J Neumsurg 75:541-544, 1991
Ventricular cerebrospinal fluid eosinophilia in children with ventriculoperitoneal shunts HOWARD TUNG, M.D., COREY RAFFEL, M.D., PH.D., AND J. GORDON MCCOMB, M.D.
Division of Neurosurgery, ChildrensHospital Los Angeles, and Department of Neurosurgery, University of Southern California School of Medicine, Los Angeles, California ~" To determine the significanceof cerebrospinal fluid (CSF) eosinophilia, the charts of 106 patients treated with shunt-related proceduresduring the calendar year 1985 were reviewed. Sixty-ninepatients presented for a shunt revision; their charts were retrospectivelyreviewed from the time of shunt insertion until January, 1988. The remaining 37 patients had a ventriculoperitonealshunt inserted during the study period and were subsequently followed to January, 1988. A total of 558 shunt-related procedures were performed on these patients during the study period, with a mean follow-upperiod of 6.9 years. The infection rate was 3.8%. Eosinophilia was diagnosed when eosinophils accounted for 8% or more of the total CSF white blood cell count. Ventricular CSF eosinophilia occurred in 36 patients sometime during their clinical course. These 36 patients required a mean of 8.5 shunt revisions,while the remaining patients required a mean of 2.5 revisions (p < 0.001). Shunt infectionswere also more frequent in patients with eosinophilia(p < 0.01). In no case was peripheral eosinophilia or a parasitic infection present. This study demonstrates that CSF eosinophilia is common in children with shunts. Children with this laboratory finding will experience more shunt failures. In addition, the new appearance of eosinophilia in the CSF of a patient with a shunt in place suggests the possibilityof a shunt infection. KEY WORDS 9 eerebrospinal fluid shunt
OSINOPHILSare not normally present in the cerebrospinal fluid (CSF) and their appearance in CSF is usually associated with parasitic infections of the central nervous system.~~ Cerebrospinal fluid eosinophilia in patients with ventriculoperitoneal (VP) shunts has only been sporadically reported.9at'14' J8.19The meaning of this laboratory finding in a shuntimplanted population remains uncertain. To determine the significance of CSF eosinophilia in children with VP shunts, the charts of 106 patients who presented at Childrens Hospital of Los Angeles for shunt-related procedures during the calendar year 1985 were reviewed. This report describes the incidence of CSF eosinophilia in these patients and relates this finding to the patients' clinical course.
E
Clinical Material and Methods
Chart reviews were performed for all patients admitted for operative shunt procedures during 1985. After identifying the patients, the entire chart for each was reviewed from the time of initial shunt insertion through January, 1988. Of 120 patients identified, 14
J. Neurosurg. / Volume 75/October. 1991
9 eosinophilia
9 ventriculoperitoneal shunt
9
were excluded because of incomplete medical records. The remaining 106 patients form the basis of this study. All patients received identical shunt hardware, consisting of a slotted ventricular catheter reservoir made of barium-impregnated silicone, a cruciform slit valve of either very low, low, medium, or normal pressure, and a 120-cm open-ended barium-impregnated silicone peritoneal catheter.* The long peritoneal catheter eliminated the need for elective shunt lengthening in these patients, and no patient in this series underwent an elective shunt revision. All patients received vancomycin (15 mg/kg) intravenously at the start of each procedure and for 24 hours postoperatively. After placement of a new ventricular catheter reservoir and collection of a ventricular CSF specimen, all patients received an intraventricular injection of 4 mg gentamicin through the catheter. The CSF samples examined were obtained either: 1) from a percutaneous shunt tap performed at the time * Catheter manufactured by Holter-Hausner International, Inc., Bridgeport,Pennsylvania. 541
H. T u n g , C. Raffel, a n d J. G. M c C o m b TABLE 1
Incidence ~f eosinophilia m 1o6 palie~lt~ receiving ventricufl~peritoneal .~'lnml,~ Etiology of Hydrocephalus
Eosinophilia
No Eosinophilia
premature birth: intraventricular hemorrhage congenital hydrocephalus myelodysplasia tumor
14
15
19
9 7 2
19 19 11
28 26 13
postinfection trauma total cases
3 1 36
4 2 70
7 3 106
Total Cases
of clinically suspected shunt malfunction or infection; or 2) in the operating room at the time of shunt insertion or revision for documented shunt occlusion. The CSF specimens were analyzed for glucose content, total protein content, cell count with differential determination, and aerobic and anerobic culture. Prior to all operative procedures, a peripheral white blood cell (WBC) count with differential analysis was obtained. No patient had a peripheral eosinophil count of greater than 4% of the total WBC count. Statistical comparisons between groups were made with the nonpaired Student t-test and the chi-square test. Results Of the 106 patients studied, 37 had their initial shunt insertion during the study period and were followed until January, 1988. Sixty-nine patients had a shunt revision during the study period and their charts were reviewed from the time of their shunt insertion until January, 1988. There were 68 boys and 38 girls in the study population. The follow-up period ranged from 2 to 20 years, with a mean of 6.9 years. Of the 106 patients, 17 had a shunt insertion only, 21 had an insertion and a single revision, 24 had two revisions, and 44 had three or more revisions. A total of 558 shunt-related procedures were performed on this group of patients. Infections occurred after 21 procedures (an incidence of 3.8%). The etiology of the hydrocephalus in this group of patients is listed in Table 1. The most common causes were intraventricular hemorrhage after premature birth, aqueductal stenosis, and myelodysplasia. A histogram showing the distribution of CSF eosinophil count as a percentage of the total CSF WBC count is presented in Fig. 1. A majority of patients exhibited no eosinophils in their CSF. A natural division in the histogram occurred at 8%. Based on this distribution, we have defined significant CSF eosinophilia to be 8% or more of the total CSF WBC count. Thirty-six (34%) of the 106 patients met this criterion for CSF eosinophilia at some time during their clinical course.
