Journal of Antimicrobial Chemotherapy (2003) 52, 651–655DOI: 10.1093/jac/dkg417Advance Access publication 1 September 2003

651. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

© The British Society for Antimicrobial Chemotherapy 2003; all rights reserved.

Factors influencing the anti-inflammatory effect of dexamethasone therapy in experimental pneumococcal meningitis

I. Lutsar*, I. R. Friedland, H. S. Jafri, L. Wubbel, A. Ahmed, M. Trujillo, C. C. McCoig

and G. H. McCracken Jr

Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas, TX, USA

Received 11 January 2003; returned 24 February 2003; revised 14 July 2003; accepted 15 July 2003

Dexamethasone (DXM) interferes with the production of tumour necrosis factor-α (TNF-α) and interleukin-1(IL-1) and can thereby diminish the secondary inflammatory response that follows initiation of antibacterialtherapy. A beneficial effect on the outcome of Haemophilus meningitis in children has been proven, but untilrecently the effect of DXM therapy in pneumococcal meningitis was uncertain. The aim of the present studywas to evaluate factors that might influence the modulatory effect of DXM on the antibiotic-induced inflam-matory response in a rabbit model of pneumococcal meningitis. DXM (1 mg/kg) was given intravenously 30min before or 1 h after administration of a pneumococcal cell wall extract, or the first dose of ampicillin. Inmeningitis induced by cell wall extract, DXM therapy prevented the increase in cerebrospinal fluid (CSF)leucocyte and lactate concentrations, but only if given 30 min before the cell wall extract. In meningitiscaused by live organisms, initiation of ampicillin therapy resulted in an increase in CSF TNF-α and lactateconcentrations only in animals with initial CSF bacterial concentrations ≥5.6 log10 cfu/mL. In those animals,DXM therapy prevented significant elevations in CSF TNF-α [median change –184 pg/mL, –114 pg/mL versus+683 pg/mL with DXM (30 min before or 1 h after ampicillin) versus controls (no DXM), respectively, P = 0.02]and lactate concentrations [median change –10.6 mmol/L, –1.5 mmol/L versus +14.3 mmol/L with DXM (30min before or 1 h after ampicillin) versus controls (no DXM), respectively, P = 0.01]. These effects were inde-pendent of the timing of DXM administration. In this model of experimental pneumococcal meningitis, anantibiotic-induced secondary inflammatory response in the CSF was demonstrated only in animals withhigh initial CSF bacterial concentrations (≥5.6 log10 cfu/mL). These effects were modulated by DXM therapywhether it was given 30 min before or 1 h after the first dose of ampicillin.

Keywords: animal models, CSF, experimental meningitis, inflammatory response, S. pneumoniae

Introduction

Pneumococcal meningitis remains a significant cause of morbidityand mortality. It is the host’s own inflammatory response that isresponsible for the central nervous system injury characteristic of thedisease.1This inflammatory response can be exacerbated by antibac-terial therapy that increases rapidly the release of proinflammatorypneumococcal cell wall products.2–5

Currently, dexamethasone (DXM) is the only anti-inflammatoryagent that has been documented to improve the outcome of bacterialmeningitis in clinical trials, most convincingly in children with Hae-mophilus influenzae meningitis.6 DXM inhibits production of TNF-αand IL-1, reverses development of brain oedema and limits theincrease in cerebrospinal fluid (CSF) lactate and leucocyte concen-trations.1,7,8 Studies have suggested that in H. influenzae meningitis

the timing of DXM in relation to the first antibiotic injection is critical.The antibiotic-induced inflammatory response is prevented only ifDXM therapy is given simultaneously with ceftriaxone; DXM given1 h later failed to modulate the antibiotic-induced inflammatoryresponse.3 A meta-analysis of 10 clinical trials in children suggestedthat the timing of DXM therapy might also be critical in pneumo-coccal meningitis. DXM therapy prevented the development ofsevere hearing loss only when it was given before or at the same timeas the first antibiotic injection (OR = 0.09; 95% CI: 0.00–0.71).6

Recently de Gans & van de Beek9 in a prospective, randomized,placebo-controlled, multicentre trial demonstrated the beneficialeffect of adjunctive DXM therapy on the outcome of pneumococcalmeningitis in adults.

