66
July 1976 TheJournalofPEDIATRICS
A controlled study of intrathecal antibiotic therapy in gram-negative enteric meningitis of infancy Report of the Neonatal Meningitis Cooperative Study Group*
Eighteen institutions collaborated in evaluating the comparative efficacy of combined ampicillin and gentamicin therapy with and without intrathecal administration of gentamicin on the clinical and bacteriologic responses of 117 infants with meningitis caused by gram-negative enteric baeteria. There was a random distribution of patients within the two study groups with regard to age on enrollment, birth weight, sex, race, number of infants greater than 30 days of age, the etiologic agent, and their antimierobial susceptibilities'. There were no statistically significant differenees (P > 0.05) in mortality, morbidity, or days" that cerebrospinal fluid cultures remained positive among the infants in the two treatment groups. The case fatality rate for all patients was 32%," that Jbr full-term infants' (18%) was significantly lower (P < 0.01) than that for low-birth-weight infants (45%) or for the patients greater than 30 days of age (48%). Fifty-one of the 80 (64%) survivors were assessed as normal on follow-up examinations performed up to four years after illness.
George H. McCraeken, Jr., M.D.,** and Susan G. Mize, Dallas, Texas
OUTCOME from neonatal gram-negative bacillary meningitis is often grave despite the use of effective chemotherapeutic agents and modern intensive care facilities. The mortality rate has been reported to range from 40% to 80%, and the majority of survivors have neurologic and developmental handicapsY :~It is likely that children judged to be normal after meningitis in infancy have *Charles A. A lford (Birmingham), Bernard Boxerbaum (Cleveland), Ernesto Calderon (Mexico City), Moses Grossman (San Francisco), Marc Gurwith (Winnipeg), William Hayden (Madison), Jerome Klein (Boston), Sheldon Korones (Memphis), Irwin Light (Cincinnati), Aarolyn Visintine and Andre Nahmias (Atlanta), James Ove:all (Salt Lake City), Herman Risemberg (Baltimore), Melvin L Marks (Montreal), Richard Michaels (Pittsburgh), Hugo Trujillo (Medellin), Catherine M. Wilfert (Durham), Harry T. Wright (Los Angeles), George McCracken (Coordinator, Dallas), Susan Mize (Statistician, Dallas). This study was supported by a grant from Schering Corporation, Bloomfield, N.J. **Reprint address: Department of Pediatrics, 5323 Harry Hines Blvd, Dallas, Texas 75235.
VoL 89, No. 1, pp. 66-72
abnormal cognitive function and/or communication disorders which become apparent only when they reach school age. 4-~ The primary goal of antimicrobial therapy for neonatal meningitis is to eradicate infection by achieving cerebrospinal fluid concentrations of antibiotic that are greater than the minimal inhibitory concentrations for the Abbreviations used CSF: cerebrospinal fluid MIC: minimal inhibitory concentration WBC: white blood count infecting organisms. In meningitis caused by grampositive bacteria, such as group B streptococci, peak CSF levels of penicillin or ampicillin are about 100-fold greater than MIC- values and CSF cultures sterilize quickly: In contrast, the peak concentrations of aminoglycosidic antibiotics in CSF approximate the MIC values of enteric bacteria, and cultures of CSF from infants with gramnegative enteric meningitis remain positive for two to five days, or longer: Because outcome from neonatal meningitis is directly correlated with the duration of positive
Volume 89 Number 1
Table I. Protocol for therapy and evaluation of infants in the meningitis cooperative study I. Therapeutic regimen based on random assignment Group A: Ampicillin 50 mg/kg every 12 hr intravenously (infants ~ 7 days) or 70 mg/kg every 8 hr intravenously (infants > 7 days) Gentamicin 2.5 mg/kg every 12 hr I.M. (infants 2'7 days) or 2.5 mg/kg every 8 hr I.M. (infants > 7 days) Group B: Same as for group A, plus gentamicin 1 mg daily intrathecally (for minimum of 3 days) II. Management Daily spinal fluid examinations plus intr~thecal gentamicin (Group B) until cultures sterilized Parenteral antibiotics continued for 2 weeks after CSF cultures sterilized or for a minimum of 3 weeks III. Follow-up examination Physical, neurologic, and developmental examinations were scheduled at 6 months after illness and every 12 months thereafter
CSF cultures,67 an antibiotic regimen which achieves rapid bacteriologic cure is essential. In September, 1971, the Neonatal Meningitis Cooperative Study Group was formed for the purpose of evaluating intrathecal antibiotic therapy in neonatal meningitis caused by gram-negative bacteria. The rationale for choosing this route of administration was that direct instillation of an antibiotic into the lumbar intrathecal space might produce greater antimicrobial activity throughout the CSF volume and result in more rapid eradication of CSF bacteria and an improved outcome from disease.
MATERIALS AND METHODS Subjects. All patients admitted to the 18 participating institutions were considered condidates for study if they were two years of age or less and had meningitis caused by gram-negative enteric bacteria as determined by Gram stain of CSF and confirmed by CSF culture. Infants who had received antibiotic therapy prior to admission were not excluded if gram-negative organisms were still present on stained smears of CSF, regardless of the culture results. Patients with meningitis caused by Hemophilus influenzae, Neisseria meningitidis, or gram-positive pathogens were excluded as were infants with meningitis complicating meningomyeloceles or other structural abnormalities of the central nervous system. Informed written consent was obtained from the parents or guardians of study patients. Therapy assignment plan. The patients were allocated at random to receive either ampicillin intravenously and gentamicin intramuscularly (Group A) or these two anti-
Intrathecal therapy in enteric meningitis
67
Table II. Number of infants seen at each participating center
City
No. infants
Atlanta Baltimore Birmingham Boston Cincinnati Cleveland Dallas Medellin, Colombia Memphis
8 2 10 0 4 3 17 16 10
City
No. infants
Montreal Salt Lake City San Francisco MexicoCity Winnipeg Durham Los Angeles Pittsburgh Madison Total
7 5 7 10 4 6 4 1 3 117
biotics plus intrathecal gentamicin (Group B). Each institution had two books for assignment of infants to a particular therapy group: one book for infants with a birth weight equal to or less than 2,500 gm and one book for infants who weighed more than 2,500 gm at birth. An additional therapy assignment book for infants greater than 30 days of age was used in Medellin, Colombia, where enteric meningitis is commonly encountered in older infants. In each book there were sealed, numbered envelopes containing cards with treatment assignment. The sequence had been determined by utilizing tables of random permutations of 16 numbers. Upon admission of an infant to the study, an envelope in numerical sequence was opened to ascertain the therapy regimen to be utilized. Drug dosage and clinical management. Table I details the drug dosage and patient management. All infants were t o be treated with the assigned regimen for a minimum of three days after which; therapy could be altered (1) if the pathogen was not susceptible to the antibiotics used, (2) if CSF cultures failed to sterilize, or (3) if the infant had not improved clinically. Therapy changes were individualized and were not directed by the study protocol. Intrathecal therapy was maintained for a minimum of three days or until CSF cultures were sterile. Ampicillin and gentamicin were continued for two weeks after sterilization of CSF cultures or for a minimum of three weeks, whichever was longer. Follow-up examinations. Survivors were scheduled for follow-up evaluations at six, 12, 24, 36, and 48 months, respectively, after illness. Complete physical, neurologic, and Denver Developmental examinations were performed at these follow-up visits. The Cattell Infant Intelligence and Gessell Fine and Gross Motor Development tests were administered to some infants 12 months of age and older.
