Clinical outcome of pneumococcal meningitis during the emergence of pencillin-resistant Streptococcus pneumoniae: an observational study

Among bacterial causes of meningitis, Streptococcus pneumoniae (the pneumococcus) is associated with the highest case-fatality and is the most likely to leave survivors with permanent sequelae. For the year 2000, the WHO's Global Burden of Disease project estimated 100,000 cases of pneumococcal meningitis occurred among children <5 worldwide, with higher than 50% case-fatality [1]. Over 95% of the pneumococcal deaths occurred in developing countries.

Prior to the identification of penicillin-resistant pneumococci, penicillin therapy was the standard treatment for pneumococcal meningitis. Recommendations of the American Academy of Pediatrics and Infectious Diseases Society of America for first-line treatment of pneumococcal meningitis were changed to a 3^rd generation cephalosporin [2–4], following reports of treatment failure and pharmacokinetic studies showing inadequate penicillin concentrations in the central nervous system to treat penicillin-resistant pneumococcal infections. However, the substantial difference in cost of third generation cephalosporins in the 1990s limited their use in developing countries. In 1997, the WHO recommended third generation cephalosporins as the first-line treatment for children ≥3 months of age with suspected bacterial meningitis in developed countries versus oily chloramphenicol and penicillin in developing countries [5], creating a double-standard [6–8]. The WHO recognized that where "significant" pneumococcal resistance to penicillin had been observed, cefotaxime or ceftriaxone were recommended, while leaving individual countries to determine what level of resistant would require a change in first-line therapy. At the time of these recommendations, few data from developing countries were available on the prevalence of penicillin-resistant pneumococci or its clinical relevance. Following availability of generic ceftriaxone after 1999, WHO recommended ceftriaxone alone as first-line antibiotic therapy for suspected bacterial meningitis in all countries [9]. However, despite availability of generic ceftriaxone, actual prices of this antibiotic may remain high and limit its use for treatment of pneumococcal meningitis in some developing countries [10].

To guide decisions about treatment of bacterial meningitis at an infectious diseases referral hospital in Salvador, Brazil, we established surveillance for penicillin-resistant pneumococcal infections in December 1995. The identification of penicillin resistance among 13% of pneumococcal isolates from meningitis cases during the first three years of surveillance led to a change in first-line antibiotic treatment to ceftriaxone [11]. Because data are limited on the clinical relevance of penicillin-resistance for outcomes of pneumococcal meningitis, we analyzed data from 10 years of surveillance for pneumococcal meningitis cases before and after introduction of empiric ceftriaxone therapy to investigate whether penicillin-resistance was associated with case-fatality and adverse outcomes of pneumococcal meningitis.

We included 548 pneumococcal meningitis case patients in this analysis. Median age was 8 years (ranging from 0 to 79 years): 230 (42%) case patients were younger than 2 years of age (35% were younger than 1 year), and 6% were adults >50 years. Males comprised 64% (137/216) of cases in children <2 years, and 64% (354/548) overall. Underlying medical conditions were present in 141 (25.7%) patients and included: alcoholism (19), hypertension (16), repetition IVAS (12), chronic otitis media (12), diabetes (8), asthma (7), sickle cell anemia (6), and HIV infection (4). The presence of underlying disease was not associated with infection with penicillin-resistant pneumococci (Odds Ratio [OR] 0.6; 95% CI, 0.3-1.1]. Thirty-four patients had a history of prior hospitalization. Thirty-one patients had a history of meningitis before the episode included in the present study.

The median duration of symptoms before hospital admission was 2 days (mean: 2.9 range: 0 to 30 days). On presentation, 34% (181/533) of the patients had a history of generalized or focal seizures and 257 (48%) patients had altered mental status; of these, 32% (83/257) were in coma. The initial CSF parameters for these patients showed a mean leukocyte (WBC) count of 4,193/mm³ (range 0 to 10,000). The mean protein and glucose concentrations were 357.3 and 26.2 mg/dl, respectively. Two hundred and thirty-five (42.8%) patients were admitted in ICU.

