April 1976 TheJournalofPEDIATRICS
Concentrations of bacteria in cerebrospinal fluid of patients with bacterial meningitis Concentrations of bacteria in cerebrospinal fluid ranged from 4.5 x 103 to 3 • 108 colony-forming units/ ml in 27patients with bacterial meningitis before antibiotic therapy and from 4 x 101 to 1.4 x 10 ~ CFU/ ml in four patients after one to two days o f antibiotic therapy. All patients with persistent positive cultures had pretreatment concentrations o f 107 C F U / m l or greater. A significant association was observed between cerebrospinal fluid lactic acid dehydrogenase activity and concentrations of bacteria (p < O.01). Large inocula ofHemophilus influenzae type b (107) increased the minimal inhibitory concentration for penicillin and ampicillin but not f o r chloramphenicol. The minimal inhibitory concentration o f each of the three antibiotics increased when group B streptococci were assayed. These data indicate that persistence o f a positive culture may be related to large initial concentrations o f bacteria. The relative "resistance" in vitro o f large inocula possibly contributes to this persistence. These observations are also consistent with the hypothesis that lactic acid dehydrogenase activity in cerebrospinal fluid is derived from bacteria.
William E. Feldman, M.D., M.S., Richmond, Va.
SEVERAL INVESTIGATORS have measured concentrations of bacteria in various body fluids? -3 Quantitation of bacteria in CSF of patients with bacterial meningitis, however, has not been previously reported. This study correlates concentrations of bacteria with levels of CSF LDH activity and with persistence of positive cultures after conventional antibiotic therapy.
SUBJECTS AND METHODS The study was conducted from April, 1974, to June, 1975. Values of CSF LDH activities from patients with meningitis were determined as described previously.4 Levels of CSF LDH activity ranged from 66 to 870 units/ ml in this study. To quantitate bacteria in CSF, samples of turbid CSF were collected aseptically and held at room temperature until they were plated. The time interval between collection and plating of the sample ranged from 30 minutes to From the Department of Pediatrics; Medical College of Virginia, Virginia Commonwealth University, Health Sciences Division. Reprint address: Department of Pediatrics, The University of Texas Southwestern Medical School at Dallas, 5323 Harry Hines Blvd., Dallas, Texas 75235.
4 hours. The majority of samples were plated by 1.5 hours after the lumbar puncture and only two samples were plated as late as 4 hours after the lumbar puncture. Two aliquots were taken from the uncentrifuged sample after it was mixed; 0.1 ml was spread evenly over the surface of a chocolate agar plate. To the second 0.2 ml aliquot, 1.8 ml of trypticase soy broth (3%, BBL) was added. Serial
See related articles, pp. 542, 553, 557, and 706. Abbreviations used CSF: cerebrospinal fluid LDH: lacticacid dehydrogenase MIC: minimal inhibitory concentration CFU: colony-forming units tenfold dilutions were made in trypticase soy broth and 0.1 ml samples from each dilution were spread on chocolate agar. The final dilution plated was 1 • 10 7. Cultures were incubated with 5% CO2 atmosphere at 35~ and were examined at 24 and 48 hours. Those with 30 to 300 colonies were used for counting. Separate CSF samples from the same lumbar puncture specimen were cultured by the bacteriology laboratory at
Vol. 88, No. 4, part 1, pp. 549-552
The Journal of Pediatrics April 1976
HEMOPHILUS GROUP B INFLUENZAE STREPTOCOCCUS
STR NEISSERIA PNEUMONIAE MENINGITIDIS
Fig. 1. Concentrations of bacteria in cerebrospinal fluid before antibiotic therapy. STR, Streptococcus. the Medical College of Virginia Hospitals and identity of bacterial growth confirmed by standard methods? Isolates of Hemophilus influenzae were typed with specific antisera. The MIC for penicillin, ampicillin, and chloramphenicol was determined by tube dilution assay using 3% trypticase soy broth. ~ Two percent Fildes supplement was added when H. influenzae isolates were tested. The MIC was the lowest concentration of antibiotic resulting in inhibition of visible growth after 24 hours at 35~ RESULTS
Bacterial quantitation was determined on 28 samples of CSF from 27 patients before antibiotic treatment (Fig. 1). Nineteen isolates were H. influenzae type b, five were group B streptococci, three were Streptococcus pneumoniae, and one was a strain of Neisseria rneningitidis group C. One patient, who is reported elsewhere, 7 had a relapse of/-/, influenzae type b meningitis: she had 2 x 10~ C F U / ml CSF initially and 8.5 x 10~ C F U / m l CSF upon relapse. The two episodes of her illness were counted separately in this series. Concentrations of bacteria ranged from 4.5 x 103 to 2.5 x 108 C F U / m l CSF (Fig. 1). Only one isolate had a concentration below l0 s C F U / m l CSF; five isolates had concentrations of 108 C F U / m l CSF or greater. There was no relationship between concentration of bacteria and species of bacteria. The number of bacteria per oil immersion field in a gram-stained specimen of uncentrifuged CSF correlated well with the concentration. Samples in which 1-50 bacteria were observed in each field bad concentrations of 107 C F U / m l or greater. Those which had 0-1 bacteria/field had concentrations between l0 s and 107 C F U / m l , and
