Development of a standardized screening method for detection of vancomycin resistant enterococci

(1)

0095-1137/94/$04.00+0

Development of

a

Standardized

Screening

Method for

Detection of

Vancomycin-Resistant Enterococci

JANAM.SWENSON,l* NANCYE C.

CLARK,'

MARYJANEFERRARO,2 DANIEL F. SAHM,3 GARYDOERN,4 MICHAELA. PFALLER, tL. BARTH RELLER,6 MELVINP. WEINSTEIN,7 RONALD J.ZABRANSKY,8

ANDFRED C.TENOVER'

HospitalInfections Program, Centers for Disease Control and Prevention, Mailstop G08, Atlanta, Georgia

30333';

Microbiology

Laboratory, MassachusettsGeneral Hospital, Boston, Massachusetts 021142; Microbiology Laboratory, JewishHospital, St. Louis, Missouri 631103;Departmentof Clinical Microbiology, UniversityofMassachusetts MedicalCenter,

Worcester, Massachusetts 016554; Departmentof Pathology, Oregon Health Sciences University, Portland, Oregon 972101-30985; ClinicalMicrobiologyLaboratory, Duke University Medical

Center, Durham, North Carolina 277106; Microbiology Laboratory, Robert Wood Johnson UniversityHospital, New

Brunswick,

NewJersey08903-00197;

and ClinicalMicrobiology, UniversityofTexasMedical Branch atGalveston, Galveston, Texas 775508

Received 3 January1994/Returned for modification8February 1994/Accepted12April 1994

Theincidence of vancomycin resistance amongenterococci is increasing in the United States and elsewhere in the world, but automated susceptibility testing methods have difficulty detecting resistance expressed by certain strains. Theagarscreening method described by Willey etal. (B. M.Willey, B. N.Kreiswirth, A. E. Simor, G. Williams, S.R.Scriver,A.Phillips, and D. E. Low, J. Clin. Microbiol.30:1621-1624, 1992)has been proposed as a reliable method forconfirming vancomycin resistance. In this study, we investigated various parameters associatedwith theagar screening method and, on the basis of the findings,established optimum testing conditions for the method. First,toevaluatemedia and vancomycin concentrations, one laboratory used Mueller-Hinton and brain heartinfusion agars supplemented with 4, 6, and 8 ,ugofvancomycin per ml to test 100geneticallycharacterizedenterococcal strains. On the basis of the results obtained, brain heart infusion agar supplemented with 6 ,ug of vancomycin per ml was selected for further study. Subsequently, eight laboratories used the medium to test both reference andclinical isolates. Therewasvery good perfonnance with the reference strains and,among 158 clinical isolates tested, the method demonstrated sensitivity and specificity of 100% and from 96to99%, respectively.

Vancomycin resistanceinenterococci wasfirst recognized in the late 1980s (5, 15). Since then, there has been an alarming increase in the incidence of resistance as reported to the National Nosocomial Infections Surveillance system: from 0.3% in 1989 to 7.9% in 1993. The most disturbing increase during this period was among isolates from patients in inten-sive care units (from 0.4 to 13.6%) (1). Furthermore, addi-tional resistant strains may have been missed because of the failure of automated susceptibility testing methods(11, 12, 14,

17) and disk diffusion (8, 13) to recognize strains with low levels ofvancomycin resistance. A recentstudy that used five isolates representative of the currently known vancomycin resistancephenotypes demonstrated problems with MicroScan

(American MicroScan, Sacramento, Calif.), Vitek

(Bio-Merieux, Hazelwood, Mo.), and disk diffusion methods (14).

Although revised breakpoints for disk diffusion (13) and

updated software for the Vitek system (16) have improved detectionby both systems, there have been few reports docu-menting the performance of most commercial systems for detection ofvancomycin resistance in enterococci.

