Cerebrospinal Fluid Cachectin/Thmor Necrosis Factor-a and Platelet-Activating Factor Concentrations and Severity of Bacterial Meningitis in Children Moshe Arditi, Kirk R. Manogue, Michael Caplan, and Ram Yogev
From the Department of Pediatrics, Northwestern University Medical School, Divisions of Infectious Diseases and Neonatology, Children's Memorial Hospital, Chicago, Illinois, and Laboratory of Medical Biochemistry, Rockefeller University, New lVrk City
The morbidity and mortality associated with bacterial meningitis have remained distressingly high decades after the introduction of antibiotics , and the advent of newer, more potent antibiotics has not substantially improved the outcome of this disease . One explanation is that the pathologic consequences of the disease beyond the leptomeninges, but within the central nervous system (CNS) , progress despite bacteriologic cure [3, 4]. During the course of bacterial meningitis, bacterial cell wall products including peptidoglycan, teichoic acid, or endotoxin (lipopolysaccharide, LPS) which are generated in vivo or released during antibiotic therapy [5, 6], induce an intense host inflammatory response in the subarachnoid space . The resulting inflammatory response includes not only the well-known cytochemical abnormalities but also activation of the complement cascade , increase in arachidonate metabolism , release of a complex network of inflammatory cytokines and other mediators [9-12], and activation of granulocytes and platelets [13, 14]. The specific pathophysiologic changes leading to cerebral dysfunction and damage during bacterial meningitis [15-17] are most probably induced by both bacterial products and the
Received 6 October 1989; revised 24 January 1990. Presented in part: annual meeting of the Society for Pediatric Research, Washington, DC, May 1989. Informed written consent was obtained from parents of patients enrolled; the study was approved by the hospital's Institutional Review Board. Reprints and correspondence: Dr. Moshe Arditi, Children's Memorial Hospital, Division of Infectious Diseases, 2300 Children's Plaza, Chicago, IL 60614. The Journal of Infectious Diseases 1990;162:139-147 © 1990 by The University of Chicago. All rights reserved. 0022-1899/90/(j20l-0022$01.00
host inflammatory response [18, 19] and are well underway by the time the diagnosis of bacterial meningitis is made. The induction and amplification of these host inflammatory responses, to control locally the infectious process, may actually exert a destructive effect on the CNS and thus may contribute to the morbidity and mortality of meningitis [18-20]. The pathophysiologic effects of LPS are mediated by endogenous factors, such as cachectin/tumor necrosis factor-a (TNFa), interleukin-l (lL-l), and platelet-activating factor (PAF) [21-26]. Although LPS-host interactions and the systemic effects of such cytokines have been extensively studied [22, 27-35], the role of locally released cytokines and other mediators has remained unclear. Recent studies have shown that local overproduction of cytokines, especially if prolonged, sustain the inflammatory response and induce consequences potentially detrimental to the host [36-39]. Evidence obtained from experimental meningitis suggests that components ofbacterial cell wall stimulate local production and release of inflammatory mediators such as TNFa and IL-l in the cerebrospinal fluid (CSF) [10, 11, 14, 40]. The complex interactions among the host and bacterial determinants of the CNS damage in bacterial meningitis are incompletely defined, and further understanding and identification of these interactions are essential before significant improvement in prognosis can be expected. Thus, the goals of the current study were (1) to measure TNFa and PAF concentrations in CSF of children with bacterial meningitis and TNFa in CSF of children with viral meningitis; (2) to analyze the quantitative relationships among quantities or concentrations of CSF bacteria, TNFa, PAF, and LPS and other inflammatory indices such as white blood cells (WBC), lac-
Downloaded from http://jid.oxfordjournals.org/ at Indiana University Library on July 7, 2015
In prospective studies, tumor necrosis factor (TNFa) was detected in cerebrospinal fluid (CSF) of 33 of 38 children with bacterial meningitis (BM) but in none of 15 with viral meningitis/encephalitis (P < .001). BM CSF TNFa «35 to >25,500 pg/ml) correlated with CSF bacterial density (P < .01), CSF protein (P < .001), endotoxin (LPS) in gram-negative disease (P < .01), and consecutive febrile hospital days (P < .(01); initial CSF TNFa >1000 pg/ml was associated with seizures (P < .05). Only 5 children with BM (13%) had detectable plasma TNFa activity on admission. A higher proportion who died had detectable plasma TNFa activity compared with survivors (3/4 vs. 2/34, P < .005). Platelet-activating factor (PAP) in CSF was higher in 19 children with Haemophilus influenzae meningitis than in 17 controls (P < .01) and correlated with bacterial density (P < .01), CSF LPS (P < .01), CSF TNFa levels (P < .01), and the Herson-Todd severity score (P < .01). Elevated CSF TNFa and PAF are often present in children with BM and are associated with seizures and severity of disease. Detectable CSF TNFa appears to distinguish BM from viral meningitis.