542
FIG. 1. Histogram showing the distribution of eosinophil counts, expressedas a percentage of total cerebrospinal fluid white blood cells, in the study population of 106 patients with ventriculoperitoneal shunts. Note the natural break between 7% and 8%. On examination of CSF at the time of shunt insertion, eosinophilia was present in only one patient. This patient developed hydrocephalus soon after a posterior fossa craniectomy and gross total resection of a cerebellar astrocytoma but has required no further shuntrelated procedure. All other patients developed CSF eosinophilia after the initial shunt insertion. Because CSF was obtained from these patients only when a shunt occlusion or infection was suspected, we were unable to assess accurately the duration of CSF eosinophilia. However, in all patients but one, CSF eosinophilia was a transient finding. While the CSF eosinophilia in a given patient may have been found in a number of serial taps, a CSF sample was eventually obtained after a period of weeks to months that contained less than 8% eosinophils. Patients with CSF eosinophilia experienced an average (_+ standard deviation) of 8.5 _+ 6.5 shunt revisions (range 0 to 24), while patients with no eosinophilia experienced 2.1 _+ 2.5 revisions (range 0 to 13) (p < 0.001). Twelve (32.4%) of the 37 patients with shunt insertion in 1985 had CSF eosinophilia during the follow-up period. These 12 patients required an average of 3.4 _+ 3.3 revisions (range 0 to 10), compared to 0.8 _+ 1.3 revisions for the remaining 25 patients (range 0 to six) (p < 0.01). Similarly, 24 (34.7%) of the 69 patients who presented for a shunt revision during 1985 had CSF eosinophilia at some time during their course. These 24 patients required 11.0 _+ 6.3 revisions (range three to 24), compared to 2.8 _ 2.6 revisions (range 0 to 13) required by the remaining patients (p < 0.001). Thirteen shunt infections occurred in the 36 patients with eosinophilia compared to eight infections in the 70 patients without eosinophilia (chi-square = 7.62, p < 0.01). Among the 12 patients with eosinophilia whose shunts were inserted in 1985, there were five infections, while in the 25 patients without eosinophilia, there were two infections (chi-square = 4.0, p < 0.05). The 24 patients with eosinophilia whose shunts were revised in 1985 suffered eight infections, while four inJ. Neurosurg. / Volume 75 / October. 1991
Cerebrospinal fluid eosinophilia in children with VP s h u n t s TABI.E 2 Association o/'shunt it!/~'ction with eosim~philia in 21 ca,ses
Case No.
Infecting Organism
1 2 3 4 5 6
Staplo'lococcus cpidermMis Staphylococcus epidermMA Proteus sp. Staph.vlococcus ~Tffdermidis l'scheJqchia coli Staph.vloc~,'ccu~aIII'CIIS
7
cullure negative Psettdomonas sp. culture negative
8 9
I0 11 12 13 14 15
16 17 18 19 20
21
Staph)'locr Staph.vloc~'cus Staphylococcus Staphylococcus Staphl,lo~vccus
ato'ett.~ attreus epidermidis aureus epidermidis
culture negative Sabnonella sp.
Staphylococcus attreus Staph)'lococcus epiderrnidis Staphylococcus epidermidL~" Kh,bswlla sp. Pseudomonas sp.
