Clinical characteristics and therapeutic outcomes of nosocomial super-infection in adult bacterial meningitis

Super-infection in adult bacterial meningitis (ABM) is a condition in which the cerebrospinal fluid (CSF) grows new pathogen(s) during the therapeutic course of meningitis [1, 2]. This uncommon situation is rarely examined solely in overview studies of ABM [3, 4]. In ABM management, early and appropriate antibiotic use is an important strategy for better outcomes [5–7]. However, super-infection may cause a dilemma in antibiotic choice because starting a new antimicrobial regimen for this complicated condition during the therapeutic course of ABM is a difficult decision. To better delineate the clinical characteristics of super-infection in ABM, this study analyzed the clinical characteristics and therapeutic outcomes of ABM patients with super-infection.

The medical and microbiological records for CSF of all adult patients with bacterial meningitis admitted to Chang Gung Memorial Hospital (CGMH)-Kaohsiung, a 2482-bed acute-care teaching hospital and the largest tertiary care center in southern Taiwan, from January 2001 to June 2010 (9.5 years) were retrospectively reviewed. The hospital's Ethics Committee approved the study (IRB 99-1897C).

Diagnosis of ABM was based on: i) age ≥17 years; ii) positive CSF culture in patients with clinical presentation of acute meningitis (i.e. fever, altered consciousness, or seizure); and iii) at least one of the following CSF parameters: leukocyte count >0.25 × 10⁹/L with predominant polymorphonuclear cells; CSF lactate concentration >3.5 mmol/L; glucose ratio (CSF glucose/serum glucose) <0.4; or CSF glucose concentration <2.5 mmol/L if no simultaneous blood glucose was determined [3, 4]. "Super-infection" in ABM was defined as a condition wherein CSF grew new pathogen(s) during the therapeutic course of an existing bacterial meningitis [1, 2].

For this study, ABM was classified as either nosocomial or community-acquired. Nosocomial meningitis was defined as positive bacterial infection not present on hospital admission, clinical evidence of infection no sooner than 48 hours after admission, or clinical evidence of meningitis within a short period of time, i.e. within one month after discharge from the hospital where the patient received an invasive procedure, especially a neurosurgical one. Otherwise, the patient was considered as having "community-acquired" infection.

Patients who developed meningitis related to head trauma with skull fractures or neurosurgical procedures were classified as "post-neurosurgical" meningitis. Otherwise, patients who demonstrated no clear distinctive disease characteristics and who had not undergone any invasive procedures were classified as "spontaneous" meningitis. "Mixed-infection" was defined as at least two bacterial organisms isolated concomitantly from CSF culture [4, 8].

During the study period, intravenous administration of vancomycin with either a 3^rd - or 4^th- generation cephalosporin (e.g. ceftriaxone, ceftazidime, or cefepime) were the initial empiric antibiotics used to treat adult patients with clinical evidence of bacterial meningitis. Further antibiotic adjustment was guided by results of pathogen identification and antibiotic susceptibility tests. Antibiotic susceptibility of the isolated pathogen was tested using the Kirby-Bauer diffusion method (BBL, Muller-Hinton II agars; Becton Dickinson Microbiology Systems, Cockeysville, MD) while bacterial susceptibility to antimicrobial agents was determined according to criteria of the National Committee for Clinical Laboratory Standards (CLSI). The therapeutic results were evaluated by using Glasgow Outcome Scale (GOS) [9]: score 1 = death, score 2 = persistent vegetative state, score 3 = severe disability, score 4 = moderate disability, score 5 = good recovery. Scores of 1-3 were classified as "poor prognosis" while scores of 4 and 5 were "good prognosis".

For comparative analysis, the patients were divided into fatal and non-fatal groups. Data such as sex, type of infection, underlying conditions, and clinical manifestations between these two patient groups were analyzed by Fisher's exact test. Age between the two groups was analyzed by Mann-Whitney U test.

There were 27 episodes of super-infection in the 21 ABM patients. The initial and subsequently implicated pathogens are listed in Table 1. Eight patients (Cases 1-8) were converted from Gram-positive (G(+)) pathogen infection to Gram-negative (G(-)) pathogen infection and two (Cases 9 and 10) were converted from G(-) pathogen to G(+) pathogen infection. Six patients (Cases 11-16) were converted from G(-) pathogen infections to another G(-) pathogen infection. Cases 17-21 had other converted states of implicated pathogens, while Case 18 had a more complicated conversion in different CSF cultures with different combinations of implicated pathogens noted in each culture.

Of the 27 episodes of super-infection, the time interval from finding the previous pathogen to detecting a new pathogen, i.e. appearance of super-infection, ranged from 3 to 90 days. Changes in clinical and laboratory evidence of super-infection were noted in 20 of the 27 episodes. Recurrent fever was the most common and found in 17 episodes (Cases 1-3, 5-8, 10, 11, 14-16, 18, 20), The other new presentations were CSF pleocytosis in 5 episodes (Cases 1, 3, 7, 19, 20), peripheral leukocytosis in 3 episodes (Cases 1, 13, 20), appearance of hydrocephalus in 3 episodes (Cases 2, 14, 19), altered consciousness in 2 episodes (Cases 5, 14), seizure in 1 episode (Case 5), scalp wound infection in 1 episode (Case 18) and appearance of ventriculitis in 1 episode (Case 19). The other 7 episodes without new clinical presentations were detected by regular CSF check-up studies.

