Fluid Prostaglandins, Interleukin 1\g=b\,and Tumor Necrosis Factor in Bacterial Meningitis

Cerebrospinal

Clinical and Laboratory Correlations in Placebo-Treated and Dexamethasone-Treated Patients Mahmoud M. Mustafa, MD; Octavio Ramilo, MD; Xavier S\l=a'\ez-Llorens,MD; Kurt D. Olsen, Ronald R. Magness, PhD; George H. McCracken, Jr, MD \s=b\ Prostaglandins (PGs), interleukin 1\g=b\ (IL-1\g=b\), and tumor necrosis factor \g=a\ (TNF\g=a\)are likely mediators of local inflammatory reactions. We measured PGE2, PGI2, IL-1\g=b\, and TNF concentrations in paired cerebrospinal fluid (CSF) samples (on admission, CSF1, and 18 to 30 hours later, CSF2) from 80 infants and children with bacterial meningitis. Forty patients received dexamethasone sodium (0.6 mg/kg per day in four intravenous doses) and 40 received an intravenous saline placebo. In CSF1, PGE2, PGI2, IL-1\g=b\, and TNF were detected in 90%, 56%, 98%, and 71% of specimens with mean (\m=+-\SEM) concentrations of 462\m=+-\65,13276\m=+-42, and 799\m=+-\227 pg/mL, respectively. Concentrations of PGE2 correlated significantly with PGI2, IL-1\g=b\, TNF, and lactate and inversely correlated with glucose concentrations in the first CSF specimens. The PGE2, PGI2, IL-1\g=b\,and TNF were still detected in 40%, 18%, 97%, and 60%, respectively, of second CSF specimens obtained from placebo-treated patients. Compared with patients who had detectable PGI2 or TNF\g=a\concentrations in CSF2 specimens, those placebo-treated patients with no detectable PGI2 or TNF\g=a\activity in CSF2 had a lower incidence of neurological sequelae. Dexamethasone\x=req-\ treated patients had significantly lower PGE2, IL-1\g=b\, and lactate concentrations and higher glucose concentrations in CSF 18 to 30 hours later, shorter duration of fever, and a lower incidence of neurological sequelae than did placebo-treated

patients. (AJDC. 1990;144:883-887)

"Dacterial meningitis in infants and

children is associated with case fa¬ tality rates of 2% to 5% and with neuro¬ logical sequelae in 20% or more of survi¬ vors.1 Because the advent of newer, Accepted for publication December 28,1989. From the Department of Pediatrics, University of Texas Southwestern Medical Center, Dallas. Reprints not available.

active antimicrobial agents for the treatment of bacterial meningitis has not resulted in improved outcome, an understanding of the pathophysiologic more

mechanisms of central nervous system injury during bacterial meningitis has become imperative to provide new ther¬ apeutic directions. Data from experimental meningitis studies have shown that endotoxin of Haemophilus influenzae type b and cell-wall components of Streptococcus pneumoniae stimulated the release of mediators of meningeal inflammation, such as interleukin Iß (IL-lß),2 cachectin (tumor necrosis factor [TNF] a),3 and prostaglandins (PGs)," and activat¬ ed the complement cascade.7 Addition¬ ally, intracisternal inoculation of rab¬ bits with IL-lß and TNFa induced cerebrospinal fluid (CSF) pleocytosis and blood-brain barrier inj ury similar to those induced by endotoxin or live bac¬ teria8 (O.R., X.S.-L., J. Mertsola, un¬

published data, 1989). Prostaglandin F^ (PGF J has been detected in CSF of five patients with viral and bacterial meningoencephalitis.9 Furthermore, IL-lß

and TNFa have been detected in the CSF of neonates10,11 and infants and chil¬ dren12 with bacterial meningitis; the CSF IL-lß concentrations correlated significantly with indexes of meningeal inflammation and with outcome from disease in those patients.10"12 Arachidonic acid metabolites have been shown to play an important role as mediators and/or modulators of acute and chronic inflammation13,14 as well as regulators of immune functions.1" Pros¬ taglandins, more specifically, have been correlated with various inflammatory disorders.15,16 Moreover, data from ex¬ perimental arthritis studies and from joint diseases in humans have shown that PGs may be involved in the genesis

