted results indicate those species which can be discriminated from other species by the indicated method. Slashed results indicate those species which are indistinguishable by the method indicated. Other Mycobacterium and Nocardia species in the MALDI-TOF MS libraries not listed in this table have specific patterns by MALDI-TOF MS, but the strain diversity represented in the library may be limited for less common species, so specificity issues may be recognized as additional strains are added to the libraries.

^bThe growth rate was used to distinguish from M. conceptionense/M. houstonense/M. senegalense.

Prospective comparison of MALDI-TOF MS versus 16S rRNA gene sequencing for the identification of Mycobacterium spp.

A comparison of MALDI-TOF MS with partial (500-bp) 16S rRNA gene sequencing for 297 mycobacterial isolates cultured from patient specimens in the clinical mycobacteriology laboratory resulted in correct identification of 162/297 (55%) isolates using the manufacturer's combined libraries and 261/297 (88%) isolates using the MCL in combination with the manufacturer's libraries (Table 2). We also examined whether lowering the acceptable score for species-level identification from ≥2.0 to ≥1.7, as has been done for other groups of organisms, would improve the identification using both libraries. Lowering the accepted cutoff score to ≥1.7 did result in an improvement in the number of correct identifications to 222/297 (75%) isolates using the manufacturer's libraries and to 269/297 (91%) isolates using the MCL. Importantly, there were no misidentifications using either the manufacturer's libraries or the MCL with either the 2.0 or 1.7 cutoff score. The MCL in combination with the manufacturer's libraries was significantly better at identifying mycobacteria to the species level than the manufacturer's libraries only (P < 0.05). Lowering the cutoff score from ≤2.0 to ≤1.7 also resulted in a significant increase in the identification of mycobacteria to the species level using the combined libraries (P < 0.05). In our laboratory, mycobacterial isolates not identified using either the manufacturer's libraries or the MCL are reflexed to partial 16S rRNA gene sequencing for identification. For closely related species (e.g., some members of the M. avium complex), additional sequencing targets are sometimes needed to differentiate members to the species level.

Of interest, identification of M. chelonae improved significantly with use of the MCL. With the manufacturer's libraries, 0/28 isolates were identified in the prospective study but with use of the MCL, identification improved to 28/29 (97%) isolates (Table 2). Similarly, identification of the M. abscessus group improved from 39/82 (48%) isolates using the manufacturer's libraries to 79/82 (96%) isolates using the MCL. M. kansasii isolates failed to be identified by either database for approximately 25% of isolates tested. Although the precise reason for this is unknown, one possible explanation is that the entire range of strain diversity for M. kansasii is not covered in the databases. M. kansasii is known to have seven subtypes with type 1 being responsible for most infections. Other subtypes are thought to be environmental but may still be found in clinical samples via water or aerosol exposures (14).

Prospective comparison of MALDI-TOF MS versus 16S rRNA gene sequencing for identification of Nocardia spp.

The comparison of MALDI-TOF MS with partial (500-bp) 16S rRNA gene sequencing for 136 Nocardia species isolates cultured from patient specimens resulted in correct identification of 62/148 (42%) isolates using the manufacturer's library and 133/148 (90%) isolates using the MCL in combination with the manufacturer's library (Table 3). Lowering the acceptable score for a species-level identification from ≥2.0 to ≥1.7 improved the identification using both libraries with 85/147 (57%) isolates identified using the manufacturer's library and 140/148 (95%) isolates identified using the MCL in combination with the manufacturer's library. There were no misidentifications using the manufacturer's library. There was 1 misidentification of Nocardia gamkensis identified by 16S rRNA gene sequencing, which was identified as Nocardia pneumoniae by MALDI-TOF MS using the MCL. A “no identification” score was given by the manufacturer's library. N. gamkensis is a rare Nocardia species that was not represented in either MALDI-TOF library. After addition of N. gamkensis to the MCL, in silico analysis of the library entry indicated that N. gamkensis can be differentiated from the closely related N. pneumoniae if isolates are encountered in the clinical laboratory in the future. The MCL in combination with the manufacturer's libraries was significantly better at identifying Nocardia to the species level than the manufacturer's libraries only (P < 0.05). Lowering the cutoff score from ≤2.0 to ≤1.7 also resulted in a significant increase in the identification of Nocardia to the species level using the combined libraries (P < 0.05). Nocardia isolates not identified using either the manufacturer's library or the MCL are reflexed to 16S rRNA gene sequencing for identification.