Eosinophilia Present no
yes yes yes yes no no no no yes yes yes yes yes yes yes* no no no yes yes
* Transient eosinophilia 5 years prior to peritoneal infection.
fections occurred in the remaining 45 patients (chisquare = 4.93, p < 0.05). Thirteen of 36 patients with CSF eosinophilia experienced shunt infections; among these, CSF eosinophilia was initially documented at the time the shunt infection was discovered in nine (69.2%). In an additional three cases, persistent CSF eosinophilia was identified from 1 week to 4 months before a shunt infection was documented. Thus, eosinophilia was closely related to the discovery of infection in 12 (92.3%) of 13 cases in which both infection and eosinophilia occurred. In the one remaining patient, a transient CSF eosinophilia was identified 5 years prior to a distal shunt infection. The organisms causing the shunt infections in the patients both with and without eosinophilia are presented in Table 2. In neither group did one organism predominate as compared to the other. Because patients with shunt infections frequently require more shunt revisions, the data for the number of revisions were reassessed after exclusion of patients experiencing infection. Patients with eosinophilia and no infection required an average of 7.0 _ 6.3 revisions (range 0 to 24), compared to an average of 2.0 _+ 2.4 revisions (range 0 to 13) in the remaining patients (p < 0.001). The number of revisions required by patients with both eosinophilia and infection averaged I 1.1 ___ 6.5 (range one to 24), while the patients without eosinophilia but with infection required an average of 3.5 _+ 3.2 revisions (range one to nine) (p < 0.01). Total protein (175 _+ 305 mg/100 ml) and glucose (53 _+ 24 mg/100 ml) values for the patients with CSF eosinophilia did not differ greatly from the values in patients without eosinophilia (128 _ 350 mg/100 ml and 76 _ 70 mg/100 ml, respectively). J. Neurosurg. / Voh~me 75 / October, 1991
Discussion
The earliest report of finding eosinophils in CSF was made in 1907 in relation to neurosyphilis. T M Since that report, CSF eosinophilia has been observed in a variety of pathological processes involving the central nervous system, the most common being a parasitic infection. Cerebrospinal fluid eosinophilia is a rare finding. Bosch and Oehmichen: reviewed 10,000 CSF specimens and, excluding patients with helminth infections, found only 94 specimens with eosinophils accounting for more than 1% of the total CSF WBC count. Our study suggests that the incidence of CSF eosinophilia in children with VP shunts is remarkably high. In the series of Bosch and Oehmichen,2 six of the 94 cases of eosinophilia occurred in children with shunts; the number of CSF samples in their series obtained from patients with shunts was not stated. Tzvetanova and Tzekov ~9reported a series of 404 children shunted for "internal" hydrocephalus and found a 6.4% incidence of CSF eosinophil count greater than 1% of the total CSF WBC count. In our series, 34% of patients with shunts developed a CSF eosinophil count of 8% or greater at some time during their clinical course. The high incidence in our series may relate to our routinely processing all CSF specimens for a differential cell count of the CSF sediment, allowing the discovery of a high percentage ofeosinophil even when the total CSF WBC count is low. In dividing our patients into those with and those without CSF eosinophilia, we identified two striking differences. First, the appearance of eosinophils in the CSF at any time during a patient's clinical course indicated the likelihood that more shunt obstructions, and thus more revisions, would occur in that patient. This increase occurred even in patients where CSF eosinophilia was a transient finding. The increase was evident in both the total population and the population remaining after removal of those patients with a shunt infection. It also applied both to those patients having their initial shunt insertion in 1985 and to those undergoing a revision in 1985. No predisposing factor, including age at insertion, sex, etiology of hydrocephalus, or number of shunt procedures before development of CSF eosinophilia, could be identified in the group of patients developing CSF eosinophilia. The role of the eosinophil in shunt obstruction is not known. Eosinophils at the periphery are attracted to areas of immunoglobulin E (lgE)-mediated hypersensitivity or to areas with lymphocytic infiltrates, as some lymphocytes can release eosinophil chemotactic factors. z3`8"~~176 The silicone used to manufacture shunt hardware is immunogenic and may induce a delayed hypersensitivity reaction. 6'~2 Sekhar, et al., n demonstrated chronic granulomatous inflammation in the tissue associated with shunts removed for culture-negative obstruction. Gower, et al.,S used scanning electron microscopy to document the chronic granulomatous response of tissue around shunts having culture-nega543
