1101991;163 (June) Concise Communications 1369
results of the present study indicate that this mechanism of entrycouldalsooperatein thepathogenesis ofHib meningitis.
References
1. Moxon ER.Haemophilus injluenwe.In:Mandell GL, Douglas RG, Ben-nett JE, eds. Principlesand practice of infectious diseases. 3rd ed. New York: Churchill Livingstone, 1990:1722-9.
2. MoxonER, Smith AL, Averill DR, Smith DH.Haemophilus influen-zaemeningitis in infantrats after intranasalinoculation. J InfectDis 1974;129:154-62.
3. Moxon ER, Winkelstein JA. Interaction ofHaemophilus injluenwewith complement. In: Cabello FC, Pruzzo C, eds. Bacteria, complement and the phagocytic cell. NATO ASI series, Vo124. Berlin: Springer-Verlag, 1988:177-86.
4. NoelGJ, MosserOM, EdelsonPI Roleof complement in murinemac-rophage binding ofHaemophilus influenzaetype b. J Clin Invest 1990;85:208-18.
5. WellerPF, SmithAL, SmithDH, AndersonP. Roleof immunity in the clearanceof bacteremiadue toHaemophilus injluenzae.J Infect Dis 1978;138:427-36.
6. MoxonER, GoldthornJF, SchwartzAD.Haemophilus influenzaeb in-fection in rats: effectof splenectomy on bloodstream and meningeal invasion after intravenous and intranasalinoculations. InfectImmun 1980;27:872-5.
7. Fothergill LD, Chandler CA, Dingle JH. The survival of virulentH. influenwein phagocytes. J Immunol 1937;32:335-9.
8. Williams AE. Relationship between intracellular survivalinmacrophages and pathogenicity ofStreptococcus suistype2 isolates. MicrobPathog 1990;8:189-96.
9. ZwahlenA, RubinLG, MoxonER. Contributionoflipopolysaccharide to pathogenicity ofHaemophilus influenzae:comparative virulence of genetically-related strains in rats. MicrobPathog 1986;1:465-73. 10. KrollJS, MoxonER. Capsulationand gene copy numberat thecap lo-cus ofHaemophilus influenwetype b. J BacterioI1988;170:859-64. 11. WeiserJ, Williams AE, MoxonER. Phase-variable lipopolysaccharide structuresenhancethe invasive capacityofHaemophilus irfiuenzae. Infect Immun 1990;58:3455-7.
12. Virji M, WeiserIN, Moxon ER. Antigenic similarities in lipopolysac-charides ofHaemophilusandNeisseriaand expression of digalac-tose structurealsopresenton humancells. MicrobPathog(in press). 13. KimuraA, HansenEJ.Antigenic andphenotypic variations of Haemophi-Iusinfluenwetype b lipopolysaccharide and their relationship to vir-ulence. Infect Immun 1986;51:69-79.
14. Williams AE, Blakemore WE Monocyte-mediated entry of pathogens into the centralnervoussystem. Neuropathol Appl Neurobiol199O; 16:377-92.
15. Williams AE, Blakemore WE Pathogenesis ofmeningitis causedby Strep-tococcus suistype 2. J Infect Dis 1990;162:474-81.
Resistance of
Staphylococcus aureus Recovered from Infected Foreign Body
In Vivo to Killing by Antimicrobials
Christian Chuard, Jean-Christophe Lucet,*Peter Rohner, Mathias Herrmann,
Raymond Auckenthaler, Francis A. Waldvogel, and DanielP. Lew
Division oflnfectiousDiseases, University Hospital, Geneva, Switzerland
Becausepersistence of infectionsassociatedwith prosthetic material despite the use of appro-priate antibioticsis a major clinicalproblem, the antimicrobialsusceptibility of bacteria respon-sible for a chronic subcutaneous tissue cage infection in rat was investigated ex vivo. Three to 6 weeks after the initiation of infection, suspensions of two strains ofStaphylococcus aureus re-covered from the foreign body surface and surrounding fluid were exposed to either oxacillin, vancomycin, fleroxacin, gentamicin, or rifampin. The MBCs of these bacteria were markedly elevated, in most cases 128 to >256 times higher than the MBC of batch culture S.aureusin either logarithmic or stationary phase. Kinetic studies showed the bacteria did not growwhen incubated for 2 h in Mueller-Hinton broth, possibly reflecting dormancy. Their killing was slow and incomplete byallantibiotics at >8 times their MIC. These data providedirect evidence of a decreasedsusceptibility of S.aureusto the killing effectof antimicrobials during chronic for-eign body infections in vivo.