The present study was conducted to evaluate the modulatoryeffect of DXM therapy given 30 min before, compared with 1 h after,

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

**Correspondence address. Clinical Development, Pfizer Ltd., Ramsgate Road ,CT13 9NJ, UK. Tel: +44-1304-645173; Fax:+44-1304-655669; E-mail: irja_lutsar@sandwich.pfizer.com

by guest on February 19, 2012http://jac.oxfordjournals.org/

Dow

nloaded from

I. Lutsar et al.

652

intracisternal injection of pneumococcal cell walls, or after the firstinjection of ampicillin, in experimental meningitis caused by livepneumococci.

Material and methods

Inocula

Pneumococcal cell wall (1 cm3 of this product was obtained from2.5 × 109 live organisms) was produced and provided by Dr ElaineTuomanen.10 Streptococcus pneumoniae (MIC and MBC of ampicillin,0.01 mg/L) was originally isolated from a patient with bacterial menin-gitis. To induce meningitis, 0.2 mL pneumococcal cell wall product, or104–105 cfu/mL of live organisms, were inoculated intracisternally.

Meningitis model and treatment

A rabbit meningitis model originally described by Dacey & Sande11

was used. In the first set of experiments, DXM (1 mg/kg) was givenintravenously 30 min before or 1 h after the administration of pneumo-coccal cell walls to six and seven animals, respectively. In the second setof experiments, 16–18 h after inoculation of live organisms, animals weretreated with ampicillin alone (75 mg/kg every 12 h) for 24 h, or with thecombination of ampicillin and intravenous DXM (1 mg/kg) given 30 minbefore or 1 h after the first ampicillin dose. Each treatment group con-sisted of 12–13 animals. No treatment was given to control animals.

CSF sample collection and analysis

CSF was collected directly from the cisterna magna under acepromazineand ketamine anaesthesia before and 2, 4, 6, 12 and 24 h after start oftherapy. For the first four CSF collections, animals were restrainedunder anaesthesia in stereotactic frames. Leucocytes were counted ina Neubauer haemocytometer. Bacterial concentrations in CSF weremeasured by plating undiluted and serial dilutions of CSF on sheepblood agar and incubating in 5% CO2 at 35°C for 24 h. The lower limitof detection was 10 cfu/mL. The remaining CSF was centrifuged andthe supernatant stored at –70°C. CSF lactate concentrations were meas-ured by a photocolorimetric assay (Behring Diagnostics Inc, MiltonKeynes, UK). TNF-α concentrations were measured by cytotoxic assay

using L929 tumorigenic murine fibroblasts.12 The standard curve forthe TNF-α assay was linear from 40–2500 pg/mL.

Statistical analysis

Normally distributed data are presented as mean ± S.D. and non-normallydistributed data as median and range. Student’s t-test was used forcomparison of parametric data and the Kruskall–Wallis analysis ofvariance for non-parametric variables.

Results

Meningitis induced by pneumococcal cell walls

Administration of pneumococcal cell walls resulted in the release ofTNF-α, an influx of leucocytes and increased lactate concentrationsin CSF (Figure 1). The elevation in TNF-α concentrations was pre-vented by DXM therapy when given 30 min before or 1 h after pneu-mococcal cell walls. The increase in CSF leucocyte and lactateconcentrations, however, was inhibited only when DXM therapy wasgiven 30 min before the cell wall products.

Meningitis induced by live organisms

The addition of DXM to ampicillin therapy resulted in a lower initialbacterial killing rate compared with ampicillin therapy alone (0.39 ±0.1 cfu/mL/h versus 0.57 ± 0.12 cfu/mL/h; P = 0.04). The changes inCSF inflammatory indices were similar in all study groups and werenot influenced by the co-administration of DXM (Figure 2).