H. Tung, C. Raffel, and J. G. McComb tive malfunctions. These authors suggested a delayed hypersensitivity response to the silicone as the cause of inflammation. Cerebrospinal fluid eosinophilia may be a marker for those patients mounting a hypersensitive response to the silicone. This immune response is hypothesized, on the basis of increased cells and protein massing around the shunt, to be responsible for the increased number of shunt failures seen in patients with CSF eosinophilia. Delayed hypersensitivity to shunt material seems to be an unlikely cause of CSF eosinophilia, because the eosinophilia does not persist in a given patient. In addition to malfunction, shunt infection was also more common in patients with eosinophilia. The appearance of the eosinophils in the CSF coincided temporally with the documentation of infection in over 90% of the patients with both eosinophilia and infection. Shunt infections may be caused by indolent organisms that are difficult to culture, such as Propionibacterium species. ~'4't5 Perhaps some of the patients with multiple shunt malfunctions in the presence of CSF eosinophilia had unidentified low-grade infections. The infection may have been cleared by the removal and replacement of hardware during revisions. Based on the data presented here, the new appearance of eosinophils in the CSF of a shunted patient serves as an indicator of possible infection, warranting a workup for infection. In two reported studies, CSF eosinophilia was associated with an increased CSF protein level. 2'~9 In this series, protein levels were elevated in patients both with and without CSF eosinophilia. The increased protein in both groups may relate to the even distribution between the groups of patients with intraventricular hemorrhage and meningitis. The normal glucose values found in both groups of our patients are in accord with findings reported elsewhere. 2't9 In conclusion, this study demonstrates that CSF eosinophilia is an important and common finding in patients with VP shunts. Children with this finding will experience a significantly increased number of shunt obstructions and shunt infections over those without eosinophilia. References
1. Beeler BA, Crowder JG, Smith JW, et al: Propionibacterium acnes: pathogen in central nervous system shunt infection. Report of three cases including immune complex glomerulonephritis. Am J Meal 61:935-938, 1976 2. Bosch I, Oehmichen M: Eosinophilic granulocytes in cerebrospinal fluid: analysis of 94 cerebrospinal fluid specimens and review of the literature. J Neurol 219: 93-105, 1978
544
3. Butterworth AE, David JR: Eosinophilic function. N Engl J Med 304:154-156, 1981 4. Everett ED, Eickhoff TC, Simon RH: Cerebrospinal fluid shunt infections with anaerobic diphtheroids (Propionibacterium species). J Neurosurg 44:580-584, 1976 5. Gower DJ, Lewis JC, Kelly DL Jr: Sterile shunt malfunction. A scanning electron microscopic perspective. J Nenrosurg 61:1079-1084, 1984 6. Heggers JP, Kossovsky N, Parsons RW, et al: Biocompatibility of silicone implants. Ann Plast Surg 11:38-45, 1983 7. Hirsch JG, Hirsch BZ: Paul Ehrlich and the discovery of the eosinophil, in Mahmoud AAF, Austen KF (eds): The Eosinophil in Health and Disease. New York: Grune & Stratton, 1980, pp 3-23 8. Hubscher T: Role of the eosinophil in the allergic reactions II. Release of prostaglandins from human eosinophilic leukocytes. J lmmunol 114:1389-1393, 1975 9. Jeanes AL: Cerebrospinal eosinophilia following Torkildsen's operation. Guys Hasp Rep 114:28-31, 1965 10. Kay AB: The eosinophil in infectious diseases, d Infect Dis 129:606-613, 1974 I I. Kessler LA, Cheek WR: Eosinophilia of the cerebrospinai fluid of non-infectious origin: report of 2 cases. Neurology 9:371-374, 1959 12. Kossovsky N, Heggers JP, Dujovny M, et al: Ventricular shunt failure: evidence of immunologic sensitization. Surg Forum 34:527-529, 1983 13. Kuberski T: Eosinophils in the cerebrospinal fluid. Ann Intern Med 91:70-75, 1979 14. Mine S, Sato A, Yamaura A, et al: Eosinophilia of the cerebrospinal fluid in a case of shunt infection: case report. Neurosurgery 19:835-836, 1986 15. Noetzel MJ, Baker RP: Shunt fluid examination: risks and benefits in the evaluation of shunt malfunction and infection. J Neurosurg 61:328-332, 1984 16. Punyagupta S, Juttijudata P, Bunnag T: Eosinophilic meningitis in Thailand. Clinical studies of 484 typical cases probably caused by Angiostrongylus cantonensis. Am J Trop Med Hyg 24:921-931, 1975 17. Sekhar LN, Moossy J, Guthkelch AN: Malfunctioning ventriculoperitoneal shunts. Clinical and pathological features. J Neurosurg 56:411-416, 1982 18. Traynelis VC, Powell RG, Koss W, et al: Cerebrospinal fluid eosinophilia and sterile shunt malfunction. Neurosurgery 23:645-649, 1988 19. Tzvetanova EM, Tzekov CT: Eosinophilia in the cerebrospinal fluid of children with shunts implanted for the treatment of internal hydrocephalus. Acta Cytol 30: 277-280, 1986 20. Walls RS, Beeson PB: Mechanism of eosinophilia. VIII. Importance of local cellular reactions in stimulating eosinophilic production. Clin Exp Immunol 12:111-119, 1972 Manuscript received October 29, 1990. Accepted in final form March 25, 1991. Address reprint requests to: Corey Raffel, M.D., Ph.D., 1300 North Vermont Avenue #906, Los Angeles, California 90027.