In light of these findings, the vancomycin screening proce-dureproposed by Willeyet al. (17) has potential as a supple-mental method for detection ofvancomycin-resistant

entero-*Corresponding author. Mailing address: CDC, 1-3361

(G08),

1600

CliftonRoad, Atlanta, GA 30333. Phone: (404) 639-0196. Fax: (404) 639-1381.

tPresent address: Department ofPathology, University of Iowa CollegeofMedicine,IowaCity, IA52242.

cocci. However, the optimal conditions regarding inoculum size, medium, drug concentration, and length of incubation have not been established for this method. The purpose of the

I presentstudywas toestablishoptimumtestconditions for the

testusing isolates that had been well characterized genetically.

MATERIALSANDMETHODS

Study design. Thestudywasconducted intwophases.Phase 1 involved the comparisonof several types of agars and three concentrations of vancomycin to optimize the screen test. Multiplelotsof brain heart infusion(BHI)mediumwerethen tested in one laboratory (Centers for Disease Control and Prevention [CDC], Atlanta,Ga.). Phase 2 involved the evalu-ation of the screen test ineightlaboratories.

Participatinglaboratories. Thelaboratoriesperforming the study were at the CDC, Duke University (Durham, N.C.),

MassachusettsGeneralHospital (Boston),JewishHospital(St.

Louis, Mo.), Oregon Health Sciences University (Portland),

Robert Wood Johnson University Hospital (New Brunswick,

N.J.), University of Massachusetts (Worcester), andthe Uni-versity ofTexasMedical Branch(Galveston).

Molecular characterization. The vanA, vanB, and vanC genotypeswere confirmedbyamplifyingthe respective genes by PCRandprobing withaspecificgeneprobe.The

oligonu-cleotide primer sequences, conditions for amplification, and gene probeswerepreviouslydescribed (2).

Organisms. The organisms usedin phase 1 werethe same 100strains usedtodevelop newparametersfor disk diffusion

1700

on May 15, 2020 by guest

http://jcm.asm.org/

(2)

testing

(13).

Before use, the strains were retested by agar dilution with Mueller-Hinton (MH) agar to determine the

susceptibility phenotype. The isolates included 24 strains for which thevancomycinMICswerec2 ,ug/ml (13Enterococcus

faecalis

strains, 10 E. faecium strains, and 1 E.

casseliflavus

strain),

9strains for which the MICswere4to 8

,ig/ml

(allE. gallinarum, vanC), 30 strains forwhich the MICs were 16 to 128 ,ug/ml (22 E. faecium and 8 E. faecalis strains [all vanB exceptfor1strain of E.faeciumfor which theteicoplaninMIC was4 ,ug/ml thatwasneithervanA, vanB, nor vanC]), and 37 strains forwhich the MICs were >128 ,ug/ml (20 E.faecium

strains

[10

vanA and 10

vanB]

and 17 E.faecalis strains [all vanB]). The medium and concentration studies involved the

testingof all 100 strains.In thetestingofmultiplelots ofBHI agarduring phase 1, only 70strainswere used(30strains for which MICs were -128 ,ug/ml were excluded because their resistance was detected without problem in the first part of

phase 1).

Inphase 2, reference and clinical strainsweretestedby all laboratories.Amongthe 10 reference strainswere onethatwas vancomycin susceptible (E.

faecalis),

five with the VanB

phe-notype

(two

E.

faecalis

and three E.faecium strains),twowith the VanA phenotype (E. faecium), and two with the VanC

phenotype

(E. gallinarum).

Inaddition,eachlaboratorytested 20 clinical strains chosen from their culture collections, of whichno morethan 10werevancomycin resistant.Allclinical strains were sent to CDC for molecular and susceptibility

characterization.

Organisms

were subcultured twice prior to

testing

iftheyhadbeen

previously

frozen.

Antimicrobial agents. Each laboratory participating in the

studyuseditsownsupplyofvancomycin powderobtained from either Eli

Lilly

& Co.

(Indianapolis, Ind.)

or

Sigma (St.

Louis,

Mo.).

