PowerPoint Presentation · •VITEK MS IVD 2.0 –91.2% sppID; Bruker RUO 88% ID •Identical identification rates for unusual or difficult to ID •85.1% correct species ID for both

11/12/2016

1

BETTER, FASTER, STRONGER:

CLINICAL MICROBIOLOGY IN THE

ERA OF MALDI-TOF MASS

SPECTROMETRY

Susan Butler-Wu, Ph.D., D(ABMM), SM(ASCP)

Associate Professor

Keck School of Medicine of USC

Los Angeles, CA

Timeline of diagnostic microbiology

1677

Van Leeuwenhoek observes

“animalcules”

1881

First bacterial culture

(Robert Koch)

1882

Gram stain

(Christian Gram)

1884

Acid-fast stain

(Paul Ehrlich)

1887

Petri Dish Invented

(William Petri)

1928

Penicillin Discovered

(Alexander Fleming)

1977

DNA Sequencing

(Sanger & Gilbert)

1983

PCR Invented

(Kary Mullis)

1949

Poliovirus grown in test tube

cultures of human tissues (Enders, Weller and

Chapman Robbins)

1995

First complete bacterial

genome sequenced (Venter, Smith & Fraser)

Disk Diffusion

(Kirby, Bauer, Sherris & Turck)

1990

Continuous monitoring blood

culture systems

1997

Automated biochemical ID

1966

2000

First launch of NGS

technology

2006

First launch of a MALDI

microbial ID system

2011

First launch of “syndromic”

infection testing

2014

First CLIA-waived

molecular test

Dingle TC & SM Butler-Wu. 2013. Clin. Lab. Med. 33(3):589-609

MALDI-TOF-MS = Matrix

Assisted Laser Desorption

Ionization Time-of-Flight Mass

Spectrometry

Apply matrix solution

1.

2.

3.

Anaerobic

bacteria

Select pure colony

Full protein

extraction

OR

Apply colony to

target plate

Overlay with

formic acid

Apply supernatant to

target plate

Gram-negative

bacteria

Select pure colony

1.

Apply colony to

target plate

2.

Direct analysis

Analyze in MALDI-TOF MS instrument

1.

2.

3.

Select pure colony

Gram-positive bacteria

and yeast

Overlay with

formic acid

Full protein

extraction

Apply

supernatantto target plate

Apply colony to

target plate

Direct analysis

OR

OR3.

Bourassa L & SM Butler-Wu. 2015. In: “Methods in Microbiology:Current and emerging technologies for the diagnosis of microbial infections”

Major advantage: Much faster TAT

Dingle TC & SM Butler-Wu. 2013. Clin. Lab. Med. 33(3):589-609

Real-world TAT data

• 952 prospective bacterial & yeast isolates1

• >85% isolates ID’d on first day of analysis with MALDI

• 9.4% isolates ID’d on first day with conventional methods

• Average time to identification reduced by 1.45 days

• >400 consecutive bacterial & yeast isolates compared

pre- & post-MALDI implementation2

FINAL ID: 60.50 h vs.

49.98 h

1Tan KE et al. J. Clin. Microbiol. 2012. 50(10), 3301-33082Charnot-Katsikas A et al. J. Med. Microbiol. 2014. 63(Pt 2):235-41

11/12/2016

2

THE MAJOR PLAYERS…

Bruker MALDI Biotyper

• CA System - FDA-cleared

• 280 species (Gram-positive,

Gram-negative, yeast &

anaerobes)

• Separate Research Use

Only (RUO) database

• >2,200 species (371 genera)

• Separate RUO

mycobacterial and fungal

databases

In the red corner…

Identification by MALDI-TOF MS: The

Bruker method

Main Spectrum (MSP)

Match score >2.0 = definitive ID

Match score >1.7 = ID to genus level

bioMerieux VITEK MS

• IVD v2.0 - FDA-cleared

• 193 species (Gram-positive,

Gram-negative, yeast &

anaerobes)

• RUO Saramis v4.12

Database (available with

VITEK MS Plus)