There is an increasing number of ABM patients with preceding neurosurgical condition, according to previous epidemiologic studies [4, 14–16]. A preceding neurosurgical state is associated with mixed, relapsing, or recurrent infectious state of ABM [3, 4, 8]. In the current study, all 21 ABM patients with super-infection had a neurosurgical condition as the preceding event and all had undergone a procedure involving the insertion of a neurosurgical device during the entire clinical course. This may imply that neurosurgical device insertion is an important event in the development of super-infection. Aside from this pre-condition, 69% (18/26) of the super-infection episodes occurred despite an antibiotic-treatment course for bacterial meningitis longer than two weeks (Table 1). Thus, the occurrence of super-infection during ABM treatment may also suggest an inadequate infectious control program of the hospital.

Due to the preceding neurosurgical condition and existing bacterial meningitis, the clinical presentation of super-infection in ABM may be obscured. In the 27 episodes of the 21 ABM patients in this study, new evidence of meningitis is noted in 74.1% (20/27). Recurrence of fever was the most common, occurring in 63.0% (17/27) of the episodes. Other new evidences of meningeal infection included CSF pleocytosis, peripheral leukocytosis, and altered consciousness. All of these presentations are not unique and can be found in other infectious states, including central nervous system infections [6]. Therefore, a high index of suspicion is important because thus far, only repeat CSF follow-up studies and bacterial culture are the most important steps to confirm the diagnosis of ABM with super-infection.

Staphylococcal species were the most common G(+) pathogens implicated in both initial ABM and ABM with super-infection in the 21 study patients (Table 1). This finding is consistent with previous reports [17–19] that have shown the high prevalence of staphylococcal infection in this specific group of patients. Most of the implicated staphylococcal pathogens shown in this study were methicillin-resistant. However, the presence of these resistant G(+) strains is not a therapeutic problem because intravenous vancomycin is one of the empiric antibiotics for ABM treatment in the hospital [4, 16].

It is known that early and appropriate antimicrobial use is of strategic importance in managing ABM [4, 5, 7]. Early diagnosis of super-infection in the therapeutic course is also important because most patients need an adjustment of antimicrobial agents after identification of the newly-implicated pathogens (Table 1). Such adjustments are important in Cases 1-10 because of the conversion from G(+) to G(-) bacterial pathogens in Cases 1-8 and from G(-) to G(+) pathogens in Cases 9 and 10.

Another important therapeutic consideration in antibiotic use is the emergence of multi-drug resistant bacterial strains in G(-) ABM [11–13, 20–25]. This problem is notable especially in patients with underlying G(-) ABM and a super-infection caused by another G(-) pathogen. Case 12 had ESBL-Proteus mirabilis infection, Cases 13 and 14 had multi-drug resistant (MDR) Acinetobacter spp. infections, and Case 16 had ceftazidime- and cefepime-resistant Pseudomonas aeruginosa infection. All four cases needed antibiotic adjustments for the corresponding newly-cultured bacterial pathogens. The emergence of Stenotrophomonas maltophilia as the implicated pathogen was noted in Cases 2, 11, and 18. Its emergence has caused a therapeutic challenge because it is usually not susceptible to traditional antibiotics used in ABM treatment [24, 25]. Presently, trimethoprime-sulfamethoxazole or quinolones are the antimicrobial agents of choice [23, 24].

In this study, another observed problem is the change of antibiotic susceptibilities during the therapeutic course of super-infection by G(-) pathogens (Table 2). MDR Stenotrophomonas maltophilia-strain emerged in Case 2 and meropenem-resistant A. Baumannii strain in Case 3. Other resistant strains that emerged during the therapeutic course of super-infection were ceftazidime-resistant A. baumannii in Case 6, ceftazidime-resistant Pseudomonas aeruginosa in Cases 9 and 12, and ceftazidime-resistant Enterobacter cloacae in Case 20. All of these changes in antibiotic susceptibilities required an adjustment of antimicrobial agents. In the super-infection of Case 18, anaerobic pathogens appeared and metronidazole was administered thereafter.

The appearance of anaerobic pathogens may also pose a therapeutic challenge because metronidazole is not a regular antimicrobial agent used in ABM treatment [3–7]. Other complicated super-infections in ABM were noted in Cases 11 and 18, in which Candida spp was the implicated pathogen. This uncommon super-infection was also noted in the report of Gelfand et al. [2]. Chronic use of broad spectrum antimicrobial agents is an important factor known to encourage the colonization of Candida organisms. Antifungal agents should be used for such an uncommon super-infection in treating ABM.

With an effort of management, the therapeutic result of super-infection ABM showed that 95.2% (20/21) of the involved patients belonged to the poor prognosis group. The mortality rate of this specific group of ABM patients is high (33.3%, 7/21) and survivors also have a high incidence of severe morbidity. Because most patients with super-infection ABM have both a preceding neurosurgical event and a pre-existing bacterial meningitis, it is difficult to assess the therapeutic outcome of this uncommon infectious state.

Super-infection in ABM is usually seen in patients with preceding neurosurgical event, including insertion of an external ventricular drainage device. Fever recurrence is the most common presentation and most new evidences are non-specific. Repeat CSF culture study is needed for diagnostic confirmation. The occurrence of super-infection in ABM has a great impact on the ABM treatment because a high percentage of the implicated bacteria strains are non-susceptible to common antibiotics used and early detection is important for adjusting the antibiotic regimen. Infections by unusual pathogens like anaerobes and fungi may appear and these deserve special attention because the choice of antimicrobials will be different. Because the pre-existing neurosurgical condition is complicated by super-infection, prognosis of super-infection in ABM remains poor. The interpretation of clinical data shown in this study is limited by its small case series and the retrospective data collection in a single hospital-based study. Further large-scale study is needed for better delineation of the presentation and prognosis.