Phase 1.(i) Comparisonofmedia and concentrations.Agar

dilution

plates

were

prepared

withBHIagarpowderobtained from Becton Dickinson

Microbiology

Systems

(BDMS)

(Cock-eysville,

Md.)

and Difco Laboratories

(Detroit, Mich.)

and withMHagar

powder (MH

II;

BDMS)

containing vancomycin

concentrations of4, 6, and8 ,ig/ml. Plateswerestoredat4to 8°C in

plastic

bags to prevent evaporation. Inoculum was

prepared by suspending growth

from an overnight trypticsoy blood agar

plate

inMHbroth. Thesuspensionwasadjustedto

equal

a0.5 McFarland standard. Plateswereinoculated with both 10- and 1-,ul spots in orderto achieve 106and 105 CFU per spot,

respectively.

The

10-pul

inoculumwasdelivered with a

micropipettor,

and the

1-pld

inoculumwas delivered with a Steers

replicator.

Plateswerereadat

18, 24,

and 48 h. Growth was

interpreted

aspositiveif

confluent,

weak ifahaze thatwas difficulttoreadwaspresent,asthe number ofcolonies if there were<10,and as

negative

ifnogrowthoccurred.Strainswere considered resistant if

growth

was

positive

orweakorifmore than one colonywasobserved.

(ii) Evaluation ofBHI agars.Agar plates containing6 ,ugof

vancomycin

per mlwere prepared with BHI media obtained from five manufacturers: Acumedia

(Baltimore, Md.),

Adams Scientific

(West Warwick, R.I.), BDMS, Difco,

and Oxoid

(Unipath, Ogdensburg, N.Y.).

Both BHI agar and modified BHI agar from BDMS were used; the modified BHI agar medium contains differentamountsof BHI andpeptic digests

of animal tissue and casein than the

regular

BHI agar medium. Three lots of DifcoBHI agarwereused, includingtheoneused in the medium and concentration study. Inoculation and incubation were thesame as in the initial studies except that inocula were delivered with 1- and

10-pd plastic

calibrated

loops.

Phase 2: multilaboratory evaluation ofscreeningplates. A

commonlot ofBHIagarplates(Acumedia) containing 6 p.g of vancomycin per ml was prepared in one laboratory (CDC) and distributed to each participant. In addition, each laboratory preparedunique lots of plates, including BHI agar from Difco (four laboratories), BDMS (two laboratories), Adams Scien-tific (one laboratory), and Oxoid (one laboratory). The refer-ence and quality control strains were tested on both the commonlot and theunique lots of plates. The clinical strains were tested on the unique lots only. Inoculum was delivered with either calibrated loops or pipets. Plates were read as described for the phase 1 studies.

The clinical strains were characterized by a standard Na-tional Committee for Clinical Laboratory Studies (NCCLS) broth microdilution method (7) using cation-adjusted MH brothtodeterminetheir susceptibility phenotypes. The strains were identified with a modified set of biochemicals recom-mendedby Facklam and Washington (4) that was included in the microdilution panels. The biochemical reactions included growth in6.5%NaCl; utilization of mannitol, sorbitol, sorbose,

arabinose, raffinose, and pyruvate; and detection of arginine

dihydrolase. In addition, the presence of an obvious yellow pigment, motility as determined with semisolid agar stabs, and the presence of L-pyrrolidonyl-aminopeptidase and leucine

aminopeptidase were investigated. When we were unable to identify a species by that method, additional biochemical reactionsweretestedby the Streptococcus LaboratoryatCDC. Qualitycontrol. TwoE.faecalis strains were included inall phases of the study: ATCC 29212 (vancomycin susceptible) and ATCC 51299, a strain that is borderline resistant to vancomycin (MIC, 16 to 32

jig/ml).

Studies to validate E.

faecalis ATCC 51299 as a quality control strain for the

screeningtestwillbe reported elsewhere (unpublisheddata).

Stability studies. Sufficientcommon lotplates forphase 2 wereprepared and storedat 4 to8°Csothat thestability of the media could be determined. Bothqualitycontrol strainswere testedone tothreetimesmonthlyfor 6 months with inocula of both105and106CFU, withreadingsmade after 24 and 48 h of incubation.

RESULTS

WecomparedBHI andMH agarscontaining4, 6, and 8

pug

ofvancomycin per mlfor their ability to detect resistance in enterococci. Strainswerecategorized asresistant or suscepti-ble(susceptible,MIC '4

jig/ml;

resistant,MIC 28

pug/ml)

on the basis ofMICcategoriesrecommendedbythe NCCLS(9).