68
McCracken and Mize
The Journal of Pediatrics July 1976
Table IlL Population characteristics of the two therapy groups Systemic therapy Systemicplus only intrathecal therapy (65 infants) (52 infants)
Male Female
36 (55%) 29 (45%)
28 (54%) 24 (46%)
Caucasian Black
53 (82%) 12 (18%)
43 (83%) 9 (17%)
21 (32%) 35 (54%)
17 (33%) 25 (48%)
9 (14%)
10 (19%)
9.80* _+ 1.95 days 125 _+ 95 days
11.88 _+ 2.57 days 214 _+ 122 days
Birth weight 2,500 gm > 2,500 gm Age > 30 days Age on enrollment 30 days > 30 days
*Mean _+ 2 SE of the mean.
Table IV. Etiology of neonatal meningitis
Organism Eseherichia coli Salmonella species Citrobacter diversus Proteus mirabilis Serratia species Klebsiella species Enterobacter species None* Moraxella species Multiple organismst Pseudomonas aeruginosa
Number
82 7 5 ~ 5 4 4 2 1 1 1
Relative frequency
70% 6% 4% 4% 4% 3% 3% 2% 1% 1% 1%
*Gram-negative rods observed on stained smears of CSF; cultures sterile. rE. coli non-Kl, Klebsiella species, and Flavobacterium meningosepticure.
An infant was considered normal if the physical, neurologic, and developmental examinations revealed no abnormalities that could be attributed to meningitis. Mild-to-moderate neurologic residua were said to be present if the infant had compensated hydrocephalus or neurologic abnormalities which were not severe enough to require special care or to prevent adequate, albeit delayed, development. Severe neurologic sequelae included failure to develop beyond the age at which the disease occurred a n d / o r requirement of custodial care. Laboratory studies. Antimicrobial susceptibilities of the pathogens were performed in each hospital laboratory. These strains were also studied in Dallas using standard plate dilution and tube dilution techniques? Serum and
CSF specimens were assayed for gentamicin activity using the micromethod of Simon and Yin" with Bacillus subtilis as the t~st organism. Patient variables and therapy evaluation. Basic identifying data such as date of enrollment, age, weight classification, sex, and race were recorded for every patient. In addition, a detailed case report form was prepared for each patient enumerating the obstetrical and medical history, treatment regimen, laboratory data, neurologic residua, findings at follow-up examinations, and a detailed summary of the course of illness and complications of disease or treatment. Autopsy reports were also used in the analysis of data. Statistical methods. Chi-square tests were used. TM When 20% or more of the expected number of observations in a contingency table were fewer than 5, Fisher's exactprobability method was utilized. Mean differences were evaluated by Student's t test. 11 Differences that had a probability of greater than 0.05 by the appropriate null hypothesis were considered not significant. Standard errors of the mean noted in this paper are expressed as the mean plus or minus 2 SE of the mean. RESULTS
Study population. During the 48-month period from September, 1971, through August, 1975, the 18 institutions enrolled 133 patients in the study. Sixteen infants were found unacceptable for analysis for the following reasons: (1) in 13 cases CSF cultures were negative or yielded nonenteric organisms; (2) two infants died before antimicrobial therapy was initiated; and (3) one infant with Pseudomorlds meningitis was treated with carbenicillin and gentamicin instead of ampicillin and gentamicin. The remaiffing 117 infants comprise the study population. Distributioh of cases among the 18 institutions is shown in Table II. Sixty-five infants were assigned to Therapy Group A and 52 infants to Therapy Group B (intrathecal gentamicin). Table III shows that there was a random distribution of patients between the two therapy groups with regard to age on enrollment, birth weight, sex, race, and number of infants > 30 days of age. Antibiotic regimens were given for the time designated by the protocol in 86% of infants. Changes in antibiotic therapy were made in ten infants given systemic therapy alone and in seven infants treated with systemic plus intrathecal therapy. In addition, nine (17%) of the 52 infants assigned to Group B failed to receive three intrathecal gentamicin doses: five patients were given only one dose and four infants received two doses. The remaining 43 infants received three to 11 (average, four) intrathecal doses.
Volume 89 Number 1
Intrathecal therapy in enteric meningitis
69
Table V. Case fatality rates for the two therapy groups Systemic therapy only (65 infants) Category
Total Birth weight 2,500 gm > 2,500 gm Age Neonates only ( 2 30 days) Infants ( > 30 days) Death due to meningitis Death after 48 hr of therapy Etiology: E. coli only Etiology: Non-E. coli Pathogens susceptible to both antibiotics*
Systemic plus intrathecal therapy (52 infants)
Dead
A live
% Dead
Dead
A live
% Dead
Total population
Chi square "'P" value
21
44
32.3
16
36
30.8
117
0.982
10 7
11 28
47.6 20.0
7 4
10 21
41.2 16.0
38 60
0.945 0.955
17
39
30.4
11
31
26.2
98
0.821
4 21 12
5 44 44
44.4 32.3 21.4
5 14 12
5 36 36
50.0 28.0 25.0
19 115 104
0.586 0.769 0.843
15 6 14
34 10 37
30.6 37.5 27.5
7 9 11
26 10 25
21.2 47.3 30.6
82 35 87
0.492 0.807 0.941
*Represents only those pathogens known to be susceptible to both ampicillin and gentamicin. Etiology. Escherichia coli was the etiologic agent in 82 (70%) of the 117 cases (Table IV). Fifty-eight (71%) of the E. coli strains possessed capsular K1 antigen. Patients with E. coli K1 meningitis were equally distributed between the two therapy groups. Six of the seven patients with Salmonella meningitis and four o f the five infants with disease caused by Serratia species were from Medellin, Colombia. Outcome. Thirty-seven (32%) infants died (Table V). Full-term infants had a significantly lower (P < 0.01) mortality rate (18%) when compared with low-birthweight infants (45%) and with infants greater than 30 days of age (48%). Case fatality rates were not significantly different in t h e two therapy groups nor were they different when subcategories were analyzed according to birth weight or age. Two infants who received intrathecal therapy died from causes unrelated to meningitis (congenital heart disease and gastrointestinal bleeding secondary to ulcerative colitis). Excluding these two patients from the analysis did not significantly influence the mortality statistics (Table V). Differences in death rates between the two treatment groups were not affected when the following variables were assessed: (1) death before completion of 48 hours of therapy; (2) meningitis caused b y E. coli strains only; (3) disease due to non-E, coli organisms; (3) susceptibility of the pathogens to ampicillin and gentamicin; (5) antibiotic therapy prior to enrollment; (6) failure to receive a m i n i m u m of three intrathecal doses; or (7) when patients from Colombia and Mexico were excluded from the
Table VI. Neurologic status of 80 infants who survived gram-negative meningitis Last examination after illness
Therapy group*
12 mo or greater
Group A Group B
16 17
8 7
4 2
28 26
6to 11 mo
GroupA Group B
6 5
2 0
0 1
8 6
Hospital discharge
Group A Group B
4 3
4 1
0 0
8 4
51
22
7
80
Total
Neurologic status? MiM-moderate Severe Normal residua residua
Total
*Group A--systemicantibiotics only; Group B systemic antibiotics plus intrathecal gentamicin. tSee text for definitions.