Z Neurosurg. / Volume 75 / October, 1991
Ventricular cerebrospinal fluid eosinophilia in children with ventriculoperitoneal shunts HOWARD TUNG, M.D., COREY RAFFEL, M.D., PH.D., AND J. GORDON MCCOMB, M.D.
Division of Neurosurgery, ChildrensHospital Los Angeles, and Department of Neurosurgery, University of Southern California School of Medicine, Los Angeles, California ~" To determine the significanceof cerebrospinal fluid (CSF) eosinophilia, the charts of 106 patients treated with shunt-related proceduresduring the calendar year 1985 were reviewed. Sixty-ninepatients presented for a shunt revision; their charts were retrospectivelyreviewed from the time of shunt insertion until January, 1988. The remaining 37 patients had a ventriculoperitonealshunt inserted during the study period and were subsequently followed to January, 1988. A total of 558 shunt-related procedures were performed on these patients during the study period, with a mean follow-upperiod of 6.9 years. The infection rate was 3.8%. Eosinophilia was diagnosed when eosinophils accounted for 8% or more of the total CSF white blood cell count. Ventricular CSF eosinophilia occurred in 36 patients sometime during their clinical course. These 36 patients required a mean of 8.5 shunt revisions,while the remaining patients required a mean of 2.5 revisions (p < 0.001). Shunt infectionswere also more frequent in patients with eosinophilia(p < 0.01). In no case was peripheral eosinophilia or a parasitic infection present. This study demonstrates that CSF eosinophilia is common in children with shunts. Children with this laboratory finding will experience more shunt failures. In addition, the new appearance of eosinophilia in the CSF of a patient with a shunt in place suggests the possibilityof a shunt infection. KEY WORDS 9 eerebrospinal fluid shunt
OSINOPHILSare not normally present in the cerebrospinal fluid (CSF) and their appearance in CSF is usually associated with parasitic infections of the central nervous system.~~ Cerebrospinal fluid eosinophilia in patients with ventriculoperitoneal (VP) shunts has only been sporadically reported.9at'14' J8.19The meaning of this laboratory finding in a shuntimplanted population remains uncertain. To determine the significance of CSF eosinophilia in children with VP shunts, the charts of 106 patients who presented at Childrens Hospital of Los Angeles for shunt-related procedures during the calendar year 1985 were reviewed. This report describes the incidence of CSF eosinophilia in these patients and relates this finding to the patients' clinical course.
E
Clinical Material and Methods
Chart reviews were performed for all patients admitted for operative shunt procedures during 1985. After identifying the patients, the entire chart for each was reviewed from the time of initial shunt insertion through January, 1988. Of 120 patients identified, 14
J. Neurosurg. / Volume 75/October. 1991
9 eosinophilia
9 ventriculoperitoneal shunt
9
were excluded because of incomplete medical records. The remaining 106 patients form the basis of this study. All patients received identical shunt hardware, consisting of a slotted ventricular catheter reservoir made of barium-impregnated silicone, a cruciform slit valve of either very low, low, medium, or normal pressure, and a 120-cm open-ended barium-impregnated silicone peritoneal catheter.* The long peritoneal catheter eliminated the need for elective shunt lengthening in these patients, and no patient in this series underwent an elective shunt revision. All patients received vancomycin (15 mg/kg) intravenously at the start of each procedure and for 24 hours postoperatively. After placement of a new ventricular catheter reservoir and collection of a ventricular CSF specimen, all patients received an intraventricular injection of 4 mg gentamicin through the catheter. The CSF samples examined were obtained either: 1) from a percutaneous shunt tap performed at the time * Catheter manufactured by Holter-Hausner International, Inc., Bridgeport,Pennsylvania. 541
H. T u n g , C. Raffel, a n d J. G. M c C o m b TABLE 1
Incidence ~f eosinophilia m 1o6 palie~lt~ receiving ventricufl~peritoneal .~'lnml,~ Etiology of Hydrocephalus
Eosinophilia
No Eosinophilia
premature birth: intraventricular hemorrhage congenital hydrocephalus myelodysplasia tumor
14
15
19
9 7 2
19 19 11
28 26 13
postinfection trauma total cases
3 1 36
4 2 70
7 3 106
Total Cases
of clinically suspected shunt malfunction or infection; or 2) in the operating room at the time of shunt insertion or revision for documented shunt occlusion. The CSF specimens were analyzed for glucose content, total protein content, cell count with differential determination, and aerobic and anerobic culture. Prior to all operative procedures, a peripheral white blood cell (WBC) count with differential analysis was obtained. No patient had a peripheral eosinophil count of greater than 4% of the total WBC count. Statistical comparisons between groups were made with the nonpaired Student t-test and the chi-square test. Results Of the 106 patients studied, 37 had their initial shunt insertion during the study period and were followed until January, 1988. Sixty-nine patients had a shunt revision during the study period and their charts were reviewed from the time of their shunt insertion until January, 1988. There were 68 boys and 38 girls in the study population. The follow-up period ranged from 2 to 20 years, with a mean of 6.9 years. Of the 106 patients, 17 had a shunt insertion only, 21 had an insertion and a single revision, 24 had two revisions, and 44 had three or more revisions. A total of 558 shunt-related procedures were performed on this group of patients. Infections occurred after 21 procedures (an incidence of 3.8%). The etiology of the hydrocephalus in this group of patients is listed in Table 1. The most common causes were intraventricular hemorrhage after premature birth, aqueductal stenosis, and myelodysplasia. A histogram showing the distribution of CSF eosinophil count as a percentage of the total CSF WBC count is presented in Fig. 1. A majority of patients exhibited no eosinophils in their CSF. A natural division in the histogram occurred at 8%. Based on this distribution, we have defined significant CSF eosinophilia to be 8% or more of the total CSF WBC count. Thirty-six (34%) of the 106 patients met this criterion for CSF eosinophilia at some time during their clinical course.