• >1300 species (>300 genera),

which also includes

mycobacteria and filamentous

fungi

In the blue corner…

Identification by MALDI-TOF MS: The

bioMerieux method

Unimodal Species

Closely Related Species

Multimodal Species

Match to superspectrum = definitive ID

Match to multiple individual spectra = definitive ID

Weighted Bin Matrix

11/12/2016

3

Head-to-head: Routine bacterial & yeast

isolates• Initial studies showed superior performance of the Bruker

Biotyper vs. VITEK MS• Identification rates of 94.4-95.3% vs. 88.8-93.4%

• Database enhancements led to improved performance 92.7-93.9% vs. 93.2-93.7%

• Performance for anaerobes• 61.3% (Biotyper) vs. 75.3% (VITEK MS) spp. identification rates3

• Bacteroides spp. > problematic for VITEK MS

• Fusobacterium spp. > problematic for Biotyper

• VITEK MS IVD 2.0 – 91.2% spp ID; Bruker RUO – 88% spp ID

• Identical identification rates for unusual or difficult to ID• 85.1% correct species ID for both systems

1Carbonelle E et al. J Microbiol Methods.2012. 89(2):133-1362Cherkaoui A et al. J. Clin. Microbiol. 2010. 48(4): 1169-11753Martiny D et al. J. Clin. Microbiol. 2012. 50(4), 1313-13254Deak E et al. DMID. 2015. 81(1), 27-335McElvania-Tekippe E et al. J. Clin. Microbiol. 2013. 51(5), 1421-1427

What about IVD cleared versions?

• Biotyper CA system (1st claim)

• 2,263 GN rods (23 genera, 61 spp.)

• 98.2% identified correctly to spp. level

• Only 71% of H. influenzae correctly identified

• H. haemolyticus misidentified as H. influenzae

• Addition of H. haemolyticus strains known to improve

performance - 99.6% accurate for H. influenzae2

• Performance of most recent claim not yet published…

• 98.9% identified to genus or species level per FDA

1Faron ML et al . PLoS One. 2015. 10(11):e01413502Bruin JP et al. EJCMID. 2014. 33(2):279-84

What about IVD cleared versions?

• VITEK MS IVD 2.0

Category No. isolates ID to spp. MisID Reference

Gram-positive 1,147 92.8% 1.6% Rychert J,

2013

Enterobacteriaecae 965 96.1% 0.6% Richter SS,

2013

Non-Enterobacteriaecae 558 77.8% 1.6% Manji R,

2014

Fastidious, aerobic 226 96% 4% Branda JA,

2014

Anaerobes 651 91.2% 3.2% Garner O,

2014

Yeast 852 96.1% 0.6% Westblade

LF, 2013

A cautionary tale…

• 62 yo male (vacationing

in Thailand) suffered a

heart attack &

hospitalized for 1 week

• Had been hiking in rural

areas

• 1 month after return to

Seattle, presented to

PCP with UTI symptoms

• Urine sent for culture

• 24 hours – NG on MAC,

growth on BAP

• 48 hours – growth on

MAC & BAP

Dingle TC et al. J Clin Microbiol. 2014. 52(9):3490-1

Image: CDC PHIL

What happened next…

Burkholderia thailandensis

(SV= 1.864)

Oxidase

Result

This was ultimately identified by the WA state PH

Laboratory as B. pseudomallei

• 21 employees exposed; both “high risk” & “low risk”

exposures

MALDI-TOF and select agents

• Bruker Biotyper: Security-Relevant database (53 spectra

from 6 select agents; RUO)

• All 18 Francisella & Brucella isolates id’d to at least genus level

• 1 B. pseudomallei id’d as B. mallei

• 1 B. pseudomallei id’d as B. pseudomallei

• “no reliable identification” for all Brucella & Francisella isolates when

tested against Biotyper reference library

• VITEK MS: IVD database has B. anthracis & Y. pestis (not

cleared)

• RUO database has others, including B. pseudomallei

Cunningham & Patel. J Clin Microbiol. 2013. 51(5): 1639

11/12/2016

4

Other misidentification issues…

• Bruker Biotyper• S. pneumoniae vs. viridans streptococci

• S. pseudopneumoniae vs. S. pneumoniae

• Shigella vs. E. coli

• N. meningitidis vs. N. polysaccharea

• VITEK MS• Shigella vs. E. coli

• Kocuria rhizophila misidentified as Corynebacterium jeikeium

• Anaerobic GP’s misidentified as coagulase-negative staphylococci

• S. pseudopneumoniae misidentified as A. schaalii

• Other issues• Salmonella – not reliable for serotyping

• Difficulty resolving members of complexes or groups (e.g. B. cepacia, Bacillus cereus)

Cunningham SA et al. J Clin Microbiol. 2014. 52(6):2270-1Van Prehn J et al. DMID. 2016.

