Introduction

The treatment of bacterial meningitis offers unique challenges to ensure appropriate outcomes. After obtaining blood cultures, and cerebrospinal fluid (CSF) via lumbar puncture, empiric antimicrobial therapy should be initiated to treat the likely infecting pathogens based on the patient’s age and other predisposing factors [1]. In almost all age groups, the choice of empiric antimicrobial therapy should include treatment for meningitis caused by Streptococcus pneumoniae. Based on current knowledge of in vitro susceptibility patterns, the approach to therapy of pneumococcal meningitis must include consideration that antimicrobial resistance is likely. In the past, pneumococci were uniformly susceptible to penicillin with a minimal inhibitory concentration (MIC) ≤ 0.06 μg/mL. However, numerous reports throughout the world, including the United States, have documented strains of pneumococci that are of intermediate susceptibility (MIC range of 0.1–1 μg/mL) or highly resistant (MIC ≥ 2.0 μg/mL) to penicillin, as a result of alteration in the structure and molecular size of penicillin-binding proteins. Although the Clinical and Laboratory Standards Institute recently changed the in vitro susceptibility breakpoints for pneumococcal isolates in patients with meningitis as either susceptible (MIC ≤ 0.06 μg/mL) or resistant (MIC ≥ 0.12 μg/mL) to penicillin [2], previous breakpoints are more useful for making decisions on use of specific antimicrobial therapy. Resistance has been reported in several pneumococcal serotypes, but the majority of resistance is seen in serotypes 6, 14, 19, and 23. Although rates of pneumococcal meningitis among children and adults have been reported to decrease since introduction of the heptavalent pneumococcal conjugate vaccine [3•], the recent increase in meningitis cases caused by serotypes not in the vaccine (specifically 19A, 22F, and 35B) is concerning and has included some strains that are not susceptible to many antimicrobial agents.

Surveillance studies from the United States have demonstrated that the prevalence of pneumococcal strains that are nonsusceptible to penicillin ranges from 25% to 50% [4], with even higher rates in many regions of the world. Although isolation of resistant organisms has been reported more commonly in patients with certain risk factors, nonsusceptible strains have also been isolated in patients without comorbid conditions [5]. This article reviews the general principles of antimicrobial therapy for treatment of bacterial meningitis, and presents data for use of specific antimicrobial agents for the treatment of pneumococcal meningitis.

Principles of Antimicrobial Therapy

Several factors influence the choice of a specific antimicrobial agent in the treatment of patients with bacterial meningitis [6–8, 9•], some of which are discussed below. The first factor is the penetration of the antimicrobial agent across the blood-brain barrier (BBB) and into the CSF. β-Lactam antibiotics (eg, penicillin) generally penetrate poorly into CSF (ie, about 0.5–2% of peak serum concentrations) when the BBB is normal. In the presence of meningeal inflammation, in which inflammatory cytokines act to separate the intercellular tight junctions and increase numbers of pinocytotic vesicles in the endothelial cells of the BBB, drug entry into CSF is enhanced. Antimicrobial entry into CSF is also enhanced by drugs with a high lipid solubility, low molecular weight, low degree of protein binding in serum, and low degree of ionization at physiologic pH. CSF penetration of antibiotics decreases as inflammation subsides, such that appropriate dosages of parenteral antimicrobial therapy should be continued throughout the course of therapy.

A second factor concerns the bactericidal activity of an antimicrobial agent in purulent CSF [6–8, 9•]. Because bacterial meningitis is an infection in an area of impaired host resistance, rapid bacterial killing is necessary. Studies in experimental animal models have demonstrated that rapid killing only occurs when CSF concentrations of β-lactams or aminoglycosides exceed the minimal bactericidal concentration (MBC) by 10-fold to 20-fold. The bactericidal activity of some antimicrobial agents may also be reduced by accumulation of lactate in CSF leading to reduced pH, elevated CSF protein concentrations which may bind free drug, removal of some drugs by an active transport mechanism that exists in the choroid plexus, or conversion of drugs to inactive metabolites.

A final factor relates to pharmacodynamics, which specifically refers to the antimicrobial effect of drug concentrations at a particular site of infection over time [6–8, 9•]. For agents that kill bacteria by time-depending killing, the bactericidal activity depends upon the time (T) that its concentration exceeds the MBC (T > MBC) as a proportion of the dosing interval; a study in experimental animals with cephalosporin-resistant pneumococcal meningitis demonstrated that T > MBC was the single most important determinant of ceftriaxone efficacy [10]. The second pattern of antimicrobial killing is concentration-dependent, and is characterized by bacterial killing over a wide range of antimicrobial concentrations and a prolonged recovery (termed the post-antibiotic effect); aminoglycosides and the fluoroquinolones exhibit concentration-dependent killing, although some fluoroquinolones also demonstrate features of time-dependent killing.

