Professional Documents
Culture Documents
BACKGROUND: First introduced into clinical microbiol- urine samples and positive blood culture bottles. The
ogy laboratories in Europe, MALDI-TOF MS is being strengths and limitations of MALDI-TOF MS applica-
rapidly embraced by laboratories around the globe. Al- tions in clinical microbiology are also addressed.
100
MALDI-TOF for the Diagnosis of Infectious Disease
Reviews
lished a study demonstrating that by constructing their demic laboratories have engaged in database building to
own database of mass spectra, they could accurately iden- enhance the performance of commercially available sys-
tify nonfermenting gram-negative bacterial isolates (6 ). tems. The Mayo Clinic Custom MALDI-TOF MS Li-
A year later, Seng et al. published a prospective evaluation brary, for example, currently contains 1599 mass spectral
of Bruker’s commercial mass spectral database for iden- entries representing organisms not adequately addressed
tification of 1660 bacteria routinely isolated in their lab- by commercially available databases.
oratory, reporting correct identification of 95% and 84%
to the genus and species levels, respectively, in 6 min/ How Does MALDI-TOF MS Work?
isolate (7 ). Over the past 5 years, upgraded analysis soft-
ware and increasingly inclusive and better curated mass In clinical microbiology, there are several preparatory
trum of the test isolate is compared to a database of ref- peaks in the test isolate not present in database entries.
erence spectra or deconvoluted spectra to determine re- According to the manufacturer’s criteria for their RUO ap-
latedness to spectra in the database; the most closely plication, a score of ⱖ2.000 represents species-level, a score
related organisms are identified with a value provided as of 1.700 –1.999 a genus-level, and a score of ⱕ1.700 no
to the level of confidence in identification. Depending on identification. System users have found that lower score cut-
how high the value is, the organism may be identified at offs may be adequate for identification of certain groups of
the family, genus, or species level. A variety of algorithms organisms; extensive verification by the end user is required
are used for database comparison. For example, the Bio- before adopting this approach. For the MALDI Biotyper
typer system uses pattern recognition and generates a CA System, a score of ⱕ1.999 represents no identification.
score ranging from 0 to 3.000 on the basis of the presence The VITEK MS RUO system database is searched using 1
or absence of mass spectral signals (peaks) in the test of 2 algorithms. One is a pattern matching process, referred
isolate matching database entries, and the presence of to as a reference spectra approach. In the second approach,
mass spectral entries are grouped together as “Super- one another (e.g., Escherichia coli and Shigella species)
Spectra,” representing spectra of conserved mass signals de- that their differentiation using available algorithms is un-
rived from multiple individual isolates of a particular organ- reliable with any of the aforementioned strategies. This is
ism group, with those signals weighted according to their a particular challenge because E. coli is one of the most
specificity for species, genera, and/or families. The output is frequent organisms encountered in clinical microbiology
a confidence value, ranging from 0% to 100%. With the laboratories; MALDI-TOF MS users must use alternate
VITEK MS system, analysis is not directly based on pat- or additional strategies (e.g., lactose fermentation, quick
tern recognition comparing mass spectra. The mass spec- indole) to distinguish E. coli from Shigella species.
tral entries of individual isolates are computationally MALDI-TOF MS does not currently accurately differ-
sorted into weighted bin matrices generating “Advanced entiate all species within certain groups of organisms,
Spectra Classifiers.” The bin pattern of test isolates is such as the Enterobacter cloacae complex (11 ), Burkhold-
compared against the commercial weighted bin matrix. eria cepacia complex, and Streptococcus bovis species
The weighting algorithm weights bins according to their group. Likewise, accurate differentiation of all species of
importance. For example, bins that are highly specific for Acinetobacter may be challenging. Neisseria polysaccharea
individual species receive a higher weight than do those may be misidentified as Neisseria meningitidis using the
that are moderately species specific. All 3 strategies func- Biotyper system (12 ). It is not yet known whether dis-
tion well. Closely related species with similar patterns and crimination of some closely related organisms will be pos-
a small number of distinguishing peaks, such as Strepto- sible with improved databases and/or software. Misiden-
coccus pneumoniae and Streptococcus mitis groups, may be tifications may occur if such details are not considered
better (though not perfectly) distinguished using the during data analysis. The aforementioned limitations
VITEK MS strategy than the others (9, 10 ). Neverthe- aside, most organisms are either correctly identified or