analysis. Excluding one patient who died more than two years after institution o f therapy (Group B), the m e a n number of days from onset of therapy to death was 33 _+ 32 days for infants in Group A and 26 + 23 days for infants in Group B. Eighty (68.4%) infants survived and have been evaluated for variable intervals after illness (Table VI). Twelve (15%) infants had hospital discharge evaluations only, fourteen (17.5%) patients have been examined between
70
McCracken and Mize
six to 11 months after illness, and 54 (67.5%) infants have been tested from one to four years after illness. There were no signifcant differences in the neurologic status of infants within the two therapy groups regardless of the time after illness when evaluations were performed. Fiftyone (64%) of the surviving infants were considered normal, 22 (27%) had mild-to-moderate neurologic sequelae, and seven (9%) infants were judged to have severe neurologic handicaps. Of these patients, 13 im Group A and eight in Group B have failed to return for current evaluation and therefore, are considered to be lost to follow-up. Age on enrollment, weight classification, race, sex, or type of organism did not appear to affect inorbidity. Days of hospitalization. Infants enrolled in Group A were hospitalized for an average of 24 _+ 4 d@s and those in Group B for 27 _+ 5 days. Cerebrospinai fluid studies. CSF cultures were positive for 3.3 _+ 0.6 days for infants in Group A and 3.9 _+ 1.2 days for infants in Group B. When only the surviving infants were considered, the average duration of positive CSF cultures was 3.3 days in both groups. The CSF white blood cell counts and protein contents were analyzed with respect to outcome from disease. When the single largest value for each patient was analyzed, the mean CSF leukocyte ~ount was significantly greater for infants who died (23.7 _+ 9.4 x 103 W B C / mm 3) than for those who survived (9.9 _+ 2.4 x 10mm0their illness. Similarly, CSF protein concentrations were significantly greater in those who died than in survivors (1,282 _+ 626 mg/dl versus 517 _+ 120 mg/dl, respectively). Special laboratory studies. Bacteria from 108 patients were studied for antimicrobial susceptibility. Only one organism, an E. coli, was resistant to gentamicin (MIC 10t~g/ml) and had an ampicillin MIC of 20 t~g/ml. In spite of this relative antibiotic resistance, CSF cultures sterilized after two days of therapy and the infant was normal 12 months after illness. Thirty (28%) of the 108 organisms were resistant to ampicillin (M1C > 10 /~g/ml). Of the 78 susceptible strains, 43 (55%) strains had MIC values of 2.5/zg/ml or less. Ampicillin-resistant strains were equally distributed between the two study groups and mortality rates for all infants with resistant and susceptible strains were not significantly different (40% and 29%, respectively). Outcome was not significantly different in the 30 infants with meningitis due to ampicillin-resistant organisms from that of 30 patients with susceptible pathogens (MIC ~ 2.5 /~g/ml) selected randomly from the other study patients. Nine (30%) infants with ampicillinresistant organisms died compared with six (20%) in the
The Journal of Pediatrics July 1976
ampicillin-susceptible group. Fifteen (50%) infants with ampicillin-resistant organisms were normal or had mildto-moderate neurologic sequelae on follow-up examinations, whereas ten infants (33%) with susceptible pathogens had similar evaluations. CSF cultures remained positive for approximately three days in both groups. In vitro antimicrobial synergism studies were not performed with these ampicillin-resistant organisms. Previous studies demonstrated that ampicillin and gentamicin act synergistically against approximately 40% of E. coli strains isolatedfrom neonates. 1~ Gentamicin assays were performed on one or more CSF samples from 43 infants. Specimens obtained at various intervals after 2.5 mg/kg doses given intramuscularly contained 0.3 to 3.7/~g/ml (mean, 1.6/~g/ml) gentamicin. Gentamicin concentrations in CSF taken at similar times after parenteral therapy and at 18 to 24 hours after 1 mg gentamicin given into the lumbar intrathecal space were 0.5 to 3.4 t~g/ml (mean, 1.6 t~g/ml). Four infants had lumbar punctures performed two to four hours after 1 mg intrathecal doses and the CSF gentamicin levels were 18.4, 25, 36.8, and > 40/~g/ml, respectively. Safety. Parenteral antimicrobial therapy was well tolerated and there was no evidence of drug-related renal abnormalities or ototoxicity. There were no acute adverse reactions due to intrathecal gentamicin therapy. Cerebrospinal fluid white blood cell counts and protein values during illness and at the time of hospital discharge were similar for the two therapy groups. One infant was first noted to have left lower extremity weakness and hyperreflexia 12 months after illness. This infant was given 1 mg gentamicin intrathecally for three days and because of continued positive CSF cultures, 1 mg polymyxin B was administered daily for four days. The patient's neurologic findings improved but were still present at three and onehalf years of age when fine motor function and mental development were considered normal. No abnormalities were observed among the other infants who received intrathecal therapy. DISCUSSION The results of this collaborative project establish that 1 mg gentamicin administered daily into the lumbar intrathecal space for three days or longer has no discernable beneficial or adverse effect on outcome from neonatal enteric meningitis when compared with systemic therapy alone. Disparity in the number of patients assigned to the two therapy groups was due to the occasional incorrect usage of the therapy assignment books, inability to initiate intrathecal therapy in a few infants, and incorrect diagnosis in 13 infants at the time of admission to the study.
