Article Text
Statistics from Altmetric.com
The development of a hydrocephalus as a typical complication after subarachnoid haemorrhage often requires implantation of a catheter for drainage of ventricular fluid. Because this is an open system it carries a risk of development of bacterial ventriculitis by contamination. This potentially life threatening or neurologically disabling disease can be diagnosed by microbiological tests on the CSF. However, microbiological examination is often not possible or shows pathological results only after a delay, probably because of antibiotic prophylaxis. Standard CSF indices are difficult to interpret, because subarachnoid haemorrhage leads to an irritation syndrome in the CSF, which can imitate infection.
Lysozyme is a basic polypeptide of 129 amino acids weighing 15 kDa, which is found in neutrophilic granulocytes or monocytes, and is released from cytoplasmic and azurophilic granules.1 It is capable of degrading bacterial proteoglycans.2 Lysozyme shows significantly higher concentrations in CSF in bacterial meningitis in comparison with abacterial meningitis, whereas in other diseases of the CNS much lower concentrations are found.3 4
We retrospectively documented the results from 146 patients who underwent external ventricular drainage after subarachnoid haemorrhage. As well as documenting the frequency and timing of infection, the medical notes were inspected for risk factors of infection.
In 64 patients (15 with microbiologically confirmed ventriculitis, and 49 with no evidence of ventriculitis) the following measurements were compared: serial CSF cytology; total protein; albumin; IgM, IgA, IgG; plasma/CSF ratio of albumin, IgM, IgA, IgG; locally produced fraction of IgM, IgA, IgG and CSF oligoclonal IgG, lactate, and glucose.
For further analysis, 61 CSF samples were available from 14 patients with microbiologically confirmed ventriculitis. As a control, we used one sample from each of 31 patients in whom no evidence of ventriculitis was found. Immediately after these CSF samples had been collected for clinical tests, the remainder of the samples had been frozen at−80°C. These CSF samples were investigated to measure their lysozyme concentrations by radial immunodiffusion. The sample is transferred to circular wells in an agarose gel; the antigen then diffuses radially from the sample into the gel, which contains the corresponding antibody. After an incubation period (18 hours) the antigen concentration can be read from a reference table (“The binding site”, Birmingham, UK).
For statistical analysis we used the χ2 test, the Wilcoxon matched pairs test, and a Kaplan-Meier survival analysis. A p value<0.05 was considered to be statistically significant.
Fifteen of 146 patients (10.3%) developed a microbiologically established ventriculitis, confirmed by detection ofpseudomonas aeruginosa (one case),staphylococcus epidermidis (12 cases), gram negative rod cells (one case), and enterococci (two cases). The period of drainage was 3–39 days (mean 10 days). Infection rate was 3.5% (5/146) within the 1st week, 5.3% (4/77) within the 2nd week, and 15.4% (6/39) after 2 weeks of drainage (p<0.02, df=2). Mean survival time (=no ventriculitis) was 26 days. The probability of drainage without ventriculitis decreased to 0.96 (SE 0.02) after 7 days, 0.86 (SE 0.045) after 14 days, 0.66 (SE 0.1) after 22 days, and 0.53 (SE 1.4) after 26 days. We identified potential risk factors for ventriculitis in eight of 15 patients with ventriculitis, and in nine of 131 patients without ventriculitis (p<0.001, df=1). The risk factors were: leakage in the drainage system (26% with ventriculitis; 8.4% without ventriculitis), disconnection (13.3%; 2.3%), blockage which was cleared by irrigation (20%; 5.3%), and dermal infection (6.6%; 0%).
Comparison of the CSF results in 15 patients who had established ventriculitis with those in 49 patients without ventriculitis showed no significant difference for most of the mentioned indices except the following. Ten of 15 patients with ventriculitis and only eight of 49 patients without ventriculitis showed oligoclonal IgG in the CSF (p<0.001, df=1). Thirteen of 15 patients with ventriculitis and 18 of 49 patients without ventriculitis showed increased CSF-IgM (>7 mg/l) concentrations (p=0.002, df=1). Thirteen of 15 patients with ventriculits and 17 of 49 patients without ventriculitis showed increased CSF lactate (>2.1 mmol/l; p<0.001, df=1).
In all 14 of the 15 patients with microbiologically confirmed ventriculitis available for analysis, maximal lysozyme concentrations of above 2.5 mg/l were found (mean 11.29 (SD 8.1) mg/l, range 2.6–25 mg/l), in 61 samples at different times (figure). In only two of the 31 control patients could lysozyme concentrations just above 2.5 mg/l be detected, with all others lying clearly below this concentration (mean 1.44 (SD 0.64) mg/l, range 0–2.6 mg/l, p<0.001, df=1). In 17 samples, the ventricular concentration of lysozyme was compared with the concentration in lumbar CSF samples taken synchronously. The mean ratio of lumbar to ventricular concentration was 2.24 (SD 0.525) (range 1.4–2.8, 2 tailed p=0.0003, Wilcoxon matched pairs test).
Mean concentrations of lysozyme concentration in ventriculitis during the 4 day periods around the detection of bacteria.
Our study shows a clear connection between the duration of drainage and the risk of ventriculitis. The indication for implanting or continuing ventricular drainage should therefore be checked daily, and alternatives such as catheter exchange should be considered at an early stage. The documented risk factors show that improving the standards of hygiene and nursing care are necessary for reduction of the risk of ventriculitis. We found the presence of oligoclonal IgG bands, CSF IgM,5 CSF lactate, and lysozyme in the CSF to be statistically relevant indices for diagnosis. The finding that lysozyme concentrations in lumbar CSF are more than twice as high as in ventricular CSF has not to our knowledge been described previously. Evidently lysozyme, as a relatively large protein molecule, is concentrated in the lumbar CSF. The interpretation and the definition of normal values must take account of these findings.
This study shows that a single, marginally pathological value has less diagnostic value than an increase in lysozyme detected in a series of CSF samples. As a pragmatic approach to early diagnosis of ventriculitis, we recommend a CSF analysis within the 1st week, including measurement of lysozyme concentration. If drainage is to be continued longer than 1 week, we would advise daily CSF analysis, in particular of lysozyme. If there is a relative increase in lysozyme concentration, ventriculitis should be assumed, even when microbiological findings are still negative.