Received 4 June 1990; revised 4 February 1991.
Financial support: SwissNationalResearch Foundation (grant 3.829-0.87; fellowship 32-27222.89 to C.C.).
Reprints or correspondence: Dr. C. Chuard, Divisionof Infectious Dis-eases, Geneva University Hospital, CH-1211 Geneva 4, Switzerland.
*
Presentaddress: IntensiveCare Unit, Bichat-Claude Bernard Hospital, Paris.TheJournalofInfectiousDIseases 1991;163:1369-1373
©1991byTheUnivenity of Chicago. All rights reserved. 0022-1899191/6306-0033$01.00
Infection is a majorcomplication of implanted devices. Pa-tients frequently do notrespond to highdoses ofantimicrobial agents even if administered for prolonged periods, and for-eign bodies usually mustberemoved to achieve cure.Itis known that susceptibility to antibiotics maybeprofoundly affectedbygrowthof microorganisms nearbiomaterials. Thus vancomycin, testedin a system mimicking infection of a peri-toneal dialysis catheter, wasunable to decreasethe viability
1370 Concise Communications 110 1991;163 (June)
of a population of biofilm-enclosedStaphylococcus
epider-midis
[1]. Similarly, a significant survival of S.epidermidis
adherent on steelwasshown whenthe bacteriawereexposed to a concentration of tobramycin 20-200 times higher than the MBC that had been measuredfor microorganisms in so-lution [2]. Othershave reported similarresultswith
Pseudo-monas aeruginosa
[3]. However, these in vitro models are characterized by an artificial environment in which bacteria grow for a short time in the absence of host factors under selected, not necessarily physiologic, conditions.Here we used an experimental model of chronic subcuta-neoustissuecageinfection in rats [4] to explore the suscepti-bilityof suspensions ofStaphylococcusaureusrecovered from the foreign body surfaceand surrounding fluid to the killing effect of various antistaphylococcal agents.
Materials
and
MethodsBacterial strainsandantimicrobial agents. Theselected S.aureus strains120and MRGR3 werebloodstream clinicalisolatesfrompa-tients with intravenous catheter infections. Both were penicillin-resistant, and MRGR3 was methicillin- and gentamicin-resistant. Standard reference preparations of all antibiotics wereobtained from their respective manufacturers.
Animal model. Polytetrafluoroethylene multiperforated tissue cages (length, 32 mm; diameter, 10 mm) containing three poly-methylmethacrylate coverslips(7 x 7 mm) were implanted sub-cutaneously into the flank of Wistar rats. At 3-6 weeks after implantation, the fluid thathad accumulated in the cageswasinocu-latedwith 1 x lOS to 1X 106cfu of S.aureus 120or MRGR3. Two weeks later, a sustained and stable infection >1 x lOS cfu/ml of tis-sue cage fluid (mean, 1X 107to 1X 108cfu/ml) was obtainedin >90% of cages [4].
Isolation of bacteria from tissuecage fluidandcoverslipsfor sus-ceptibility testing. Threeto 6 weeks after infection, 0.1 ml of tis-sue cage fluid was collected.Itwas centrifuged and the pellet was resuspended in PBS(GmCO, Paisley, UK)with0.1 % TritonX-100 (Merck, Darmstadt, FRG) for5minto disrupthostcells. Afterwash-ing,bacteriaweresonicated for 1minat 60 W (model 2200; Brand-son UltraBrand-soncis, Branbury, CT) to avoid clumping and then further diluted appropriately. In control experiments, the procedure was harmless for bacteria regarding their ability to multiply and their susceptibility to antibiotics. Clumping, monitored by microscopy, was very low.
Bacteriadirectlyadherentto coverslips werealso studied. Cover-slips were rinsedthreetimes in PBS by gentle agitation, then ex-posed to bovine trypsin (Serva, New York) 6 units/ml in PBS for 20min, and sonicated for 1minat 60 W; this method was shown microscopically to efficiently detach surface-bound S.aureus. De-tached bacteriawere washed and declumped. Controls proved this procedure did not affect microorganisms. The numberof adherent bacteria recovered fromthreecoverslips was 5-10 times less than that recovered from 0.1 ml of simultaneously collected tissuecage fluid; therefore notall susceptibility testsappliedto tissuecagefluid bacteria were possible with coverslip bacteria as the final inocula in the latter were sometimes too low.