Volume 89 Number I
Even so, characteristics of the two groups at the onset of treatment where comparable and no significant differences in mortality and long-term morbidity rates were observed. Outcome in these infants was better than that reported by others I-~ for neonatal gram-negative meningitis (mortality rates of 40% to 80%). The probable explanation for this improved outcome relates more to intensive management of the respiratory and metabolic problems commonly encountered in these patients than to the specific antibiotic regimens employed. Intrathecal therapy was chosen for study with the anticipation that direct instillation of drug into CSF would result in higher gentamicin concentrations and improved bactericidal activity. Althoug~ gentamicin concentrations of 18 to > 40 /~g/ml were observed in lumbar CSF several hours afterqntrathecal administration, the levels were only 1 to 3~/~g/ml 18 to 24 hours later. These latter values are similar to those measured in CSF from infants treated with parenteral therapy only. Rahal and associates la found lumbar CSF gentamicin concentrations of 20 to 50 /~g/ml two to eight hours after 4 mg intrathecal doses in adults. The half-life of gentamicin in CSF was five hours. The relatively rapid disappearance of gentamicin from CSF of patients in these two studies may be explained by a dilutional effect secondary to diffusion of drug in the CSF space, transport of drug out of the CSF, or inactivation of drug. The third possibility is unlikely because gentamicin is extremely stable and such factors as reduced pH, protein-binding, or ionic composition of CSF would be counteracted by the conditions of the bioassay. Spector 1~ has demonstrated that gentamicin is transported out of CSF of rabbits by an active process within the choroid plexus. Whether this efflux mechanism is operative in the regulation of CSF gentamicin concentrations in human beings is unknown. In order for intrathecal therapy to be bacteriologically effective, the antibiotic must diffuse throughout the entire subarachnoid space and, perhaps also, into the ventricles. Kaiser and McGee 1~have recently studied the pharmacokinetics of gentamicin and tobramycin in lumbar, cisternal, and ventricular fluids after lumbar intrathecal administration to adults with gram-negative meningitis. Peak gentamicin and tobramycin concentrations of 27 to 80/~g/ ml were noted in lumbar CSF during the first six hours after 5 to 10 mg intrathecal doses. The antibiotics disappeared rapidly from CSF (average half-life, six hours). The aminoglycosides failed to enter the cerebral ventricles and were poorly distributed to other parts of the CSF space. *~ These data are consistent with the known unidirectional flow of CSF out of the ventricles. It is likely that delayed bacteriologic cure results from an inability to deliver adequate antibiotic concentrations
Intrathecal therapy in enteric meningitis"
71
throughout the central nervous system. One possible inaccessible site of infection is the ventricular system. In the present study, ventriculitis was documented in eight of 12 infants at postmortem examination and in five of seven patients in whom ventricular taps were done during the course of illness. Berman and Banker 1~ demonstrated ventriculitis on postmortem examination of 23 of 25 neonates with meningitis and Salmon 17 encountered venfricular infection in eight of 12 similarly affected babies. Although the total number of patients evaluated is small, these studies indicate that ventriculitis occurs commonly in neonates with meningitis, particularly when there is delayed sterilization of CSF cultures. The inadequate delivery of aminoglycoside to the ventricles after intralumbar therapy was illustrated in an infant with Proteus morganii meningitis reported by Moelloring and Fischer. 1~ Despite parenteral and intrathecal gentamicin CSF cultures failed to sterilize and ventriculitis was diagnosed on the seventeenth day of therapy. Pharmacologic studies demonstrated transiently elevated lumbar CSF gentamicin concentrations and very low ventricular levels after intralumbar administration. Intraventricular gentamicin therapy produced rapid and uniform distribution of antibiotic throughout the CSF space and prompt sterilization of ventricular and lumbar CSF cultures. ~r Similar pharmacokinetic data were reported by Kaiser and McGee? ~ When 5 mg of gentamicin was injected into the ventricles of adults, concentrations of 15 to 45 ~g/ml were found in ventricular fluid six hours later and values of 10 to 20/~g/ml were measured in lumbar fluid. During most of the remaining 24-hour period, concentrations of 5/~g/ml persisted in both sites. These gentamicin levels are within the therapeutic range for susceptible gram-negative bacteria. SPECULATION The intrathecal antibiotic regimen e m p l o y e d in this study proved to be no better than systemic therapy alone for gram-negative enteric meningitis of infancy. Clinical and autopsy data indicate that ventricular infection occurs commonly in this disease and that ventriculitis persists despite parenteral a n d intralumbar antibiotic therapy. Pharn?acologic studies offer an explanation for these observations. Aminoglycosides diffuse poorly throughout the CSF space after intrathecal administration and subtherapeutic concentrations are present in ventricular fluid. On the other hand, direct instillation of antibiotic into the ventricles produces concentrations of aminoglycoside that are many fold higher than the minimal bactericidal titers for most enteric pathogens. On the basis of these findings and the continued high mortality and morbidity rates associated with neonatal
72
McCracken and Mize
gram-negative meningitis, a new approach to antimicrobial therapy in this condition is necessary. Ideally, a ventricular tap should be performed at the onset of therapy in order to define the incidence of ventriculitis and to select the m o d e of antimicrobial therapy, More practically, a ventricular tap must be performed in an infant with positive lumbar CSF cultures for two or three days after intrathecal a n d / o r parenteral ~tntibiotic therapy. If ventriculitis is diagnosed, intraventricular aminoglycoside therapy should be used. This may be accomplished by daily taps or by instilling antibiotic through an implanted reservoir. A more direct approach to the diagnosis and treatment of ventricular infection should be evaluated. Merely altering the conventional antibiotic regimens for neonatal meningitis is not likely to improve outcome from this disease and may result in adverse effects if dosages are increased to obtain higher C S F concentrations. We thank the many pediatric faculty members, residents, fellows, and nurses at the 18 institutions who managed the study patients. Without their cooperation and interest this study could not have been performed. Norma Threlkeld, R.N., coordinated data collection and follow-up reports. REFERENCES
1. Haggarty RJ, and Ziai Mohsen: Acute bacterial meningitis, Adv Pediatr 13:129, 1964. 2. Overall JC: Neonatal bacterial meningitis, J PEI~IATR 76:499, 1970. 3. Groover RV, Sutherland JM, and Landing BH: Purulent meningitis of newborn infants, N Engl J Med 264:1115, 1961. 4. Fitzhardinge PM, Kazemi M, Ramsay M, and Stern L: Long-term sequelae of neonatal meningitis, Dev Med Clin Neurol 16:3, 1974. 5. Sell SHW, Merrill RE, Doyne EO, and Zimsky EP: Long-
The Journal of Pediatrics July 1976
6.
7.
8.
9. 10.
11. 12.
13.
14.
15.
16.
17. 18.
term sequellae of Hemophilus influenzae meningitis, Pediatrics 49:206, 1972. McCracken GH: The rate of bacteriologic response to antimicrobial therapy in neonatal meningitis, Am J Dis Child 123:547, 1972. McCracken GH, Sarff LD, Glode MP, Mize SG,. Schiffer MS, Robbins JB, Gotschlich EC, J~'rskov I, and ~rskov F: Relation between Escherichia coli K1 capsular polysaccharide antigen and clinical outcome in neonatal meningitis, Lancet 2:246, 1974. Haltalin KC, Markley AH, and Woodman E: Agar plate dilution method for routine antibiotic susceptibility testing in a hospital laboratory, Am J Clin Pathol 60:384, 1973. Simon HJ, and Yin EJ: Microbioassay of antimicrobial agents, Appl Microbiol 19:573, 1970. Siegel S: Non-parametric statistics for the behavioral sciences, New York, 1956, McGraw-Hill, Book Company, Inc., pp 96 and 197-200. Ostle B: Statistics in research, Ames, Iowa, 1963, The Iowa State University Press, p 119. Kaplan JM, McCracken GH, Horton LJ, Thomas MS, and Davis N: Pharmacologic studies in neonates given large dosages of ampicillin, J PEOIATR84:571, 1974. Rahal JJ, Hyams PJ, Simberhoff MS, and Rubinstein E: Combined intrathecal and intramuscular gentamicin for gram-negative meningitis, N Engl J Med 290:1394, 1974. Spector R: The transport of gentamicin in the choroid plexus and cerebrospinal fluid, J Pharmacol Exp Ther 194:83, 1975. Kaiser AB, and McGee ZA: Aminoglycoside therapy of gram negative bacillary meningitis, N Engl J Med 293:1215, 1975. Berman PH, and Banker BQ: Neonatal meningitis: A clinical and pathologic study of 29 cases, Pediatrics 38:6, 1966. Salmon JH: Ventriculitis complicating meningitis, Am J Dis Child 124:35, t:972. Moellering RC, and Fischer EG: Relationship of intraventricular gentamicin levels to cure of meningitis, J PEOIAXR 81:534, 1972.