For further analysis, animals were divided in two groups based onCSF bacterial concentrations at the start of therapy (≤5.5 log10 or ≥5.6log10 cfu/mL) (Figure 3). After the first dose of ampicillin, animalswith high initial bacterial concentrations demonstrated significantlygreater changes in CSF TNF-α [median ∆+683 pg/mL (quartiles+246 to +758) versus ∆–16.3 pg/mL (quartiles –6 to –29)], whiteblood cells (WBC) [median ∆+2175 cells/mL (quartiles 437 to 6987)versus ∆–325 cells/mL (quartiles –787 to +25)] and lactate [median∆+14.3 mmol/L (quartiles –7.6 to –14.6) versus ∆–8.5 mmol/L(quartiles –5.1 to –16.4)] concentrations compared with animals withlower initial bacterial concentrations.

Figure 1. Concentrations of WBC (cells/mL; left), TNF-α (pg/mL; middle) and lactate (mmol/L; right) in the CSF. Meningitis was induced by the intracisternaladministration of pneumococcal cell walls. Animals were treated with DXM given 30 min before (open squares) or 1 h after (solid squares) administration of cell walls.No treatment was given to the control animals (solid triangles). *P = 0.02 versus controls or those treated with DXM 1 h after introduction of cell walls.

by guest on February 19, 2012http://jac.oxfordjournals.org/

Dow

nloaded from

Dexamethasone in pneumococcal meningitis

653

In those with high initial bacterial concentrations (≥5.6 log10 cfu/mL), DXM therapy prevented elevations in CSF TNF-α [median∆–184 pg/mL (quartiles –116 to –258) or ∆–114 pg/mL (quartiles –39 to –175) versus ∆+683 pg/mL (quartiles 246 to +758) with DXMgiven 30 min before or 1 h after ampicillin versus without DXM,respectively; P = 0.02] and lactate concentrations [median ∆–10.6mmol/L (quartiles 7.6 to 17.4) or ∆–1.5 mmol/L (quartiles –19.7 to –1.3) versus ∆+14.3 mmol/L (quartiles –7.6 to –14.6) with DXMgiven 30 min before or 1 h after ampicillin and without DXM,respectively; P = 0.01]. These effects were not significantly differentas a result of the timing of DXM administration, although there was atrend indicating that changes in TNF-α and lactate values were lowerin animals given DXM before ampicillin therapy (Figure 4). The

changes in leucocyte concentrations were not affected by DXM ther-apy. In animals with low CSF bacterial concentrations (≤5.5 log10cfu/mL), there were no differences in TNF-α, WBC or lactate con-centrations between those treated with or without DXM (data notshown). There was no correlation between changes in TNF-α, leuco-cyte and lactate concentrations in CSF, and the degree of bacterialkilling after introduction of ampicillin therapy (r = 0.06; r = 0.47 andr = 0.46, respectively; P > 0.05).

Discussion

In this model of pneumococcal meningitis, we demonstrated that theantibiotic-induced secondary inflammatory response, as evidenced

Figure 2. Concentrations of WBC (cells/mL), TNF-α (pg/mL), lactate (mmol/L) and S. pneumoniae (log10 cfu/mL) in CSF in animals with pneumococcalmeningitis. Animals were treated with ampicillin alone (solid triangles), withDXM given 30 min before (open squares) or 1 h after ampicillin therapy (solidsquares). No treatment was given to the control (open triangles) animals, andthey died after 6–10 h. Error bars demonstrate lower and upper quartile.*P = 0.04 versus those not receiving DXM therapy.

Figure 4. Concentrations of WBC (cells/mL, left), TNF-α (pg/mL, middle), lactate (mmol/L, right) in CSF in animals with initial CSF bacterial concentrations ≥5.6log10 cfu/mL. DXM therapy was given 30 min before (open squares) or 1 h after the first dose of ampicillin (solid squares). Control animals (solid triangles) weretreated with ampicillin only. Error bars demonstrate lower and upper quartiles. *P < 0.05 versus those treated with ampicillin and DXM.

