●

alS(●

●

w(0GG(Tri●

etwnPo

A

aEab

PID

rI

IS

7

The In vitro Impact of Moxifloxacin andGatifloxacin Concentration (0.5% vs 0.3%)and the Addition of Benzalkonium Chloride

on Antibacterial Efficacy

REGIS P. KOWALSKI, MS, [M]ASCP, BRITTANY R. KOWALSKI,ERIC G. ROMANOWSKI, MS, FRANCIS S. MAH, MD,

PAUL P. THOMPSON, BMEDSC, AND Y. JEROLD GORDON, MD

tC●

eBipts7

SnwbctfluttaagfopGn

ZGa

PURPOSE: Varied concentrations of moxifloxacin (MOX)nd gatifloxacin (GAT) and the addition of 0.005% benza-konium chloride (BAK) were evaluated for eliminatingtaphylococcus aureus (SA), Pseudomonas aeruginosaPA), and coagulase-negative Staphylococcus (CNS).

DESIGN: In vitro laboratory investigation.METHODS: The time-kill survival of SA, PA, and CNSere tested at one, two, three, six, eight, and 24 hours to:

1) Mueller-Hinton broth, (2) BAK, (3) 0.5% MOX, (4).5% GAT, (5) 0.3% MOX, (6) 0.3% GAT, (7) 0.3%AT plus BAK, (8) 0.5% MOX plus BAK, (9) 8 �g/mlAT, and (10) 8 �g/ml MOX. Antibiotic interactions

GAT and BAK) were determined by checkerboard testing.he outcome measures were (1) time-to-kill, (2) killing-

ates, and (3) fractional inhibitory concentration (FIC)ndices.

RESULTS: MOX and GAT at either 0.5% or 0.3% hadquivalent antibacterial effects. BAK alone or the addi-ion of BAK to either antibiotic eliminated SA and CNSithin one hour, whereas 0.3% GAT plus BAK elimi-ated bacteria faster than 0.5% MOX (P � .016). ForA, BAK alone had no antibacterial effect. The kill ratesf MOX and GAT were equivalent. FIC indices indicated

ccepted for publication Jun 1, 2006.From The Charles T. Campbell Ophthalmic Microbiology Laboratory

t the University of Pittsburgh Medical Center (R.P.K.); and the UPMCye Center, Ophthalmology and Visual Science Research Center, Eyend Ear Institute, Department of Ophthalmology, University of Pitts-urgh School of Medicine, Pittsburgh, Pennsylvania.Supported by the Eye and Ear Foundation of Pittsburgh, Pittsburgh,

ennsylvania, has provided salary support. Research to Prevent Blindness,nc, New York, New York has provided financial support to theepartment of Ophthalmology.The authors are paid consultant fees and have performed contract

esearch for Allergan, Inc (Irvine, California) and Alcon Laboratories,nc (Ft Worth, Texas).

Inquiries to Regis P. Kowalski, MS, [M]ASCP, The Eye and Ear

wnstitute Bldg., Ophthalmic Microbiology, Room 642, 203 Lothroptreet, Pittsburgh, PA 15213; e-mail: kowalskirp@upmc.edu

© 2006 BY ELSEVIER INC. A30

hat GAT and BAK were indifferent against SA andNS, but antagonistic to PA.CONCLUSION: As a preservative, MOX and GAT have

quivalent antibacterial activity with similar killing rates.AK appears to independently complement GAT for elim-

nating SA and CNS, but has no effect on PA. The in vitroredictive clinical effect due to varied antibiotic concentra-ion and the addition of BAK requires confirmatory clinicaltudies for validation. (Am J Ophthalmol 2006;142:30–735. © 2006 by Elsevier Inc. All rights reserved.)