542
FIG. 1. Histogram showing the distribution of eosinophil counts, expressedas a percentage of total cerebrospinal fluid white blood cells, in the study population of 106 patients with ventriculoperitoneal shunts. Note the natural break between 7% and 8%. On examination of CSF at the time of shunt insertion, eosinophilia was present in only one patient. This patient developed hydrocephalus soon after a posterior fossa craniectomy and gross total resection of a cerebellar astrocytoma but has required no further shuntrelated procedure. All other patients developed CSF eosinophilia after the initial shunt insertion. Because CSF was obtained from these patients only when a shunt occlusion or infection was suspected, we were unable to assess accurately the duration of CSF eosinophilia. However, in all patients but one, CSF eosinophilia was a transient finding. While the CSF eosinophilia in a given patient may have been found in a number of serial taps, a CSF sample was eventually obtained after a period of weeks to months that contained less than 8% eosinophils. Patients with CSF eosinophilia experienced an average (_+ standard deviation) of 8.5 _+ 6.5 shunt revisions (range 0 to 24), while patients with no eosinophilia experienced 2.1 _+ 2.5 revisions (range 0 to 13) (p < 0.001). Twelve (32.4%) of the 37 patients with shunt insertion in 1985 had CSF eosinophilia during the follow-up period. These 12 patients required an average of 3.4 _+ 3.3 revisions (range 0 to 10), compared to 0.8 _+ 1.3 revisions for the remaining 25 patients (range 0 to six) (p < 0.01). Similarly, 24 (34.7%) of the 69 patients who presented for a shunt revision during 1985 had CSF eosinophilia at some time during their course. These 24 patients required 11.0 _+ 6.3 revisions (range three to 24), compared to 2.8 _ 2.6 revisions (range 0 to 13) required by the remaining patients (p < 0.001). Thirteen shunt infections occurred in the 36 patients with eosinophilia compared to eight infections in the 70 patients without eosinophilia (chi-square = 7.62, p < 0.01). Among the 12 patients with eosinophilia whose shunts were inserted in 1985, there were five infections, while in the 25 patients without eosinophilia, there were two infections (chi-square = 4.0, p < 0.05). The 24 patients with eosinophilia whose shunts were revised in 1985 suffered eight infections, while four inJ. Neurosurg. / Volume 75 / October. 1991
Cerebrospinal fluid eosinophilia in children with VP s h u n t s TABI.E 2 Association o/'shunt it!/~'ction with eosim~philia in 21 ca,ses
Case No.
Infecting Organism
1 2 3 4 5 6
Staplo'lococcus cpidermMis Staphylococcus epidermMA Proteus sp. Staph.vlococcus ~Tffdermidis l'scheJqchia coli Staph.vloc~,'ccu~aIII'CIIS
7
cullure negative Psettdomonas sp. culture negative
8 9
I0 11 12 13 14 15
16 17 18 19 20
21
Staph)'locr Staph.vloc~'cus Staphylococcus Staphylococcus Staphl,lo~vccus
ato'ett.~ attreus epidermidis aureus epidermidis
culture negative Sabnonella sp.
Staphylococcus attreus Staph)'lococcus epiderrnidis Staphylococcus epidermidL~" Kh,bswlla sp. Pseudomonas sp.