July 1976 TheJournalofPEDIATRICS
A controlled study of intrathecal antibiotic therapy in gram-negative enteric meningitis of infancy Report of the Neonatal Meningitis Cooperative Study Group*
Eighteen institutions collaborated in evaluating the comparative efficacy of combined ampicillin and gentamicin therapy with and without intrathecal administration of gentamicin on the clinical and bacteriologic responses of 117 infants with meningitis caused by gram-negative enteric baeteria. There was a random distribution of patients within the two study groups with regard to age on enrollment, birth weight, sex, race, number of infants greater than 30 days of age, the etiologic agent, and their antimierobial susceptibilities'. There were no statistically significant differenees (P > 0.05) in mortality, morbidity, or days" that cerebrospinal fluid cultures remained positive among the infants in the two treatment groups. The case fatality rate for all patients was 32%," that Jbr full-term infants' (18%) was significantly lower (P < 0.01) than that for low-birth-weight infants (45%) or for the patients greater than 30 days of age (48%). Fifty-one of the 80 (64%) survivors were assessed as normal on follow-up examinations performed up to four years after illness.
George H. McCraeken, Jr., M.D.,** and Susan G. Mize, Dallas, Texas
OUTCOME from neonatal gram-negative bacillary meningitis is often grave despite the use of effective chemotherapeutic agents and modern intensive care facilities. The mortality rate has been reported to range from 40% to 80%, and the majority of survivors have neurologic and developmental handicapsY :~It is likely that children judged to be normal after meningitis in infancy have *Charles A. A lford (Birmingham), Bernard Boxerbaum (Cleveland), Ernesto Calderon (Mexico City), Moses Grossman (San Francisco), Marc Gurwith (Winnipeg), William Hayden (Madison), Jerome Klein (Boston), Sheldon Korones (Memphis), Irwin Light (Cincinnati), Aarolyn Visintine and Andre Nahmias (Atlanta), James Ove:all (Salt Lake City), Herman Risemberg (Baltimore), Melvin L Marks (Montreal), Richard Michaels (Pittsburgh), Hugo Trujillo (Medellin), Catherine M. Wilfert (Durham), Harry T. Wright (Los Angeles), George McCracken (Coordinator, Dallas), Susan Mize (Statistician, Dallas). This study was supported by a grant from Schering Corporation, Bloomfield, N.J. **Reprint address: Department of Pediatrics, 5323 Harry Hines Blvd, Dallas, Texas 75235.
VoL 89, No. 1, pp. 66-72
abnormal cognitive function and/or communication disorders which become apparent only when they reach school age. 4-~ The primary goal of antimicrobial therapy for neonatal meningitis is to eradicate infection by achieving cerebrospinal fluid concentrations of antibiotic that are greater than the minimal inhibitory concentrations for the Abbreviations used CSF: cerebrospinal fluid MIC: minimal inhibitory concentration WBC: white blood count infecting organisms. In meningitis caused by grampositive bacteria, such as group B streptococci, peak CSF levels of penicillin or ampicillin are about 100-fold greater than MIC- values and CSF cultures sterilize quickly: In contrast, the peak concentrations of aminoglycosidic antibiotics in CSF approximate the MIC values of enteric bacteria, and cultures of CSF from infants with gramnegative enteric meningitis remain positive for two to five days, or longer: Because outcome from neonatal meningitis is directly correlated with the duration of positive
Volume 89 Number 1
Table I. Protocol for therapy and evaluation of infants in the meningitis cooperative study I. Therapeutic regimen based on random assignment Group A: Ampicillin 50 mg/kg every 12 hr intravenously (infants ~ 7 days) or 70 mg/kg every 8 hr intravenously (infants > 7 days) Gentamicin 2.5 mg/kg every 12 hr I.M. (infants 2'7 days) or 2.5 mg/kg every 8 hr I.M. (infants > 7 days) Group B: Same as for group A, plus gentamicin 1 mg daily intrathecally (for minimum of 3 days) II. Management Daily spinal fluid examinations plus intr~thecal gentamicin (Group B) until cultures sterilized Parenteral antibiotics continued for 2 weeks after CSF cultures sterilized or for a minimum of 3 weeks III. Follow-up examination Physical, neurologic, and developmental examinations were scheduled at 6 months after illness and every 12 months thereafter
CSF cultures,67 an antibiotic regimen which achieves rapid bacteriologic cure is essential. In September, 1971, the Neonatal Meningitis Cooperative Study Group was formed for the purpose of evaluating intrathecal antibiotic therapy in neonatal meningitis caused by gram-negative bacteria. The rationale for choosing this route of administration was that direct instillation of an antibiotic into the lumbar intrathecal space might produce greater antimicrobial activity throughout the CSF volume and result in more rapid eradication of CSF bacteria and an improved outcome from disease.
MATERIALS AND METHODS Subjects. All patients admitted to the 18 participating institutions were considered condidates for study if they were two years of age or less and had meningitis caused by gram-negative enteric bacteria as determined by Gram stain of CSF and confirmed by CSF culture. Infants who had received antibiotic therapy prior to admission were not excluded if gram-negative organisms were still present on stained smears of CSF, regardless of the culture results. Patients with meningitis caused by Hemophilus influenzae, Neisseria meningitidis, or gram-positive pathogens were excluded as were infants with meningitis complicating meningomyeloceles or other structural abnormalities of the central nervous system. Informed written consent was obtained from the parents or guardians of study patients. Therapy assignment plan. The patients were allocated at random to receive either ampicillin intravenously and gentamicin intramuscularly (Group A) or these two anti-
Intrathecal therapy in enteric meningitis
67
Table II. Number of infants seen at each participating center
City
No. infants
Atlanta Baltimore Birmingham Boston Cincinnati Cleveland Dallas Medellin, Colombia Memphis
8 2 10 0 4 3 17 16 10
City
No. infants
Montreal Salt Lake City San Francisco MexicoCity Winnipeg Durham Los Angeles Pittsburgh Madison Total
7 5 7 10 4 6 4 1 3 117
biotics plus intrathecal gentamicin (Group B). Each institution had two books for assignment of infants to a particular therapy group: one book for infants with a birth weight equal to or less than 2,500 gm and one book for infants who weighed more than 2,500 gm at birth. An additional therapy assignment book for infants greater than 30 days of age was used in Medellin, Colombia, where enteric meningitis is commonly encountered in older infants. In each book there were sealed, numbered envelopes containing cards with treatment assignment. The sequence had been determined by utilizing tables of random permutations of 16 numbers. Upon admission of an infant to the study, an envelope in numerical sequence was opened to ascertain the therapy regimen to be utilized. Drug dosage and clinical management. Table I details the drug dosage and patient management. All infants were t o be treated with the assigned regimen for a minimum of three days after which; therapy could be altered (1) if the pathogen was not susceptible to the antibiotics used, (2) if CSF cultures failed to sterilize, or (3) if the infant had not improved clinically. Therapy changes were individualized and were not directed by the study protocol. Intrathecal therapy was maintained for a minimum of three days or until CSF cultures were sterile. Ampicillin and gentamicin were continued for two weeks after sterilization of CSF cultures or for a minimum of three weeks, whichever was longer. Follow-up examinations. Survivors were scheduled for follow-up evaluations at six, 12, 24, 36, and 48 months, respectively, after illness. Complete physical, neurologic, and Denver Developmental examinations were performed at these follow-up visits. The Cattell Infant Intelligence and Gessell Fine and Gross Motor Development tests were administered to some infants 12 months of age and older.
