Introduction

Ventriculitis associated to external ventricular drainages (EVDs) is one of the most important complications related to the use of these devices. EVD-associated ventriculitis has a variable incidence, ranging from 5 to 20 %, and entails a high morbidity and mortality [1, 2]. Biofilm formation on EVDs has been postulated as the main pathogenic mechanism of ventriculitis [3].

Studies have been performed in order to assess the advisable time of EVD use and the effectiveness of different materials to avoid biofilm formation. Routine change of the EVD catheter has not demonstrated a clear benefit on ventriculitis prevention and is not currently recommended [4, 5]. Silver-impregnated EVDs have shown a decrease of EVD-related ventriculitis, although the available scientific evidence is still scarce [6, 7]. Similar results have been found for antibiotic-impregnated EVDs [811]. This type of EVD successfully avoided biofilm formation in vitro, but this effect has not been corroborated in vivo [12]. As Gram-positive bacteria are the microorganisms more frequently involved in ventriculitis, rifampicin, clindamycin, and minocycline have been the most frequently used antibiotics. However, the emergence of multidrug-resistant Gram-negative bacteria, such as Acinetobacter baumannii, as a common etiology of ventriculitis could limit the efficacy of these devices [13].

In order to assess the prevalence of ventriculitis and biofilm formation on EVDs, and to analyze the influence of antibiotic-impregnated EVD on the risk of biofilm formation and ventriculitis, we prospectively collected and processed all EVDs placed in neurocritical patients in our intensive care unit (ICU).

Methods

Patients

This study was approved by our institutional ethics board and an informed consent was obtained from the patients or their relatives. Over a period of 12 months, we prospectively recorded all patients that were consecutively admitted to our ICU and required the insertion of an EVD. We did not include the same patient if a second EVD was inserted. Recommended maneuvers to prevent EVD-related ventriculitis were performed in all our patients [1].

External ventricular drainages

Neurosurgeons could choose between conventional EVD (silicone) or impregnated EVD (Bactiseal®, Codman Johnson & Johnson, Raynham, MA, USA). Bactiseal® is an EVD with antibiotics impregnated throughout the silicone material, with 0.15 % clindamycin and 0.054 % rifampicin [14].

Definitions

Biofilm

Clusters or microcolonies of bacteria, encased in an extracellular matrix and adherent to the EVD surface [15].

Colonization

A positive cerebrospinal fluid (CSF) or EVD culture, in the absence of abnormal CSF findings or clinical signs and symptoms of infection [5].

EVD-related ventriculitis

A positive CSF culture accompanied by abnormal CSF findings and appropriate clinical signs and symptoms [5].

Variables and outcome measures

Clinical variables were collected daily starting on the day of admission. We recorded the following data using a standardized form: (a) relative to the patient: demographics (age, sex), comorbidities (heart failure, cerebrovascular, neoplastic, renal, or liver diseases), immunological state, cause of EVD insertion, antibiotic treatment, admission and release dates, dates of initiation and end of EVD, diagnosis of ventriculitis and other hospital-acquired infections; (b) relative to CSF samples (microbial isolates in CSF ); (c) relative to EVDs: type of EVD, microbial isolated on EVD culture, and presence and extension of biofilm on EVDs.

CSF samples

We performed surveillance sampling of CSF twice a week. The sample was analyzed (cytobiochemistry) and qualitatively cultured. If ventriculitis was suspected, we obtained an extra CSF sample before antibiotic administration (when possible).

Processing of EVD

Upon extraction, we collected EVDs, avoiding external contamination. After rinsing with sterile saline to eliminate CSF or blood, a 1-cm-long cross-sectional intracranial segment was divided into two portions for both electron microscopy and aerobic/anaerobic cultures. Biofilm was obtained by EVD sonication under sterile conditions. Samples were cultured using standard bacteriologic techniques. For scanning electron microscopy (SEM), the section of the EVD was fixed in a mixture of 4 % paraformaldehyde–2.5 % glutaraldehyde in 0.1 M phosphate buffer (pH 7.0) for 1 h at room temperature. All samples were maintained at 4 °C and, previous to SEM examination, they were sequentially processed: washed and dehydrated with crescent ethanol concentrations, dried with liquid CO2, mounted, sputter-coated, and examined under a Hitachi (S-4100) scanning electron microscope (SCSIE, University of Valencia, Valencia, Spain).