Eosinophilia Present no
yes yes yes yes no no no no yes yes yes yes yes yes yes* no no no yes yes
* Transient eosinophilia 5 years prior to peritoneal infection.
fections occurred in the remaining 45 patients (chisquare = 4.93, p < 0.05). Thirteen of 36 patients with CSF eosinophilia experienced shunt infections; among these, CSF eosinophilia was initially documented at the time the shunt infection was discovered in nine (69.2%). In an additional three cases, persistent CSF eosinophilia was identified from 1 week to 4 months before a shunt infection was documented. Thus, eosinophilia was closely related to the discovery of infection in 12 (92.3%) of 13 cases in which both infection and eosinophilia occurred. In the one remaining patient, a transient CSF eosinophilia was identified 5 years prior to a distal shunt infection. The organisms causing the shunt infections in the patients both with and without eosinophilia are presented in Table 2. In neither group did one organism predominate as compared to the other. Because patients with shunt infections frequently require more shunt revisions, the data for the number of revisions were reassessed after exclusion of patients experiencing infection. Patients with eosinophilia and no infection required an average of 7.0 _ 6.3 revisions (range 0 to 24), compared to an average of 2.0 _+ 2.4 revisions (range 0 to 13) in the remaining patients (p < 0.001). The number of revisions required by patients with both eosinophilia and infection averaged I 1.1 ___ 6.5 (range one to 24), while the patients without eosinophilia but with infection required an average of 3.5 _+ 3.2 revisions (range one to nine) (p < 0.01). Total protein (175 _+ 305 mg/100 ml) and glucose (53 _+ 24 mg/100 ml) values for the patients with CSF eosinophilia did not differ greatly from the values in patients without eosinophilia (128 _ 350 mg/100 ml and 76 _ 70 mg/100 ml, respectively). J. Neurosurg. / Voh~me 75 / October, 1991
Discussion
The earliest report of finding eosinophils in CSF was made in 1907 in relation to neurosyphilis. T M Since that report, CSF eosinophilia has been observed in a variety of pathological processes involving the central nervous system, the most common being a parasitic infection. Cerebrospinal fluid eosinophilia is a rare finding. Bosch and Oehmichen: reviewed 10,000 CSF specimens and, excluding patients with helminth infections, found only 94 specimens with eosinophils accounting for more than 1% of the total CSF WBC count. Our study suggests that the incidence of CSF eosinophilia in children with VP shunts is remarkably high. In the series of Bosch and Oehmichen,2 six of the 94 cases of eosinophilia occurred in children with shunts; the number of CSF samples in their series obtained from patients with shunts was not stated. Tzvetanova and Tzekov ~9reported a series of 404 children shunted for "internal" hydrocephalus and found a 6.4% incidence of CSF eosinophil count greater than 1% of the total CSF WBC count. In our series, 34% of patients with shunts developed a CSF eosinophil count of 8% or greater at some time during their clinical course. The high incidence in our series may relate to our routinely processing all CSF specimens for a differential cell count of the CSF sediment, allowing the discovery of a high percentage ofeosinophil even when the total CSF WBC count is low. In dividing our patients into those with and those without CSF eosinophilia, we identified two striking differences. First, the appearance of eosinophils in the CSF at any time during a patient's clinical course indicated the likelihood that more shunt obstructions, and thus more revisions, would occur in that patient. This increase occurred even in patients where CSF eosinophilia was a transient finding. The increase was evident in both the total population and the population remaining after removal of those patients with a shunt infection. It also applied both to those patients having their initial shunt insertion in 1985 and to those undergoing a revision in 1985. No predisposing factor, including age at insertion, sex, etiology of hydrocephalus, or number of shunt procedures before development of CSF eosinophilia, could be identified in the group of patients developing CSF eosinophilia. The role of the eosinophil in shunt obstruction is not known. Eosinophils at the periphery are attracted to areas of immunoglobulin E (lgE)-mediated hypersensitivity or to areas with lymphocytic infiltrates, as some lymphocytes can release eosinophil chemotactic factors. z3`8"~~176 The silicone used to manufacture shunt hardware is immunogenic and may induce a delayed hypersensitivity reaction. 6'~2 Sekhar, et al., n demonstrated chronic granulomatous inflammation in the tissue associated with shunts removed for culture-negative obstruction. Gower, et al.,S used scanning electron microscopy to document the chronic granulomatous response of tissue around shunts having culture-nega543