INCE THE INTRODUCTION OF COMPETING PRODUCTS,

Vigamox® (moxifloxacin [MOX] 0.5%) and Zymar®

(gatifloxacin [GAT] 0.3% plus 0.005% benzalko-ium chloride [BAK]), much controversy has arisen as tohich product is more effective in the topical treatment ofacterial ocular infections. Although no independent,omparative clinical data have been reported to evaluatehe respective efficacy of these two fourth-generationuoroquinolones, in vitro microbiologic data have beensed to separate the potential benefit of one antibiotic overhe other. On the in vitro testing level, minimum inhibi-ory concentration (MIC) data have determined that GATnd MOX were more potent than the earlier fluoroquinolonentibiotics (ciprofloxacin, ofloxacin, and levofloxacin) forram-positive bacteria but demonstrated no real advantageor gram-negative bacteria.1–3 In a head-to-head comparisonf MOX with GAT, MIC data indicated that MOX was moreotent than GAT against gram-positive bacteria, whereasAT was generally more potent than MOX against gram-egative bacteria.1–3

Two key areas of difference between Vigamox® andymar® are antibiotic concentration (0.5% MOX vs 0.3%AT), and the addition of BAK (a biocide) to Zymar®. It

ppears logical that a higher concentration of antibiotic

ould be more beneficial for self-preservation and provide

LL RIGHTS RESERVED. 0002-9394/06/$32.00doi:10.1016/j.ajo.2006.06.006

atbscrttb

aoicqfaM(b

●

d(tf0ciC(a8mdw

e(t0((0tcc

VWMac

cw0

po5oCmrctcbowp

(bcctsaa

�rU(at

●

c(awtanaodecow

●

sia

V

more efficient clinical effect. It also seems possible thathe addition of BAK may not only act as a preservative,ut also may contribute to a clinical effect. Time-killtudies can determine the in vitro advantage of antibioticoncentration and the addition of BAK to topical prepa-ations. In addition, in vitro synergy testing can determinehe interaction of BAK and GAT. The results of in vitroesting still require clinical studies to establish any clinicalenefit in the treatment of ocular infections.In this in vitro laboratory study, we tested the key

ntibiotic active components (MOX, GAT, and BAK)f Vigamox® and Zymar® for antibiotic efficacy, and thenteractive effects between GAT and benzalkoniumhloride. The goals of this study were to answer threeuestions: (1) Does increasing an antibiotic concentrationrom 0.3% to 0.5% for MOX and GAT provide anyntibacterial advantage? (2) Does the addition of BAK toOX or GAT provide any antibacterial advantage? and

3) Is there any interactive effects between GAT andenzalkonium chloride?

METHODS

TIME-KILL STUDIES: Standard time-kill studies wereetermined with five isolates each of Staphylococcus aureusSA), Pseudomonas aeruginosa (PA), and coagulase-nega-ive Staphylococcus (CNS). The bacterial isolates wererozen stocks (�80°C) (University of Pittsburgh, IRB No.506122) that were initially isolated from culture-provenases of endophthalmitis and keratitis. The minimumnhibitory concentrations (�g/ml, MICs) of SA, PA, andNS were predetermined to MOX and GAT using E-tests

AB Biodisk, Piscataway, New Jersey, USA) to assure thatll isolates were susceptible to antibiotics concentrations of.00 �g/ml. The MICs of these same isolates were deter-ined to benzalkonium chloride using the standard broth

ilution method to determine whether 0.005% (50 �g/ml)as sufficient to inhibit the isolates.4Bacteria survivals were tested at one, two, four, six,

ight, and 24 hours to: (1) Mueller-Hinton broth onlyvehicle-growth medium), (2) BAK (0.005% concentra-ion of benzalkonium chloride contained in Zymar®), (3).5% MOX (concentration of moxifloxacin in Vigamox®),4) 0.5% GAT, (5) 0.3% MOX, (6) 0.3% GAT plus BAKconcentrations of gatifloxacin and BAK in Zymar®), (7).5% MOX plus BAK, (8) 8 �g/ml GAT (lowest concen-ration denoting resistance by CLSI standards)5 (thisoncentration was used in a previous study for killing rateomparisons),6 and (9) 8 �g/ml MOX.

It is important to note that commercial products ofigamox® and Zymar® were not used in this study.orking concentrations of MOX (moxifloxacin HCl-5794, LKT Laboratories, Inc, St Paul, Minnesota, USA)

nd GAT (gatifloxacin, G0278, LKT Laboratories) were

onfirmed by a broth dilution bioassay.4,7 Benzalkonium b

IMPACT OF ANTIBIOTIC CONOL. 142, NO. 5

hloride (B6295, Sigma, St Louis, Missouri, USA) waseighed, dissolved in sterile distilled water, and diluted to.005% (50 �g/ml) in each time-kill test solution.The initial bacterial inoculum was determined by sus-