Figure 3. Changes in CSF concentrations of WBC (cells/mL), TNF-α (pg/mL),lactate (mmol/L) and S. pneumoniae (log10 cfu/mL) in pneumococcal meningi-tis after introduction of ampicillin therapy. The CSF bacterial concentrations atthe introduction of ampicillin therapy were ≤5.5 log10 cfu/mL (open squares) or≥5.6 log10 cfu/mL (solid squares). Error bars demonstrate lower and upper quar-tile. *P < 0.05

by guest on February 19, 2012http://jac.oxfordjournals.org/

Dow

nloaded from

I. Lutsar et al.

654

by an elevation of CSF TNF-α and lactate concentrations, occurredonly in animals with high bacterial concentrations before the startof ampicillin therapy. The increase in CSF lactate and TNF-α con-centrations were inhibited by adjunctive DXM therapy regardlessof whether it was given 30 min before or 1 h after the first dose ofampicillin.

Liberation of free endotoxin or cell-wall components by cell-wallactive antibiotics has been demonstrated in vitro and in experimentalmeningitis.3,4,5,13 This is associated with enhanced inflammation inthe subarachnoid space, as evidenced by an increase in leucocyte,TNF-α and lactate concentrations.3,14 Our study showed that the anti-biotic-induced inflammatory burst did not occur in all animals andwas seen only in those with greater CSF bacterial concentrations(≥5.6 log10 cfu/mL). Although this has been intimated previously,data were not provided.14 In experimental meningitis, the magnitudeof the inflammatory response in CSF depends on the concentration ofinoculated cell walls10 and thus it is not surprising that animals withhigher bacterial counts demonstrate a more pronounced host inflam-matory response after antibiotic therapy.

There is concern that rapid bacterial killing by antibiotics couldresult in an enhanced inflammatory response and a worsening of theclinical outcome in meningitis.15,16 The results of this and previousstudies, however, do not support these speculations. On the contrary,clinical and experimental studies have demonstrated that rapid effec-tive bactericidal therapy protects against the development of deaf-ness and other neurological disabilities.17,18 Moreover, this andprevious studies4 found no correlation between CSF inflammatorymarkers and the magnitude of bacterial killing. Uncontrolled bac-terial growth eventually results in a much greater release of cell wallcomponents—and consequent enhancement of the inflammatoryresponse—than that induced by antibacterial therapy.4,13 These find-ings suggest that early antibacterial therapy and rapid clearance ofbacteria from CSF outweigh the potential adverse effects caused bythe antibiotic-induced inflammatory burst.

DXM, as an adjunct to antibacterial therapy in bacterial meningi-tis, has been shown to be beneficial in H. influenzae meningitis inchildren.6 In pneumococcal meningitis, however, its modulatoryeffect was uncertain because of the relatively small numbers of chil-dren enrolled in prospective controlled trials.19 In adults with pneu-mococcal meningitis, an unfavourable outcome was seen in 26% ofpatients receiving adjunctive DXM compared with 52% among thosereceiving placebo.9 The results of our study clearly supported previ-ous findings by Tuomanen et al.8 and showed that, similar to H. influ-enzae meningitis, DXM therapy prevents the antibiotic-inducedrelease of TNF-α and lactate concentrations in CSF in pneumococcalmeningitis.3

Early institution of DXM therapy has been suggested because ofits delayed onset of action.20 In our study, the timing of DXMappeared to be critical only in meningitis induced by the pneumo-coccal cell wall. In meningitis induced by live microorganisms, how-ever, once CSF inflammation was established, administration ofDXM before the dose of ampicillin did not appear to have a signifi-cantly greater salutary effect than 1 h after ampicillin; DXM adminis-tration that was delayed further was not assessed. These resultscontrast with those obtained in experimental H. influenzae meningi-tis, where the antibiotic-induced inflammatory response was modu-lated only if DXM was given before or simultaneously withceftriaxone therapy.3 The effectiveness of early DXM administrationwas also highlighted in a recent meta-analysis of 10 randomized con-trolled clinical trials.6