H. Tung, C. Raffel, and J. G. McComb tive malfunctions. These authors suggested a delayed hypersensitivity response to the silicone as the cause of inflammation. Cerebrospinal fluid eosinophilia may be a marker for those patients mounting a hypersensitive response to the silicone. This immune response is hypothesized, on the basis of increased cells and protein massing around the shunt, to be responsible for the increased number of shunt failures seen in patients with CSF eosinophilia. Delayed hypersensitivity to shunt material seems to be an unlikely cause of CSF eosinophilia, because the eosinophilia does not persist in a given patient. In addition to malfunction, shunt infection was also more common in patients with eosinophilia. The appearance of the eosinophils in the CSF coincided temporally with the documentation of infection in over 90% of the patients with both eosinophilia and infection. Shunt infections may be caused by indolent organisms that are difficult to culture, such as Propionibacterium species. ~'4't5 Perhaps some of the patients with multiple shunt malfunctions in the presence of CSF eosinophilia had unidentified low-grade infections. The infection may have been cleared by the removal and replacement of hardware during revisions. Based on the data presented here, the new appearance of eosinophils in the CSF of a shunted patient serves as an indicator of possible infection, warranting a workup for infection. In two reported studies, CSF eosinophilia was associated with an increased CSF protein level. 2'~9 In this series, protein levels were elevated in patients both with and without CSF eosinophilia. The increased protein in both groups may relate to the even distribution between the groups of patients with intraventricular hemorrhage and meningitis. The normal glucose values found in both groups of our patients are in accord with findings reported elsewhere. 2't9 In conclusion, this study demonstrates that CSF eosinophilia is an important and common finding in patients with VP shunts. Children with this finding will experience a significantly increased number of shunt obstructions and shunt infections over those without eosinophilia. References
1. Beeler BA, Crowder JG, Smith JW, et al: Propionibacterium acnes: pathogen in central nervous system shunt infection. Report of three cases including immune complex glomerulonephritis. Am J Meal 61:935-938, 1976 2. Bosch I, Oehmichen M: Eosinophilic granulocytes in cerebrospinal fluid: analysis of 94 cerebrospinal fluid specimens and review of the literature. J Neurol 219: 93-105, 1978
544
3. Butterworth AE, David JR: Eosinophilic function. N Engl J Med 304:154-156, 1981 4. Everett ED, Eickhoff TC, Simon RH: Cerebrospinal fluid shunt infections with anaerobic diphtheroids (Propionibacterium species). J Neurosurg 44:580-584, 1976 5. Gower DJ, Lewis JC, Kelly DL Jr: Sterile shunt malfunction. A scanning electron microscopic perspective. J Nenrosurg 61:1079-1084, 1984 6. Heggers JP, Kossovsky N, Parsons RW, et al: Biocompatibility of silicone implants. Ann Plast Surg 11:38-45, 1983 7. Hirsch JG, Hirsch BZ: Paul Ehrlich and the discovery of the eosinophil, in Mahmoud AAF, Austen KF (eds): The Eosinophil in Health and Disease. New York: Grune & Stratton, 1980, pp 3-23 8. Hubscher T: Role of the eosinophil in the allergic reactions II. Release of prostaglandins from human eosinophilic leukocytes. J lmmunol 114:1389-1393, 1975 9. Jeanes AL: Cerebrospinal eosinophilia following Torkildsen's operation. Guys Hasp Rep 114:28-31, 1965 10. Kay AB: The eosinophil in infectious diseases, d Infect Dis 129:606-613, 1974 I I. Kessler LA, Cheek WR: Eosinophilia of the cerebrospinai fluid of non-infectious origin: report of 2 cases. Neurology 9:371-374, 1959 12. Kossovsky N, Heggers JP, Dujovny M, et al: Ventricular shunt failure: evidence of immunologic sensitization. Surg Forum 34:527-529, 1983 13. Kuberski T: Eosinophils in the cerebrospinal fluid. Ann Intern Med 91:70-75, 1979 14. Mine S, Sato A, Yamaura A, et al: Eosinophilia of the cerebrospinal fluid in a case of shunt infection: case report. Neurosurgery 19:835-836, 1986 15. Noetzel MJ, Baker RP: Shunt fluid examination: risks and benefits in the evaluation of shunt malfunction and infection. J Neurosurg 61:328-332, 1984 16. Punyagupta S, Juttijudata P, Bunnag T: Eosinophilic meningitis in Thailand. Clinical studies of 484 typical cases probably caused by Angiostrongylus cantonensis. Am J Trop Med Hyg 24:921-931, 1975 17. Sekhar LN, Moossy J, Guthkelch AN: Malfunctioning ventriculoperitoneal shunts. Clinical and pathological features. J Neurosurg 56:411-416, 1982 18. Traynelis VC, Powell RG, Koss W, et al: Cerebrospinal fluid eosinophilia and sterile shunt malfunction. Neurosurgery 23:645-649, 1988 19. Tzvetanova EM, Tzekov CT: Eosinophilia in the cerebrospinal fluid of children with shunts implanted for the treatment of internal hydrocephalus. Acta Cytol 30: 277-280, 1986 20. Walls RS, Beeson PB: Mechanism of eosinophilia. VIII. Importance of local cellular reactions in stimulating eosinophilic production. Clin Exp Immunol 12:111-119, 1972 Manuscript received October 29, 1990. Accepted in final form March 25, 1991. Address reprint requests to: Corey Raffel, M.D., Ph.D., 1300 North Vermont Avenue #906, Los Angeles, California 90027.
Z Neurosurg. / Volume 75 / October, 1991