ending a few colonies of bacteria that were grownvernight on a trypticase soy agar plate supplemented with% sheep’s blood (BBL, Sparks, Maryland, USA), in 10 mlf Mueller-Hinton broth (Hardy Diagnostics, Santa Maria,alifornia, USA). The turbidity of the suspension waseasured spectrophotometrically at 650 nm. The turbidity

eading corresponded to a predetermined viable bacteriaount and an inoculum was determined from this readinghat resulted in an approximate colony count of 1 � 106

fu/ml in each time-kill testing tube (10 ml). The finalacterial concentration was confirmed with standard col-ny counts using trypticase soy agar plates supplementedith 5% sheep’s blood. Bacterial survival at each timeoint was also determined using standard colony counts.The main outcome measures were (1) time to kill and

2) killing rates. The time (hours) to kill 100% of eachacteria in all time-kill test tubes was determined for theombinations of MOX, GAT, and BAK from the colonyounts. The data were analyzed nonparametrically withhe Kruskal-Wallis test with Duncan’s Multiple Compari-ons (True Epistat, Richardson, Texas, USA) to determineny differences between antibiotic concentration and theddition of BAK.

The killing rate of each bacterial isolate to MOX (8g/ml) and GAT (8 �g/ml) were determined with a

egression analysis (Minitab, State College, Pennsylvania,SA) by plotting the log10 (colony counts) vs time

hours). The killing rates of each bacterial group to MOXnd GAT were statistically compared using a two-sample-test (Minitab, State College, Pennsylvania, USA).

TESTING FOR ANTIBIOTIC INTERACTION: Whenombining antibiotics, four interactions are possible:1) synergy, (2) additive effect, (3) antagonistic effect,nd (4) indifference.8 Synergism is a positive effect inhich the combination of two antibiotics are more active

han when either antibiotic is used independently. Andditive effect is also a positive effect in which the combi-ation of two antibiotics is more active, but not morective than any one antibiotic used independently. Antag-nism is a negative effect in which the combination of bothrugs is effectively less than the separate independentffects of both antibiotics. An indifferent interaction indi-ates that both antibiotics react independently from eachther.8 These antibiotic interactions can be determinedith (1) time-kill studies and (2) checkerboard testing.

TIME-KILL ANTIBIOTIC INTERACTION: Time-killtudies can determine antibiotic interaction. A synergisticnteraction between two antibiotics is indicated with andditional 2-log decrease in colony counts with the anti-

iotic combination compared with the most effective

CENTRATION AND BAK 731

sacweaa

oBPt

●

ttctA

(utGnrcTfTcial(msG

Fgll

7

ingle antibiotic. An antagonistic interaction between twontibiotics is indicated with an additional 2-log increase inolony counts with the antibiotic combination comparedith the most effective single antibiotic. An indifferentffect between two antibiotics is indicated with an equiv-lent colony count between the antibiotic combinationnd the most effective single antibiotic.

As described previously for the time-kill studies, antibi-tic interactions were determined for 0.3% GAT andAK, and 0.5% MOX and BAK against all isolates of SA,A, and CNS. Time-kill studies were not performed forhe MICs of GAT, MOX, and benzalkonium chloride.

CHECKERBOARD PROCEDURE: Antibacterial interac-ions between GAT and benzalkonium chloride wereested against all isolates of SA, PA, and CNS using theheckerboard technique.8 The technique was similar tohat for determining MICs by the broth dilution method.

IGURE 1. The in vitro laboratory results of the time-killatifloxacin (GAT) to Staphylococcus aureus, Pseudomonas aereft) Time-to-kill (median hours) bacteria by 0.5% MOX, 0.3%eft, right) Kill-rates of bacteria to 8 �g/ml of MOX or GAT a

ll testing was performed in 96 well round bottom plates G

AMERICAN JOURNAL OF32

CoStar-3799, Corning Inc, Corning, New York, USA)sing Mueller-Hinton broth and a final bacterial concen-ration of 5 � 105 cfu/ml. The isolates were tested withAT at a range from 16 to 0.0019 �g/ml and benzalko-ium chloride at a range from 512 to 0.06 �g/ml. The firstow of the 96-well plate did not contain any benzalkoniumhloride, and the first column did not contain any GAT.his allowed for the MIC to be determined on each plate

or each isolate against GAT and benzalkonium chloride.he remaining wells contained varying concentrationombinations of GAT and benzalkonium chloride. Thenoculated plates were incubated for 24 hours at 35°C inn air incubator. The plates were examined under indirectight for bacterial overgrowth (pellets). The MIC of GATMICgat) and benzalkonium chloride (MICbak) were deter-ined from the lowest concentration that did not demon-

trate bacterial growth. An inhibitory concentration ofAT(G) was determined by the lowest concentration of

ies by varying concentrations of moxifloxacin (MOX) andosa (PA), and coagulase-negative Staphylococcus (CNS). (TopT plus 0.005% BAK, and 0.005% BAK. (Top right, Bottom

termined by regression analysis.