Alby K et al. J Clin Microbiol. 2015. 53(1):360-1

A special word on the Actinomycetaceae…

• 158 strains, isolated over 8 years

• Tested with both VITEK MS IVD v2.0 and Bruker Biotyper IVD

Ferrand J et al. J Clin Microbiol. 2016. 54(3):782-4

The diversity of Actinomyces…

• 115 patients with invasive

Actinomyces infections

• 41% correct spp ID by VITEK MS

IVD 2.0 vs. 16S rDNA

• 13% misidentified with “excellent

ID->99% probability”• Clostridium perfringens

• Escherichia coli

• Kocuria kristinae

• Kytococcus sedentarius

• Arcanobacterium haemolyticum

• Streptococcus mitis

• Geobacillus thermoglucosidasius

• Paracoccus yeei

• Staphylococcus warneri

• Cedecea neteri

Lynch T et al. J Clin Microbiol. 2016 Mar;54(3):712-7

The bottom line…

• MALDI-TOF-MS generally displays equivalent or superior

performance compared with traditional methodologies

System No. Type

Period of

Isolate Collection

Genus

Level

Species

LevelCountry Comparator

Comparator

PerformanceReference

Bruker

Biotyper1013 Bacteria 2 mo 99% 97% France

Phoenix, API,

Biochemical

93.2%

species, 98.4% genus

Bessede,

2011

Bruker

Biotyper468 Bacteria 3 mo 97% 92% Japan

MicroScan,

API, Phoenix

91.5%

species

Sogawa,

2011

Bruker

Biotyper

2781

(927 x3)

Bacteria 1 mo 96% 85% AustraliaVITEK2, API,

BiochemicalNot provided

Neville,

2011

Vitek MS 767 Bacteria 6 wk 95% 87% France VITEK2 Not providedDubois,

2012

Bruker

Biotyper/

Vitek MS

986 Bacteria 3 mo 96%/94% 93%/93% Belgium

Bruker

Biotypercompared to

Vitek MS

N/AMartiny,

2012

Adapted from Patel R, 2013. Clin Inf Dis. 57(4):564-72

MALDI-TOF-MS is cost-effective for

routine organism ID

• Estimated annual cost savings of $102,413 per year1

• 54% less than standard protocol

• Estimated annual reagent cost savings of €26,771 ($34,385)2

• Estimated reagent savings of $69,108 per year3

• 87.8% less than standard protocol

UWMC Pre-MALDI Post-MALDI

Reagents $41,035 $4,376

Labor $87,637 $54,306

Sequencing $29,700 $5,700

Total $158,372 $64,382

• Annual savings of $94,000/year

• Reduced number of routine

bacteria & yeast isolates requiring

sequencing from 300 to 60 per

year

1Tan KE et al. J. Clin. Microbiol. 2012. 50(10):33012Martiny D et al. EJCMID. 2014 33(5):7453Tran A et al. J. Clin. Microbiol. 2015. 53(8):2473-9

11/12/2016

5

Benefits to patients

• Direct from blood culture ID• 44.6% of pts with improved time to appropriate therapy1

• Decreased all-cause 30-day mortality (21% vs. 8.9%) and annual hospital

cost savings of $2.4 million2

• Blood culture colony identification • MALDI ID of colonies from positive blood cultures3:

• improved time to optimal antibiotic therapy (90.3 vs. 47.3 hours)

• decreased mortality (20.3% vs. 14.5%)

• decreased length of intensive care unit stay (14.9 vs. 8.3 days)

• Both strategies require the involvement of engaged and

empowered antimicrobial stewardship!