68
McCracken and Mize
The Journal of Pediatrics July 1976
Table IlL Population characteristics of the two therapy groups Systemic therapy Systemicplus only intrathecal therapy (65 infants) (52 infants)
Male Female
36 (55%) 29 (45%)
28 (54%) 24 (46%)
Caucasian Black
53 (82%) 12 (18%)
43 (83%) 9 (17%)
21 (32%) 35 (54%)
17 (33%) 25 (48%)
9 (14%)
10 (19%)
9.80* _+ 1.95 days 125 _+ 95 days
11.88 _+ 2.57 days 214 _+ 122 days
Birth weight 2,500 gm > 2,500 gm Age > 30 days Age on enrollment 30 days > 30 days
*Mean _+ 2 SE of the mean.
Table IV. Etiology of neonatal meningitis
Organism Eseherichia coli Salmonella species Citrobacter diversus Proteus mirabilis Serratia species Klebsiella species Enterobacter species None* Moraxella species Multiple organismst Pseudomonas aeruginosa
Number
82 7 5 ~ 5 4 4 2 1 1 1
Relative frequency
70% 6% 4% 4% 4% 3% 3% 2% 1% 1% 1%
*Gram-negative rods observed on stained smears of CSF; cultures sterile. rE. coli non-Kl, Klebsiella species, and Flavobacterium meningosepticure.
An infant was considered normal if the physical, neurologic, and developmental examinations revealed no abnormalities that could be attributed to meningitis. Mild-to-moderate neurologic residua were said to be present if the infant had compensated hydrocephalus or neurologic abnormalities which were not severe enough to require special care or to prevent adequate, albeit delayed, development. Severe neurologic sequelae included failure to develop beyond the age at which the disease occurred a n d / o r requirement of custodial care. Laboratory studies. Antimicrobial susceptibilities of the pathogens were performed in each hospital laboratory. These strains were also studied in Dallas using standard plate dilution and tube dilution techniques? Serum and
CSF specimens were assayed for gentamicin activity using the micromethod of Simon and Yin" with Bacillus subtilis as the t~st organism. Patient variables and therapy evaluation. Basic identifying data such as date of enrollment, age, weight classification, sex, and race were recorded for every patient. In addition, a detailed case report form was prepared for each patient enumerating the obstetrical and medical history, treatment regimen, laboratory data, neurologic residua, findings at follow-up examinations, and a detailed summary of the course of illness and complications of disease or treatment. Autopsy reports were also used in the analysis of data. Statistical methods. Chi-square tests were used. TM When 20% or more of the expected number of observations in a contingency table were fewer than 5, Fisher's exactprobability method was utilized. Mean differences were evaluated by Student's t test. 11 Differences that had a probability of greater than 0.05 by the appropriate null hypothesis were considered not significant. Standard errors of the mean noted in this paper are expressed as the mean plus or minus 2 SE of the mean. RESULTS
Study population. During the 48-month period from September, 1971, through August, 1975, the 18 institutions enrolled 133 patients in the study. Sixteen infants were found unacceptable for analysis for the following reasons: (1) in 13 cases CSF cultures were negative or yielded nonenteric organisms; (2) two infants died before antimicrobial therapy was initiated; and (3) one infant with Pseudomorlds meningitis was treated with carbenicillin and gentamicin instead of ampicillin and gentamicin. The remaiffing 117 infants comprise the study population. Distributioh of cases among the 18 institutions is shown in Table II. Sixty-five infants were assigned to Therapy Group A and 52 infants to Therapy Group B (intrathecal gentamicin). Table III shows that there was a random distribution of patients between the two therapy groups with regard to age on enrollment, birth weight, sex, race, and number of infants > 30 days of age. Antibiotic regimens were given for the time designated by the protocol in 86% of infants. Changes in antibiotic therapy were made in ten infants given systemic therapy alone and in seven infants treated with systemic plus intrathecal therapy. In addition, nine (17%) of the 52 infants assigned to Group B failed to receive three intrathecal gentamicin doses: five patients were given only one dose and four infants received two doses. The remaining 43 infants received three to 11 (average, four) intrathecal doses.
Volume 89 Number 1
Intrathecal therapy in enteric meningitis
69
Table V. Case fatality rates for the two therapy groups Systemic therapy only (65 infants) Category
Total Birth weight 2,500 gm > 2,500 gm Age Neonates only ( 2 30 days) Infants ( > 30 days) Death due to meningitis Death after 48 hr of therapy Etiology: E. coli only Etiology: Non-E. coli Pathogens susceptible to both antibiotics*
Systemic plus intrathecal therapy (52 infants)
Dead
A live
% Dead
Dead
A live
% Dead
Total population
Chi square "'P" value
21
44
32.3
16
36
30.8
117
0.982
10 7
11 28
47.6 20.0
7 4
10 21
41.2 16.0
38 60
0.945 0.955
17
39
30.4
11
31
26.2
98
0.821
4 21 12
5 44 44
44.4 32.3 21.4
5 14 12
5 36 36
50.0 28.0 25.0
19 115 104
0.586 0.769 0.843
15 6 14
34 10 37
30.6 37.5 27.5
7 9 11
26 10 25
21.2 47.3 30.6
82 35 87
0.492 0.807 0.941
*Represents only those pathogens known to be susceptible to both ampicillin and gentamicin. Etiology. Escherichia coli was the etiologic agent in 82 (70%) of the 117 cases (Table IV). Fifty-eight (71%) of the E. coli strains possessed capsular K1 antigen. Patients with E. coli K1 meningitis were equally distributed between the two therapy groups. Six of the seven patients with Salmonella meningitis and four o f the five infants with disease caused by Serratia species were from Medellin, Colombia. Outcome. Thirty-seven (32%) infants died (Table V). Full-term infants had a significantly lower (P < 0.01) mortality rate (18%) when compared with low-birthweight infants (45%) and with infants greater than 30 days of age (48%). Case fatality rates were not significantly different in t h e two therapy groups nor were they different when subcategories were analyzed according to birth weight or age. Two infants who received intrathecal therapy died from causes unrelated to meningitis (congenital heart disease and gastrointestinal bleeding secondary to ulcerative colitis). Excluding these two patients from the analysis did not significantly influence the mortality statistics (Table V). Differences in death rates between the two treatment groups were not affected when the following variables were assessed: (1) death before completion of 48 hours of therapy; (2) meningitis caused b y E. coli strains only; (3) disease due to non-E, coli organisms; (3) susceptibility of the pathogens to ampicillin and gentamicin; (5) antibiotic therapy prior to enrollment; (6) failure to receive a m i n i m u m of three intrathecal doses; or (7) when patients from Colombia and Mexico were excluded from the
Table VI. Neurologic status of 80 infants who survived gram-negative meningitis Last examination after illness
Therapy group*
12 mo or greater
Group A Group B
16 17
8 7
4 2
28 26
6to 11 mo
GroupA Group B
6 5
2 0
0 1
8 6
Hospital discharge
Group A Group B
4 3
4 1
0 0
8 4
51
22
7
80
Total
Neurologic status? MiM-moderate Severe Normal residua residua
Total
*Group A--systemicantibiotics only; Group B systemic antibiotics plus intrathecal gentamicin. tSee text for definitions.