studugin

GAs de

AT that inhibited growth when combined with the

OPHTHALMOLOGY NOVEMBER 2006

diwcdctesb1“e“

nc

T

wMwMwMTcB(C

h0d(0Gtsaheai(0bh

hhT0

tBb(((0

atTMaw

gMiM0

bwfaBibt

F(Pwcadc2i

V

ifferent concentrations of benzalkonium chloride. Annhibitory concentration of benzalkonium chloride (B)as determined by lowest concentration of benzalkoniumhloride that inhibited growth when combined with theifferent concentrations of GAT. The fractional inhibitoryoncentration (FIC) index was determined by substitutinghe four parameters (MICgat, MICbak, G, and B) in thequation: FIC index � G/MICgat � B/MICbak. In thistudy, a FIC index of 0.5 or less would suggest “synergy”etween GAT and benzalkonium chloride (for example,/4 plus 1/4 equals 0.5); �0.5 to 1.0 would indicate anadditive” effect; �1.0 to 2.0 would indicate “indiffer-nce”; and an index greater than 2.0 would indicateantagonism.”

Antibiotic interaction between MOX and benzalko-ium chloride was not tested because the two are notombined in the commercial product (Vigamox®).

RESULTS

HE MIC RANGES (�G/ML) OF SA (n � 5) TO MOX AND GAT

ere 0.032 to 0.047, and 0.064 to 0.064, respectively. TheIC ranges (�g/ml) of CNS (n � 5) to MOX and GATere 0.032 to 0.094, and 0.047 to 0.125, respectively. TheIC ranges (�g/ml) of PA (n � 5) to MOX and GATere 0.19 to 0.5, and 0.125 to 0.25, respectively. All of theICs for these bacterial groups were well under 8 �g/ml.he MIC ranges of SA, CNS, and PA to benzalkoniumhloride were 1 to 8, 0.5 to 2, and 64 to 256, respectively.ased on the concentration of BAK in Zymar®, 50 �g/ml0.005%) would only be effective in eliminating SA andNS, but not PA.Figure 1 (Top left) summarizes the time-to-kill (median

ours to total kill) of different bacteria to 0.5% MOX,.3% GAT plus BAK and BAK alone. The observedifferences are statistically-significant. For SA, 0.3% MOXtwo hours) statistically eliminated bacteria faster than.5% MOX (four hours), 0.3% GAT (24 hours), and 0.5%AT (eight hours). The fact that 0.3% was more effective

han 0.5% can be explained as a paradoxical effect thatometimes appears when high concentrations of antibioticre tested.9 The addition of BAK to either 0.3% GAT (oneour) or 0.5% MOX (one hour) was statistically moreffective than the 0.3% or 0.5% concentrations of eitherntibiotic alone. BAK alone (one hour) was equally effectiven reducing bacteria in comparison to 0.3% GAT plus BAKZymar®) or 0.5% MOX plus BAK. BAK (one hour) and.3% GAT plus BAK (Zymar®) (one hour) eliminatedacteria statistically faster than 0.5% MOX (Vigamox®) (fourours).For CNS, 0.3% MOX (two hours), 0.5% MOX (three

ours), 0.3% GAT (eight hours), and 0.5% GAT (sixours) demonstrated an equivalent reduction of bacteria.he addition of BAK to either 0.3% GAT (one hour) or