Strains for which MICswereintermediate(8to16

jig/ml)

were grouped with resistant strains. For all parts of the study, E.

gallinarumstrainswerecategorized asresistant because of the presence of thevanCgene evenif the MICswereless than 8

jig/ml

(6).

Strains of E.

casseliflavus

were

categorized

on the

basis of MICs only. The sensitivity (percentage of resistant strains correctly categorized) and specificity (percentage of

susceptiblestrainscorrectlycategorized) of each combination areshown in Tables1 to3.

Overall, thetest performedvery wellwith all combinations of medium, time, and inoculum. At 4

jig/ml

(Table 1), the errorsthat occurredweremainlyinspecificity when BHI agar andthe

106-CFU

inoculumwereused. Most of thesusceptible strains that grew at this concentration and inoculum were

reported

as

having

weak

growth.

The one strain of E.

casselifla-vusincluded in thisphasefor which the MICwas2

jig/ml

grew wellwith the larger inoculumon BHI agar but didnot grow with the 105-CFU inoculumor onMH agar.At 6

jg/ml

(Table 2), errors were fewer and were mainlyin sensitivity with the MH agar plates failing to grow most of the strains of E.

on May 15, 2020 by guest

http://jcm.asm.org/

(3)

TABLE 1. Sensitivity and specificity of vancomycin resistance detection with BHI agar screen platescontaining

4pgofvancomycin perml'

Inoculum Time Sensitivity (%) Specificity (%)

(CFU) (h) BHI1 BHI2 MHA BHII BHI2 MHA

105 18 100 99 99 100 100 100

24 100 100 99 100 100 100

48 100 100 100 100 100 100

10" 18 100 100 100 79 71 92

24 100 100 100 83 79 96

48 100 100 100 88 88 100

TABLE 3. Sensitivityandspecificityofvancomycinresistance detection with BHI agar screen plates containing

8 ,ugofvancomycinperml'

(CFU) (h) BHII BHI2 MHA BHI1 BHI2 MHA

105 18 97 89 92 100 100 100

24 95 91 89 100 100 100

48 96 92 89 100 100 100

106 18 100 100 97 100 100 100

24 100 100 96 96 100 100

48 lOO 100 93 100 100 100

"Atotal of 100enterococci were tested, including 24 strains for which MICs "SeeTable 1,footnotea,for details anddefinitions of abbreviations. were.2(13 E.facciumstrains, 10E. faecalisstrains, and1E. casseliflavus strain),

9strains for whichMICswere 4 to 8,ug/ml (E.gallinarum),and 67 strains for which MICs were.16 pug/ml(42E. faecium and 25 E. faecalis strains). See text

for definitions of resistance. BHI1,BDMS BHI agar; BHI2, Difco BHI agar; tories (including BHI agars from three different

manufactur-MHA,BDMS MHagar. ers).

One hundredfifty-seven clinical strainswerestudiedduring

phase2.Three strainswereexcludedfrom theanalysis because

gallinarum.

There were also

specificity

errors at 6

j.g/ml,

all

they

were determined not to be enterococci. Thevancomycin with the 10"-CFU inoculum, with

growth

of some strains on MIC for 115

strains

was '4

,ug/ml

at 24

h,

including

79 E.

oneof the lots ofBHI agar

reported

asweak. These

specificity

faecalis,

15 E.

faecium,

13 E.

gallinarum,

4E.

casseliflavus,

2 E.

errorsallwereresolvedat48h,whentheresultswere

correctly

raffinosus,

and 2

E'nerococcus

sp. strains. MICs were 8 to 16 readas

negative.

At 8

pg/ml

(Table 3),

_gl/ml

theerrorswere

mostly

at 24 h for three

strains,

including

two E.

gallinarum

in

sensitivity

when the 105-CFU inoculum was used and

they

strainsand one

E. raffinosus

strain

(which

contained thevanA occurred with all three media tested. Most errors occurred

gene).