Serotypes included in the 10-valent and 13-valent pneumococcal conjugate vaccine accounted for 271 (49%) and 371 (68%) of 548 isolates from meningitis cases, respectively, and 80 (87%) versus 87 (95%) of 92 penicillin-resistant isolates (Figure 1). All three highly penicillin-resistant isolates were serotype 14. Among children <2 years, 152 (70%) and 120 (79%) of 216 isolates were serotypes included in the 10-valent or 13-valent vaccine, respectively.

During the study period, we observed significantly declines in the proportion of pneumococcal meningitis cases treated initially with ampicillin (including 47 patients treated with ampicillin and chloramphenicol and few patients treatedwith penicillin G, χ² for trend = 155.7, p < 0.0001). Ceftriaxone alone became the first-line treatment for suspected bacterial meningitis, with vancomycin use rising and falling depending upon concerns about resistance to ceftriaxone (Figure 2). Systematic corticosteroids were administered on the day of admission to 412 (75%) of the patients. Most case patients received at least one antibiotic on the day of admission to which their isolate was susceptible (Table 1).

The overall case-fatality among hospitalized patients with pneumococcal meningitis was 37%: 57 patients (28% of deaths) died within 24 hours of admission, 55 (27%) died between 24 and 48 hours of admission, and a total of 91 patients (45%) died beyond 48 hours after hospital admission. Table 1 shows case-fatality proportions for meningitis cases stratified by age and time period. Case-fatality was highest during the first three years of the study, when empiric treatment for suspected bacterial meningitis included ampicillin (with or without chloramphenicol). However, 526 (96%) of 548 case patients received an initial antibiotic regimen to which their pneumococcal isolate was susceptible; only 22 (4%) initially received discordant therapy and of these, antibiotic therapy was changed within 48 hours of admission for 14 (64% of 22) patients. Children <5 years with penicillin-resistant infections tended to have longer duration of symptoms (median, 3 days) prior to hospitalization than those with penicillin-susceptible isolates (median, 2 days), although differences were not significant (p = 0.055). Among 7 patients with penicillin-resistant pneumococci who received discordant therapy, 4 died; 3 within 36 hours of admission and one patient 11 days after admission. One of the three patients who recovered had sequelae at discharge. Among the 10 patients who received discordant therapy and died, 6 died before 36 hours of admission; and 50% (5/10) were younger than 1 year old. Among 332 patients who received a neurological evaluation prior to discharge, 120 (36%) had a neurological deficit. The median duration of hospitalization was 15 days (range 3 to 90).

In this large series of pneumococcal meningitis cases prospectively identified over a ten year period of active surveillance at a public infectious disease referral hospital in Brazil, we found an association between case-fatality and penicillin resistance among pneumococcal meningitis cases. This association is unlikely to be due to decreased sensitivity to penicillin alone, because few patients with penicillin resistant isolates received initial penicillin treatment. The small number of patients who received discordant therapy, or antibiotics to which their isolate was resistant, resulted in low statistical power to detect an independent effect of discordant therapy on outcome. In addition, initial penicillin therapy was changed to ceftriaxone once results of antibiotic susceptibility testing showed that the isolate had reduced sensitivity to penicillin. During the study period, ceftriaxone replaced penicillin as first-line therapy for suspected bacterial meningitis, yet we did not find that initial treatment with ceftriaxone reduced case-fatality. Young children, with the highest case-fatality, were more likely to have penicillin-resistant pneumococcal isolates. Following this initial finding, the hospital purchased limited amounts of cefriaxone for empiric treatment of young children with suspected bacterial meningitis. Because the majority of deaths occurred soon after admission, initial ceftriaxone therapy may not significantly have improved survival; patients with penicillin resistant pneumococcal infections may have presented with more severe disease. Alternatively, although we controlled for age in the analysis, young age may still have confounded the association between case-fatality and infection with penicillin-resistant pneumococci.