.5% MOX (one hour) was statistically more effective a

IMPACT OF ANTIBIOTIC CONOL. 142, NO. 5

han the 0.3% or 0.5% concentrations of either antibiotic.AK alone (one hour) was equally effective in reducingacteria in comparison to 0.3% GAT plus BAK (Zymar®)one hour) or 0.5% MOX plus BAK (one hour). BAKmedian one hour) and 0.3% GAT plus BAK (Zymar®)one hour) eliminated bacteria statistically faster than.5% MOX (four hours) (Figure 1, Top left).For PA, 0.3% GAT (one hour), 0.5% GAT (one hour),

nd 0.5% MOX (one hour) were statistically more effec-ive in eliminating bacteria than 0.3% MOX (two hours).he addition of BAK to 0.3% GAT (one hour) or 0.5%OX (one hour) did not demonstrate any additional

ntibacterial effect. BAK alone (more than eight hours)as not effective in eliminating PA (Figure 1, Top left).Figure 1 (Top right, Bottom left and right) represents

raphically the regression analysis of the time-kills forOX and GAT at concentrations (8 �g/ml) to five

solates each of SA, PA, and CNS. The killing rates forOX and GAT were statistically equivalent for SA (P �

.7), PA (P � 0.96), and CNS (P � 0.75).The time-kill studies indicated that the interactions

etween 0.3% GAT and BAK, and 0.5% MOX and BAKere indifferent. The FIC indices were calculated and

ound to range from 1.5 to 2.0 for SA, 2.0 to 2.0 for CNS,nd 3.0 to 9.0 for PA. The interactions between GAT andAK for SA and CNS were deemed “indifferent”; indicat-

ng no interaction between the two drugs. The interactionsetween GAT and BAK for PA were deemed “antagonis-ic.” The addition of BAK to GAT increased the GAT

IGURE 2. Antagonistic interaction of benzalkonium chlorideBAK) (�g/ml) with gatifloxacin (GAT) (�g/ml) againstseudomonas aeruginosa (PA). Note increase of GAT MICith addition of BAK to GAT. “G” is the lowest inhibitoryoncentration of GAT with different concentrations of BAKnd “B” is the lowest inhibitory concentration of BAK withifferent concentrations of GAT. The fraction inhibitory con-entration (FIC) index � G/MICgat � B/MICbak � 1/0.25 �56/256 � 5. FIC greater than two denotes an antagonisticnteraction.

ntibiotic concentration required to inhibit the growth of

CENTRATION AND BAK 733

Ptc

T

acpoab

dtdaaio(tMmsp

0oGapbhspcgOcatats

Pwakdde

tg

bdmibdcboibetwaBfatswcaiai

ifisrhBBa2vAAspoadscBBBfdt

7

A. Figure 2 depicts the actual results of checkerboardesting for PA. No additive effects were noted in theheckerboard testing.

DISCUSSION

HE GOALS OF THE CURRENT IN VITRO STUDY WERE TO

nswer three questions: (1) Does increasing an antibioticoncentration from 0.3% to 0.5% for MOX and GATrovide any antibacterial advantage? (2) Does the additionf BAK to MOX or GAT provide any antibacterialdvantage? and (Q3) Is there any antibacterial interactionetween GAT and benzalkonium chloride?The answers to all three questions turned out to be

ependent on the bacterial species tested. For question 1,he increase of antibiotic concentration from 0.3% to 0.5%id not enhance the in vitro antibacterial activity of MOXnd GAT against SA and CNS. An increase in the in vitroctivity of MOX was noted for PA but not GAT. Thencrease was necessary for MOX to be as effective as 0.3%r 0.5% GAT. With regard to a possible clinical advantagefrom other studies), an increase of antibiotic concentra-ion in the ocular tissues was noted previously for 0.5%

OX in comparison to 0.3% GAT,7,11 but these observationsay also be explained, in part, by differences in chemical

tructure that may also promote tissue penetration (for exam-le, charge, size, steric hindrance, etc.).

For question 2, the addition of BAK to 0.3% GAT or.5% MOX, and BAK alone all decreased the killing timef SA and CNS by three hours when compared with 0.3%AT and 0.5% MOX alone. BAK demonstrated no

ntibacterial effect against PA. For SA and CNS, a realreservative effect would only be realized if commercialottles were contaminated between uses (less than threeours) and contamination was at a level that threateneduperinfection. For PA, GAT and MOX act as self-reservatives in the commercial preparations. Our resultsonfirm a previous report that 0.005% (50 �g/ml) BAK isenerally not sufficient to inhibit the growth of PA.10

verall, the preservative effect of increasing antibioticoncentration and the addition of BAK may be minimal,s fortified antibiotics (nonpreserved) and nonpreservedopical drugs have been utilized for decades without posingmajor infectious threat. Finally, a proven clinical advan-

age of adding BAK to GAT still remains to be demon-trated in a clinical setting.

