International Journal of Antimicrobial Agents 24 (2004) 178–180

In vitro antagonism between�-lactam and macrolide inStreptococcuspneumoniae: how important is the antibiotic order?

Mar Ortega∗, Francesc Marco, Alex Soriano, Julià Gómez, Manel Almela, Josep Mensa

Institute of Infectious Diseases and Immunology, IDIBAPS, Hospital Cl´ınic, Escalera 9, 4◦ piso, Villarroel 170, Barcelona 08036, Spain

Received 22 December 2003; accepted 8 February 2004

Abstract

We found that the in vitro interaction between penicillin or cefotaxime and erythromycin againstStreptococcus pneumoniaevaries dependingon the order of antibiotic exposure. Time-kill experiments were performed with penicillin, cefotaxime, erythromycin and different ordercombinations of both�-lactams with erythromycin. The mean difference between the colony count at 0 and 6 h for penicillin, cefotaxime anderythromycin tested separately was 3.5 log cfu/mL, 2.4 and 1.5 respectively for susceptible strains. The mean difference for the combinationof �-lactam and erythromycin studied simultaneously was 1.8 log cfu/mL for these strains. The association of penicillin or cefotaxime witherythromycin added two hours later showed an activity similar to those of�-lactam alone (mean difference was 3.0 for this association withpenicillin and 2.5 with cefotaxime). Therefore, the antagonistic effect of macrolide activity could be less important if erythromycin wasadministrated after�-lactam.© 2004 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.

Keywords:Combined treatment; Community-acquired pneumonia; Time-kill curves

1. Introduction

Gleason et al. [1] observed that initial antimicro-bial regimens with a second-generation cephalosporinplus a macrolide, a non pseudomonal third-generationcephalosporin plus a macrolide or a fluoroquinolone alonewere associated with lower 30-day mortality among patientshospitalized with pneumonia compared with those who re-ceived a non-pseudomonal third-generation cephalosporinalone. Stahl et al.[2] studied the impact of antibiotic choiceon length of stay and mortality in general medical patientshospitalized with community-acquired pneumonia. Theyconcluded that the use of a macrolide as part of the initialtherapeutic regimen appeared to be associated with shorterlength of stay. Data from our hospital suggest that adding amacrolide to an empirical�-lactam-based antibiotic treat-ment is associated with lower mortality rates in patientswith bacteraemic pneumococcal pneumonia[3].

The combination of a�-lactam antibiotic with a macrolidemay be conflicting in treating pneumococcal infectionbecause the bacteriostatic effect of the macrolide could

∗ Corresponding author. Tel.:+34-93-2275529; fax:+34-93-4514438.E-mail address:marortegaromero@yahoo.es (M. Ortega).

antagonize the bactericidal activity of the�-lactam. Jo-hansen et al.[4] demonstrated the antagonism betweenpenicillin and erythromycin againstStreptococcus pneu-moniaein vitro and in vivo. In a mouse peritonitis model,erythromycin was given 90 min after bacterial inoculationand penicillin 60 min later in the combination treatmentgroup. Mortality was significantly higher in the mice treatedwith erythromycin before penicillin than in those treatedwith penicillin alone.

Considering that initial empirical treatment of acute pneu-monia of unknown aetiology includes a combination of acephalosporin and a macrolide[5], we conducted this studyto investigate if the result of the association in vitro coulddepend on the sequence of antibiotic administration.

2. Material and methods

2.1. Bacterial strains

Thirty clinically significant S. pneumoniaeisolates re-covered from respiratory samples (n = 20) and blood cul-tures (n = 10) were used in the study. Minimal inhibitoryconcentrations (MICs) were determined with a commer-cialized microdilution method (Sensititre, Trek Diagnostic

0924-8579/$ – see front matter © 2004 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved.doi:10.1016/j.ijantimicag.2004.02.030

M. Ortega et al. / International Journal of Antimicrobial Agents 24 (2004) 178–180 179

Systems, UK) using cation adjusted Mueller-Hinton brothsupplemented with 5% lysed horse blood. The strains werecategorized as susceptible, intermediate or resistant ac-cording to break points proposed in the NCCLS doc-ument [6]. After MIC determinations were performed,the isolates were stored at−70◦C in double-strengthskim milk and cultured in Columbia agar plus 5%sheep blood at least twice before additional studies wereconducted.

