ANTIMICROBIAL AGENTs AND CHEMOTHERAPY, Jan. 1978, p. 61-69Copyright © 1978 American Society for Microbiology

Vol. 13, No. 1Printed in U.S.A.

Evaluation of the Micro-Media System for QuantitativeAntimicrobial Drug Susceptibility Testing:

a Collaborative StudyA. L. BARRY,1 * R. N. JONES,2 AND T. L. GAVAN3

Clinical Microbiology Laboratories, University ofCalifornia (Davis)-Sacramento Medical Center,Sacramento, California 958171; Department ofPathology, Kaiser Foundation Hospital Laboratories,

Portland, Oregon 972172; and Department ofMicrobiology, The Cleveland Clinic Found:ation,Cleveland, Ohio 441063

Received for publication 5 April 1977

Micro-Media Systems (MMS) has developed a procedure by which microdilu-tion trays can be filled with dilutions of antimicrobial agents, frozen, anddistributed to clinical laboratories. The trays are prepared in various distribu-tion centers throughout the United States to supply clinical laboratories in thevicinity of each center. For use, trays are removed from the freezer, allowed tothaw, and then inoculated with inocula prepared as for any other susceptibilitytest, using a convenient disposable inoculator (ca. 5 ,ul per well). A collaborativestudy was planned to evaluate microdilution trays prepared in three Micro-Media Systems distribution centers. Microdilution minimal inhibitory concen-trations (MICs) were compared to standard tube dilution tests (the internationalcollaborative study group method). With gram-positive cocci, the two techniquesgave essentially equivalent results. With gram-negative bacilli, the microdilu-tion MICs were generally one doubling dilution lower than the standard tubedilution MICs. Similar results were seen with microdilution trays preparedwith a Cooke Dynatech MIC 2000. Inter- and intralaboratory reproducibilitywith the macro- and microdilution techniques were quite satisfactory, i.e., atleast 96% of the end points were within a range of + 1 log2 dilution intervals.

Quantitative broth dilution susceptibilitytests can be performed efficiently with a micro-dilution technique (1). By using the semiauto-mated equipment now available, the wells inmicrodilution trays can be filled and thenstored frozen for at least several weeks (2, 6).Each well in a thawed tray may be inoculatedwith a multiple-inoculum replicator. After ap-propriate incubation, the minimal inhibitoryconcentration (MIC) may be determined foreach antimicrobial agent. Microdilution tech-nology adapts the standard broth dilution testsby decreasing the total volume of antimicrobialagent-containing broth from 1 or 2 ml in stan-dard test tubes to 0.1 ml in wells in a disposableplastic tray. Other investigators have demon-strated that it is possible to obtain essentiallyequivalent results with microdilution and stan-dard macrodilution techniques (6, 7, 8).Many clinical laboratories find it impractical

or impossible to maintain stock solutions ofantimicrobial agents, to fill microdilution trayson a regular basis, and to maintain all of the

necessary controls. Micro-Media Systems(MMS) has developed a number of distributioncenters across the United States, which preparedilutions of antimicrobial agents, fill microdi-lution trays, and distribute the trays, frozen,to clinical laboratories in the near vicinity ofeach center. In a central laboratory, concen-trated stock solutions of each drug are pre-pared, frozen, and distributed to the prepara-tion centers where they are further diluted anddispensed into microdilution trays. Once filled,the trays are frozen at -70°C and then held at-20°C for as long as 60 days. They can bedelivered frozen to contracting laboratorieswhere they are held at -20°C, ready for use.