Determination of MBC and MIC by broth macrodilution. The broth macrodilution method recommended by the National Com-mitteefor Clinical LaboratoryStandards [5] wasstrictlyappliedto determine the MIC and MBC (>99.9% killing) of oxacillin,van-comycin, fleroxacin, gentamicin, and rifampin forS.aureus 120and vancomycin, fleroxacin, and rifampin for S.aureus MRGR3. Tissue cage fluid bacteria were compared with logarithmic- (3 h) and stationary-phase (18-20h) culturesgrownin Mueller-Hinton broth (MHB; Difco, Detroit), designated in vitro bacteria. The final in-oculumwas4
x
lOS to 2x
1()6 cfu/mlfor bothex vivoand in vitro microorganisms. Carryover problems weresolved at up to 256/Lg/ml withoxacillin, fleroxacin, and gentamicin and at 2.56 /Lg/ml for ri-fampin. Coverslip-adherent bacteria could not be studied by this method.For S.aureus MRGR3, the MIC and MBC of vancomycin were also measured after incubation for 20 h of batchculturestationary-phase bacteria in fibronectin (25 /Lg/ml).
Determination of MIC and MBC by agar dilution. MICs and MBCsweredetermined entirelyon solidmediausinga {3-lactamase neutralization procedure for oxacillin on S.aureus 120and a mem-brane transfer technique for vancomycin on S.aureus MRGR3.
In the {3-lactamase neutralization procedure [6, 7], the inoculum is plated on oxacillin-containing Mueller-Hinton agar(MHA; Difco), a thin agar overlay is placedon the bacteria, the oxacillin is inacti-vatedwitha penicillinase solution (penase concentrate; Difco) after determination of the MIC at 24 h, and the MBC is read after 48 h of further incubation. Penase concentrate(1ml) wasfound to inac-tivate 128/Lg/ml oxacillin in 20mlofMHA. Theinoculum was0.1 ml of a solution containing 2 x lOS to 4 X 105cfu/ml.
Themembrane transfer technique wasderived fromsimilarproce-dures previously reported [8, 9]. An inoculum (50 /Ll) of 6 X 10" to 1X 106cfu/ml was plated on 4.7-cm-diameter, 0,45-/Lm pore-sizecellulose filters (Millipore, Bedford, MA)lyingon MHAplates containing vancomycin. After24 h ofincubation, the MICwasread; filters were subsequently transferredthree times for 60 min onto MHAplateswithoutantibiotic to eliminate vancomycin and finally incubated on MHA for 48 h for MBC determination. Vancomycin concentrations~256/Lg/ml couldbe reliably washed away withthis method.
Parts of colonies originating from bacteria that had survived an exposure to 256 /Lg/ml vancomycin wereremoved from membranes when the MBC was read. They were used in a suspension of 5X lOS cfu/ml, which was immediately plated on membranes and retested for MBC.
Killing andgrowth kinetic studies. Bacterial kinetics werestud-ied by growth and time-kill curves. Experiments weredone in con-stantly agitated glasstubescontaining 10ml of MHB. Samples were platedon MHAwitha spiralplater(SpiralSystem, Cincinnati) and plates werereadafter48 h incubation. Oxacillin, vancomycin, flerox-acin, andgentamicin weretestedat 8timestheirMIC; forrifampin, the concentration was200 timesthe MICto approximate the tissue levels measured during treatment in humans. The limit for detec-tion of viablebacteriawas 1.3loglO cfu/mlfor the first four antibi-otics and 2.3 loglO cfu/ml for rifampin (solution diluted to avoid carryover). All plateswithoutvisiblegrowth wereconsidered to be at the detection limit.
Tissue cage fluid and in vitro bacteriawere exposed to each an-timicrobial agent (inoculum, 1 X 105to 3 X lOS cfu/ml for S.
JIO 1991;163 (June) Concise Communications 1371
aureus 120and 1X 106to 3X 106cfu/ml for S.aureus MRGR3).
Susceptibility of coverslip bacteria was assessed with oxacillin for S.aureus 120 and vancomycin for S. aureus MRGR3 (inoculum, 2 X l()4 to 5 X l()4cfu/ml).
Tissue cage fluid bacteria were also studied after they had been preincubated in MHBat 3rC for 4 h beforeexposure to antibiotics (S. aureus 120 with oxacillin and S. aureus MRGR3 with van-comycin).
Finally, a stationary-phase batchcultureof S.aureus MRGR3 that was grown in broth containing fibronectin (25 Ilg/ml) for 20 h was tested with vancomycin.