2.2. Time-kill curves

Time-kill curves were performed using Todd-Hewitt brothplus 0.5% of yeast extract as culture medium. Antibioticconcentrations were as follows: penicillin and cefotaxime10 mg/L and erythromycin 1 mg/L. Ten tubes with a totalvolume of 10 mL in each one were prepared for every strain.The following schedule was used: control (C), penicillin(P), erythromycin (E), cefotaxime (CTX), penicillin anderythromycin simultaneously (P+ E), cefotaxime and ery-thromycin simultaneously (CTX+ E), penicillin followedby erythromycin two hours later (P+ 2hE), cefotaxime fol-lowed by erythromycin two hours later (CTX+ 2hE), anderythromycin followed by penicillin or cefotaxime two hourslater (E+ 2hP, E+ 2hCTX).

Appropriate dilutions were performed at 0, 2, 4 and 6 hin order to ascertain the number of viable bacteria (cfu/mL)in each tube. At 2 h, the second antibiotic in tubes P+ 2hE,CTX + 2hE, E + 2hP and E+ 2hCTX was added. Alltime-kill experiments were performed in duplicate. Thevariation in colony counts at the same time points forrepeated experiments was<0.5 log cfu/mL. Antagonismwas defined as a significantly decreased killing effect (i.e.>0.5 log cfu/mL from 1 to 5 h after adding drugs)[4].

The results of time-kill experiments were analyzed inthree groups depending on the susceptibilities to penicillinand erythromycin (PS-ES: penicillin and erythromycinsusceptible strains; PI-ES: penicillin intermediate and ery-thromycin susceptible strains; and PR-ER: penicillin anderythromycin resistant strains).

Table 2Mean difference and standard deviation (S.D.) between log cfu/mL at 0 and 6 h for every time-kill experiment

Penicillin S,erythromycin S (n = 10)

Penicillin I,erythromycin S (n = 10)

Penicillin R,erythromycin R (n = 10)

Control −1.4 (0.4) −1.8 (0.6) −2.0 (0.8)Penicillin 3.5 (0.2) 3.0 (0.3) 2.4 (0.5)Cefotaxime 2.4 (0.2) 2.5 (0.2) 2.0 (0.2)Erythromycin 1.5 (0.1) 1.5 (0.2) −1 (0.1)Penicillin + erythromycin 1.8 (0.3) 1.6 (0.2) 1.1 (0.3)Cefotaxime+ erythromycin 1.8 (0.1) 1.9 (0.3) 1.3 (0.3)E + 2hP 1.7 (0.4) 1.4 (0.3) −0.9 (0.2)E + 2hCTX 1.6 (0.4) 1.7 (0.5) −0.8 (0.1)P + 2hE 3.0 (0.1) 2.8 (0.2) 2.3 (0.2)CTX + 2hE 2.5 (0.1) 2.6 (0.1) 1.9 (0.3)

P + 2hE, penicillin and two hours later, erythromycin; CTX+ 2hE, cefotaxime and two hours later erythromycin; E+ 2hP, erythromycin and two hourslater penicillin; E+ 2hCTX, erythromycin and two hours later cefotaxime. S: susceptible, I: intermediate and R: resistant.

Table 1Penicillin, erythromycin and cefotaxime MIC of the strains used in thestudy

Group No. ofstrains

PenicillinMIC

CefotaximeMIC

ErythromycinMIC

PS-ES 10 <0.03 <0.06 <0.25PI-ES 10 1 0.25 <0.25PR-ER 10 2 0.5–1 >32

MIC in mg/L. PS-ES: penicillin and erythromycin susceptible strains;PI-ES: penicillin intermediate and erythromycin susceptible strains;PR-ER: penicillin and erythromycin resistant strains.