less, spectral profiles of some organisms are so similar to yield a low score/percent, indicating that identification
has not been achieved; the latter typically implies no ing. For some isolates, multiple species had scores
“match” in the database, but can occur due to technically ⱖ2.000; these included Corynebacterium aurimucosum
poor preparation. In the next section, select studies of and Corynebacterium minutissimum, Corynebacterium
MALDI-TOF MS for bacterial and fungal identification simulans and Corynebacterium striatum, Lactobacillus gas-
are reviewed, with a focus on those recently published. seri and Lactobacillus johnsonii, and L. monocytogenes, Lis-
teria ivanovii, and Listeria innocua. Reducing the species
Performance of MALDI-TOF MS for cutoff to 1.700 increased the percentage of isolates iden-
Identification of Aerobic Bacteria tified. These investigators also compared identification of
215 clinical isolates of gram-positive bacilli by MALDI-
MALDI-TOF MS performs at least as well as, if not TOF MS using on-plate formic acid testing and a species
vant” database enables Brucella species and F. tularensis and Mycobacterium intracellulare or Mycobacterium mu-
identification (8 ). cogenicum and Mycobacterium phocaicum. Although My-
cobacterium abscessus and Mycobacterium massiliense may
Performance of MALDI-TOF MS for be challenging to differentiate, Teng et al. did so using
Identification of Anaerobic Bacteria cluster analysis of spectra generated by the Biotyper sys-
tem (30 ). MALDI-TOF MS is easier, less expensive,
MALDI-TOF MS has become the method of choice for faster, and more accessible to routine clinical microbiol-
identification of anaerobic bacteria, replacing 16S rRNA ogy laboratories than traditional strategies for mycobac-
gene sequencing and gas–liquid chromatography. Many terial identification, which will likely make it the method
clinical microbiology laboratories have not had tools of choice for their identification in the near future.
yeast isolates using on-plate formic acid testing (37 ). spectively (44 ). De Respinis et al. developed an in-house
VITEK MS correctly identified 95% of the isolates, database and performed MALDI-TOF MS with an
whereas the Biotyper correctly identified 83% of the iso- AXIMA Confidence mass spectrometer (Shimadzu Bio-
lates using a species-level cutoff of ⱖ1.700. Hamprecht tech) and MALDI-TOF MS Launchpad 2.8 software
et al. compared the Biotyper (v3.0 software, v3.0.10.0 (Shimadzu Biotech) for identification of dermatophytes;
database, species-level cutoff ⱖ2.000) and VITEK MS of 141 clinical isolates tested, 96% were correctly identi-
(V2.0 knowledge base) systems for identification of 210 fied (45 ).
yeasts using on-plate formic acid testing (38 ). The
VITEK MS system identified 96% of the isolates; the Laboratory Work Flow and Cost Associated
Biotyper system identified 91% of isolates. De Carolis et with MALDI-TOF MS
and have so far reduced the number of tubed biochem- increasing protein masses and leading to mass shifts of
ical set– based identifications from 4668 to 987/year. their corresponding peaks in the profiled spectra (52 ).
To date, we have halved the number of isolates requir- Further details on detection of antibiotic resistance
ing 16S rRNA gene sequencing; through progressive using MALDI-TOF MS can be found in the recent
database improvements, this number continues to de- review by Hrabak et al. (53 ).
crease. Following implementation of MALDI-TOF
MS for yeast identification, the associated supply costs Direct Testing of Clinical Samples
decreased from $30 525 to $5400/year as a result of
the elimination of germ tube and rapid assimilation of Although direct testing of clinical samples is not generally
trehalose and API (analytical profile index) strip tests; feasible because of the high limit of detection of MALDI-
with ⬎1.6 scores in 5 mixed cultures in the “Top 10 ony amounts to target plates. Instrument cost must be
matched pattern choices” (67 ). considered; laser, software, hardware, or mass spectrom-
MALDI-TOF MS performed on positive blood cul- eter failure can occur, necessitating an appropriate service
ture bottles can rapidly identify probable contaminants plan. Although MALDI-TOF MS is generally reproduc-
or suggest complementary diagnostic testing in the case ible, there are sources of variability, including the mass
of pathogen detection (68 ). By testing of positive blood spectrometer (different types as well as individual instru-
culture bottles using MALDI-TOF MS, time to organ- ments), matrix and solvent composition, preparation
ism identification is reduced by a day or more (69 ). A methods, technologist training and competence, culture
limitation is that antimicrobial susceptibility is not pro- conditions (such as media, colony age, and temperature),
vided. To overcome this limitation, MALDI-TOF MS and biological variability; QC strategies are being de-
analysis and interpretation of data; (b) drafting or revising the article for intel- Honoraria: None declared.
lectual content; and (c) final approval of the published article. Research Funding: None declared.