MICsfortheE.

casseliflavus

strains tested

ranged

from with the E.gallinarum strains.However,twoof the errors were 2 to 4

Rg/ml;

for the E.

gallinarum

strains, MICs were 2 to with twoE.

faecalis

strainsfor which the MICs were 16 to 32 >512

,ug/ml.

ThemostcommonMIC for the E.

gallinarum

and

i.g/ml.

Because it gave the fewest errors with both inocula, E.

casseliflavus

strains

was 4

ug/ml.

Thirty-nine

strains were

BHIagar

containing

6

jig

of

vancomycin

per mlwaschosenfor characterized as resistant,

requiring

MICs of

.32

~..gml,

furtherstudy.

including

23 vanA (19 E.

faecium

and 4

E. faecalis)

strains,

14

The resultsof the

comparison

ofBHI agarsobtained from vanB

(7

E.

faecium

and 7 E.

faecalis)

strains,

1 strain that different sources are shown in Table 4. All media except contained both thevanA and the vanC genes

(F.

gallinarum

for modified BHI agar from BDMS performed well. Overall errors

were mainly in specificity and were more frequent with the

106-CFUinoculum. One strain of E.

casseliflavus

(vancomycin TABLE 4. Sensitivityandspecificityofvancomycin resistance MIC= 2,ug/ml)waspartially responsibleforspecificityerrors detection with multiple lots of BHI agarcontaining with three of the media(Adams, BDMS modified, and Oxoid). 6 ,ugof vancomycin permla

Data from the testing of the 10 reference strains are not I and Sensitivity Specificity shown. The only errors were with the one susceptible strain

medium

18h 24 h 48 h 18h 24h 48h that was reported as resistant on both the common and the medium_ Ix_h_ 24_h_ 48_h_ 18_h_ 24_h_ 48_h unique lots in one laboratory and with one strain of E. 105 CFU

gallinarumreportedassusceptibleontheunique agar lot only. Acumedia 100 100 100 100 100 100 Resistance of one VanB strain (E. faecium) in two laboratories Adams 100 100 100 100 100 96 andoneVanC strain(E.gallinarum)inonelaboratorywasnot BDMS 100 98 100 100 100 100 detected until 48 h on theunique lots tested in thoselabora- BDMS mod 100 100 100 83 79 75

Difco 1 96 98 100 100 100 100

Difco 2 100 100 100 100 100 100

Difco 3 98 100 100 100 100 100

llrlATl TT- I_.!!^_,. z _ !__:^Oxoidr _ n _:100 100 100 96 96 96

TABLE

2. Sensitivity and

speciticity of

vancomycin resistance

detection withBHIagarscreenplates containing

6 jigof vancomycin per ml'

(CFU) (h) BHI1 BHI2 MHA BHI1 BHI2 MHA

105 18 100 97 95 100 100 100

24 100 99 91 100 100 100

48 100 100 90 100 100 100

10"CFU

Acumedia 100 100 100 100 100 100

Adams 100 100 100 83 96 88

BDMS 100 100 100 100 100 100

BDMSmod 100 100 100 71 67 58

Difco1 100 98 100 100 100 100

Difco 2 100 100 100 100 100 100

Difco 3 100 100 100 100 100 100

Oxoid 100 100 100 92 96 92

106 18 100 100 100 100 92 100 "Atotal of 70 enterococci weretested, including 24 strains for which MICs

24 100 100 100 100 92 100 were.2(13 E.faecium strains,IOE.faecalisstrains, and I E. casseliflavusstrain),

48 100 100 100 100 100 100 9strains for which MICswere4to8ptg/mI (E.gallinarum),and 37 strains for

which MICs were.16,ug/ml(24 E.faeciumand 13 E.faecalisstrains). See text

"See Table 1,footnote a, for experimental details and definitions of abbrevi- fordefinitions of resistance. BDMS mod, BDMSmodified BHI agar; Difco 1, 2,

ations. and 3, three different lots of Difco BHI agar.

on May 15, 2020 by guest

http://jcm.asm.org/

(4)

TABLE 5. Performance of screen test with 6

jig

of vancomycinper ml and unique lots of BHI agar reportedby susceptibility andspeciesa