1Tamma PD et al. Inf. Cont. Hosp. Epidemiol. 2013. 34(9):9902Perez KK et al. J Infect. 2014. 69(3):2163Huang AM et al . Clin. Inf. Dis. 2013. 57 (9): 1237

Image: www.merivaara.com

• Identification of mycobacteria

• Detection of antimicrobial

resistance

Why do we need to accurately

identify Mycobacterium spp ?

• The Mycobacterium genus consists of >170 species

• Molecular strategies for the direct detection of M.tuberculosis

are not perfect

• Xpert MTB/Rif had a pooled sensitivity of 89% over 22 studies1

• 2x MTB/Rif = 91.1% sensitive for pulmonary TB2

• Culture isn’t going anywhere…

• Important treatment differences for Non-Tuberculous

Mycobacteria (NTM)

• e.g. M. fortuitum vs. M. chelonae

• e.g. M. abscessus subsp abscessus vs. M. abscessus subsp bolletii

• ATS/IDSA recommends to ID clinically significant NTM

isolates to the species level whenever possible3

1Steingart KR et al. Cochrane Database Syst Rev. 2014.1:CD0095932Luetkemeyer AF et al. Clin Inf Dis. 2016. [Epub ahead of print]3Griffith DE et al. Am. J. Respir. Crit. Care Med. 2007. 175:367

How are Mycobacterium spp identified?

• Phenotypic

• Growth characteristics

• Biochemical phenotype

• Pigment production

• Sub-optimal accuracy

• Weeks, Labor intensive

HPLC:

• complicated extraction, not widely available

• Hours, $$

• Molecular methods – Days, $$$

• Mycobacteria

• TB, MAC, M. kansasii and M. gordonae probes

• Sequencing: rpoB, hsp65, 16S rDNA, etc.

Micro Lab 2016

Staff Retreat

Why use MALDI-TOF-MS to identify

mycobacteria?

• Possible advantages over traditional methodologies

• potential for rapid diagnosis after culture positivity

• inexpensive (<$1 per specimen)

• relatively specific

• But some other considerations…

• organism inactivation – TB!!!

• cell wall

• reproducibility/robustness

• amount of biomass required

• databases

Doern & Butler-Wu. 2016. J. Mol. Diagn. 18(6): 789–802

11/12/2016

6

• Transfer 1 µl mycobacterial biomass to 300 µl deionized

water and approx. 200 µL of 1 mm silica beads

• Heat at 95oC for 30 min

• Add 900 µl of 100% ethanol and horizontal vortex for 10

min

• Transfer the supernatant to a new tube using a pipette

• Centrifuge 2 min at 13,000 x g, aspirate off as much

ethanol as possible - repeat

• Air dry 10 min

• Add 10 µl of 70% formic acid and mix using the pipette tip

or vortex

• Add 10 µl of 100% acetonitrile and ololvortex for 20-30

sec

• Centrifuge 1 min at 10,000 x g

• Transfer 1 µl supernatant to target plate

UW Extraction Protocol

Mather C A et al . J. Clin. Microbiol. 2014;52:130 Mather C A et al . J. Clin. Microbiol. 2014;52:130

Similar findings in subsequent studies• 157 Mycobacterium isolates from Washington University,

Saint Louis

Wilen CB et al. J. Clin. Microbiol. 2015; 53(7):2308-15

Antimicrobial resistance detection

by MALDI-TOF MS

Image: (Super germ via Shutterstock)

The growing burden of antimicrobial

resistance…

Images: CDC

Threat level of urgent

• Clostrdium difficile

• CRE

• Drug-resistant

Neisseria gonorrhoeae

Threat level of serious:

12 microorganisms listed,

including:

• MDR Acinetobacter &

Pseudomonas

aeruginosa

• Vancomycin-resistant

Enterococci (VRE)

• MRSA

• Drug-resistant

tuberculosis…

The curious story of MRSA & MALDI…

Conflicting earlier reports

regarding ability of MALDI-

TOF MS to distinguish

MRSA and MSSA isolates

Identical spectral profiles

between isogenic SCCmec+

and SCCmec-neg mutants

SzabadosF et al. J Infect. 2012. 65(5):400-5

11/12/2016

7

Don’t throw the baby out with the

bathwater...