Identification of N. brasiliensis improved significantly with use of the MCL. With the manufacturer's library, 2/25 (8%) isolates were identified in the prospective study, but with use of the MCL, identification improved to 25/25 (100%) isolates (Table 3). Similarly, identification of Nocardia cyriacigeorgica improved from 19/29 (66%) isolates using the manufacturer's library to 29/29 (100%) isolates using the MCL.

Prospective comparison of MALDI-TOF MS versus 16S rRNA gene sequencing for identification of aerobic actinomycetes other than Nocardia spp.

MALDI-TOF MS was compared with partial (500-bp) 16S rRNA gene sequencing for 61 aerobic actinomycete isolates (other than Nocardia species) cultured from patient specimens. Streptomyces spp. comprised 46% (28/61) of the isolates tested. MALDI-TOF MS analysis resulted in correct identification of 9/61 (15%) isolates using the manufacturer's library and 31/61 (51%) isolates using the MCL (Table 4). Lowering the acceptable score for an identification from ≥2.0 to ≥1.7 improved the identification using both libraries with 15/61 (25%) isolates identified using the manufacturer's library and 42/61 (69%) isolates identified using the MCL. There were no misidentifications using either the manufacturer's library or the MCL. The MCL in combination with the manufacturer's libraries was significantly better at identifying aerobic actinomycetes to the species level than the manufacturer's libraries only (P < 0.05). Lowering the cutoff score from ≤2.0 to ≤1.7 also resulted in a significant increase in the identification of aerobic actinomycetes to the species level using the combined libraries (P < 0.05).

TABLE 4.

Comparison of MALDI-TOF MS identification and partial (500-bp) 16S rRNA gene sequencing for 61 aerobic actinomycete isolates (other than Nocardia spp.) using the manufacturer's library (BDAL) and the manufacturer's library combined with a custom library (MCL)

16S rRNA gene sequencing identification (no. of isolates tested)	No. of isolates identified by MALDI-TOF MS at cutoff score level of:
	≥2.0		≥1.7		<1.7 (no ID)
	BDAL	BDAL + MCL	BDAL	BDAL + MCL	BDAL	BDAL + MCL
Actinomadura madurae (n = 1)	0	0	0	0	1	1
Actinomadura verrucosospora (n = 1)	0	0	0	0	1	1
Actinomadura sp. (n = 1)	0	0	0	1	1	0
Aerobic actinomycete, not able to further identify by sequencing (n = 3)	0	0	0	0	3	3
Amycolatopsis pigmentata (n = 1)	0	0	0	0	1	1
Gordonia aichiensis (n = 2)	0	2	0	2	2	0
Gordonia bronchialis (n = 4)	0	4	0	4	4	0
Gordonia otitidis (n = 2)	0	2	0	2	2	0
Gordonia sputi (n = 2)	2	2	2	2	0	0
Gordonia terrae (n = 1)	0	1	0	1	1	0
Gordonia sp. (n = 1)a	0	0	0	1	1	0
Kitasatospora sp. (n = 1)	1	1	1	1	0	0
Kroppenstedtia eburnea (n = 2)	0	0	0	0	2	2
Rhodococcus corynebacterioides (n = 1)	1	1	1	1	0	0
Rhodococcus equi (n = 2)	2	2	2	2	0	0
Saccharothrix sp. (n = 1)	0	0	0	0	1	1
Saccharopolyspora sp. (n = 2)	0	0	0	0	2	2
Streptomyces sp. (n = 28)	3	12	8	20	20	8
Tsukamurella pulmonis (n = 4)	0	3	0	4	4	0
Tsukamurella sp. (n = 1)b	0	1	1	1	0	0
No. correct	9	31	15	42	46	19
% correct	14.8	50.8	24.6	68.9	75.4	31.1

^aMALDI-TOF MS identification was G. terrae.

^bMALDI-TOF MS identification was T. inchonensis.