No.(%)ofcorrectresults

Identification Vancomycin MIC 105CFU 106CFU

(p.g/ml)__

24h 48 h 24 h 48 h

E.faecalis,E. faecium,E.raffinosus,or Enterococcus sp. <4 99 98 (99) 97 (98) 95 (96) 95(96)

E.

casseliflavus

2-4 4 1(25) 1(25) 1(25) 1(25)

E.gallinarum 2-8 14 14(100) 14(100) 14(100) 14(100)

E.faecalis,E.faecium,E.gallinarum,orE.raffinosus >l16b 40 40 (100) 40 (100) 40 (100) 40 (100)

aAtotal of 158 clinical isolates of enterococci were tested. For organisms other than E. gallinarum, resistance was categorizedbyusingNCCLS breakpoints(8). All strains of E.gallinarumwereclassifiedasresistant.

6Includes one strain of E.gallinarumthatcontainedboth the vanA and the vanC genes.

which the vancomycin MIC was 512 ,ug/ml and the teicoplanin MIC was 32 ,ug/ml), and 1 strain that was negative for both vanA and vanB(anE.faeciumstrain for which thevancomycin andteicoplanin MICswere >512

p.g/ml).

The results of the testing of the clinical strains with the screening plates are shown in Table 5 as the percentages of strains correctly categorized. The strains are separated into four groupsonthe basis of MICand/or identification. Strains for which the MICs were .4 ,ug/ml that wereE. faecalis, E. faecium, E. raffinosus,orundetermined Enterococcus spp. were

categorized correctly as susceptible 96 to 99% of the time

(specificity). Three susceptible strains from three different laboratories grewonly when the

106-CFU

inoculum was used. Of the foursusceptibleE.faecalis strains(from four different

laboratories)for which the MICwas4 ,ug/mlat24h,onegrew on the screenplate with both inocula andat both times. For thisstrain, thevancomycinMICat48 hwas8,ug/ml; however, it contained neither the vanA nor thevanB gene. The other three E.faecalis strainsfor which thevancomycinMICwas4 ,ug/ml did not grow with either inoculum at either period. Fifteen strains of E. gallinarum and three strains ofE. cas-seliflavustested grewatboth times with bothinocula, including oneE.gallinarum strain for which the MICwas2

p,g/ml

atboth 24 and 48 h. All of the strains for which the MICswere .16 ,ug/ml were classified asresistant (100% sensitivity) under all conditions.

Qualitycontrol. Duringthe testingofmultiplelots ofBHI agarduring phase 1, itwas noted that thesusceptible control

strain,E.faecalisATCC29212,grew100% ofthe timeonthe screen plates preparedwith themodified BHI agar manufac-tured by BDMS. Two of the other manufacturers' agars also

occasionallyallowedgrowthof thesusceptible control. BDMS modifiedBHIagarwas notincluded inphase2studies because of the specificity problems with the susceptible control and some of the test strains. More thorough examination of the

qualitycontrolstudies is documented elsewhere (unpublished data).

Stabilitystudies. The susceptible control strain (E. faecalis

ATCC29212)wasinhibitedby thescreenplates until the fifth month of testing, when growth was detected at 48 h (good growth with the

106-CFU

inoculum and poorgrowth with the 105-CFU inoculum). The resistant control (E. faecalis ATCC

51299) grew wellatall times.

DISCUSSION

A screening test to aid in the detection of vancomycin

resistancewasevaluatedbyWilleyetal. in 1992and foundto be100% sensitiveandspecificfor the detection ofvancomycin resistance in155strains ofE.faecium and1strain ofE.faecalis

(17). Thirty-one of the 103 resistant strains in that study were susceptibletoteicoplanin(VanBphenotype). Willey et al. used MH agar andaninoculum of 1 ,ul of a suspension equal toa0.5 McFarland standard delivered with a Steers replicator (final

inoculum, 105CFU perspot). Inaddition, they usedasterile swabtodeliver the inoculumtothe agar surface andfound the results to be equivalent to those obtained with the replicator delivery. They noted that further evaluation of thescreening

plates including more strains for which MICs were 8 to 16 ,ug/mlwasprobably necessary. Ourmultilaboratoryevaluation wasprompted by their preliminary study and by the fact that despite the increased awareness of vancomycin resistance among enterococci and the new vancomycin disk diffusion breakpoints developedspecificallyfor enterococci(13),clinical laboratories continue to have problemsdetectingvancomycin

resistance (14). This is true mainly for strains for which MICs are in the range of 8 to 64 ,ug/ml (11, 12, 14). Some of the detectionproblems could be attributed to the shortincubation

times thatareusedbysomesystems(3, 14)or tothefailure of laboratories to apply the newdisk diffusionbreakpoints (3).