samples which had less than 0-1 bacteria/field had concentrations below l0 s C F U / m l CSF. Samples of CSF were obtained after antibiotic therapy from 22 of the 27 patients. The duration of treatment was 24 hours except for Patient B. M. who was treated for 48 hours before CSF was obtained. Bacterial growth was detected in four samples (Table I). Isolates were susceptible in vitro to the antibiotics which were given in recommended dosages. Concentrations of bacteria decreased from approximately one hundred to a million fold after initiation of antibiotic therapy. One sample of ventricular fluid was obtained from a 3-week-old male with Escherichia coli ventriculitis. He had 4,000 E. coli/ml after 72 hours of therapy with intravenous ampicillin (300 mg/kg/day) and intramuscular gentamicin (5 m g / k g / day). The MIC of the pretreatment isolate was 8 #g/ml for ampicillin and for gentamicin, respectively; the MIC of the isolate after therapy was 16 and 8/zg/ml, respective-
ly. A significant statistical relationship by analysis of variance was observed when concentrations of bacteria were compared to the level of CSF LDH activity (p < 0.01) (Table II). Concentrations of bacteria were not significantly related to other CSF measurements. All isolates were susceptible in vitro to penicillin, ampicillin, and chloramphenicol by the disc method done by the bacteriology laboratory (penicillin and ampicillin, 10/~g; chloramphenicol, 30/zg). Low MIC for ampicillin and penicillin were observed using a small inocuhim of H. influenzae or group B streptococci (Table III). One isolate of H. influenzae type b and one of group B streptococcus had a MIC for chloramphenicol of 4/zg/ml. Another isolate of group B streptococcus required 8 t~g
Volume 88 Number 4, part 1
chloramphenicol/ml. All other isolates were inhibited by 2/~g chloramphenicol/ml or less. Increase of the inoculum 1,000-fold to 107, which more closely reflected in vivo concentrations, increased the MIC for penicillin and ampiciJlin for each organism (Table llI). A similar inoculum effect was observed with chloramphenicol against group B streptococci, but not against H. influenzae type b isolates except in the case of the chloramphenicolresistant strain, for which the MIC increased from 4/~g/ml with 10' CFU to > 128/~g/ml using 107 CFU. Addition of a subinhibitory concentration of 0.6 /~g chloramphenicol/ml reduced the MIC for penicillin and ampicillin from > 128/Lg/ml to 0.5-32/~g/ml in 4 of 12 H. influenzae type b strains using a 107 inoculum. The MIC's of penicillin and ampicillin for the group B streptococci, using a 107 inoculum, were unaffected by addition of chloramphenicol. Addition of 3 /~g chloramphenicol/ml reduced the MIC of penicillin and ampicillin to < 0.25 /~g/ml for 12 strains of H. influenzae type b. Three/~g chloramphenicol/ml did not affect the MIC of penicillin or ampicillin for the chloramphenicol-resistant strain of H. influenzae type b. DISCUSSION These data show that patients with bacterial meningitis have large concentrations of bacteria in their CSF by the time they are hospitalized. Levels in several cases approached the maximum growth potential of 108-6 achievable in broth culture. These concentrations are generally larger than those reported for patients with bacterial infections other than meningitis,l-:' As a result of several factors, the data are only estimates of the in vivo concentration. For example, the interval between collection and quantitation of the sample is difficult to control in a clinical setting. During this time, the bacterial population may increase or decrease depending upon the species of bacteria, and the temperature, pH, presence of active phagocytes, etc. The correlation between the number of bacteria/oil immersion field by Gram stain made immediately after the lumbar puncture, and subsequent measurements of concentration, indicates that this measurement is reasonably accurate. Antibiotic therapy was effective despite large concentrations of bacteria. Although persistently positive cultures were detected in four of 22 patients, the population had decreased at least 100-fold as compared to pretreatment levels. All positive cultures after start of therapy were from CSF of patients who had large initial concentrations (greater than 10~ CFU/ml). I speculate that part of such large populations of bacteria were in the stationary phase of growth and were relatively "resistant" to antibiotics. This report provides a basis for previous observations that positive cultures can persist for one or
Concentrations o f bacteria in C S F
Table I. Effect of antibiotic therapy on concentrations of bacteria in cerebrospinal fluid Colony-forming units per ml CSF
Patient B.M. S.S. S.L. C.B.