In the clinical laboratory, test strains aresubcultured to a broth medium and the inoculaare standardized as for any other susceptibilitytest. The trays are then inoculated with aconvenient disposable inoculator capable of de-livering about 5 ,ul to each well. After overnightincubation, the trays are examined with theaid of a viewing box, and the MICs are deter-

61

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

62 BARRY, JONES, AND GAVAN

mined as the lowest concentration of each drugthat completely inhibits visible growth of thetest organism.The procedure is essentially an adaptation of

the standard broth dilution technique to a moreconvenient microdilution tray, by miniaturiza-tion. With the MMS system, the antimicrobialagent is subjected to several freeze-thaw cycles,and the filled trays are stored for as long as 60days. Because deterioration of antimicrobialdrugs may alter the final MIC, especially atthe lower concentrations, quality control organ-isms must be included with each batch of tests.A collaborative study was planned to evalu-

ate samples of MMS antimicrobial microdilu-tion trays prepared in three distribution cen-ters (Campbell, Calif.; Portland, Oreg.; andCleveland, Ohio). The microdilution MICs werecompared to standard broth macrodilution testsas described by Ericsson and Sherris (5) in areport from an international collaborativestudy group (the ICS method). Limited studieswere also undertaken to compare microdilutionMICs obtained with the MMS trays to thoseprepared with a Dynatech MIC 2000 (CookeEngineering Co., Alexandria, Va.).

MATERIALS AND METHODS

Test strains. Sixteen bacterial isolates were eachtested by two methods on 3 separate days in each ofthe three participating laboratories. Tables 1 and 2list the test strains and the MIC range of eachantimicrobial drug tested by both the ICS techniqueand microdilution method. Strains designated withprefix K were obtained from Kaiser FoundationHospital Laboratory (Oregon region), and all otherswere derived from ATCC stock cultures. These mi-croorganisms were selected to maximize the numberof strains with end points that fell within the sevendilution-step range of concentrations provided inthe MMS microdilution trays. Only end points inwells 2 through 6 would be expected to give "on-scale" end points if the technical variability in MICswas ±1 log2 dilution step. For each antimicrobialagent, at least three on-scale end points were ob-tained, and for all but three drugs (clindamycin,methicillin, and trimethoprim/sulfamethoxazole) atleast one strain gave an end point at the lowerconcentrations (well 5 or 6). This is important sincethe lower concentrations of the antimicrobial drugmight be more likely to lose activity during pro-longed storage and reflect minor errors in the prep-aration of serial dilutions. In unsupplementedMueller-Hinton broth, the currently recommendedcontrol strain of Pseudomonas aeruginosa (ATCC27853) gave MICs with gentamicin, tobramycin, andkanamycin which were all "off scale." These resultswere not included in the analysis. A total of 63drug-organism combinations were analyzed, withnine microdilution MICs and nine macrodilution

ANTIMICROB. AGENTS CHEMOTHER.

MICs for each combination, i.e., each of three labo-ratories tested each strain on 3 separate days.

ICS macrodilution tests. The standard tube dilu-tion reference method was the broth dilution tech-nique outlined by Ericcson and Sherris (5). Each ofthe participating laboratories received Mueller-Hin-ton broth (Difco Laboratories, Detroit, Mich., con-trol no. 628274) for the macrodilution tests. Also,all three participants received frozen portions ofconcentrated stock solution of each antimicrobialagent to be tested (most were 1,280 jAg/ml; cephalo-thin and kanamycin were prepared at 2,560 ,ug/ml,and carbenicillin was 20,480 jAg/ml). These stocksolutions were prepared at Micro-Media Laborato-ries (Campbell, Calif.) from standard powders pro-vided by the Food and Drug Administration asreference standards for assay work. Before distri-bution, each stock solution was tested by the bioas-say technique of Bennett et al. (4), and the investi-gators were instructed to dilute the solutions by theassayed value. Reproducibility of end points in thethree laboratories confirmed the absence of signifi-cant deterioration of the stock solutions in transit.All three participants prepared serial dilutions ofthe antimicrobial agents by the same dilution pro-tocol, as outlined in Table 21 (p. 66) of Ericsson andSherris (5). To prepare the inoculum, logarithmic-phase broth cultures were adjusted to give a turbid-ity matching that of a MacFarland 0.5 standardand were then further diluted 1:500 in Mueller-Hinton broth (0.05 ml + 25 ml of Mueller-Hintonbroth). The final inoculum density was about 1 x105 colony-forming units per ml, as confirmed bycolony counts; the inoculum is comparable to thatused with the microdilution tests. The tests wereincubated at 35°C for 16 to 18 h and then read forthe presence or absence of turbidity.