Statistical analyses

All statistical analyses were performed with the SPSS v.15 software. The χ2 test was used for categorical variables, and the Student’s t or Mann–Whitney tests were used for continuous variables. We considered p < 0.05 (95 % confidence interval) as being statistically significant.

Results

We included 32 patients in our study. The demographic and clinical data are depicted in Table 1. More than half of the EVDs were impregnated with rifampicin and clindamycin (n = 18; 56 %). The mean duration of use was 8 days (interquartile range, IQR: 6–12). The reasons for EVD withdrawal were resolution of hydrocephalus (p = 13; 40 %), insertion of a new EVD (n = 7; 22 %), or death (n = 12 (38 %).

Table 1 Demographic data and clinical data

There was evidence of EVD colonization in 12 cases. Seven surveillance CSF cultures were positive and the microorganism isolated was coagulase-negative staphylococci (CNS) in six of them and Escherichia coli in the remaining patient, but in any patient, it predicted the development of ventriculitis. Five EVD catheter cultures were also positive to CNS (n = 3), Propionibacterium acnes (n = 1), or Candida spp. (n = 1) in the absence of signs or symptoms of infection.

EVD-related ventriculitis was diagnosed in 6 patients (29 %). The mean time from EVD insertion to ventriculitis diagnosis was 12.5 days (IQR: 11–16.5). Five cases of ventriculitis were due to A. baumannii and one case was due to Klebsiella pneumoniae. In four of these six cases, the same microorganism was isolated from the EVD catheter culture (67 %).

Biofilm on EVDs

Biofilm formation was present in 75 % (n = 24) of the EVDs. Biofilm extension varied from scarce formations (n = 13; 54 %) to dispersed clusters (n = 5; 21 %) or confluent abundant biofilm matrices extended on the inner surface of EVDs (n = 6; 25 %) (Table 2) (Fig. 1a–d). When biofilm was observed, microorganisms could frequently be observed on the surface of the biofilm (Fig. 1b, d).

Table 2 Biofilm analysis and microbial findings
Fig. 1
figure 1

a Antibiotic-impregnated external ventricular drainage (EVD) (Bactiseal®), showing antibiotic release in the inner surface (13,000×). b Biofilm on EVD related to coagulase-negative staphylococci (CNS) colonization (15,000×). c Biofilm on EVD related to Acinetobacter baumannii ventriculitis (40×). d Biofilm on EVD related to A. baumannii ventriculitis (2,500×)

Biofilm was more frequent on EVDs originating from patients with ventriculitis and from colonized patients (100 % and 75 %, respectively vs. 64 %; p = 0.217). EVDs used for ≥7 days (n = 14; 44 %) had a higher incidence of ventriculitis (35 % vs. 0 %; p = 0.011) and biofilm was more frequently observed (88 % vs. 67 %; p = 0.094).

Influence of impregnated vs. non-impregnated EVDs on the risk of ventriculitis, colonization, and biofilm formation

We collected 18 impregnated EVDs (Bactiseal®) and 14 non-impregnated EVDs. The clinical characteristics of patients in both groups were similar (Table 1). EVD duration until removal or ventriculitis diagnosis was 9 days (IQR: 6–17) for non-impregnated EVDs and 7 days (IQR: 6–11) for impregnated EVDs (p = 0.239). There were no differences in the reasons for EVD withdrawal between impregnated and non-impregnated EVDs (Table 3).

Table 3 Outcomes related to impregnated EVDs and non-impregnated EVDs

Colonization by CNS was similar between the two groups. Although not statistically significant, biofilm was more frequent on non-impregnated EVDs (86 %) than on antibiotic-impregnated EVDs (67 %) (p = 0.217). Among EVDs removed within the first 7 days, the prevalence of biofilm was 50 % for impregnated EVDs and 80 % for non-impregnated EVDs (p = 0.264). The prevalence of biofilm on those EVDs removed after 7 days was 88 % and 89 % for impregnated and non-impregnated EVDs, respectively (p = 0.929).