AlthoughMH agar wasusedbyWilleyetal.,wefound that resultsonBHI agarwereeasiertointerpret. UsingBHI agar also makes the vancomycin screen test consistent with the proposed enterococcalaminoglycoside screen tests, which also useBHIagar(9).

Although sensitivity and specificity were excellent when 4 ,ug/mlanda105-CFU inoculumwereused with bothBHIand MH agars (Table 1), we felt that that concentrationwas too closeto thebreakpoint; indeed,when the 106-CFU inoculum wasused, specificity decreased. Conversely, at8 ,ug/ml, sensi-tivity and specificityweregood with thelarger inoculum, but

sensitivity decreased when the 105-CFU inoculum was used

(Table 3). At a concentration of 6 ,ug/ml, there was little difference in performance between the two inocula and the three media. However, therewas aslight decrease insensitivity

withMH agar and the

105-CFU

inoculum.

Evaluation of lots of BHI agar from all manufacturers

(Table4) showed good performance except with themodified BHI agarmanufacturedby BDMS. The specificity errors for the Adams andOxoidBHIagarswith the

106-CFU

inoculum wereduetoreadings of growththatwasreportedasweakwith

one strain of

E. casseliflavus

used

during

that

phase

or to readingsofgrowth thatwasreportedasweakorquestionable.

When the

E. casseliflavus

strain is excluded from the

analysis,

specificitywas 100% at 24 h for all media with both inocula. The performance of the screen test with the reference and clinical strainswasexcellent except for theclinical strains ofE.

casseliflavus tested. MICs for these strains ranged

from 2 to 4 ,ug/ml, which would categorize the strains as susceptible.

However, the three strains for which the vancomycin broth

on May 15, 2020 by guest

http://jcm.asm.org/

(5)

microdilution MIC was 4 Fig/ml grew on the screen plates,

which may be because of the increased inoculum or better

growth onthe richer medium. All of theE.gallinarum strains grew onthe screenplates, including the one strain for which

the MICwas2 fg/ml.

In conclusion, we recommend that the agar screen test

proposed byWilley etal. (17) be usedtoconfirm detection of vancomycin resistance. We have shown that this screening method, using BHI agar, 6 pLgofvancomycin perml, and an

inoculum of 10- to 106 CFU per spot, is reliable for the detection of enterococcal strains with low-level vancomycin resistance. Preliminary readingscanbe madeat18h, but final readings shouldnotberecorded until24hof incubation. These recommendations were accepted by the NCCLS and are part

of the most recent documents on disk diffusion and dilution

testing (9, 10).

ACKNOWLEDGMENTS

Wethank the following for their excellent work in performing the phase 2 studies: Jean Spargo and William Lemos, Massachusetts General Hospital; Laurie Free, Jewish Hospital; Michael Blocker, University of Massachusetts; Chris Grant, Janet Rhine-Chalberg, and Virginia Morthand, Oregon Health Sciences University; Jackie Thorpe, Duke University; Karen Jones and Judy Rothberg, Robert Woods Johnson UniversityHospital; and Anthony R.DiNuzzo, The University of Texas Medical BranchatGalveston. Wealso thank Nan Pigott in the Streptococcus Laboratory at CDC for help in the identification ofsomeof theclinical strains.

REFERENCES

1. Centers forDisease Control. 1993. Nosocomial enterococci

resis-tant tovancomycin-United States, 1989-1993. Morbid. Mortal. Weekly Rep. 42:597-599.