Microdilution susceptibility tests. Samples fromone randomly selected production lot of microdilu-tion trays were provided by the MMS distributioncenter closest to each participating laboratory, andall tests were completed within 3 weeks after man-ufacture. Additional samples from 10 different pro-duction lots of trays were tested 1 to 2 weeks afterproduction and again after 60 days of storage at-20°C or less. One investigator (T.L.G.) includedtests with microdilution trays prepared with aCooke Dynatech MIC 2000 (Alexandria, Va.). Thesetrays were inoculated with a Cooke semiautomatedinoculator using a fixed dilution (0.3 ml in 20 ml ofwater) of brain heart infusion broth culture (5 to 6h, 0.5 ml), following the principle described forstandardizing the inoculum for the agar overlaydisk technique (3). The inoculum for the MMSmicrodilution tests was prepared from the same cellsuspension used to inoculate the macrodilutiontests. A logarithmic-phase culture was adjusted tomatch a MacFarland 0.5 turbidity standard andthen diluted 1:50 in 25 ml of sterile water with0.02% Tween 80. Each tray was then inoculatedwith a separate disposable inoculator that deliversca. 5 ;L to each well. About 1 x 104 viable cellswere delivered to each well, which contained ca. 0.1ml of broth; the final concentration of cells would

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

VOL. 13, 1978

thus be equivalent to that achieved with the ICSmethod.

After 16 to 18 h of incubation at 35°C, the trayswere examined on an MMS viewing box, and theMIC was recorded as the lowest concentration thatcompletely inhibited growth of the test organism.Both micro- and macrodilution end points wereindependently read by two microbiologists, andwhen a discrepancy occurred, a third reader arbi-trated. Discrepancies between readers rarely oc-curred.One investigator (R.N.J.) estimated the actual

volume ofbroth in microdilution trays by aspiratingthe contents of each well, using a 50-Al Unimetricsyringe, which was calibrated against 50-ul dispos-able glass capillary pipettes, to measure wthin ±3%at 100 pul.

RESULTS

Tube versus microdilution. The modes andranges of MICs determined by the two dilutiontests are recorded in Tables 1 and 2. Eachmacrodilution MIC was compared directly tothe matching microdilution MIC, and the dif-ferences were expressed as an MIC ratio (ICStube method/MMS microdilution method). Ifthe MIC values of the two methods were iden-tical, the ratio was 1; if the ICS method gavelarger MIC values, the ratio was 2, 4, 8, etc.;and if the ICS method gave lower MIC values,the ratio was 0.5, 0.25, etc. Data submitted byall three participants are summarized in Table3. In 31 of the tests, one of the MIC values wasbeyond the range of concentrations tested;these data were excluded from the analysissince a valid comparison could not be made. Ifthe off-scale end points were included in thecalculations, the conclusions would not havebeen altered appreciably, i.e., if an MIC s0.25Ag/ml was considered to be 0.25 Ag/ml or if anMIC >16 ,ug/ml was considered to be 32 Ag/ml.With the gram-positive cocci, a total of 210pairs of MIC values could be compared; 96.2%of the strains gave MICs that were the same orthat differed by ± 1 log2 dilution step. Therewas a definite trend for the ICS method to giveMICs about 1 log2 higher than the MMS micro-dilution trays; 96.5% of the strains gave MICratios of 1, 2, or 4, i.e., a ratio of 2 ± 1 dilutioninterval. Somewhat greater disparity was seenwith trays tested in California than with thosestudied in Ohio.A simple experiment was performed to help

answer the question of whether the differencesbetween macro- and microdilution MICs withgram-negative bacilli reflect differences in ac-tual potency of the antimicrobial agents in theMMS trays or whether they represent a funda-mental characteristic of the microdilution tech-