The killing rates of MOX and GAT against SA, CNS, andA, respectively, were identical at 8 �g/ml. In comparisonith the earlier fluoroquinolones (ciprofloxacin, levofloxacin,nd ofloxacin), GAT and MOX (0.7) demonstrated fasterilling rates for SA (0.38)6 and CNS (0.45) (unpublishedata), and equivalent kill rates for PA (3.02 vs 3.12).6 Theseata suggest that there is no predicted clinical advantage for

ither MOX or GAT, although both of these fourth-genera- M

AMERICAN JOURNAL OF34

ion fluoroquinolones may be more effective than earlierenerations of fluoroquinolones.

The answer to Q3 is that there was no synergismetween GAT and BAK demonstrated in vitro, but someifferences were observed based on the bacterium andethod tested. For SA and CNS, both GAT and BAK,

ndependently, were highly effective in eliminating bothacteria, although it was not possible to obtain a 2-logecrease in the time-kill studies for GAT plus BAKompared with either GAT or BAK alone. The checker-oard testing for SA and CNS indicated indifference basedn the FIC indices (1.5 to 2). For PA, the time-kill studiesndicated that GAT was highly effective, but BAK had noactericidal effect at 0.005% (50 �g/ml). An antagonisticffect was not demonstrated between GAT and BAK inhe time-kill studies because no 2-log increase in growthas observed for GAT plus BAK compared with GATlone. In contrast, the checkerboard studies indicated thatAK was antagonistic to GAT for PA (FIC indices � 3.0

or all the isolates). Figure 2 demonstrates this antagonisms an increase in the GAT MICs after the addition of BAKo GAT for PA. Clinical inferences from this study woulduggest that the antibacterial effects of GAT and BAKere independent and that there would be no directlinical benefit on the bacteria by combining the antibioticnd preservative. Another clinical inference, based on then vitro data, would be that BAK may interfere with thentibacterial activity of GAT in PA infections. Thesenferences will need to be confirmed with in vivo studies.

The issue of drug synergism in combination therapy is anmportant one. Because synergism has been precisely de-ned in specific in vitro assays (for example, time-killtudies, checkerboards [see Methods]), this term should beeserved for those situations in which the necessary criteriaave been met. Preliminary reports (that is, abstracts) bylondeau and associates have reported that combiningAK with GAT decreases the GAT MIC (Blondeau andssociates, ARVO abstract, Invest Ophthalmol Vis Sci005:4880; Blondeau and associates, ARVO abstract, In-est Ophthalmol Vis Sci 2006:1903; Borsos and associates,RVO abstract, Invest Ophthalmol Vis Sci 2006:1892).lthough these data might suggest proof of synergism to

ome, appropriate caution is warranted until peer-reviewedublication of the final data occurs. A closer examinationf the available data from the abstracts indicates that theccepted definition of synergism has not been met and theata sets are incomplete. In these abstracts, there are notandard time-kill studies or checkerboard testing foromparison. The preliminary studies completely lack theAK MIC control and, in general, the concentration ofAK appeared to be 10 times the GAT concentration.londeau and associates data indicate, based on the results

rom the present in vitro study, that reduced GAT MIC isue solely to the concentration of BAK, and the antibac-erial effect of GAT is unknown. Any reduction in GAT

IC by benzalkonium chloride must be supported with a

OPHTHALMOLOGY NOVEMBER 2006

cn

aevsWeBtrgIe

cdbpvmaPttiaib

1

1

V

omparison to the MICs of GAT and BAK separately, andot with GAT alone.A brief comment should be interjected on the value of

ny in vitro data (including our own) to predict clinicalffect. The predictive value of in vitro data can only bealidated in hindsight after the completion of well de-igned clinical studies that either confirm or refute it.

hen confronted with published in vitro data, caution inxtrapolating the results to the clinic is most appropriate.ecause the results of most in vitro studies will never be

ested in clinical trials because of cost, priorities, and othereasons, the validation of the predictive clinical value of aiven in vitro data set may never be satisfactorily resolved.n that situation, the clinician is left with his or her ownmpirical experience as a guide.