Therefore, familiarity with the pharmacologic principles of the commonly used antimicrobial agents, which includes penetration of drug into the CSF and the drug’s activity in the purulent CSF, helps to determine which antimicrobial agents are appropriate for the therapy of bacterial meningitis, and may assist in defining the potential efficacy of any given antimicrobial agent in the treatment of bacterial meningitis.

Specific Antimicrobial Agents in the Treatment of Pneumococcal Meningitis

β-Lactams have been the most common class of antimicrobial agents used in the treatment of bacterial meningitis, because high systemic doses are generally well tolerated and can achieve CSF concentration that are well above the MIC of sensitive pathogens in the presence of meningeal inflammation [9•]. However, given the emergence of penicillin-resistant pneumococcal strains (see above), adequate concentrations of penicillin or ampicillin do not attain appropriate CSF concentrations following parenteral administration to eradicate strains that are not susceptible to penicillin. Therefore, penicillin is not recommended for empiric therapy of patients with suspected pneumococcal meningitis. Recommended empiric therapy is the combination of vancomycin plus a third-generation cephalosporin (either cefotaxime or ceftriaxone) [1]; this recommendation is based on data demonstrating that this combination was synergistic in an experimental rabbit model of highly penicillin-resistant pneumococcal meningitis. The efficacy of this combination, as well as that of other antimicrobial agents, is discussed in more detail below. Some of these agents have been evaluated in clinical trials, whereas others have only been evaluated in experimental animal models of infection.

Cephalosporins

The third-generation cephalosporins (ie, cefotaxime and ceftriaxone) have emerged as the agents of choice in the therapy of pneumococcal meningitis caused by strains that are of intermediate susceptibility to penicillin (MIC, 0.1–1.0 μg/mL) [1]. However, treatment failures with the third-generation cephalosporins have been reported, and pneumococcal strains resistant to these agents (MIC ≥ 2.0 μg/mL) have also been described. When the organism’s MIC to a third-generation cephalosporin is ≤1.0 μg/mL, some patients have been successfully treated with high-dose cefotaxime or ceftriaxone; one study also found that high-dose cefotaxime had reliably sufficient CSF bactericidal activity against cephalosporin-resistant pneumococci [11].

Cefepime, a fourth-generation cephalosporin with excellent in vitro activity against a variety of meningeal pathogens including penicillin-resistant S. pneumoniae, has also been studied in experimental animal models of bacterial meningitis. Its efficacy against penicillin-resistant S. pneumoniae was previously elucidated in two experimental studies using the rabbit model. One of these studies showed that the superior bacterial killing of cefepime compared with ceftriaxone was statistically significant in vivo despite similar antimicrobial activity in vitro [12]. Cefepime monotherapy was also proven to be as effective as the combination of vancomycin plus ceftriaxone against penicillin-resistant S. pneumoniae isolates with induced fluoroquinolone resistance [13]; although cefepime had similar bacterial killing rates to that of vancomycin and ceftriaxone, the fluoroquinolone-resistant strain was killed more slowly by cefepime and ceftriaxone compared to the killing of the fluoroquinolone-susceptible strain. The superior pharmacokinetics of cefepime against S. pneumoniae were also studied in hospitalized patients with external ventricular drains, in which cefepime had a higher probability than ceftriaxone of achieving 50% and 100% T > MBC in CSF against S. pneumoniae [14]; although this model was favorable for cefepime, it was derived from noninflamed meninges in patients with hydrocephalus. Cefepime has been shown to be efficacious in a study of pediatric patients with bacterial meningitis with comparable clinical efficacy when compared to ceftriaxone and cefotaxime [15].

Glycopeptides

Since the introduction of vancomycin into clinical practice, its use was limited to situations in which the infecting organism was resistant to the customarily used β-lactam agents, when therapy with a first-line agent had failed, or when a patient had a history of penicillin allergy [16]. As a result of the emergence of pneumococcal strains that are resistant to penicillin, vancomycin has gained a significant role in the treatment of pneumococcal meningitis. In experimental animal model studies, vancomycin was found to be synergistic when combined with ceftriaxone in animals with cephalosporin-resistant pneumococcal meningitis [17], and when combined with gentamicin against penicillin-resistant pneumococci in vitro and in a rabbit model of pneumococcal meningitis [18].