MMS MICRODILUTION MICs 63

nique. Further tests were performed with twostrains of Escherichia coli, which demon-strated marked differences between macro- andmicrodilution MICs with cephalothin and kan-amycin. Serial dilutions of both antibioticswere prepared in Mueller-Hinton broth andthen distributed into 1.0-ml tubes (13 by 100mm) and 0.1 ml went into each of five separatewells in a microdilution tray. In this way bothmicro- and macrodilution tests could be per-formed as before, but this time the antibioticsolutions were exactly the same. The resultingMICs are recorded in Table 4. The macrodilu-tion ICS method consistently gave MIC valuesof about 1 log2 dilution step greater than thatobtained with the microdilution technique, aswas observed with the MMS microdilutiontrays.MMS versus Cooke trays. One of the inves-

tigators (T.L.G.) routinely uses microdilutiontrays prepared with the Cooke Dynatech MIC2000. MICs were determined with the Cooketrays at the same time the MMS trays weretested. The method of standardizing the inocu-lum for the two types of trays differed, but thefinal inoculum density was about the same.Table 5 presents a direct comparison of MICswith the two microdilution techniques and ofthe ICS macrodilution tests with both microdi-lution techniques. The two microdilution tech-niques gave essentially the same result withgram-negative bacilli, and both tended to giveMICs that were about 1 log2 dilution step lowerthan those with the macrodilution ICS tech-nique. With gram-positive bacteria, both mi-crodilution methods gave MICs that were es-sentially the same as those with the macrodi-lution method, and the two microdilution testsgave comparable results.

Reproducibility. Because each microorgan-ism was tested on 3 separate days in each ofthe laboratories, we were able to estimate theinter- and intralaboratory variability of thetwo dilution techniques. The first, second, andthird determination made in each laboratorywere compared directly, and the range of differ-ences between the three laboratories is sum-marized in Table 6. Interlaboratory reproduci-bility ofresults with both techniques was essen-tially comparable; if anything, the microdilu-tion test was a little more reproducible thanthe macrodilution test. The same conclusionswould be drawn whether or not off-scale endpoints were included in the tally.

Table 7 summarizes the variability betweentriplicate values obtained within each of thethree independent laboratories on separatedays. In all three laboratories, at least 95% of

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

c M eqeq,-

c*w Xtoe e9-

7-

1-9-

ioV-

eq

010

~044C) ka*- _;o UX

cqeq CIA13oz , R

e4 _- Lo ko V- LiC4 -

I- 1010- ^ 0

"I VI slw C" I" C 4 VI01e 'I0V1I

coceqV cC'q 1

10q

eqcC4 e

-4 eq~

C4 UZ4Ci

3101e eq

A A

A A

Co -,to

Cs e1eq c 4 tv " t co

eq

j A IC4 C C-4C4 3C4

-4

cq

- V- 4-0t '40010 010

eq*.

1N

o

eq

-_C;Uoq o

1010co Goeq eq- -

eq cqCO CO

IT e- -eq -

V-

10

eq

-- o e_- _- _q_' ; _ o

1-0 C40

_0104010

uz1

eqI4 _

u1

eqR1i01'O

.RR

z1z0 t_

eq

,w q coeq 10coC-00

eq~~~~~~~c

r4 A. 4 4 4

64

2r.

I

co9I

as8

u

_4

..4su

laIs:9

6 !. mI

:3 4)

z

._

.,0

.~

0

1a.",a._C)

4)

CS

0

-

-2

01

0

.o 4

- 4

4)

Ii

0l

Ic

cI

Ic

I

c

4IcI

I0

41

X, w 11 w

co 14 CA ao 4,-II.., -

m v m v m v

-1.-

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

MMS MICRODILUTION MICs 65

F- - F- F

---

Co W

IiI"00 eqI-,1- -1I-

1 ,

Ce q eq C

CM Cq CC

A ACqZs _

1~ ~ ~ ~ _

4eq ~ LO10Lo

~ ~ *

e1 qo0 &o_0

r- 7q r-4 c4 ", "eq eqecq I" I"

w t to0000 ;5 o

10 1

U13 U13eq eq

*0Lo Lo

(6 C;eq

eq eq

*; 6

zz

_1 ,

e- eq

F- E-z z ttzz

CQ C14

C14 t- m cqt

Ce C. CO

eq~~~~~~~~~e

VOL. 13, 1978

I-,cc wCOto -

C- A- -

0

'I

CO

0

C.).