analysis. Excluding one patient who died more than two years after institution o f therapy (Group B), the m e a n number of days from onset of therapy to death was 33 _+ 32 days for infants in Group A and 26 + 23 days for infants in Group B. Eighty (68.4%) infants survived and have been evaluated for variable intervals after illness (Table VI). Twelve (15%) infants had hospital discharge evaluations only, fourteen (17.5%) patients have been examined between
70
McCracken and Mize
six to 11 months after illness, and 54 (67.5%) infants have been tested from one to four years after illness. There were no signifcant differences in the neurologic status of infants within the two therapy groups regardless of the time after illness when evaluations were performed. Fiftyone (64%) of the surviving infants were considered normal, 22 (27%) had mild-to-moderate neurologic sequelae, and seven (9%) infants were judged to have severe neurologic handicaps. Of these patients, 13 im Group A and eight in Group B have failed to return for current evaluation and therefore, are considered to be lost to follow-up. Age on enrollment, weight classification, race, sex, or type of organism did not appear to affect inorbidity. Days of hospitalization. Infants enrolled in Group A were hospitalized for an average of 24 _+ 4 d@s and those in Group B for 27 _+ 5 days. Cerebrospinai fluid studies. CSF cultures were positive for 3.3 _+ 0.6 days for infants in Group A and 3.9 _+ 1.2 days for infants in Group B. When only the surviving infants were considered, the average duration of positive CSF cultures was 3.3 days in both groups. The CSF white blood cell counts and protein contents were analyzed with respect to outcome from disease. When the single largest value for each patient was analyzed, the mean CSF leukocyte ~ount was significantly greater for infants who died (23.7 _+ 9.4 x 103 W B C / mm 3) than for those who survived (9.9 _+ 2.4 x 10mm0their illness. Similarly, CSF protein concentrations were significantly greater in those who died than in survivors (1,282 _+ 626 mg/dl versus 517 _+ 120 mg/dl, respectively). Special laboratory studies. Bacteria from 108 patients were studied for antimicrobial susceptibility. Only one organism, an E. coli, was resistant to gentamicin (MIC 10t~g/ml) and had an ampicillin MIC of 20 t~g/ml. In spite of this relative antibiotic resistance, CSF cultures sterilized after two days of therapy and the infant was normal 12 months after illness. Thirty (28%) of the 108 organisms were resistant to ampicillin (M1C > 10 /~g/ml). Of the 78 susceptible strains, 43 (55%) strains had MIC values of 2.5/zg/ml or less. Ampicillin-resistant strains were equally distributed between the two study groups and mortality rates for all infants with resistant and susceptible strains were not significantly different (40% and 29%, respectively). Outcome was not significantly different in the 30 infants with meningitis due to ampicillin-resistant organisms from that of 30 patients with susceptible pathogens (MIC ~ 2.5 /~g/ml) selected randomly from the other study patients. Nine (30%) infants with ampicillinresistant organisms died compared with six (20%) in the
The Journal of Pediatrics July 1976
ampicillin-susceptible group. Fifteen (50%) infants with ampicillin-resistant organisms were normal or had mildto-moderate neurologic sequelae on follow-up examinations, whereas ten infants (33%) with susceptible pathogens had similar evaluations. CSF cultures remained positive for approximately three days in both groups. In vitro antimicrobial synergism studies were not performed with these ampicillin-resistant organisms. Previous studies demonstrated that ampicillin and gentamicin act synergistically against approximately 40% of E. coli strains isolatedfrom neonates. 1~ Gentamicin assays were performed on one or more CSF samples from 43 infants. Specimens obtained at various intervals after 2.5 mg/kg doses given intramuscularly contained 0.3 to 3.7/~g/ml (mean, 1.6/~g/ml) gentamicin. Gentamicin concentrations in CSF taken at similar times after parenteral therapy and at 18 to 24 hours after 1 mg gentamicin given into the lumbar intrathecal space were 0.5 to 3.4 t~g/ml (mean, 1.6 t~g/ml). Four infants had lumbar punctures performed two to four hours after 1 mg intrathecal doses and the CSF gentamicin levels were 18.4, 25, 36.8, and > 40/~g/ml, respectively. Safety. Parenteral antimicrobial therapy was well tolerated and there was no evidence of drug-related renal abnormalities or ototoxicity. There were no acute adverse reactions due to intrathecal gentamicin therapy. Cerebrospinal fluid white blood cell counts and protein values during illness and at the time of hospital discharge were similar for the two therapy groups. One infant was first noted to have left lower extremity weakness and hyperreflexia 12 months after illness. This infant was given 1 mg gentamicin intrathecally for three days and because of continued positive CSF cultures, 1 mg polymyxin B was administered daily for four days. The patient's neurologic findings improved but were still present at three and onehalf years of age when fine motor function and mental development were considered normal. No abnormalities were observed among the other infants who received intrathecal therapy. DISCUSSION The results of this collaborative project establish that 1 mg gentamicin administered daily into the lumbar intrathecal space for three days or longer has no discernable beneficial or adverse effect on outcome from neonatal enteric meningitis when compared with systemic therapy alone. Disparity in the number of patients assigned to the two therapy groups was due to the occasional incorrect usage of the therapy assignment books, inability to initiate intrathecal therapy in a few infants, and incorrect diagnosis in 13 infants at the time of admission to the study.