One important concern in the use of vancomycin is its decreased and unreliable CSF penetration in the presence of dexamethasone, which is recommended as adjunctive therapy in adult patients with suspected or proven pneumococcal meningitis [1]. In one study of 11 adult patients with meningitis caused by pneumococcal strains of intermediate susceptibility to penicillin, vancomycin was associated with clinical failure in four patients, compared to no failures in 14 subsequent patients treated with ceftriaxone [19]. In addition to the fact that the concomitant administration of dexamethasone may have been associated with poor CSF penetration of vancomycin and contributed to the negative outcome, the dosage of vancomycin was only 15 mg/kg per day, below standard recommendations. These results contrast with a second study in children with bacterial meningitis who were treated with vancomycin at a dose of 60 mg/kg per day along with ceftriaxone and dexamethasone [20]; this higher dosage resulted in acceptable CSF concentrations of vancomycin and may have contributed to the favorable results. This outcome was also demonstrated in a recent, prospective, multicenter, observational study in 14 adults with suspected pneumococcal meningitis who received vancomycin administered in a continuous infusion of 60 mg/kg per day (following a loading dose of 15 mg/kg), combined with cefotaxime and dexamethasone; mean serum and CSF concentrations of vancomycin were 25.2 μg/mL and 7.9 μg/mL, respectively, and follow-up CSF analysis revealed negative CSF cultures and marked improvement in CSF parameters in all patients [21•]. These concepts were demonstrated in an experimental rabbit model of penicillin- and cephalosporin-resistant pneumococcal meningitis, in which different dosage regimens of vancomycin (20 mg/kg per day and 40 mg/kg per day) plus dexamethasone were compared [22]. Rates of clearance of bacteria from CSF were similar with both dosages, but coadministration of dexamethasone significantly reduced the CSF vancomycin concentration and bacterial clearance from animals receiving the lower dosage. These results suggest that the effects of corticosteroids on vancomycin penetration into CSF may be circumvented by using larger vancomycin doses.

A second concern regarding use of vancomycin in patients with S. pneumoniae meningitis is the emergence of vancomycin tolerance in some pneumococcal isolates [23, 24]. Tolerance reduces the rate of death on exposure to an antimicrobial agent and allows for a resumption of growth after its removal, effectively changing the antimicrobial activity from bactericidal to bacteriostatic, with attenuated killing at the defined MIC for the given isolate. This phenomenon is now thought to be present in a significant number of pneumococcal serotypes. In one study of 215 nasopharyngeal swabs of healthy vaccinated infants and 113 pneumococcal isolates from patients with meningitis, tolerance to vancomycin was found in 3.7% of swabs and in 10.6% of invasive isolates [25]; patients with these tolerant isolates also had a worse estimated 30-day survival (46% vs 86%; P = 0.048).

Despite the concerns elucidated above, vancomycin is an important agent in the treatment of pneumococcal meningitis, and should be used in empiric regimens in appropriate dosages while awaiting isolation of the causative organism and in vitro susceptibility testing. When used for the treatment of bacterial meningitis, vancomycin should be administered to maintain serum vancomycin trough concentrations of about 15 to 20 μg/mL to achieve adequate CSF concentrations [1, 26•]; intrathecal administration may be necessary in patients not responding to parenteral administration.

Other glycopeptides have been evaluated for their efficacy in the treatment of bacterial meningitis. Teicoplanin was studied in experimental animal models of pneumococcal meningitis. When used alone in treatment of pneumococcal meningitis, it resulted in effective bacterial killing and the addition of dexamethasone did not alter this result [27]. Despite in vitro synergy, the addition of ceftriaxone did not show a significant improvement in bacterial killing in vivo.

Oritavancin has been studied in an experimental animal model of penicillin-susceptible pneumococcal meningitis and exhibited comparable bactericidal activity to ceftriaxone despite low CSF penetration (1–5%) [28]. Another experimental meningitis study in rabbits comparing treatment of disease caused by cephalosporin-resistant pneumococci with oritavancin alone or in combination with ceftriaxone revealed that combination therapy led to a more rapid decrease in bacterial concentrations, although this result was not statistically significant [29]; bacterial killing was not affected by administration of dexamethasone. Despite these positive results, however, the safety and efficacy of oritavancin in bacterial meningitis has not been defined in clinical trials.

Telavancin, a lipoglycopeptide that is a semisynthetic derivative of vancomycin, has bactericidal activity against gram-positive bacteria, including drug-resistant strains. An experimental study in rabbits with meningitis caused by penicillin-resistant pneumococci revealed CSF/MIC ratios from 30 to 63 for the pneumococcal strain, and CSF penetration was 2% through inflamed meninges [30]; the combination of vancomycin and ceftriaxone was less efficacious than telavancin against this resistant strain.