I-60

C.).

Co

L.

Co

Co

Co

Co

Co

.0

Co

C~o.0

eq

2i

r.

co1E0i

0

0

At

C.)

0

0

04

WE.w

_._

C)

2

*3

aSCo

Q

0

oo

0

CoCoCoCo

03o00>0CO0

Cos

CoZ

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

66 BARRY, JONES, AND GAVAN

TABLz 3. Comparison ofmacrodilution (ICS) and microdilution (MMS) MICs, matched pairs oftestsperformed in three separate laboratories

MIC ratio (% of strains)PTesting laboratory

0.25 0.5 1 2 4 8

Gram-negative strainsCleveland 0 1.8 26.8 55.6 15.7 0Portland 0 4.4 21.2 55.7 17.7 0.9Sacramento 0 0 20.8 45.8 30.2 3.1

Total 0 2.2 23.0 52.7 20.8 2.5

Gram-positive strainsCleveland 0 22.2 54.2 23.6 0 0Portland 0 5.6 72.2 18.0 4.2 0Sacramento 0 6.1 35.8 51.5 6.0 1.5

Total 0 11.4 54.3 30.5 3.3 0.5a MIC ratio, ICS/MMS. Data based on 117 MIC determinations with gram-negative bacilli and 72 tests

with gram-positive cocci. A total of 31 off-scale end points were excluded from these data; inclusion of suchdata would not appreciably affect the results.

TABLE 4. Macro- versus microdilution MICs with the same dilutions ofantibiotic distributed into tubes andtrays and inoculated with dilutions of the same standardized cell suspension

w S~~~~~~~~~~~~~~MC (ZAgI)MIC method E. coli K 380 ATCC 25922

Cep" Kanf Cep Kan

Macrodilution 32 8 16 4Microdilutionb 16 (8-16) 6c (2-8) 8 (8-8) 2 (1-2)

a Cep, Cephalothin; Kan, kanamycin.b Values represent the mode of five separate determinations on one tray; range is noted in parentheses.c Mode of 6 = same number of values at 4 and at 8.

TABLE 5. Comparison ofmacrodilution (ICS) and microdilution (MMS and Cooke) MICs, matched pairs oftests performed at the Cleveland Clinical Foundationa

MIC ratio (% of testsPMethods compared (no. oftests) 0.12 0.25 0.5 1 2 4 8

Gram-negative strainsICS/MMS (108) 1.9 26.9 56.5 14.8ICS/Cooke (104) 4.8 8.7 24.0 43.3 18.3 1.0MMS/Cooke (100) 5.0 6.0 16.0 51.0 22.0

Gram-positive strainsICS/MMS (72) 23.6 52.8 23.6ICS/Cooke (58) 5.2 32.8 36.2 22.4 3.4MMS/Cooke (59) 1.7 28.8 49.2 15.3 5.1a Microdilution trays were either provided by MMS (Cleveland) or prepared at the Cleveland Clinic

Foundation with a Cooke MIC 2000 dispenser.b Excluding off-scale values.

the triplicate MIC determinations were repro-ducible within 1 log2 dilution interval, and allbut one MIC value fell within a range of 2dilution steps (mode, + 1 dilution).To further test lot-to-lot reproducibility of

MMS microdilution trays, additional samplesof trays from 10 different production lots weretested in triplicate 1 to 2 weeks after prepara-tion and again after 60 days of storage at - 70°C(3 lots) or at -20°C (2 lots). Five lots of trays

ANTIMICROB. AGENTS CHEMOTHZR.