USA100

USA300

This peak is observed only when

direct cell analysis is performed,

and absent if cells undergo

extraction

VRE & MALDI-TOF MS

1Griffin PM et al.J Clin Microbiol. 2012. 50(9):2918-312Nakano S et al. Int J Antimicrob Agents. 2014. 44(3):256-9

vanB+

vanB-

67 consecutively isolated vanB+

E. faecium isolates analyzed1

• Identified 5 peaks of interest

• Prospectively evaluated a

Support Vector Machine (SVM)

model based on these peaks

• Sensitivity of 96.7% and a

specificity of 98.1%

Similar approach for vanA-positive

E. faecium showed 87.5%

sensitivity, 97.3% specificity2

Detection of a blaKPC-containing plasmid

• 2011 outbreak at NIH with

Klebsiella pneumoniae harboring

a pKpQIL plasmid with blaKPC

• 11,109 Da protein present in

blaKPC+ strains from the outbreak1

• Top-down proteomics identified

the 11,109 Da protein as

pKpQIL_p019 (p019)

• p019 detected by MALDI-TOF

MS in 25/26 strains positive for

the pKpQIL plasmid2

Spectra from 18 recultured

blaKPC-positive & blaKPC-negative K. pneumoniae

isolates1Lau AF et al. J Clin Microbiol. 2014. 52(8):2804-122Youn JH et al. J Clin Microbiol . 2016. 54(1):35-42

Detection of enzymatic activity

The diversity of β-lactamasesCategory Enzymes Features

Broad Spectrum TEM-1, TEM-2, SHV-1 Penicillinases - limited activity vs. cephs

Extended Spectrum TEM, SHV variants Active against extended cephs, penicillins

Inhibited by Clavulanic acid, etc.

Extended Spectrum CTX-Ms As above, but < active vs. ceftazidime

Inhibitor-resistant Mostly TEM-1, TEM-2 variants R to inhibitor combos

Serine carbapenemases KPC

SME, NMC-A, IMI-1

KPC – plasmid

• Detection issuesSME – S. marcescens (chomosomal)

NMC - E. cloacae (chromosomal)

Metallo-β-lactamase

(carbapenemases)

VIM – plasmid, P. aer & now enterics

IMP- plasmid, enterics & non-enterics

NDM-1 - plasmid

Others - chromosomal

R to carbapenems!!!

• Detection issues

Cephalosporinase Many – CMY, LAT, MIR, FOX, etc. Chromosomal

• not inhibited by clavulanic acid, etc.• Inducible – e.g. Enterobacter, Citrobacter

Plasmid

Oxacillin-hydrolyzing OXA Integrons, plasmids

Resistance to variety of penicillinsESBL-phenotype in P.aer = OXA-11, OXA-14, OXA-20

Some can hydrolyze carbapenems e.g. OXA 24-40,

OXA-48

Detection of carbapenemase activity by

MALDI-TOF-MS

Burckhardt & Zimmerman. JCM. 2011. 49 (9): 3321-3324

• Cheap

• <$1.50/reaction

• Fast

• 1-2 hours,

depending on

enzyme

• BUT only capable of

detecting hydrolysis-

based resistance

• Subsequent studies have also adapted the method to cephalosporins and tested

a variety of carbapenemases and beta-lactamases

• Detection of cefotaxime-resistance by MALDI directly from blood cultures

showed 100% sensitivity and 92% specificity for Enterobacteriaceae**Jung et al. JCM. 2014. 52(3): 924–930

11/12/2016

8

Issue: Lack of standardization between

studies

Mirande C et al. Eur J Clin Microbiol Infect Dis. 2015. 34(11):2225-34

(Imipenem)

(+ Cephalosporinase)

(+ ESBL)

(+ OXA-48)

(+ KPC)

(+ NDM)

300 Da254 Da

Lasserre C et al. J Clin Microbiol. 2015. 53(7):2163-71

1-μl loopful of

bacteria suspended

in 20 μl of 0.5 mg/ml

imipenem solution -

20 min incubation,

37°C

[M/(M + I)] calculated

for 223 clinical strains

(77 carbapenemase-

positive)