Statistics. For time-kill curves, the Wilcoxon testfor pairedrank-able scores (two-tailed) wasused to compareat each time point the differences in mean count decreases between the three conditions tested for all antibiotics on both strains.
Results
MICs. No significant difference between MICs againstin vitro (logarithmic- and stationary-phase) and ex vivo (tissue cagefluid and coverslips) bacteria wasobserved. Broth mac-rodilution and agar dilution methods yielded comparable results (table 1).
MBCs. For in vitro bacteria, the MBC-to-MIC ratio was 1-2:1 for all antibiotics testedexceptrifampin, which wasnot bactericidal against logarithmic- and stationary-phase S. aureus 120 and stationary-phase S. aureus MRGR3. Except for rifampinwith S.aureus MRGR3, there was perfect
con-cordancebetweenthe MBCsfor exponential culturesand 18-to 20-h cultures.
The difference between in vitro bacteria and bacteria as-sociatedwith the foreign body was striking, with a MBC-to-MIC ratio >128-256:1 in all conditions for the latter. Similar resultswere obtainedfor bacteria recovered from tissuecage fluid and coverslips.
The agar dilutionmethod, whichis notprone to artifactual results (falsely elevated MBC)occasionally linkedto the ini-tial broth incubation phase of classic procedures, confirmed the resultsobtainedwiththe broth macrodilution method(ta-ble 1).
Daughter cells originating from tissue cage fluid bacteria that survived exposure to a high concentration of antibiotic had the same MBC values as in vitro bacteria.
Preincubation of stationary-phase batchculture bacteriain fibronectin did not change their MBC.
Kineticstudies. Withbacteria in logarithmic phase, a de-creaseof viablecounts of more than 3 10glO (mean,4.22; SD, 0.61; range, 3.37-5.03) after 24 h, that is, bactericidal activ-ity,wasobservedinallcasesexceptfor rifampinon S.aureus 120 (mean, 1.68). This decrease is underestimated since the detectionlevelwasreachedin most cases. The 99.9% killing level was usually seen 8 h after antibioticexposure. Control populations grown without antimicrobial agents showed a mean increase of 1.38
±
0.22 10glO within the first 2 h of in-cubation.Table1. MIC and MBC (ug/ml) of antimicrobial agents studied against Staphylococcus aureus strains 120 and MRGR3.
MIC MBC
Ex vivo* Ex vivo
In vitro, In vitro
logarithmic/ Tissue Tissue
stationaryt cage fluid Coverslips Logarithmic Stationary cage fluid Coverslips S.aureus120 Broth rnacrodilution Oxacillin 0.6 0.3 NO 0.8 0.7 >256 NO Vancomycin 1 1 NO 1.5 1 >256 NO Fleroxacin 0.9 0.9 NO 1 1.2 >256 NO Gentamicin 1.4 1 NO 1.5 1.3 ~256 NO Rifampin 0.02 0.01 NO >2.56 >2.56 >2.56 NO Agar dilution Oxacillin 0.25 0.25 0.5 0.75 0.75 >128 >128 S.aureusMRGR3 Broth macrodilution Vancomycin 0.9 1.4 NO 2 1.9 >256 NO Fleroxacin 0.7 1.1 NO 1.1 1 256 NO Rifampin 0.01 0.01 NO 0.02 >2.56 >2.56 NO Agar dilution Vancomycin 1.1 1.6 1.6 1.5 1.5 >200 >200
NOTE. MICs and MBCs were determined by classic broth macrodilution and by~-Iactamaseinhibition or a membrane transfer agar dilution technique. Results are mean of~3determinations. NO= not done.
tIdentical results were obtained with in vitro cultures in logarithmic and stationary phases.
1372 Concise Communications JID 1991;163 (June)
The killing was slightly but significantly
(P=
.028) less
for stationary-phase bacteria but exceeded a 3 10gIO decrease
in inoculum (mean, 3.90
±
0.44; range, 3.32-4.54) after 24
h with allantibiotics butrifampin (mean, 1.76). Growth curves
did notdiffer fromlogarithmic-phase bacteria control curves.
The decrease in viability of tissue cage fluid bacteriawas
significantly
(P=
.028) slower than that of in vitrobacteria,
whether tested in logarithmic or in stationary phase;
rifam-pin, forwhich therewas tolerancein vitro,wasan exception.