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

MMS MICRODILUTION MICs 67

for gram-positive organisms and five lots oftrays for gram-negative organisms were usedto test the appropriate reference strains. Thetrays were all collected from two different MMSdistribution centers. We were unable to docu-ment any significant differences in the MICend points with different lots of trays, i.e., allMICs were within + 1 dilution interval from

the expected mode. Furthermore, after 60 daysof storage, the trays performed the same as 1-to 2-week-old trays (±1 dilution). Two lots oftrays were also tested 2 weeks after the expira-tion date; a significant loss of activity was notdemonstrated by MIC end points.

Filling accuracy. One investigator (R.N.J.)estimated the volume of broth in the microdi-

TABLz 6. Interlaboratory variation in broth dilution MIC determinationsaMIC range (as 2x dilutions)'

Determination0 Dilutions 1 Dilution 2 Dilutions 3 Dilutions

Microdilution (MMS)lst test 19 (19)" 37 (35) 5 (3) 2 (0)2nd test 24 (24) 27 (26) 11 (7) 1 (1)3rdtest 17 (17) 31 (27) 12 (8) 3 (1)

Total no. 60 (60) 95 (88) 28 (18) 6 (2)Total% 32 (36) 50 (52) 15 (11) 3 (1)

Macrodilution (ICS)lst test 20 (20) 31 (30) 10 (8) 2 (2)2ndtest 18 (17) 31 (29) 12 (12) 2 (1)3rd test 9 (9) 38 (37) 13 (11) 3 (3)

Total no. 47 (46) 100 (96) 35 (31) 7 (6)Total % 25 (26) 53 (54) 18 (17) 4 (3)

a Range of MIC values obtained from three separate laboratories with each organism tested on 3different days.

b Expressed as the number of strains in each category. Strains with end points off scale were consideredto have MICs at the lowest concentration tested if there was no growth in that well or at 1 dilution abovethe highest concentration tested if growth appeared in all wells.

c Numbers in parentheses represent the total number of strains in each category when all data with oneor more values off scale are excluded.

TABLz 7. Intralaboratory variation in broth dilution MIC determinationsaMIC range (as 2x dilutions)'

Testing laboratory0 Dilutions 1 Dilution 2 Dilutions 3 Dilutions

Microdilution (MMS)Cleveland 42 (41)" 18 (15) 3 (1) 0 (0)Portland 27 (27) 32 (31) 3 (3) 1 (0)Sacramento 27 (27) 33 (25) 3 (3) 0 (0)

Total no. 96 (95) 83 (71) 9 (7) 1 (0)Total % 51 (55) 44 (41) 5 (4) 1 (0)

Macrodilution (ICS)Cleveland 33 (33) 30 (30) 0 (0) 0 (0)Portland 30 (30) 30 (30) 3 (2) 0 (0)Sacramento 31 (29) 29 (29) 2 (2) 1 (1)

Total no. 94 (92) 89 (89) 5 (4) 1 (1)Total % 50 (49) 47 (48) 3 (2) 1 (1)

a Range ofMIC values between triplicate tests in each of three separate laboratories." Expressed as the number of strains in each category. Strains with end points off scale were considered

to have MICs at the lowest concentration tested if there was no growth in that well or at 1 dilution abovethe highest concentration tested if growth appeared in all wells.

c Numbers in parentheses represent the total number of strains in each category when all data with oneor more values off scale are excluded.