ROC analysis: Cut-off

of 0.82 = 99%

sensitive, 100%

specific for 266

clinical isolates (102

carbapenemase-

positive)

Where are we going? The case for

moving beyond peak detection…

Vancomycin-intermediate S. aureus

• Vancomycin is a mainstay in the treatment of serious

MRSA infections

• CLSI MIC breakpoints:

• S ≤2µg/mL, I=4-8µg/mL, R ≥16µg/mL

• VISA is associated with both mutation & transcriptional

changes

• plethora of genes

• VISA strains are associated with

• Thickened cell wall

• Decreased autolytic activity

• Decreased cross-linking between murein monomers

• Increased proportions of non-amidated muropeptides

The hVISAproblem

• “h” is for heterogeneous

• resistant population present at frequency of ≤10-5-10-6

• but susceptibility testing is performed using an inoculum of 5x104

CFU/ml

• Population analysis profile PAP is considered the gold

standard for hVISA detection

• hVISA PAP/AUC ≥0.9 vs. control strain (Mu3)

• time-consuming & labor-intensive

Does hVISA matter clinically?

• Retrospective, multicenter study – 2004-2012• 122 pts with MRSA BSI – 61 matched pairs (hVISA and VSSA)

• 11x greater odds of

vancomycin failure

• Increased LOS by

median of 8 days

Casapao AM et al. AAC. 2013. 57:4252

11/12/2016

9

Comparison of VISA and VSSA by

MALDI-TOF MS

• 16 VISA, 12 VSSA strains

• Identified peaks present in

≥75% of either group (i.e.

VISA or VSSA) with a

coefficient of variance (CV)

<50% within that group

• 22 peaks that maximized the

separation of VISA and VSSA

isolates

Mather CA et al. J Clin Microbiol. 2016. [Epub ahead of print]

Extended analysis to 21 VISA & 38 VSSA

isolates

Forward stepwise regression

performed on these 60 peaks

identified 14 peaks that best

separated the VISA and VSSA

isolates

Mather CA et al. J Clin Microbiol. 2016. [Epub ahead of print]

Reliable separation of VISA and VSSA

using SVM analysis

• Overall classification accuracy

of 98%

Mather CA et al. J Clin Microbiol. 2016. [Epub ahead of print]

• Inclusion of hVISA in SVM

algorithm: Overall

classification accuracy of 89%

What about antimicrobial susceptibility?

• MBT-ASTRA

• Quantitative MALDI

• Internal quantitative standard in combination with normalization to

maximum peak allows direct comparison of peak intensities –

“relative growth values”

• 108 Klebsiella isolates – 97.3% sensitive & 93.5% specific

compared with E-test (MIC method)

• Media already frequently used in clinical lab (BHI)Susceptible strain Resistant strain

Lange C et al. J Clin Microbiol. 2014 52:4155–4162

MBT-ASTRA: MIC of Acinetobacter

baumannii isolates vs. E-test MIC

• 3.5 hour

incubation

• [Tobramycin]

range tested =

2–256 μg/mL

• BUT assay has

to be optimized

for every bug-

drug combo…

Sparbier K et al. Methods. 2016. [Epub ahead of print]

Conclusions

• MALDI-TOF MS is revolutionizing the practice of clinical

microbiology

• Allows us to see microorganism species we never knew

we were actually seeing!

• MALDI-TOF is not infallible – not a substitute for skilled

microbiologists

• Identification of mycobacteria by MALDI-TOF will reduce

the need for sequence-based identification

• Detection of resistance & MALDI-TOF: Watch this

space…

11/12/2016

10

Acknowledgements

• University of Washington

• Cheryl Mather, M.D.

• Brian Werth, Pharm.D.

• Susan Turner

• Lauren Curtis

The UW extraction protocol effectively

inactivates Mycobacterium spp.• Inoculated Middlebrook 7H11 agar and VersaTREK Myco

bottles with 100 µl of sample after the inactivation step

• Tested the following:• M. tuberculosis (5 strains)

• M. chelonae

• M. kansasii

• M. abscessus

• M. fortuitum

• Result: No growth i.e. inactivation achieved