Expert Testimony: None declared.
Authors’ Disclosures or Potential Conflicts of Interest: Upon man- Patents: None declared.
uscript submission, all authors completed the author disclosure form. Dis-
closures and/or potential conflicts of interest: Acknowledgments: I thank Dr. Nancy L. Wengenack, Scott A. Cun-
ningham, and Brenda L. Dylla and Jill M. Mainella for their thoughtful
Employment or Leadership: None declared. reviews of this manuscript. I thank Sherry M. Ihde and Sherri L. Wohl-
Consultant or Advisory Role: R. Patel, Curetis and Thermo Fisher. fiel for providing the cost, volume, and turnaround-time data for Mayo
Stock Ownership: None declared. Clinic Rochester.
References
spectrometry for routine identification of yeast. J Clin 48. Sparbier K, Schubert S, Weller U, Boogen C, Kostrzewa ferential centrifugation protocol and validation crite-
Microbiol 2011;49:1614 – 6. M. Matrix-assisted laser desorption ionization-time of rion for enhancing mass spectrometry (MALDI-TOF)
35. Lacroix C, Gicquel A, Sendid B, et al. Evalua- flight mass spectrometry-based functional assay for results in microbial identification using blood culture
tion of two matrix-assisted laser desorption rapid detection of resistance against beta-lactam anti- growth bottles. Eur J Clin Microbiol Infect Dis 2013;32:
ionization-time of flight mass spectrometry (MALDI-TOF biotics. J Clin Microbiol 2012;50:927–37. 699 –704.
MS) systems for the identification of Candida species. 49. Hoyos-Mallecot Y, Cabrera-Alvargonzalez JJ, 62. Leli C, Cenci E, Cardaccia A, Moretti A, D’Alo F, Pa-
Clinical Microbiol Infect 2014;20:153– 8. Miranda-Casas C, Rojo-Martin MD, Liebana-Martos gliochini R, et al. Rapid identification of bacterial and
36. Westblade LF, Jennemann R, Branda JA, Bythrow M, C, Navarro-Mari JM. MALDI-TOF MS, a useful instru- fungal pathogens from positive blood cultures by
Ferraro MJ, Garner OB, et al. Multicenter study evaluat- ment for differentiating metallo-beta-lactamases in MALDI-TOF MS. Int J Med Microbiol 2013;303:205–9.
ing the Vitek MS system for identification of medically Enterobacteriaceae and Pseudomonas spp. Lett Appl 63. Tadros M, Petrich A. Evaluation of MALDI-TOF mass
important yeasts. J Clin Microbiol 2013;51:2267–72. Microbiol 2014;58:325–9. spectrometry and Sepsityper Kit for the direct identifi-
37. Pence MA, McElvania TeKippe E, Wallace MA, Burnham 50. Alvarez-Buylla A, Picazo JJ, Culebras E. Optimized cation of organisms from sterile body fluids in a Cana-
CA. Comparison and optimization of two MALDI-TOF method for Acinetobacter species carbapenemase de- dian pediatric hospital. Can J Infect Dis Med Microbiol
trometry. J Clin Microbiol 2013;51:603– 6. Drissner D. Beyond the matrix-assisted laser desorption in clinical microbiology practice. Ann Lab Med 2014;
75. Josten M, Reif M, Szekat C, Al-Sabti N, Roemer T, Spar- ionization (MALDI) biotyping workflow: in search of 34:111–7.
bier K, et al. Analysis of the matrix-assisted laser desorp- microorganism-specific tryptic peptides enabling dis- 78. Razavi M, Johnson LD, Lum JJ, Kruppa G, Anderson NL,
tion ionization-time of flight mass spectrum of Staphy- crimination of subspecies. Appl Environ Microbiol Pearson TW. Quantification of a proteotypic peptide
lococcus aureus identifies mutations that allow 2014;80:4234 – 41. from protein C inhibitor by liquid chromatography-free
differentiation of the main clonal lineages. J Clin Micro- 77. Mutters NT, Hodiamont CJ, de Jong MD, Overmeijer SISCAPA-MALDImassspectrometry:application to identifi-
biol 2013;51:1809 –17. HP, van den Boogaard M, Visser CE. Performance of cation of recurrence of prostate cancer. Clin Chem
76. Gekenidis MT, Studer P, Wuthrich S, Brunisholz R, Kiestra total laboratory automation combined with MS 2013;59:1514 –22.