In summary, although differences in the clinical out-ome based on treatment by Vigamox® or Zymar® are notistinguished, many inferences based on in vitro data areeing used by the pharmaceutical industry to marketroduct. In this in vitro study, the preservative effect ofarying antibiotic concentration of GAT and MOX wasinimal and the addition of BAK to GAT had a small

dvantage with SA and CNS, but no advantage againstA. MOX and GAT had similar killing rates against theest bacteria, and both appeared to be more effective thanhe earlier fluoroquinolones. There was no synergisticnteraction between GAT and BAK against SA and CNS,nd testing indicated antagonism against PA. Any clinicalnferences based on the in vitro data of this study need toe proven with clinical studies.

REFERENCES1. Mather R, Karenchak LM, Romanowski EG, Kowalski RP.

Fourth-generation fluoroquinolones: new weapons in thearsenal of ophthalmic antibiotics. Am J Ophthalmol 2002;133:463–466.

2. Kowalski RP, Dhaliwal DK, Karenchak LM, Romanowski

EG, et al. Gatifloxacin and moxifloxacin: an in vitro

IMPACT OF ANTIBIOTIC CONOL. 142, NO. 5

susceptibility comparison to levofloxacin, ciprofloxacin,and ofloxacin using bacterial keratitis isolates. Am JOphthalmol 2003;136:500 –505.

3. Kowalski RP, Yates KA, Romanowski EG, et al. An ophthal-mologist’s guide to understanding antibiotic susceptibilityand minimum inhibitory concentration (MIC) data. Oph-thalmology 2005;112:1987–1991.

4. National Committee for Clinical Laboratory Standards.Methods for dilution antimicrobials susceptibility tests forbacteria that grow aerobically, ed. 4. Approved standard.Villanova, PA: National Committee for Clinical LaboratoryStandards, 2003, document M7-A6, vol. 20, No. 2.

5. Clinical and Laboratory Standards Institute. Performancestandards for antimicrobial susceptibility testing; fifteenthinformational supplement. Wayne, PA: Clinical and Labo-ratory Standards Institute, 2005, document M100-S15, vol.24, No. 1.

6. Kowalski RP, Pandya AN, Karenchak LM, et al. An in vitroresistance study of levofloxacin, ciprofloxacin, and ofloxacinusing keratitis isolates of Staphylococcus aureus and Pseudo-monas aeruginosa. Ophthalmology 2001;108:1826–1829.

7. Kim DH, Stark WJ, O’Brien TP, Dick JD. Aqueous penetra-tion and biological activity of moxifloxacin 0.5% ophthalmicsolution and gatifloxacin 0.3% solution in cataract surgerypatients. Ophthalmology 2005;112:1992–1996.

8. Pillai SK, Moellering RC, Eliopoulos GM. Antimicrobial com-binations. In: Lorian V, editor. Antibiotics in laboratory med-icine. 5th ed. Philadelphia, Pennsylvania: Lippincott Williams& Wilkins, 2005:365–440.

9. Holm SE, Tornqvist IO, Cars O. Paradoxical effects ofantibiotics. Scand J Infect Dis Suppl 1990;74:113–117.

0. Lambert RJ. Comparative analysis of antibiotic and anti-microbial biocide susceptibility data in clinical isolates ofmethicillin-sensitive Staphylococcus aureus, methicillin-resistant Staphylococcus aureus and Pseudomonas aeruginosabetween 1989 and 2000. J Appl Microbiol 2004;97:699 –711.

1. Solomon R, Donnenfeld ED, Perry HD, et al. Penetration oftopically applied gatifloxacin 0.3%, moxifloxacin 0.5%, andciprofloxacin 0.3% into the aqueous humor. Ophthalmology

2005;112:466–469.

CENTRATION AND BAK 735

RPai

7

Biosketch

egis P. Kowalski, MS, [M]ASCP, is an Assistant Professor of Ophthalmology at the University of Pittsburgh, Pittsburgh,ennsylvania, an Associate Clinical Medical Director at the Charles T. Campbell Ophthalmic Microbiolog Laboratorynd a Clinical Ophthalmic Microbiologist at the University of Pittsburgh Medical Center. Dr Kowalski’s research interestncludes clinical microbiology testing, antibiotic discovery, and testing.

AMERICAN JOURNAL OF OPHTHALMOLOGY35.e1 NOVEMBER 2006