DISCUSSION

This study provides a comprehensive description of the utility of MALDI-TOF MS for the identification of Mycobacterium spp., Nocardia spp., and other aerobic actinomycetes. MALDI-TOF MS is an efficient and effective method for the identification of these groups of organisms. However, analysis is not as straightforward as that reported for other groups of bacteria and for yeasts. While direct spotting or on-plate extraction is feasible for many groups of organisms, the mycobacteria, Nocardia spp., and other aerobic actinomycetes require additional up-front processing steps both for safety reasons and for effective lysing of the organism to make proteins available for analysis. Our work confirms the results of an earlier report by Dunne et al. (12) that a 10-min incubation in 70% (vol/vol) ethanol is sufficient to render mycobacteria, including the M. tuberculosis complex and the aerobic actinomycetes, nonviable. The bead-beating step performed after ethanol treatment is an additional lysis step that adds a secondary layer of safety to the process, and in our facility, work is conducted in a BSL3 laboratory until completion of the bead-beating step. Finally, additions of acetonitrile, formic acid, and matrix are also anticipated to render organisms nonviable, providing redundant layers of safety to prevent laboratory staff exposure to viable organisms using this processing approach. We selected a 1-μl inoculating loop size to “pick” isolates for the extraction procedure in order to reduce the potential for overinoculation of tubes during the ethanol step, but our data also demonstrate that even with a 10-fold increase in the inoculum, it will still be rendered nonviable and safe for MALDI-TOF MS analysis.

In our preprocessing viability studies, we found a single isolate of Streptomyces sp. that was still viable after the ethanol incubation step. A possible explanation is the potential overinoculation of the tube. Unlike the mycobacteria, Streptomyces is a filamentous organism, which makes it difficult to obtain a small amount since the inoculating loop tends to pick up the organism as a mat/mass of filaments. Since Streptomyces is a BSL2 organism and since there is an additional bead-beating lysis step after the ethanol step, performing subsequent MALDI-TOF MS analysis under BSL2 conditions on this genus is reasonable in our assessment. The processing method described is accomplished in a single tube and does not include any heat or washing steps. One-tube processing is advantageous for the laboratory technologists for workflow organization, time, and avoidance of potential ergonomic issues from opening and closing tubes repeatedly. In addition, we utilized a speed vacuum to remove any remaining ethanol, thus avoiding washing steps and ensuring that the ethanol does not inhibit spectra acquisition.

Previous studies have reported that additional entries to the manufacturer's libraries (BDAL 5627 and Mycobacteria Library v. 2.0) are needed in order to successfully identify Mycobacterium spp., Nocardia spp., and other aerobic actinomycetes to the species level (10, 15). Along with the addition of type strains from culture collections such as the ATCC or DSMZ, a number of well-characterized clinical isolates that were identified using 500-bp 16S rRNA gene sequencing or nucleic acid hybridization probes were also added. The clinical isolates helped to account for strain diversity which often is not provided through the use of type strains alone. In the case of some of the aerobic actinomycetes, such as Streptomyces spp., the addition of 46 clinical strains still did not ensure reliable identification for all isolates in the genera during the head-to-head comparison arm of the study because of the tremendous diversity present in these genera. MALDI-TOF MS is able to differentiate some mycobacteria species that are not able to be differentiated by the 500-bp 16S rRNA gene sequencing and that have previously required additional supplemental tests such as the use of additional molecular targets (e.g., M. chelonae/M. abscessus) or a photochrome test (e.g., M. ulcerans/M. marinum and M. kansasii/M. gastri). Confirming the results of an earlier report by Verroken et al. (11), we found that using only the manufacturer's MALDI-TOF MS library resulted in poor rates of identification of Nocardia species. Identification of Nocardia sp. isolates improved upon use of the custom library such that 90% of Nocardia spp. were identified to the species level. Nocardia veterana and Nocardia elegans, and Nocardia higoensis and Nocardia shimofusensis are not able to be separated to the species level using 500-bp 16S rRNA gene sequencing, but they can be reliably differentiated by MALDI-TOF MS using the supplemented library. With use of both the manufacturer's library and the custom library, only 51% of non-Nocardia aerobic actinomycetes were identified using MALDI-TOF MS in the head-to-head study, and this likely is due to the reduced representation of strain diversity in both libraries. In some instances, only a single strain was added to the custom library because only a single strain was able to be acquired, but addition of multiple strains is preferred to enhance the robustness of the library. As the libraries continue to expand, we anticipate that the rate of success for aerobic actinomycete isolate identification by MALDI-TOF MS will continue to improve, and the need to reflex to sequencing will continue to decrease over time. Compared to sequencing, MALDI-TOF MS is faster by approximately 1 day, is more cost-effective, and is less labor-intensive. Since implementing MALDI-TOF MS for these organisms, we have decreased our sequencing volume for mycobacteria and aerobic actinomycetes by approximately 88%.