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 05/24/2015

and maintenance of synovial inflamma¬

tion." The finding of PGE2 in the CSF of

rabbits with experimental meningitis induced with S pneumoniaeu> and H in¬ fluenzae type b (M.M.M., unpublished observation, 1989) and of PGF2a in the CSF of patients with viral and bacterial meningoencephalitis9 prompted the present study in which we measured the concentrations of IL-lß, TNF, PGE2, and, instead of the highly unstable prostanoid PGI2, its stable metabolite, 6-keto-PGFle in 160 paired CSF speci¬ mens from 80 infants and children with bacterial meningitis and correlated these concentrations with each other and with clinical and CSF laboratory findings. Because half of these patients received dexamethasone therapy at the time of diagnosis, we evaluated the ef¬ fects of dexamethasone administration on the concentrations of PGs, IL-lß, and TNF in CSF specimens obtained 18 to 30 hours after dexamethasone and antibiotic therapy. PATIENTS AND METHODS Between May 1984 and April 1988, a total of 260 infants and children with cultureproved bacterial meningitis treated at Chil¬ dren's Medical Center or Parkland Memorial Hospital, Dallas, lex, were enrolled in three

prospective, double-blinded, placebo-con¬ trolled studies of dexamethasone therapy.

protocols for those studies were ap¬ proved by the University's Institutional Re¬

The

view Board and informed, written consent obtained for all study patients. The re¬ sults of these studies and the salutary effects of dexamethasone treatment of infants and children with bacterial meningitis have been described elsewhere.18,19 Aliquots of CSF ob¬ tained at the time of diagnosis (CSF,) and again 18 to 30 hours later (CSF2) were avail¬ able from 80 of these patients, 50 of whom were enrolled in the third meningitis study and were not part of our previously published study of IL-lß and TNF.12 Data derived from

was

these 80 patients are the basis of this report. The CSF samples were immediately centrifuged and frozen at 70°C within 15 minutes of the lumbar puncture. The neurological and audiological evaluations performed on all pa¬ tients during hospitalization and on repeated follow-up examinations have been previously -

described.18,19

PGI2 (Prostacyclin) Assay Samples of CSF (0.1 mL) were assayed for the stable metabolite of prostacyclin, 6-ketoPGFla, using a previously described and vali¬ dated radioimmunoassay.20,21 The sensitivity of the assay was consistently less than or equal to 2 pg per tube. Intra-assay and interassay coefficients of variation at 250 pg/mL and 1000 pg/mL were 6.7% and 11.3% and 8.7% and 12.5%, respectively.

PGE2 Assay The assay for PGE2 was the same as that described above for the 6-keto-PGFlo except that duplicate samples were mixed with 0.1 mL of antiserum (1:14 000 titer) and 0.1 mL of 16 500 disintegrations per minute of tritiated-hydrogen-PGE2 (New Research Prod¬ uct, Boston, Mass). Intra-assay and interassay coefficients of variation were 10.8% and 13.8% at 250 pg/mL and 7% and 17.7% at 1000 pg/mL, respectively. The sensitivity of this assay was 2 pg per tube.

IL-1 ß Assay

Cerebrospinal fluid samples were assayed for IL-lß using an enzyme-linked immunosorbent assay (Cistron Biotechnologies, Pine Brooks, NJ).22 The lower limit of sensitivity of the assay was 20 pg/mL at the 95% confi¬ dence level (CI).

Cytolytic Assay for TNF Tumor necrosis factor activity was deter¬ mined by modification of a previously de¬ scribed cytotoxic assay.23 Equivalent concen¬ trations of human recombinant TNF (rTNF) were determined for experimental samples by interpolation of the human rTNF stan¬ dard curve run simultaneously (0.1 pg/ mL -1 µg/mL). Using this assay, we could reliably detect as little as 10 pg of rTNF/mL. The samples were assayed in quadruplicate, and an SD within 10% of the mean was ob¬ served. The L929 cells are unaffected by ILlß. Representative CSF specimens contain¬ ing TNF were preincubated with anti-TNFa antibody, which, when tested in parallel with the CSF samples before incubation, showed complete elimination of the cytotoxicity. Tu¬ mor necrosis factor is stable for long periods when stored at -70°C.