Arditi et al.
tate, and protein; and (3) to determine whether the initial CSF or blood concentrations of TN Fa or PAF correlate with clinical presentation or with outcome.
Materials and Methods
aged. The lower limit of detection of endotoxin was 0.01 ng/mI. Intraand interassay variability were ~1O% of the mean. TNFa assays. TNFa was measured in all CSF and plasma samples using a sandwich-type enzyme-linked immunosorbent assay (ELISA) described previously , with a typical lower limit of detection of 35 pg/ml. Equivalent concentrations of recombinant human TNFa were determined for study samples by interpolation of the recombinant human TNFa standard curve run on each assay plate (25,500-35 pg/ml in a 3x dilution series). All samples were run in triplicate with recombinant human TNFa and monoclonal anti-TNFa antibodies (gifts of Chiron, Emeryville, CA). The intraassay variability is typically 10 cells/mm') with a predominance of polymorphonuclear leukocytes (PMNL), or any cell count associated with positive CSF Gram's stain or bacterial latex particle agglutination, were enrolled. Children who had received intravenous antibiotics before their initial lumbar puncture and those whose CSF remained sterile were excluded from our analysis. CSF and serum/plasma samples were collected on admission from all children and stored at -70°C for later assay. A second lumbar puncture was performed in five children within 4 h and in seven children 5-48 h after initiation of intravenous antibiotic therapy (ceftriaxone, 10 mg/kg initialloding dose, 50 mg/kg every 12 h thereafter). CSF WBC, glucose, protein, lactate, and lactic dehydrogenase (LDH) concentrations were determined by routine methods. CSF bacterial concentrations (colony-forming units per milliliter) were determined from overnight incubation of CSF serially diluted in PBS and plated on appropriate culture media. The scoring system of Herson and Todd  was used on admission to assess the clinical severity of Haemophilus injluenzae meningitis. The number of consecutive febrile hospital days (a day with at least one rectal temperature >38.l°C) was recorded for each patient. Hearing was assessed by auditory brain stem responses (ABR) or age-appropriate behavioral hearing tests on all patients at hospital discharge and again at 6 weeks after discharge if their first ABR was abnormal. Control samples. Control CSF and plasma samples were obtained from three groups of children. The afebrile controls were 10 hydrocephalic children (aged 3-60 months; median, 15) admitted for suspected ventriculoperitoneal shunt malfunction that required extraventricular drainage. All 10 CSF samples were sterile, had 0 WBC/mm3 , and had normal protein and glucose concentrations. The febrile controls were 10 children (aged 1-36 months; median, 8) who were evaluated in the emergency room for high fever and whose CSF findings were completely normal, including sterile cultures. The third group (aseptic, presumed viral, meningitis or encephalitis) was 15 children (aged 1-26 months; median, 6) whose CSF findings were consistent with nonbacterial disease (CSF pleocytosis with sterile bacterial CSF and blood cultures in the absence of previous antibiotic therapy). Eight had culture-proven enteroviral meningitis, 5 had varicella encephalitis, 1 had measles encephalitis, and 1 had presumed herpes encephalitis. Endotoxin assay. A commercially available quantitative chromogenic limulus lysate assay (Whittaker M.A. Bioproducts, Walkersville, MD) was used to determine the total CSF endotoxin level as previously described . A standard curve was created using reference Escherichia coli 0111:B4 endotoxin with each assay. Final endotoxin concentrations were expressed as the amount of standard E. coli 0111:B4 endotoxin producing activity equivalent to that of the sample. All samples were run in triplicate and the values aver-
JID 1990;162 (July)
CSF TNFa and PAF in Meningitis
JID 1990;162 (July)
Table 1. Admission TNFa levels in CSF of 38 patients with bacterial meningitis.
Haemophilus injluenzae Streptococcus pneumoniae Neisseria meningitidis Escherichia coli Streptococcus agalactiae Total
22 9 3 2 2 38
TNF detected, >35 pg/ml, no. (%) 19 8 2 2 2 33
(86) (89) (66) (100) (100) (87)
Geometric mean, pg/ml