The mean time from EVD insertion to ventriculitis diagnosis was similar between non-impregnated and impregnated EVDs. Ventriculitis was more frequent in non-impregnated EVDs (43 %) than in impregnated EVDs (21 %), without reaching statistical significance (p = 0.354). Bacteria causing ventriculitis were resistant to antibiotics enclosed in Bactiseal® EVDs. The results related to impregnated and non-impregnated EVDs are summarized in Table 3.

Discussion

Our results show a high prevalence of biofilm formation on EVDs. All EVDs originating from patients with ventriculitis showed biofilm, but it was also present on EVDs removed for other reasons. We observed a delay in biofilm formation in impregnated EVDs but, after 7 days in use, the presence of biofilm was similar. Due to the etiology of our infections (Gram-negative bacteria), antibiotic-impregnated EVDs could not avoid ventriculitis development. Another interesting finding of our study is the lack of efficacy of surveillance CSF cultures in detecting bacterial colonization preceding the infection.

Colonization and biofilm formation is nearly a universal phenomenon on invasive devices, such as endotracheal tubes (ETTs) and indwelling central venous catheters (CVCs) [1619]. This is the first time that EVDs have been systematically studied in order to analyze the presence of biofilm. Like for other devices, we have found a high prevalence of biofilm (74 %). Although our sample size was small and the results did not reach statistically significant differences, impregnated EVDs reduced biofilm formation during the first 7 days of use. This result is consistent with previous findings from several studies. In an experimental model of antibiotic-impregnated endovascular catheters, it was found that there was a loss of minocycline concentration by day 7 [20]. A decrease in the zone of inhibition against different bacteria in rifampicin-/minocycline-impregnated catheters by day 7 has also been described [21].

Concerning ventriculitis, the failure of impregnated EVDs in avoiding infection could be explained because the pathogens involved were resistant to clindamycin and rifampicin (A. baumannii and K. pneumoniae). Some studies analyzed the efficacy of antibiotic-impregnated EVDs [811] and found a reduction in bacterial colonization and ventriculitis due to staphylococci with EVDs loaded with potent antistaphylococcal and antibiofilm agents, such as minocycline and rifampicin [811]. The increasing presence of multidrug-resistant Gram-negative bacteria in device-related infections has made necessary the evaluation of new biomaterials impregnated with antibiotics active against these bacteria, such as colistin-impregnated beads to prevent multidrug-resistant A. baumannii implant-associated osteomyelitis [22]. The same strategy could be adopted in EVD ventriculitis to design new EVDs pointed towards these bacteria [13].

We have found a high prevalence of ventriculitis related to an A. baumannii outbreak in our ICU during this study. Preventive measures have reduced the presence of these bacteria and have diminished significantly the incidence of nosocomial infections. Even in this specific epidemiological context, surveillance cultures of CSF (at least with the planned periodicity) did not anticipate the diagnosis and etiology of the ventriculitis.

Our study has several limitations. The main drawback of our study is the number of patients included; however, this is the first time that in vivo biofilm formation on EVDs has been systematically studied. The fact that there was an outbreak of multidrug-resistant Gram-negative did not allow us to evaluate the effectiveness of antistaphylococcal impregnated EVDs. As described in previous studies, we analyzed only a 1-cm cross-section of the intracranial portion of the EVD, so we can only estimate the real percentage of occupation of the whole catheter; moreover, the absence of biofilm in this section cannot assure the complete integrity of the EVD (at a low probability rate).

Conclusions

In conclusion, biofilm is a common phenomenon on external ventricular drainages (EVDs). In agreement with other studies in different invasive devices, impregnated EVDs delay biofilm formation in the first 7 days of use. However, they could not significantly reduce the rate of ventriculitis because the pathogens were resistant to the antibiotics loaded into the EVDs. In the future, it would be necessary to develop EVDs loaded with antibiotics active against multidrug-resistant pathogens that are increasing worldwide. In addition, surveillance cultures were not able to predict ventriculitis due to Gram-negative bacteria.