2. Clark,N.C.,R.C. Cooksey,B. C.Hill,J. M.Swenson,and F.C.

Tenover. 1993. Characterization ofglycopeptide-resistant

entero-cocci from U.S. hospitals. Antimicrob. Agents Chemother. 37: 2311-2317.

3. CollegeofAmerican Pathologists. 1992. 1992CAPsurveys,

spec-imen D-lb. Final critique. College of American Pathologists, Northfield, ILL.

4. Facklam, R. R.,andJ.A.WashingtonII. 1991.Streptococcus and

related catalase-negative Gram-positive cocci,p. 238-257.In A.

Balows, W. J. Hausler, Jr., K. L. Herrmann, H.D.Isenberg,and

H. J. Shadomy (ed.), Manual of clinical microbiology, 5th ed. AmericanSocietyforMicrobiology, Washington, D.C.

5. Leclercq, R., E. Derlot, J. Duval,and P.Courvalin. 1988. Plasmid-mediated resistance to vancomycin andteicoplanin in Enterococ-cusfaecium.N. Engl.J. Med. 319:157-161.

6. Leclercq,R., S.Dutka-Malen, J.Duval, and P. Courvalin. 1992. Vancomycin resistance gene vanC is specific to Enterococcus gallinarum.Antimicrob. Agents Chemother. 36:2005-2008. 7. National Committee for Clinical Laboratory Standards. 1990.

Methods fordilution antimicrobialsusceptibilitytestsfor bacteria that growaerobically.Approved standardM7-A2. National Com-mitteefor Clinical LaboratoryStandards,Villanova, Pa. 8. National Committee for Clinical Laboratory Standards. 1991.

Performance standards for antimicrobial susceptibility testing; third informational supplement. Document M100-S3. National Committeefor ClinicalLaboratoryStandards,Villanova, Pa. 9. National Committee for Clinical Laboratory Standards. 1993.

Methods for dilutionantimicrobialsusceptibilitytestsforbacteria that growaerobically. Approvedstandard M7-A3. National Com-mittee for ClinicalLaboratory Standards, Villanova,Pa. 10. National Committee for Clinical Laboratory Standards. 1993.

Performance standardsforantimicrobial disksusceptibility tests. Approved standard M2-A5. National Committee for Clinical Laboratory Standards, Villanova, Pa.

11. Sahm,D.F., J.Kissinger, M. S. Gilmore, P.R.Murray,R.Mulder, J.Solliday,and B.Clarke. 1989.Invitrosusceptibility studies of vancomycin-resistant Enterococcus faecalis. Antimicrob. Agents Chemother.33:1588-1591.

12. Sahm, D. F., andL.Olsen. 1990.Invitro detection of enterococcal vancomycinresistance. Antimicrob. Agents Chemother. 34:1846-1848.

13. Swenson, J. M., M. J. Ferraro, D. F. Sahm, P. Charache, The NationalCommittee for Clinical LaboratoryStandards Working Group on Enterococci, and F. C. Tenover. 1992. Newvancomycin disk diffusion breakpoints for enterococci. J. Clin. Microbiol. 30:2525-2528.

14. Tenover, F.C., J. Tokars, J. Swenson, S. Paul, K. Spitalny, andW.

Jarvis.1993.Abilityof clinicallaboratoriestodetectantimicrobial agent-resistant enterococci.J. Clin. Microbiol.31:1695-1699. 15. Uttley,A. H., C. H.Collins, J. Naidoo, and R. C. George. 1988.

Vancomycin-resistant enterococci. Lancet i:57-58.(Letter.) 16. Willey,B.M., B. N.Kreiswirth, A.E.Simor, Y. Faur, M. Patel, G.

Williams, and D. E. Low. 1993.Identification andcharacterization ofmultiplespeciesofvancomycin-resistant enterococci, including

an evaluation of Vitek software version 7.1. J. Clin. Microbiol. 31:2777-2779.

17. Willey, B. M., B. N. Kreiswirth, A. E. Simor, G. Williams, S. R. Scriver, A.Phillips,and D. E.Low.1992.Detection ofvancomycin resistance in Enterococcus species. J. Clin. Microbiol. 30:1621-1624.