VOL. 13, 1978

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

68 BARRY, JONES, AND GAVAN

lution wells after they thawed for 20 to 30 minat room temperature (Table 8). Samples oftrays obtained from three MMS preparationcenters contained 64 to 104 JA of broth in eachwell. The control wells (Hi and 2) containedthe least amount of broth. Cooke trays wereprovided by E. H. Gerlach (St. Francis Hospi-tal, Wichita, Kans.). They were prepared witha Cooke MIC 2000 dispenser and shipped, fro-zen, to Portland, Oreg. The wells in these trayscontained 66 to 123 ul of broth. One tray fromeach lot was then incubated for 15 to 18 h at350C, covered with a blank tray. As much as25% of the volume was lost from the wellsalong the outer edges of the tray, whereas only1 to 3% of the broth medium was lost from thewells in the center of the tray. By storing thetrays at -20°C without sealing tape, 7.2 to9.8% of the volume was lost by evaporation,but, when the trays were covered with sealingtape, the moisture loss was negligible. Thelatter estimate was made by sampling twotrays from each of five different lots; one of thetrays was covered with sealing tape at thetime of manufacture. Both trays were wrappedin plastic bags, frozen at -70°C, and thenstored at -20°C for 9 to 49 days. The amount ofmoisture lost was not related to the length ofstorage.To estimate the effect of altering the volume

of broth over a wide range, special trays wereprepared with 50, 100, 150, and 200 ,ul of brothcontaining the various antimicrobial agents.The actual concentration of drug was notchanged, but the volume was. These trays wereinoculated with the appropriate test strains,and the MICs were determined. In all cases,the MICs varied no more than 1 dilution inter-val from the median, regardless of the volumeof broth in the wells. Finally, sample trays

TABLz 8. Estimated volume of broth inmicrodilution trays prepared in four different

laboratories

Prepn No. of wells Mean vol per well (pL)center sampled M

MMSCalif. 100 93.0 (80-99POreg. 100 90.7 (64-104)Ohio 60 86.0 (74-96)

CookeKans. 80 103.7 (66-123)a From each preparation center, two trays from

each of three to five lots were sampled and 10 wellswere sampled from each tray, following a prescribeddiagonal pattern.

b Numbers in parentheses represent range.

(uncovered) were placed in an incubator at35°C for 4.5 h, at which time about 40% of thevolume was lost by evaporation and the drugconcentration increased almost twofold. Thesetrays gave MICs that were equal to or 1 dilutionstep lower than that obtained with the controltrays that had not been subjected to such exces-sive evaporation.

DISCUSSIONMicrodilution susceptibility tests are per-

formed routinely in an increasing number ofclinical laboratories. If each laboratory under-takes the responsibility for preparing stocksolutions of antimicrobial agents, filling micro-dilution trays, and controlling the storage con-ditions with adequate quality control proce-dures, there should be a reasonable degree ofinter- and intralaboratory reproducibility.Even greater reproducibility should be ex-pected if the antimicrobial dilutions are pre-pared in a central laboratory and dispensedinto microdilution trays in large batches, withmore stringent controls, than those that can beapplied in the average clinical laboratory. Inthis sense, the MMS system offers an importantstep in the direction of standardization; eachcontracting laboratory would theoretically re-ceive trays with drugs of nearly identical po-tency. In the present study, different lots ofmicrodilution trays prepared in three geo-graphically separate distribution centers per-formed essentially the same. In fact, there wasless interlaboratory variation with the MMSmicrodilution trays than with the standard ICSmacrodilution technique; even though all tubedilution tests were prepared from the sameconcentrated stock solutions of antimicrobialagents. Intralaboratory variation in MICs wasquite satisfactory with both dilution tech-niques. Both the MMS microdilution trays andthose prepared with a Cooke Dynatech MIC2000 gave essentially the same end points,confirming the acceptability of the MMS micro-dilution trays.With the gram-positive cocci, the microdilu-

tion MICs were essentially equivalent to thoseobtained with the standard ICS macrodilutiontechnique. However, with the gram-negativebacilli, the microdilution MICs were about 1dilution step lower than those obtained withthe standard tube method. There was no cleartendency for such discrepancies to occur withone particular group of drugs or microorga-nisms: it appears to be a general phenomenonfor most gram-negative bacilli. It would bedifficult to believe that the discrepancies withtrays designed for testing gram-negative organ-

ANTimicRoB. AGENTS CHEMOTHZR.