Antibiotic GBBS*Penicillin HITBt Ampicillin HITB Ampicillin HITB Penicillin Chloramphenicol
1.1 x 1.8 x 8.6 x 1.8 x
1.7 x 1.4 x 2.2 x 4.0 x
108 l0s 107 107
10~ 10~ 10-' 101
*GBBS; GroupB fl hemolyticstreptococcus. tHITB; Hemophilus influenzae type b.
Table II. Correlation of concentrations of bacteria with other cerebrospinal fluid measurements (23 patients) Measurement
Red blood cell count White blood cell count Polymorphonuclear leukocytes Total protein concentration Glucose concentration Lactic acid dehydrogenase activity
0.34 0.95 0.07 0.48 0.10 < 0.01
Table III. The effect of inoculum on minimal inhibitory concentrations of 13 isolates of Hemophilus influenzae and 5 isolates of group B fl streptococci Hemophilus influenzae type b lnoculum Ampicillin Penicillin Chloramphenicol
10~ < 0.25-0.5* < 0.25-1.0 0.5-2.0t
Group B fl streptococci 104
< 0.25 > 128 > 128 > 128 < 0.12 > 128 1.0-2.0t 0.5-8.0 32-> 128
*/~g/ml. tOne additionalisolatewas resistant(see text). two days regardless of eventually effective antibiotic therapy. 8-1~ The concentrations of bacteria are roughly 1,000-fold greater than the currently recommended 6 inoculum for in vitro antibiotic sensitivity testing of 104 or 105. All isolates tested were "resistant" using large inocula to ampicillin and penicillin. Isolates of group B streptococci, but not of H. influenzae type b isolates, were also "resistant" to chloramphenicol using large inocula. In this respect, chloramphenicol has a theoretic advantage over ampicillin in the treatment of H. influenzae type b meningitis. This observation may be related to the clinical impression
that persistently positive cultures are more frequent with ampicillin than with chloramphenicol therapy. The Committee on Infectious Diseases of the A m e r i c a n Academy of Pediatrics recently r e c o m m e n d e d the use of penicillin or ampicillin with chloramphenicol in areas where ampicillin-resistant strains of H. influenzae type b are endemic. 13 A subinhibitory concentration of chloramphenicol was effective in combination with penicillin or ampicillin against four of 12 strains of H. influenzae type b using inocula likely to reflect conditions in C S F during initial antibiotic therapy. Thus these preliminary data support their recommendation. Previous reports have shown no correlation of C S F L D H activity with the CSF white blood cell count or with total protein or glucose concentration?, 14-~6 The significant association between levels of bacteria in CSF and the level of CSF L D H activity is consistent with the hypothesis that CSF L D H activity is derived from bacteria. Thus, determination of the C S F L D H activity may be useful to the clinician if a patient suspected of having bacterial meningitis has received inadequate oral antibiotic therapy. The concentration o f bacteria in C S F is likely related to the prognosis of the patient. I f this relationship is confirmed, determination of the C S F L D H activity m a y be used to predict outcome. In this regard, preliminary results14.16 show a higher CSF L D H activity in patients who had neurologic sequelae or died as compared to those who recovered completely although concentrations o f bacteria were not measured. ADDENDUM
Pretreatment C S F concentrations of bacteria from an additional 15 patients with H. influenzae type b and one with N. meningitidis group c meningitis were significantly related to C S F concentrations of bacterial antigen, as measured by counte_rcurrent immunoelectrophoresis (polynomial regression analysis, p < 0.017). Other workers 17-2~have shown that the C S F concentration of bacterial antigen is related to the prognosis of the patient. A prospective study is underway to determine whether concentrations, of bacteria in CSF can also be used to predict the outcome. I wish to thank William E. Laupus, M.D., and Herbert Welsheimer, Ph.D., for their invaluable support and guidance. REFERENCES
1. Dietzman DE, Fischer GW, and Schoenknecht FD: Neonatal Escherichia coli septicemia-bacterial counts in blood, J PEDIATR85:128, 1974. 2. Lorber B, and Swenson RM: Bacteriology of aspiration pneumonia: A prospective study of community- and hospital-acquired cases, Ann Intern Med 81:329, 1974.