Prior to the availability of low cost third-generation cephalosporins, developing countries faced a dilemma in dealing with the problem of rising penicillin resistance among S. pneumoniae: implement empiric therapy with cephalosporins at substantial increased cost to the health system, or maintain empiric therapy with penicillin (and chloramphenicol) [6]. Treatment of resistant pneumococcal infections with chloramphenicol and penicillin risked treatment failure and adverse outcomes, even at high concentrations of penicillin [16–18]. Pharmacokinetic studies indicated poor penetration of penicillin into cerebrospinal fluid, which resulted in concentrations below those required for sterilization of CSF for pneumococcal isolates with MIC > 0.06 μg/mL [19]. In addition, failure of chloramphenicol treatment has been described in patients infected with penicillin-resistant pneumococci [17]. In settings in which antimicrobial susceptibility testing was available, patients with penicillin-resistant pneumococcal infections could receive appropriate treatment after results of susceptibility testing were known. In practice, this strategy leads to delay in administration of appropriate treatment, which is associated with delayed sterilization of CSF and worse outcomes [20]. Delays in diagnosis have also been associated with poor outcomes in bacterial meningitis [21] and likely contributed to the high case-fatality in our setting. An alternative strategy used successfully in Papua New Guinea initiated empiric therapy with ceftriaxone and switched to a less expensive antibiotic (chloramphenicol) once isolates were identified as fully susceptible [22]. In our setting, switching from ceftriaxone to penicillin for penicillin-sensitive isolates was uncommon due to a preference for administration of ceftriaxone at 12-hour intervals versus 4- or 6-hour intervals for penicillin G or ampicillin, respectively.

Our findings underscore the importance of antimicrobial susceptibility testing, both to guide treatment as well as to provide local data on prevalence of antimicrobial resistance. Following identification in 1996 of a cluster of penicillin-resistant pneumococcal meningitis cases with high case-fatality [11], the infectious diseases hospital increased its purchase of ceftriaxone, at the time ten times the cost of penicillin therapy. Initially, ceftriaxone was prioritized for treatment of children due to higher proportions of penicillin resistant pneumococcal meningitis among children. One of the objectives of monitoring outcomes of pneumococcal meningitis cases at the study hospital was to justify the increased expense of empiric ceftriaxone therapy. However, despite evidence of increasing resistance to penicillin from several settings in Brazil [23, 24], the Brazilian Ministry of Health did not change recommended first-line therapy for suspected bacterial meningitis or pneumococcal meningitis to ceftriaxone due to concerns about costs and risk of stock-outs of the more expensive antibiotic. The prevalence of penicillin resistant infections was considered low enough not to warrant revision of national recommendations, acknowledging the need for rapid identification of resistant infections and local decisions regarding choice of empiric therapy. One result of the WHO recommendations was that Ministries of Health in developing countries were left to make their own decisions regarding levels of penicillin resistance considered acceptable for use of penicillin as first-line therapy for pneumococcal meningitis [5]. However, few developing countries had representative data on antimicrobial susceptibility patterns among invasive pneumococcal isolates.

The 2008 revision of Clinical and Laboratory Standards Institute susceptibility breakpoints for Streptococcus pneumoniae brought laboratory definitions in line with treatment recommendations [15]. Prior to the 2008 revision, invasive pneumococcal isolates with penicillin MIC > 0.06 μg/mL were classified as penicillin-nonsusceptible: penicillin MIC from 0.1 to 1.0 μg/mL was classified as intermediate and MIC > 1.0 μg/ml as resistant [25]. Observational studies suggested that high dose penicillin therapy was effective for treatment of invasive pneumococcal infections outside the central nervous system [26]. Together with the limited clinical data on outcomes of penicillin-resistant pneumococcal meningitis treated with penicillin, this classification created uncertainty regarding the treatment of pneumococcal meningitis with intermediate resistance to penicillin. The low prevalence of pneumococcal isolates with MIC > 1.0 μg/mL in studies throughout Brazil [27] was considered by the Brazilian Ministry of Health not to justify replacement of penicillin as first-line therapy for suspected bacterial meningitis.