At the time of diagnosis, ∼40% of patients with pneumococcalmeningitis have CSF bacterial concentrations <106 cfu/mL21, and ourfindings suggest that antibiotic-induced enhanced inflammation maynot occur in such patients. Whether these patients benefit from DXMtherapy, and how they can be identified at diagnosis, requires furtherclarification. Detection of bacteria in Gram-stained specimens ofCSF may be useful in identifying patients with large bacterial loads.22

One possible detrimental effect of DXM therapy, also demon-strated in this study, is decreased bacterial clearance from CSF.23,24

DXM decreases the permeability of the blood–brain barrier, resultingin decreased concentrations of hydrophilic antibiotics in the CSF.25

Also, high concentrations of DXM (400 µg/mL) inhibit phagocytosisby CSF leucocytes.26The clinical relevance of these findings, how-ever, has not been demonstrated.6,9

In this model of pneumococcal meningitis, CSF bacterial concen-trations at the start of therapy appeared to be more important than thetiming of DXM therapy in influencing the antibiotic-induced inflam-matory response. It is likely that there is a time beyond which DXMloses its effectiveness, but this point has not been clearly defined.

Acknowledgements

I. Lutsar was a recipient of a fellowship award from the EuropeanSociety for Paediatric Infectious Diseases, supported by Lederle-Praxis Biologicals. Part of this study was presented at the 1997 annualmeeting of Infectious Diseases Society of America (IDSA), SanFrancisco. The study has followed animal experimentation guide-lines and was approved by the Institutional Animal Care andResearch Advisory Committee of the University of Texas.

References

1. Spellerberg, B. & Tuomanen, E. I. (1994). Pathophysiology ofpneumococcal meningitis. Annals of Medicine 26, 411–18.

2. Nau, R., Zysk, G., Schmidt, H. et al. (1997). Trovafloxacin delaysthe antibiotic-induced inflammatory response in experimental pneumo-coccal meningitis. Journal of Antimicrobial Chemotherapy 39, 781–8.

3. Mustafa, M. M., Ramilo, O., Mertsola, J. et al. (1989). Modulation ofinflammation and cachectin activity in relation to treatment of experimen-tal Haemophilus influenzae type b meningitis. Journal of InfectiousDiseases 160, 818–25.

4. Friedland, I. R., Jafari, H., Ehrett, S. et al. (1993). Comparison ofendotoxin release by different antimicrobial agents and the effect oninflammation in experimental E. coli meningitis. Journal of InfectiousDiseases 168, 657–62.

5. Fischer, H. & Tomasz, A. (1984). Production and release of pepti-doglycan and wall teichoic acid polymers in pneumococci treated withbeta lactam antibiotics. Journal of Bacteriology 157, 507–13.

6. McIntyre, P. B., Barkey, C. S., King, S. M. et al. (1997). Dexame-thasone as adjunctive therapy in bacterial meningitis: a meta-analysis ofrandomized clinical trials since 1988. Journal of the American MedicalAssociation 278, 925–31.

7. Täuber, M. G., Khayam-Bashi, H. & Sande, M. A. (1985). Effectsof ampicillin and corticosteroids on brain water content, cerebrospinalfluid pressure, and cerebrospinal fluid lactate levels in experimentalpneumococcal meningitis. Journal of Infectious Diseases 151, 528–34.

8. Tuomanen, E., Hengstler, B., Rich, R. et al. (1987). Nonsteroidalanti-inflammatory agents in the therapy for experimental pneumococcalmeningitis. Journal of Infectious Diseases 155, 985–90.

9. de Gans, J. & van de Beek, D. (2002). Dexamethasone in adultswith bacterial meningitis. New England Journal of Medicine 347, 1549–56.