Noneof the antimicrobial agents, on either S.
aureus 120 or
MRGR3, led to a 3 10gIO decrease in initial inoculum after
24 h ofincubation (mean, 1.91
±
0.80; range, 0.17 for
rifam-pin on S.
aureus 120 to 2.48 for gentamicin).
Inthe absence
of antibiotics, bacteriaassociated with the foreign body also
behaved differently fromin vitromicroorganisms, sincethere
wasa 2-h delay for initiation of population growth (figure 1).
Bacteria directly detached from coverslips were studied with
oxacillin for S.
aureus 120 and vancomycin for S. aureus
MRGR3. Therewas a strictparallelism between curves
oftis-suecagefluid andcoverslip bacteria, whetheror notexposed
to antibiotics (withoxacillin, decrease from 5.31
±
0.22 to
2.90
±
0.36 10gIO cfu/ml after 24 h [difference, 2.41] for
cagefluid and from4.72
±
0.90 to 2.19
±
0.27 log
10cfu/ml
[difference, 2.53] for coverslips).
Whentissuecagefluid bacteria werepreincubated in MHB
for 4 h before they were tested, their kill curves and growth
kinetics werecomparable to thoseof batchculturebacteria.
Stationary-phase batchculture bacteria grown
inbrothcon-taining fibronectin didnotshow a significant alteration
ofkill-ing and growth patterns.
Discussion
This studyfound a markedly decreased susceptibility of S.
aureus associated with foreign body infection in vivo to the
killing effect of antibiotics when compared with batch
cul-turebacteria ofthe same straingrown underconventional
con-ditions. MBCs determined
bytheclassical brothmacrodilution
method, which has been criticized [10], were confirmed
byan agar dilution technique and
bykinetic studies.
Resistance of microorganisms to the killing actionof
bac-tericidal antimicrobial agents due to their physiologic
condi-Oxacillin
4 8 12162024 Time (h)Gentamicin
9 81';>
- 7I.
~
6 IU 5
8' 4 ;
~ 3 \ ~-t---i 2 !'~~~~[
o
4 8 12 16 20 24 Time (h)Vancomycin
4 8 12162024 Time (h)Rifampin
9 8l;>
-71.
~
6 Iu5
8'4
~3 2.r/~,/ /'////."/~,/ 1 0~'.,LL~fLL-~;LL,o
4 8 12 16 20 24 Time (h)Fleroxacin
4 8 12162024 Time (h)•.•.•.•.• in vitro log. phase, withantibiotic - - . - - in vitro stat. phase, withantibiotic
~ tissue cage fluid, withantibiotic •.•.-o .•.•in vitro log. phase, control - - -0- _. in vitro stat. phase, control - - 0 -tissue cage fluid, control
0/&
area under detection limitFigure 1. Killing kineticstudiesonStaphylococcus aureus120withfive antibiotics: oxacillin(4I£g/ml),vancomycin, fleroxacin, gentami-cin (8 I£g/ml each), and rifampin (2 I£g/ml). Bacteriagrownin vitroto logarithmic and stationary phasesweretestedas werethoserecovered from tissue cage fluid. Mean and SD forthreeexperiments for each condition are shown. Inocula at time0were standardized to mean inoculum value ofallexperiments (5.32 10gIO cfu/ml; SD, 0.17; range, 5.05-5.52).
1ID 1991;163 (Iune) Concise Communications 1373
tionor to a particularenvironment (phenotypic tolerance) has been recognized since the early use of penicillin. In foreign bodyinfections several determinants of phenotypic tolerance are likely to be met. Slowgrowth is the most classic condi-tion that leads to a decreasedsusceptibility of bacteria to an-tibioticsin fluids [11] and on surfaces[12].Itis thoughtlikely thatbacteriaresponsible for infections withprosthetic devices have low metabolic activity. In our experiments, there was a 2-h lag time before the number of cage fluid bacteria in-creased when bacteria were incubated in MHB without an-tibiotic; this suggests possibledormancy that differs fromthe mere stationary phase of an overnight culture. When these bacteriawereexposed to antibiotics in a growing phase,after being in MHB for 4 h, their susceptibility to drugs was re-stored. Nutrientlimitation[13] maybe implicated in the gen-esis of rgen-esistance to killing in vivo.
A role for the biofilm (aggregates of microorganisms and gross extracellular material)as a barrier againstthe penetra-tion of antimicrobial agents has been suggested [14]. It ap-pears this is not the unique component that determines the resistanceof bacteriaassociatedwithforeign body infections to killing, since in our model, resistance was observed for organisms in tissuecagefluidand for bacteriadetachedfrom coverslips after disruptionof the biofilm. The coatingofbac-teria by matrix proteins, such as fibronectin and fibrinogen, could also have some importance in vivo; however, incuba-tion of batch culture bacteria in fibronectin in vitro did not alter their susceptibility to killing.