Additional data are needed on potential benefits and risks of adjunct dexamethasone therapy for pneumococcal meningitis in developing countries, especially in settings where antibiotic resistance is prevalent. In a 2010 Cochrane review, adjunctive treatment with dexamethasone was associated with lower case-fatality in pneumococcal meningitis and lower incidence of severe hearing loss in children with meningitis due to Haemophilus influenzae, but no benefit of dexamethasone treatment was found for bacterial meninigitis patients in low-income countries [28]. One randomized, controlled trial conducted in six Latin American countries showed that treatment with oral glycerol reduced neurological sequellae in children with meningitis, suggesting alternatives to dexamethasone treatment [29]. Brazilian Ministry of Health guidelines for treatment of bacterial meningitis provide no recommendation regarding use of corticosteroids [12]. In our analysis, we observed no confounding effect of adjunct corticosteroid therapy on the association between case-fatality and penicillin resistance. Although the use of steroids was frequent among patients with pneumococcal meningitis, dexamethaxone was not used systematically and was indicated according to physician preference. Therefore, the lack of association between initial dexamethasone therapy and outcome of pneumococcal meningitis in this series should be interpreted with caution, as with other observational studies [30]. Concern has been expressed that treatment with steroids would reduce the vancomycin concentrations in cerebrospinal fluid in adults and therapeutic failures have been described [30]. In our hospital, the pneumococcal meningitis cases treated with ceftriaxone and vancomycin were mainly pediatric; most also received dexamethaxone.

The availability of low cost ceftriaxone therapy in developing countries makes the findings of this study less relevant for decisions regarding choice of empiric therapy for suspected bacterial meningitis or pneumococcal meningitis. However, surveys of essential medicines in developing countries show wide variability in the actual price of ceftriaxone and its availability in public hospitals, suggesting that penicillins may still be used for first-line treatment in some countries [10]. Pneumococcal resistance to cephalosporins needs to be monitored [31, 32], although cefotaxime nonsusceptibility has not been associated with increased mortality of pneumococcal meningitis treated with cefotaxime [33–35].

The high mortality and burden of pneumococcal meningitis among Brazilian children were among the principal reasons for the Brazilian Ministry of Health's decision to introduce pneumococcal conjugate vaccine in the routine infant immunization schedule in 2010. Brazil was one of the first countries to introduce a novel pneumococcal conjugate vaccine including 10 serotypes into its national immunization program. Based on data from this surveillance as well as laboratory-based surveillance in Brazil [36], the 10-valent vaccine formulation includes the most prevalent serotypes in pneumococcal meningitis cases in young children, including those serotypes accounting for the majority of penicillin-resistant isolates. Because the 10-valent vaccine was licensed based on immunogenicity data, its effectiveness against invasive pneumococcal disease due to vaccine serotypes, as well as its effect on potentially cross-reactive serotypes (such as 6A and 19A) needs to be determined. Antibiotic use may exert selective pressure leading to expansion of resistant clones with non-vaccine type capsular polysaccharides. Continued surveillance following vaccine introduction is warranted to monitor antibiotic resistance among invasive pneumococcal infections to guide treatment decisions.

In this observational study of pneumococcal meningitis outcomes, patients infected with pneumococci with reduced susceptibility to penicillin had higher case-fatality than patients with penicillin-sensitive isolates, even when initial antibiotic therapy included ceftriaxone. Appropriate antibiotic therapy may not have been initiated early enough in the course of infection to prevent death. Although ceftriaxone has become more available in developing countries since the time of this study, these findings are relevant where penicillin antibiotics are still used due to lack of information about penicillin resistance or limited availability of ceftriaxone. These findings support the use of third generation cephalosporin antibiotics for treatment of suspected pneumococcal meningitis even at low prevalence of pneumococcal resistance to penicillins.