Rifampin

The various pharmacodynamic properties of rifampin, including good CSF penetration and in vitro activity against many meningeal pathogens, make it a favorable antimicrobial agent in the treatment of bacterial meningitis. Rifampin has been studied in many experimental models, usually in combination with other antimicrobial agents (eg, ceftriaxone and/or vancomycin), because of the development of rapid resistance if rifampin is used alone. In an experimental rabbit model of penicillin- and cephalosporin-resistant pneumococcal meningitis, the combination of rifampin and ceftriaxone was studied; the combination resulted in more rapid bacterial killing than ceftriaxone alone, with similar results whether or not the animals also received dexamethasone [31]. In another experimental study, rifampin—used in combination with vancomycin in a model of penicillin-resistant S. pneumoniae meningitis that included administration of dexamethasone—was able to somewhat reduce the negative effect of dexamethasone on CSF penetration of vancomycin [32], likely a result of enhanced metabolism of dexamethasone in the presence of rifampin. Two other experimental rabbit studies of penicillin- and cephalosporin-resistant pneumococcal meningitis compared the use of ceftriaxone and rifampin combination therapy to regimens that included vancomycin, and revealed that the addition of rifampin to ceftriaxone was comparable to the combination of ceftriaxone and vancomycin. One showed equal efficacy [33] and the other showed superior efficacy [34]; in the latter study, the three-drug combination (vancomycin, ceftriaxone, and rifampin) had similar efficacy to the combination of ceftriaxone and rifampin.

In patients with pneumococcal meningitis caused by penicillin-nonsusceptible strains, the addition of rifampin to vancomycin, with or without a third-generation cephalosporin, has been recommended by some authorities, although clinical data are lacking. Rifampin should be added only if the organism is susceptible in vitro, the expected clinical or bacteriologic response is delayed, or the pneumococcal isolate has a cefotaxime or ceftriaxone MIC > 4 μg/mL [1].

Carbapenems

The carbapenems (specifically imipenem, ertapenem, and meropenem) have been studied in the treatment of bacterial meningitis. Imipenem was shown to be effective in pneumococcal meningitis caused by penicillin- and cephalosporin-resistant strains and has excellent in vitro activity against many other important meningeal pathogens, although a high frequency of seizures in children receiving imipenem for meningitis has largely precluded its use for this infection [35]. Ertapenem was found to be effective in an experimental study of pneumococcal meningitis in rabbits caused by penicillin-sensitive and penicillin-resistant strains [36]; ertapenem showed sufficient CSF penetration and was successful in sterilizing the CSF, although no clinical data are available on its efficacy.

Meropenem is the carbapenem that has been most studied in patients with bacterial meningitis. It is less neurotoxic with less seizure tendency compared to imipenem. An experimental study evaluated meropenem as monotherapy in meningitis caused by cephalosporin-susceptible and cephalosporin-resistant S. pneumoniae strains using two different animals, rabbits and guinea pigs [37]; bactericidal activity was excellent against the cephalosporin-susceptible strain in both animal species and against the cephalosporin-resistant strain in guinea pigs, but therapeutic failure was observed in the rabbits inoculated with the cephalosporin-resistant strain.

Meropenem was evaluated in patients with bacterial meningitis and shown to have clinical and microbiologic outcomes similar to that of cefotaxime and ceftriaxone [38–40]; meropenem has also been used successfully in the treatment of patients with pneumococcal meningitis caused by highly penicillin- and cephalosporin-resistant strains [40, 41]. However, in one in vitro study of 20 cefotaxime-resistant pneumococcal strains, four were intermediate and 13 were resistant to meropenem, suggesting that meropenem is not a useful alternative as monotherapy for treatment of pneumococcal meningitis caused by these highly resistant strains [42].

Fluoroquinolones

Newer fluoroquinolones (eg, levofloxacin, trovafloxacin, gatifloxacin, moxifloxacin, clinafloxacin, gemifloxacin, and garenoxacin) have good CSF penetration, excellent in vitro activity against S. pneumoniae, and have been studied in the experimental animal models of pneumococcal meningitis [9•]. Moxifloxacin, in particular, has excellent CSF penetration (50–85%) and has shown efficacy in experimental animal models of pneumococcal meningitis [43, 44]. One study in healthy humans demonstrated good CSF penetration of moxifloxacin after a single oral dose of 400 mg [45], attaining a CSF concentration to achieve a satisfactory bactericidal effect against penicillin-resistant S. pneumoniae.