Volume 89 Number I
Even so, characteristics of the two groups at the onset of treatment where comparable and no significant differences in mortality and long-term morbidity rates were observed. Outcome in these infants was better than that reported by others I-~ for neonatal gram-negative meningitis (mortality rates of 40% to 80%). The probable explanation for this improved outcome relates more to intensive management of the respiratory and metabolic problems commonly encountered in these patients than to the specific antibiotic regimens employed. Intrathecal therapy was chosen for study with the anticipation that direct instillation of drug into CSF would result in higher gentamicin concentrations and improved bactericidal activity. Althoug~ gentamicin concentrations of 18 to > 40 /~g/ml were observed in lumbar CSF several hours afterqntrathecal administration, the levels were only 1 to 3~/~g/ml 18 to 24 hours later. These latter values are similar to those measured in CSF from infants treated with parenteral therapy only. Rahal and associates la found lumbar CSF gentamicin concentrations of 20 to 50 /~g/ml two to eight hours after 4 mg intrathecal doses in adults. The half-life of gentamicin in CSF was five hours. The relatively rapid disappearance of gentamicin from CSF of patients in these two studies may be explained by a dilutional effect secondary to diffusion of drug in the CSF space, transport of drug out of the CSF, or inactivation of drug. The third possibility is unlikely because gentamicin is extremely stable and such factors as reduced pH, protein-binding, or ionic composition of CSF would be counteracted by the conditions of the bioassay. Spector 1~ has demonstrated that gentamicin is transported out of CSF of rabbits by an active process within the choroid plexus. Whether this efflux mechanism is operative in the regulation of CSF gentamicin concentrations in human beings is unknown. In order for intrathecal therapy to be bacteriologically effective, the antibiotic must diffuse throughout the entire subarachnoid space and, perhaps also, into the ventricles. Kaiser and McGee 1~have recently studied the pharmacokinetics of gentamicin and tobramycin in lumbar, cisternal, and ventricular fluids after lumbar intrathecal administration to adults with gram-negative meningitis. Peak gentamicin and tobramycin concentrations of 27 to 80/~g/ ml were noted in lumbar CSF during the first six hours after 5 to 10 mg intrathecal doses. The antibiotics disappeared rapidly from CSF (average half-life, six hours). The aminoglycosides failed to enter the cerebral ventricles and were poorly distributed to other parts of the CSF space. *~ These data are consistent with the known unidirectional flow of CSF out of the ventricles. It is likely that delayed bacteriologic cure results from an inability to deliver adequate antibiotic concentrations
Intrathecal therapy in enteric meningitis"
71
throughout the central nervous system. One possible inaccessible site of infection is the ventricular system. In the present study, ventriculitis was documented in eight of 12 infants at postmortem examination and in five of seven patients in whom ventricular taps were done during the course of illness. Berman and Banker 1~ demonstrated ventriculitis on postmortem examination of 23 of 25 neonates with meningitis and Salmon 17 encountered venfricular infection in eight of 12 similarly affected babies. Although the total number of patients evaluated is small, these studies indicate that ventriculitis occurs commonly in neonates with meningitis, particularly when there is delayed sterilization of CSF cultures. The inadequate delivery of aminoglycoside to the ventricles after intralumbar therapy was illustrated in an infant with Proteus morganii meningitis reported by Moelloring and Fischer. 1~ Despite parenteral and intrathecal gentamicin CSF cultures failed to sterilize and ventriculitis was diagnosed on the seventeenth day of therapy. Pharmacologic studies demonstrated transiently elevated lumbar CSF gentamicin concentrations and very low ventricular levels after intralumbar administration. Intraventricular gentamicin therapy produced rapid and uniform distribution of antibiotic throughout the CSF space and prompt sterilization of ventricular and lumbar CSF cultures. ~r Similar pharmacokinetic data were reported by Kaiser and McGee? ~ When 5 mg of gentamicin was injected into the ventricles of adults, concentrations of 15 to 45 ~g/ml were found in ventricular fluid six hours later and values of 10 to 20/~g/ml were measured in lumbar fluid. During most of the remaining 24-hour period, concentrations of 5/~g/ml persisted in both sites. These gentamicin levels are within the therapeutic range for susceptible gram-negative bacteria. SPECULATION The intrathecal antibiotic regimen e m p l o y e d in this study proved to be no better than systemic therapy alone for gram-negative enteric meningitis of infancy. Clinical and autopsy data indicate that ventricular infection occurs commonly in this disease and that ventriculitis persists despite parenteral a n d intralumbar antibiotic therapy. Pharn?acologic studies offer an explanation for these observations. Aminoglycosides diffuse poorly throughout the CSF space after intrathecal administration and subtherapeutic concentrations are present in ventricular fluid. On the other hand, direct instillation of antibiotic into the ventricles produces concentrations of aminoglycoside that are many fold higher than the minimal bactericidal titers for most enteric pathogens. On the basis of these findings and the continued high mortality and morbidity rates associated with neonatal
72
McCracken and Mize
gram-negative meningitis, a new approach to antimicrobial therapy in this condition is necessary. Ideally, a ventricular tap should be performed at the onset of therapy in order to define the incidence of ventriculitis and to select the m o d e of antimicrobial therapy, More practically, a ventricular tap must be performed in an infant with positive lumbar CSF cultures for two or three days after intrathecal a n d / o r parenteral ~tntibiotic therapy. If ventriculitis is diagnosed, intraventricular aminoglycoside therapy should be used. This may be accomplished by daily taps or by instilling antibiotic through an implanted reservoir. A more direct approach to the diagnosis and treatment of ventricular infection should be evaluated. Merely altering the conventional antibiotic regimens for neonatal meningitis is not likely to improve outcome from this disease and may result in adverse effects if dosages are increased to obtain higher C S F concentrations. We thank the many pediatric faculty members, residents, fellows, and nurses at the 18 institutions who managed the study patients. Without their cooperation and interest this study could not have been performed. Norma Threlkeld, R.N., coordinated data collection and follow-up reports. REFERENCES
1. Haggarty RJ, and Ziai Mohsen: Acute bacterial meningitis, Adv Pediatr 13:129, 1964. 2. Overall JC: Neonatal bacterial meningitis, J PEI~IATR 76:499, 1970. 3. Groover RV, Sutherland JM, and Landing BH: Purulent meningitis of newborn infants, N Engl J Med 264:1115, 1961. 4. Fitzhardinge PM, Kazemi M, Ramsay M, and Stern L: Long-term sequelae of neonatal meningitis, Dev Med Clin Neurol 16:3, 1974. 5. Sell SHW, Merrill RE, Doyne EO, and Zimsky EP: Long-
The Journal of Pediatrics July 1976
6.
7.
8.
9. 10.
11. 12.
13.
14.
15.
16.
17. 18.
term sequellae of Hemophilus influenzae meningitis, Pediatrics 49:206, 1972. McCracken GH: The rate of bacteriologic response to antimicrobial therapy in neonatal meningitis, Am J Dis Child 123:547, 1972. McCracken GH, Sarff LD, Glode MP, Mize SG,. Schiffer MS, Robbins JB, Gotschlich EC, J~'rskov I, and ~rskov F: Relation between Escherichia coli K1 capsular polysaccharide antigen and clinical outcome in neonatal meningitis, Lancet 2:246, 1974. Haltalin KC, Markley AH, and Woodman E: Agar plate dilution method for routine antibiotic susceptibility testing in a hospital laboratory, Am J Clin Pathol 60:384, 1973. Simon HJ, and Yin EJ: Microbioassay of antimicrobial agents, Appl Microbiol 19:573, 1970. Siegel S: Non-parametric statistics for the behavioral sciences, New York, 1956, McGraw-Hill, Book Company, Inc., pp 96 and 197-200. Ostle B: Statistics in research, Ames, Iowa, 1963, The Iowa State University Press, p 119. Kaplan JM, McCracken GH, Horton LJ, Thomas MS, and Davis N: Pharmacologic studies in neonates given large dosages of ampicillin, J PEOIATR84:571, 1974. Rahal JJ, Hyams PJ, Simberhoff MS, and Rubinstein E: Combined intrathecal and intramuscular gentamicin for gram-negative meningitis, N Engl J Med 290:1394, 1974. Spector R: The transport of gentamicin in the choroid plexus and cerebrospinal fluid, J Pharmacol Exp Ther 194:83, 1975. Kaiser AB, and McGee ZA: Aminoglycoside therapy of gram negative bacillary meningitis, N Engl J Med 293:1215, 1975. Berman PH, and Banker BQ: Neonatal meningitis: A clinical and pathologic study of 29 cases, Pediatrics 38:6, 1966. Salmon JH: Ventriculitis complicating meningitis, Am J Dis Child 124:35, t:972. Moellering RC, and Fischer EG: Relationship of intraventricular gentamicin levels to cure of meningitis, J PEOIAXR 81:534, 1972.