Independent of these conditions, it is possiblethat surface growthper se is in part responsible for tolerance due to the changes it induces in the ultrastructure of bacteria [15].
In conclusion,
S.
aureus
recovered from an experimental subcutaneous tissuecageinfection show a markedresistance to the killingeffect of antimicrobial agents actingat different bacterialsites. Thisphenomenon is likely to contribute to the frequent failure of antibiotic therapy of infected implanted devices.Asystematic and detailed studyof the pathophysiol-ogyof bacteriaassociated withforeign bodyinfections in vitro and in vivo shouldallowbetter understanding of the mecha-nisms of this complex and important clinical problem.Acknowledgment
WethankManuela Bento and IvanaRatti for excellent technical assistance and PierreVaudaux andJean-Claude Pechere for helpful suggestions.
References
1. EvansRC,HolmesCI.Effect ofvancomycin hydrochloride on Staphylo-coccusepidermidisbiofilm associated with silicone elastomer. An-timicrob Agents Chemother 1987;31:889-94.
2. GristinaAG, Hobgood CD,Webb LX, MyrvikQN. Adhesive coloniza-tion of biomaterials and antibiotic resistance. Biomaterials 1987; 8:423-6.
3. NickelIC,Ruseska I, Wright lB, CostertonIW. Tobramycin resistance ofPseudomonas aeruginosacells growing as a biofilm on urinary cathetermaterial. Antimicrob Agents Chemother 1985;27:619-24. 4. LucetIC,Herrmann M, Rohner P,Auckenthaler R, Waldvogel FA,Lew DP. Treatment of experimental foreign body infection causedby
methicillin-resistantStaphylococcus aureus.Antimicrob Agents Che-mother 1990;34:2312-7.
5. National Committee for Clinical Laboratory Standards. Methods for determining bactericidal activityof antimicrobial agents. Document M26-P. Villanova, PA: NCCLS, 1987.
6. Woolfrey BF, LallyRI,Ederer MN. Evaluation of oxacillin tolerance inStaphylococcus aureusbya novel method. Antimicrob Agents Che-mother 1985;28:381-8.
7. Woolfrey BR, Gresser-Burns ME, LallyRI.Effect of temperature on inoculum as a potential sourceof error in agar dilution plate count bactericidal measurements. Antimicrob Agents Chemother 1988;32: 513-7.
8. Courvalin P, Goldstein F,Philippon A, SirotI. Determination de la con-centration minimale bactericide en milieu solide.In:Courvalin P, Goldstein F,Philippon A, SirotI,eds.L'antibiogramme. Paris: MPC-VIDEOM, 1985:195-8.
9. FernandesCI,Stevens DA,GrootObbinkOJ, Ackerman VP. A replica-tor methodfor the combined determination of minimum inhibireplica-tory concentration andminimum bactericidal concentration. I Antimicrob Chemother 1985;15:53-60.
10. Thylor PC, Schoenknecht FD, SherrisIC,Linner EC. Determination ofminimum bactericidal concentrations ofoxacillin for Staphylococ-cusaureus:influence andsignificance oftechnical factors. Antimicrob Agents Chemother 1983;23:142-50.
11. Cozens RM, Thomanen E, Tosch W,Zak0,SuterI, Tomasz A. Evalua-tionof thebactericidal activity of I3-laetam antibiotics on slowly grow-ing bacteriaculturedin a chemostat. Antimicrob Agents Chemother 1986;29:797-802.
12. Evans OJ, Brown MRW, AllisonDG, Gilbert P. Susceptibility ofbac-terial biofilm to tobramycin: role of specificgrowthrate and phase in the division cycle. I Antimicrob Chemother 1990;25:585-91. 13. Brown MRW, Williams P. Influence of substrate limitation andgrowth
phase on sensitivity to antimicrobial agents. I Antimicrob Chemo-ther 1985;15(suppl A):7-14.
14. CostertonIW,ChengKJ, Geesey GO, et al. Bacterial biofiims in nature and disease. Annu Rev Microbiol 1987;41:435-64.
15. Lorian V,Zak0,Suter I, BruecherC. Staphylococci, in vitro and in vivo. Diagn Microbiol Infect Dis 1985;3:433-44.