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from

MMS MICRODILUTION MICs 69

isms resulted from the use of excessively highconcentrations of drugs at all three MMS prep-aration centers because the gram-positivepanels did not show the same phenomenon. Inaddition, the data reported in Tables 5 and 8support the fact that we observed a generalphenomenon common to the microdilution tech-nique, regardless ofhow the trays are prepared(MMS, Cooke MIC 2000, etc.).We can suggest two factors that might con-

tribute to such discrepancies between the twomethods. At concentrations just below the MIC,bacterial growth was often reduced, but notcompletely inhibited. In the test tube, a faint,barely visible turbidity was often seen at oneconcentration of drug, and complete inhibitionoccurred at the next higher log2 dilution. Be-cause of the optics involved in reading endpoints in microdilution trays, the barely visibleturbidity might not be seen, and thus themicrodilution MICs would be 1 log2 dilutionstep lower than the tube dilution MIC. Inaddition, minor differences in the actual inoc-ula densities might contribute some variabilityin end points. The absolute number of viablecells delivered to the microdilution trays wasabout one-tenth of that delivered to the tubes,but the final concentration of cells should becomparable.From a practical point of view, most tech-

niques that involve serial twofold dilutions areconsidered satisfactorily controlled if the re-sults vary no more than ± 1 dilution. Themacro- and microdilution MICs were equiva-lent within the acceptable range of + 1 dilutionwith 85% of the tests (74% of gram-negativestrains and 96% of the gram-positive strains).Both inter- and intralaboratory reproducibilitywith both methods also showed an acceptable

range of variation (+ 1 dilution). We concludethat the MMS microdilution tests were entirelysatisfactory and the MICs were essentiallyequivalent to those obtained with a standardbroth dilution technique.

ACKNOWLEDGMENTWe express our gratitude to L. J. Effinger, R. E. Badal,

R. Packer, M. J. Telenson, S. Diedrich, K. McHale, M.Teplitz, and P. Lieberman for their invaluable assistancein collecting the data summarized in this report.

LITERATURE CITED1. Barry, A. L. 1976. The antimicrobic susceptibility test:

principles and practices, p. 95-99. Lea & Febiger,Philadelphia.

2. Barry, A. L., L. J. Effinger, and R. E. Badal. 1976.Short-term storage of six penicillins and cephalothinin microdilution trays for antimicrobial susceptibilitytests. Antimicrob. Agents Chemother. 10:83-88.

3. Barry, A. L., F. Garcia, and L. D. Thrupp. 1970. Animproved single-disk method for testing antibioticsusceptibility of rapidly growing pathogens. Am. J.Clin. Pathol. 53:149-158.

4. Bennett, J. V., J. L. Brodie, E. J. Benner, and W. M.M. Kirby. 1966. Simplified, accurate method for anti-biotic assay of clinical specimens. Appl. Microbiol.14:170-177.

5. Ericsson, H. M., and J. C. Sherris. 1971. Antibioticsensitivity testing report of an international collabo-rative study. Acta Pathol. Microbiol. Scand. Sect. B.(Suppl.) 217, 90 p.

6. Gavan, T. L., and D. A. Butler. 1974. An automatedmicrodilution method for antimicrobial susceptibilitytesting, p. 88-93. In A. Balows (ed.), Current tech-niques for antibiotic susceptibility testing. Charles CThomas, Springfield, ill.

7. Gavan, T. L., and M. A. Town. 1970. A microdilutionmethod for antibiotic susceptibility testing: an eval-uation. Am. J. Clin. Pathol. 53:880-885.

8. Gerlach, E. H. 1977. Dilution test procedures for sus-ceptibility testing, p. 45-51. In A. Bondi, J. T. Bar-tola, and J. E. Prier (ed.), The clinical laboratory asan aid in chemotherapy of infectious disease. Univer-sity Park Press, Baltimore.

VOL. 13, 1978

on June 23, 2017 by guesthttp://aac.asm

.org/D

ownloaded from