3. Results

Table 1 shows penicillin, erythromycin and cefotaximeMICs of strains studied.

The mean inoculum at time 0 h was 7.54 log cfu/mL(S.D.: 0.6).Table 2 show the mean difference (S.D.) be-tween log cfu/mL at 0 and 6 h for every experiment and forevery group of strains. The greater activity was observedwhen penicillin was tested alone. The mean differencebetween the colony count at 0 and 6 h for susceptible, in-termediate and resistant penicillin strains was 3.5 (0.2), 3.0(0.3) and 2.4 (0.5), respectively (P = 0.06). Cefotaximeshowed lower killing rate by time than penicillin and thekilling rate was slightly less for resistant strains. The exper-iments with erythromycin alone showed slower reductionof colony count by time than with�-lactam antibiotics.

When the combination of�-lactam (penicillin or cefo-taxime) and erythromycin was tested, the association showedantagonism with the difference between the activity of�-lactam alone and the combination being >0.5 log cfu/mL.This reduction in killing occurred in all groups, even instrains resistant to erythromycin.

The combination of penicillin plus erythromycin addedtwo hours later, showed an activity similar to penicillinalone. The mean difference between the colony count at 0and 6 h for susceptible, intermediate and resistant penicillinstrains with the combination P+ 2hE was: 3.0 (0.1), 2.8(0.2) and 2.3 (0.2), respectively.

180 M. Ortega et al. / International Journal of Antimicrobial Agents 24 (2004) 178–180

The combination of cefotaxime plus erythromycin addedtwo hours later showed an activity similar to cefotaximealone (difference between colony count at time 0 and 6 hof study for susceptible, intermediate and resistant strainswith the combination CTX+ 2hE was: 2.5 (0.1), 2.6 (0.1)and 1.9 (0.3), respectively).The activity of the combinationerythromycin plus�-lactam, either penicillin or cefotaxime,added two hours later, was similar to erythromycin alone;the differences did not exceed 0.5 log cfu/mL.

4. Discussion

Penicillin showed the highest killing rate in the threegroups of strains. However, its activity was slower in caseof resistant penicillin strains. Cefotaxime demonstrated aslower killing rate than penicillin in all groups of strains.Erythromycin showed the slowest killing rate during thesix hours of exposure. Time-kill assays with�-lactam anderythromycin simultaneously demonstrated antagonism be-tween these drugs and the resulting curve was similar tothe slow inhibitory activity of erythromycin tested alone. Iferythromycin was added two hours later than�-lactam, theactivity of the combination was similar to that of penicillinor cefotaxime alone.

The inhibitory activity of the macrolide on growth or celldivision of bacteria is the probable reason for antagonismagainst penicillin, which can only act on bacteria that arein the growth phase and producing a cell wall. Johansenet al.[4] demonstrated that the bacteriostatic activity of ery-thromycin and its inhibitory effect on penicillin were clearlyapparent in vitro and in vivo experiments. Other authorswrote about the lack of synergy of erythromycin combinedwith penicillin or cefotaxime againstS. pneumoniaein vitro[7]. The clinical significance of the antagonism between abactericidal drug such as penicillin or cefotaxime and a bac-teriostatic protein synthesis inhibitor has been demonstratedin previous studies[8].

Recent studies suggest that the combination of a�-lactamplus a macrolide is probably the better regimen for theinitial treatment of community-acquired pneumonia[1,2].Houck et al. [9] looked to see if there was an associa-tion between empirical antibiotic therapy and mortality in10 069 Medicare beneficiaries who were hospitalized withcommunity-acquired pneumonia during the years 1993,1995 and 1997. Their conclusion was that inclusion of amacrolide or a fluoroquinolone in initial empirical treat-ment was associated with improved survival. Mufson andStanek[10] conducted a surveillance study of bacteraemicpneumococcal pneumonia from 1978 to 1997 to investigatecase-fatality rates and antibiotic usage. They concluded thattreatment regimens that included a macrolide resulted inthe lowest fatality rate.