The Journal of Pediatrics April 1976
3. Werner AS, Cobbs CG, Kaye D, and Hook EW: Studies on the bacteremia of bacterial endocarditis, JAMA 202:127, 1967. 4. Feldman WE: Cerebrospinal fluid lactic acid dehydrogenase activity: Levels in untreated and partially antibiotictreated meningitis, Am J Dis Child 129:77, 1975. 5. Bailey WR, and Scott EG: Diagnostic microbiology: A textbook for the isolation and identification of pathogenic microorganisms, ed 3, St. Louis, 1970, The C.V. Mosby Co, p 115. 6. Washington JA, and Barry AL: Dilution test procedures, in Lennett EH, Spaulding EH, and Truant JP, editors: Manual of clinical microbiology, ed 2, Washington, D.C., 1974, American Society for Microbiology, p 410. 7. Feldman WE, Laupus WE, and Ledaal P: Relapse of Haemophilus influenzae type b meningitis after combined antibiotic therapy: Report of a case, Pediatrics (in press). 8. Barrett FF, Taber LH, Morris CR, Stephenson WB, Clark DJ, and Yow MD: A 12-year review of the antibiotic management of Haemophilus influenzae meningitis: Comparison of ampicillin and conventional therapy including chloramphenicol, J PEDIATR81:370, 1972. 9. Haltalin KC, and Smith JB: Reevaluation of ampicillin therapy for Hemophilus influenzae meningitis: An appraisal based on a review of cases of persistent or recurrent infection, Am J Dis Child 122:328, 1971. 10. Mathies AW Jr, Leedom JM, Thrupp LD, Ivler D, Portnoy B, and Wehrle PF: Experience with ampicillin in bacterial meningitis, Antimicrob Agents Chemother -1965, p 610, 1966. 11. Shackelford PG, Bobinski JE, Feigin RD, and Cherry JD: Therapy of Haemophilus influenzae meningitis reconsidered, N Engl J Med 287:634, 1972. 12. Wilson HD, and Haltalin KC: Ampicillin in Haemophilus influenzae meningitis: Clinicopharmacologic evaluation of intramuscular vs intravenous administration, Am J Dis Child 129:208, 1975. 13. Committee on Infectious Diseases: Ampicillin-resistant strains of Haemophilus influenzae type b, Pediatrics 55:145, 1975. 14. Beatty HN, and Oppenheimer S: Cerebrospinal fluid lactic dehydrogenase and its isoenzymes in infections of the central nervous system, N Engl J Med 279:1197, 1968. 15. Neches W, and Platt M: Cerebrospinal fluid LDH in 287 children, including 53 cases of meningitis of bacterial and nonbacterial etiology, Pediatrics 41:1097, 1968. 16. Lending M, Slobody LB, and Mestern J: Cerebrospinal fluid glutamic oxaloacetic transaminase and lactic dehydrogenase activities in children with neurologic disorders, J PEDIATR65:415, 1964. 17. Coonrod JD, and Rytel M: Determination of aetiology of bacterial meningitis by counterimmunoelectrophoresis, Lancet 1:1154, 1972. 18. Hoffman TA, and Edwards EA: Group-specific polysaccharide antigen and humoral antibody response in disease due to Neisseria meningitidis, J Inf Dis 126:636, 1972. 19. Ingram DL, O'Reilly RJ, Peter G, Anderson P, and Smith DH: Systemic capsular antigen, clinical course and antibody response in Hemophilus influenzae b meningitis, Pediatr Res 7:369~ 1973. 20. Shackleford PG, Campbell J, and Feigin RD: Countercurrent immunoelectrophoresis in the evaluation of childhood infections, J PEDIATR85:478, 1974.