10. Tuomanen, E., Liu, H., Hengstler, B. et al. (1985). The induction ofmeningeal inflammation by components of the pneumococcal cell wall.Journal of Infectious Diseases 151, 859–68.

by guest on February 19, 2012http://jac.oxfordjournals.org/

Dow

nloaded from

Dexamethasone in pneumococcal meningitis

655

11. Dacey, R. G. & Sande, M. A. (1974). Effect of probenecid oncerebrospinal fluid concentration of penicillin and cephalosporin derivates.Antimicrobial Agents and Chemotherapy 6, 437–41.

12. Flick, D. A. & Gifford, G. E. (1984). Comparison of in vitro cell cyto-toxic assays for tumor necrosis factor. Journal of Immunological Methods68, 167–75.

13. Stuertz, K., Schmidt, H., Eiffert, H. et al. (1998). Differentialrelease of lipoteichoic and teichoic acids from Streptococcus pneumo-niae as a result of exposure to β-lactam antibiotics, rifamycins, trova-floxacin, and quinupristin–dalfopristin. Antimicrobial Agents andChemotherapy 42, 277–81.

14. Pfister, H.-W., Fontana, A., Täuber, M. G. et al. (1994). Mecha-nism of brain injury in bacterial meningitis: workshop summary. ClinicalInfectious Diseases 19, 463–79.

15. Tuomanen, E. I., Austrian, R. & Masure, H. R. (1995). Pathogene-sis of pneumococcal infection. New England Journal of Medicine 332,1280–4.

16. Täuber, M. G., Shibl, A. M., Hackbarth, C. J. et al. (1987). Anti-biotic therapy, endotoxin concentration in cerebrospinal fluid, and brainedema in experimental Escherichia coli meningitis in rabbits. Journal ofInfectious Diseases 156, 456–62.

17. Winter, A. J., Comis, S. D., Osborne, M. P. et al. (1998). Ototoxicityresulting from intracochlear perfusion of Streptococcus pneumoniae inthe guinea pig is modified by cefotaxime or amoxycillin pretreatment.Journal of Infection 36, 73–7.

18. Lebel, M. H. & McCracken, G. H., Jr (1989). Delayed cerebro-spinal fluid sterilization and adverse outcome of bacterial meningitis ininfants and children. Pediatrics 83, 161–7.

19. Arditi, M., Mason, E. O., Bradley, J. S. et al. (1998). Three-yearmulticenter surveillance of pneumococcal meningitis in children: clinicalcharacteristics, and outcome related to penicillin susceptibility and dexa-methasone use. Pediatrics 102, 1087–97.

20. Nakamura, H., Mizushima, Y., Seto, Y. et al. (1985). Dexametha-sone fails to produce antipyretic and analgesic actions in experimentalanimals. Agents and Actions 16, 542–7.

21. Bingen, E., Lambert-Zechovsky, N., Mariani-Kurkdjian, P. et al.(1990). Bacterial counts in cerebrospinal fluid of children with meningitis.European Journal of Clinical Microbiology and Infectious Diseases 9,278–81.

22. Feldman, W. E., Ginsburg, C. M., McCracken, G. H., Jr et al.(1982). Relation of concentrations of Haemophilus influenzae type b incerebrospinal fluid to late sequelae of patients with meningitis. Journal ofPediatrics 100, 209–12.

23. Schaad, U. B., Kaplan, S. L. & McCracken, G. H. (1994). Steroidtherapy for bacterial meningitis. Clinical Infectious Diseases 20, 685–90.

24. Cabellos, C., Martinez-Lacasa, J., Tubau, F. et al. (2000). Evalua-tion of combined ceftriaxone and dexamethasone therapy in experimentalcephalosporin-resistant pneumococcal meningitis. Journal of AntimicrobialChemotherapy 45, 315–20.

25. Lutsar, I., McCracken, G. H. & Friedland, I. R. (1998). Antibioticpharmacodynamics in cerebrospinal fluid. Clinical Infectious Diseases27, 1117–29.

26. Weitbrecht, W. U. (1980). Influence of milieu and dexamethasoneon in vitro phagocytosis of cerebrospinal fluid phagocytes. PathologicalResearch and Practice 167, 393–9.

by guest on February 19, 2012http://jac.oxfordjournals.org/

Dow

nloaded from