Measurement of CSF Lactate, Protein, and Glucose Concentrations Cerebrospinal fluid samples were ana-

lyzed for

lactate with a kinetic enzymatic method that used the reaction of lactate with oxidized nicotinamide-adenine dinucleotide ß in the presence of lactate dehydrogenase to produce the reduced form of nicotinamideadenine dinucleotide and pyruvate. The pro¬ duction of the reduced form of nicotinamideadenine dinucleotide was quantitatively monitored at a 340-nm absorbance on a Gil¬ ford 2600 spectrophotometer (Gilford, Oberlin, Ohio) with a deterium source. The preci¬ sion of the method was tested by running a known standard of 40 mg/dL (Sigma UV test 826-UV, Sigma, St Louis, Mo) 10 times dur¬ ing the course of 1 day. The coefficient of variation was 2.2%. Protein concentrations were determined using the turbidity method. The turbidity produced by sulfosalicylic acid when added to a solution containing protein is proportional to the concentration of the protein in the solution. The turbidity was measured using a spectrophotometer set at 420 nm. Sigma pro¬ tein standard (No. P-7656, Sigma) as a con¬ trol was assayed with each run. Glucose con¬ centrations were determined using a commercially available colorimetrie test (Ek-

tachem, Kodak, Rochester, NY).

Statistical Analysis The strength of the relationship between PGI2, PGE2, IL-lß, and TNF concentrations

and duration of illness, duration of fever, and CSF concentrations of white blood cell (WBC) counts, glucose, lactate, and protein was as¬ sessed with Spearman's rank correlation coef¬ ficients. Repeated measures of analysis of variance (ANOVA) on log-transformed data were used to examine the effects of dexameth¬ asone treatment on CSF concentrations of nu¬ merical variables and 2 contingency analysis (or Fisher's Exact Test) to compare propor¬ tions with detectable activity. For the purpose of analysis, PGE2, PGI2, IL-lß, and TNF con¬ centrations below the lower limit of sensitivity of the assay were considered equal to 0. All numerical values were given as mean (±SEM). Pss.10 was considered significant when repeated measures of ANOVA were used. For other statistical tests P=s.05 was considered significant.

RESULTS Paired CSF specimens were available from 80 of the original 260 study pa¬ tients. There were 50 male and 30 fe¬ male patients aged 3 to 44 months. For¬ ty-five patients were white, 27 black, and 8 Hispanic. The age, sex, race, and clinical characteristics on admission of the 80 patients in this study were simi¬ lar to those of the entire 260 infants and children enrolled originally in the three studies.18,19 Additionally, the CSF WBC

Downloaded From: http://archpedi.jamanetwork.com/ by a New York University User on 05/24/2015

counts and

protein and glucose concen¬ trations in both CSF samples obtained at diagnosis and again 18 to 30 hours later were not statistically different for these 80 patients and the original 260 study patients.

In this series H influenzae type b accounted for 71 (89%) of 80 cases of bacterial meningitis, S pneumoniae for 6 cases (7%), and Neisseria meningitidis for 3 cases (4%). These percentages reflect closely the distribution of cases in the entire series of 260 patients. Fif¬ ty-eight patients were treated with cefuroxime sodium and 22 patients with ceftriaxone sodium. Dexamethasone was administered at diagnosis to 40 pa¬ tients and placebo to 40 patients. The effects of dexamethasone and placebo on the outcome of patients treated with cefuroxime in the first study18 were not significantly different from those who received ceftriaxone in the second study,19 which provided justification for combining data for statistical analyses.24 The number of patients with pathogens recovered from blood and CSF cultures was similar in both groups, and the per¬ centages of patients with positive CSF cultures after 24 hours of therapy were similar for those treated with cefurox¬ ime or ceftriaxone and those treated with dexamethasone or placebo.

PE2 Concentrations in CSF

Prostaglandin E2, PGI2, IL-lß, and TNF assays were performed on 160 paired CSF specimens from all 80 pa¬ tients. At the time of diagnosis, 72 pa¬ tients (90%) had detectable PGE2 activity in CSF. The mean (±SEM) PGE2 concentration was 462 ± 65 pg/mL (range,

Cerebrospinal fluid prostaglandins, interleukin 1 beta, and tumor necrosis factor in bacterial meningitis. Clinical and laboratory correlations in placebo-treated and dexamethasone-treated patients.

Prostaglandins (PGs), interleukin 1 beta (IL-1 beta), and tumor necrosis factor alpha (TNF alpha) are likely mediators of local inflammatory reactions...
857KB Sizes 0 Downloads 0 Views