If we consider the results of these studies, we mustconclude that the combination of�-lactam and macrolideis a suitable empirical treatment for community-acquiredpneumonia. But components of this regimen have demon-

strated that their activity is antagonistic in vivo and in vitro[4,7]. However, our results demonstrate that the existenceof antagonism depends on the order in which the antibioticsare combined. When the first antibiotic was the�-lactamand two hours after erythromycin was added, the activityof the association was similar to that of�-lactam alone.The antagonism of cefotaxime is minor, probably becausethe slower activity of cefotaxime compared with penicillin.Our results suggest that when the association of�-lactamand macrolide is chosen for the empirical treatment ofcommunity-acquired pneumonia, it is advisable to begin theantibiotic administration with the�-lactam.

Another interesting finding of our study is that the ery-thromycin resistant strains showed slower growth comparedwith the control culture when they were exposed to lowerythromycin concentrations. If we consider that the ery-thromycin concentration in epithelial lining fluid is higherthan the concentrations used in our study is reasonable tothink that the in vivo activity of the macrolide could behigher than the activity observed in vitro. These results agreewith those of Amsden[11] who suggested that the macrolideMIC values need to be compared with in vivo antibioticpharmacokinetics and pharmacodynamics.

References

[1] Gleason PP, Meehan TP, Fine JM, Galusha DH, Fine MJ. Associ-ations between initial antimicrobial therapy and medical outcomesfor hospitalized elderly patients with pneumonia. Arch Intern Med1999;159:2562–72.

[2] Stahl JE, Barza M, Desjardin J, Martin R, Eckman MH. Effectof macrolides as part of initial empiric therapy on length of stayin patients hospitalized with community-acquired pneumonia. ArchIntern Med 1999;159:2576–80.

[3] Martı́nez JA, Horcajada JP, Almela M. Adding a macrolide to a�-lactam-based empiric antibiotic regimen is associated with lessin-hospital mortality in patients with bacteremic pneumococcal pneu-monia. Clin Infect Dis 2003;36:389–95.

[4] Johansen HK, Jensen TG, Dessau RB, Lundgren B, Frimodt-MøllerN. Antagonism between penicillin and erythromycin againstStrep-tococcus pneumoniaein vitro and in vivo. J Antimicrob Chemother2000;46:973–80.

[5] Barlett JG, Dowell SF, Mandell LA, File TM, Musher DM, FineMJ. Practice guidelines for the management of community-acquiredpneumonia in adults. Clin Infect Dis 2000;31:347–82.

[6] National Committee for Clinical Laboratory Standards. Performancestandards for antimicrobial susceptibility testing: twelfth informa-tional supplement. Document M100-S12. Wayne, PA: 2002.

[7] Lin E, Stanek RJ, Mufson MA. Lack of synergy of erythromycincombined with penicillin or cefotaxime againstStreptococcus pneu-moniaein vitro. Antimicrob Agents Chemother 2003;47:1151–3.

[8] Mathies AW, Leedom JM, Ivler D, Wehrle PF, Portnoy B. Antibioticantagonism in bacterial meningitis. Antimicrob Agents Chemother1967;7:218–24.

[9] Houck PM, MacLehose RF, Niederman MS, Lowery JK. Empiric an-tibiotic therapy and mortality among Medicare pneumonia inpatientsin 10 Western States. 1993, 1995 and 1997. Chest 2001;119:1420–6.

[10] Mufson MA, Stanek RJ. Bacteremic pneumococcal pneumonia inone American city: a 20-year longitudinal study, 1978–1997. Am JMed 1999;107(1A):34S–43S.

[11] Amsden GW. Pneumococcal macrolide resistance—myth or reality?J Antimicrob Chemother 1999;44:1–6.