Several experimental studies in rabbit models have also demonstrated the synergistic effects of fluoroquinolones with other antimicrobial agents used to treat bacterial meningitis. The combination of levofloxacin and meropenem was shown to increase the efficacy of levofloxacin against penicillin-resistant pneumococci in vitro, and was more efficacious than the combination of vancomycin and ceftriaxone [46]. The combination of levofloxacin and cefotaxime was shown to have higher bactericidal activity than monotherapy, and twice that of the combination of vancomycin and ceftriaxone [47]; combination therapy almost completely diminished levofloxacin-induced resistance of the test strain. In another study, the combination of levofloxacin and ceftriaxone prevented the emergence of levofloxacin resistance in the pneumococcal strain [48].

The only multicenter randomized trial to date that has evaluated a fluoroquinolone in the treatment of bacterial meningitis compared trovafloxacin to ceftriaxone, with or without vancomycin, in 311 children from 11 countries with bacterial meningitis [49]; 27% of cases were caused by S. pneumoniae. No significant differences were found in CSF sterilization (94% for trovafloxacin vs 96% for the comparator) or clinical success (75% for trovafloxacin vs 82% for the comparator) at the end of treatment. However, reports of liver toxicity with trovafloxacin have largely limited its use, despite the fact that it showed great promise for the treatment of bacterial meningitis caused by S. pneumoniae. The other fluoroquinolones with in vitro activity against S. pneumoniae have not been evaluated in clinical trials, but use of a newer fluoroquinolone (eg, moxifloxacin), combined with a third-generation cephalosporin or vancomycin, can be considered as an alternative regimen in patients not responding to standard therapy.

Other Agents

Daptomycin, a cyclic lipopeptide agent, has potent antibacterial activity against resistant gram-positive bacteria, but there is little experimental data or clinical evidence on its use in bacterial meningitis. In a rabbit model of meningitis caused by penicillin- and quinolone-resistant S. pneumoniae, daptomycin had a CSF penetration of 6%, and demonstrated comparable efficacy when compared to the combination of cefotaxime and levofloxacin, and superior efficacy when compared to the combination of vancomycin and ceftriaxone [50]. Daptomycin is also a nonbacteriolytic agent that may be beneficial in bacterial meningitis by preventing release of proinflammatory mediators (teichoic acid and lipoteichoic acid in pneumococci) following antimicrobial-induced lysis. In a rabbit model of penicillin-resistant pneumococcal meningitis, release of [3H] choline, a main component of teichoic acid and lipoteichoic acid, was only mildly elevated with no morphologic changes in the pneumococcal cell wall after treatment with daptomycin, compared to drastic increases in [3H] choline and cell wall morphologic changes after treatment with ceftriaxone [51]. In another study comparing ceftriaxone to daptomycin in infant rats with pneumococcal meningitis, 7 of 28 animals treated with ceftriaxone had cortical brain damage compared to none of the 30 animals treated with daptomycin [52]. Despite these positive results, no data exist on the efficacy of daptomycin in patients with pneumococcal meningitis.

Linezolid, a novel antimicrobial of the oxazolidinone class, was reported to be effective in meningitis caused by penicillin-nonsusceptible S. pneumoniae. Although no clinical trials have compared linezolid to standard antimicrobial therapy for pneumococcal meningitis, an experimental rabbit study showed inferior killing rates of linezolid against penicillin-sensitive and penicillin-resistant pneumococci compared with vancomycin plus ceftriaxone despite linezolid’s good CSF penetration (∼38%) [53]; antibacterial activity was more pronounced against the penicillin-resistant than the penicillin-sensitive strain.

Conclusions

Pneumococcal meningitis carries a significant morbidity and mortality despite the availability of effective antimicrobial therapy. Vaccination has significantly reduced the incidence of pneumococcal meningitis in areas of the world where routine vaccination has been implemented, but invasive disease (including meningitis) continues to occur and infection caused by penicillin- and cephalosporin-resistant strains is a concern. Based on these trends, penicillin is not recommended as empiric therapy for presumed pneumococcal meningitis. Instead, the combination of vancomycin and a third-generation cephalosporin (either cefotaxime or ceftriaxone) should be used. Meropenem is not a good option for monotherapy for treatment of highly penicillin- or cephalosporin-resistant strains. Use of a fluoroquinolone with in vitro activity against S. pneumoniae is an option for patients not responding to standard therapy, but if used, should be combined with a third-generation cephalosporin or vancomycin given the absence of clinical data on monotherapy with one of these agents. Newer glycopeptides, daptomycin, and linezolid require further study to determine their efficacy in patients with pneumococcal meningitis.