MALDI-TOF MS does have limitations for the identification of mycobacteria, Nocardia spp., and other aerobic actinomycetes in the clinical microbiology laboratory. The MALDI-TOF MS workflow is less complex and has a faster turnaround time than sequencing. However, unlike sequencing where only scant growth is needed for identification because of a PCR amplification step that occurs prior to sequencing, MALDI-TOF MS requires a greater amount of organism for analysis because there is no preamplification step. While this is rarely an issue for the rapidly growing mycobacteria, it can be a problem for some slowly growing mycobacteria and might delay the turnaround time for identification. In our hands, MALDI-TOF MS can often be used to identify rapidly growing mycobacteria directly from positive MGIT broth tubes as long as several wash steps are done to remove interfering substances prior to analysis. Unfortunately, the use of MALDI-TOF MS on positive MGIT broths did not work as well in our experience when we analyzed slowly growing mycobacteria. Although the reasons for lower success with slowly growing mycobacteria are not clear, we suspect that the low organism burden combined with the interfering substances present in MGIT tubes reduces the ability to generate sufficient levels of protein peaks to allow for identification. Another limitation in the use of MALDI-TOF MS is the potential for no identification or, theoretically, incorrect identification if nonisolated colonies from a culture plate containing more than one species of organism are being tested. To obtain enough biomass for MALDI-TOF MS analysis, the technologist may have to sweep from a “hazy,” thin layer of growth on the first quadrant of a culture plate or may select several pinpoint colonies from a quadrant, assuming that they are isolated colonies of a single organism. However, these may be mixed cultures that will be recognized as the colonies on the culture plate mature. Mycobacterium lentiflavum, Mycobacterium xenopi, Mycobacterium avium complex, and Mycobacterium gordonae are examples of species which can appear initially as a film on a culture plate. In addition, the presence of a rapidly growing mycobacterium may mask the growth of a more slowly growing mycobacterium underneath. In our experience, these situations lead to no identification rather than a misidentification, but continued vigilance during the extended incubation of culture plates is necessary to ensure that a pure colony or culture was tested and that no additional morphologies consistent with a second organism are observed over time. Although both the manufacturer's libraries and the MCL performed well for the identification of these groups of organisms, the use of an extraction method that differed from the manufacturer's recommended extraction method may account for the enhanced performance of the custom library relative to the manufacturer's library. Use of the manufacturer's extraction procedure might potentially improve the identification rate obtained with the manufacturer's library, but this gain must be balanced by the increased preanalysis processing work that needs to be performed in the laboratory.

In instances where multiple species have scores higher than the cutoff threshold by MALDI-TOF MS, a “10% rule difference” is applied to determine whether to report the species with the top score (16). Our data suggest that this is required for <1% of all Mycobacterium species and Nocardia species analyzed using MALDI-TOF MS and the manufacturer's databases combined with the MCL database. In the rare instance that two scores are within 10% of each other, the isolate is reflexed to DNA sequencing for identification. Finally, some closely related mycobacteria are not able to be easily identified to the species or subspecies level by MALDI-TOF MS and should be reported at the complex level or have species identification done using an alternative method such as DNA sequencing. Notable examples of organisms that we report at the complex level are members of the M. avium complex and members of the M. abscessus complex.

In summary, this study demonstrated that with a custom, enhanced library and a streamlined extraction procedure, the laboratory can use MALDI-TOF MS to reliably and rapidly identify approximately 88% of Mycobacterium species, 90% of Nocardia species, and 51% of other aerobic actinomycetes encountered in routine clinical practice at a tertiary medical center/reference laboratory. Since not all laboratories have the resources necessary to generate a custom mycobacterial or aerobic actinomycete library, we also provided information about the ability of the manufacturer's library to identify these groups of organisms and about the effects of lowering the accepted cutoff score from ≥2.0 to ≥1.7. As the manufacturer continues to expand its database, we anticipate that many laboratories will have the ability to identify many of the isolates they routinely encounter using MALDI-TOF MS. An expanded custom library may ultimately be the most useful tool for identification of the uncommon species encountered most often in a reference laboratory setting.