Reports of HCAVM from multiple studies range from 0.3 to 23.2% of all patients who underwent EVD insertion [20]. This is in contrast to our institution’s experience of a prevalence of 30.4% among those who underwent the said procedure. The wide range had been attributed to factors such as differences by which the devices had been handled, the presence of IVH during placement, and the frequency of EVD manipulation [3, 8, 12]. The risk also increases if catheterization duration reaches more than five days with the risk reaching its peak at days nine to 11 post-insertion [3, 8, 12]. In our study, the range of EVD placement prior to the suspicion of HCAVM was zero to 26 days, with its mean at six days. However, analysis of potential risk factors contributing to the higher prevalence of HCAVM in our institution was beyond the scope of this study.
The CSF qualitative and quantitative results may also be influenced by factors such as (1) the primary neurologic indication for EVD insertion, (2) the presence of foreign devices in the CNS, and (3) one’s prior exposure to antibiotics.
CSF pleocytosis had been documented in 2 to 13% of cases of AIS but this was often self-limited and bore no diagnostic significance [21, 22]. This abnormality had also been documented in approximately 3% of cases of intraparenchymal hemorrhage [22]. Cases of SAH can also cause elevated CSF RBC count, hyperproteinorrachia, and hypoglycorrhachia [23]. On the other hand, IVH may result to a picture of chemical or aseptic meningitis as activated leukocytes enter the CSF to phagocytose the RBCs [7]. CSF evaluation of patients with primary CNS tumors such as gliomas, lymphomas, and meningiomas had also been shown to display pleocytosis in 16 to 30% of cases on top of hyperproteinorrachia and hypoglycorrhachia. Pleocytosis was more frequently observed among tumors localized near the ventricular system as this facilitated shedding of cells to the CSF [22, 24].
There may also be leukocyte infiltration as a response to foreign bodies within the CNS such as the EVD [6]. The yield of CSF cultures may also be lowered by active antimicrobials which patients receive for other ongoing systemic infections. In one study, almost half of patients which eventually developed HCAVM had been on antibiotics even prior to the suspicion for the said nosocomial CNSI [4]. In this study, the proportion was higher with 93.9% of those suspected and eventually diagnosed with HCAVM receiving intravenous antibiotics. More than three-fourths (76.1%) of this subset were receiving the medications because of a pre-existing systemic infection. Because of prior antimicrobial exposure, CS studies can be negative in 23% up to 78% of patients with HCAVM despite it being the reference standard [12]. In our sample, 57.4% had culture negative results. Among those with CSF positive results, Gram negative pathogens were the predominant growth at 85.7%.
The interpretation of a prior antibiotic exposure and a negative CSF CS result becomes more complex in the subgroup whom HCAVM was suspected but eventually ruled out. In this study, all (100%) of those under this subset had been receiving intravenous antibiotics. Almost all (90.2%) were on antimicrobials because of a pre-existing systemic infection. The diagnosis of HCAVM was ultimately ruled out after careful consideration of their over-all CSF picture and after a different focus of infection was detected.
Given the effect of antibiotic exposure on CSF CS yield, one may intuitively think that the same intervention may also affect other CSF parameters. However, recent studies showed that in pediatric and adult patients with HCAVM who received antibiotics, serum and CSF WBC did not significantly differ from those who did not receive treatment prior to CSF sampling [25]. The same results regarding CSF WBC were also observed among children with community-acquired bacterial meningitis [26]. In addition, there were no significant differences in CSF WBC count regardless of the duration of prior antibiotic exposure [26]. It is however, beyond the scope of this study to characterize the effects of the duration of antibiotic exposure in the CSF parameters and CI of the included patients.
Currently, there are no defined guidelines on how to factor in the various CSF changes brought about by the pre-existing CNS pathology, the post-operative state, and one’s prior antimicrobial exposure. CI had been proposed as a measure to take into account changes in the CSF composition from RBC contamination secondary to the existing CNS pathology or the operative procedure. The hypothesis is that ventricular blood simply dilutes the CSF with peripheral blood; hence, CSF WBC and RBC should exist in similar proportion as that of the peripheral blood.
The utility of CI was first documented in 2004 after patients who developed HCAVM showed a significant increase in the said parameter at least 3 days prior to conventional diagnosis. A limitation, however, was the study’s limited sample size of only 13 [7]. A follow-up study with 39 patients evaluated its utility but results did not attain statistical significance [2]. Subsequent studies showed positive results although they identified various cut-off points as compared to the 5-fold increase cut-off as originally proposed [9]. One study with 34 samples showed that a cut-off of 2.9 had a sensitivity of 95.0%, specificity of 92.9%, and AUC-ROC of 0.982 [1]. Another study with a sample size of 111 utilized a cut-off of 4.3 which showed an AUC-ROC of 0.825 [4]. The latest study included a sample size of 95 patients and utilized a cut-off of 10.4 revealing a sensitivity of 80.0%, specificity of 70.5%, and AUC-ROC of 0.727 [6]. An important observation, however, is that these previous studies only included patients ≥ 16 years old. The patients included also predominantly had ICH as the main neurologic indication for EVD insertion. Hence, the utility of the said parameter has not yet been previously studied in the pediatric age group and in other neurologic indications.
In our study, using the originally proposed cut-off value of 5, a sensitivity of 18.4%, specificity of 95.5%, and AUC-ROC of 0.599 were acquired. Following the optimal cut-off value of ≥ 1.21 improved the sensitivity to 30.6% but decreased specificity to 86.4% as an effect. However, AUC-ROC did not differ significantly at 0.585. As compared to previous studies which revealed higher sensitivity-specificity and fair to excellent diagnostic accuracy for CI following AUC-ROC analysis [27], our study showed lower sensitivity and AUC-ROC with the latter indicating that CI failed to adequately diagnose HCAVM.
Subgroup analysis done for age showed that CI offers a higher sensitivity at 52.9% although with a lower specificity at 66.7% in the pediatric population. However, its AUC-ROC of 0.598 indicates that the parameter is also unable to adequately diagnose HCAVM in this age group. This is also true for the adult population wherein AUC-ROC was only 0.516. Analyzing the utility of the parameter in terms of the neurologic indication for EVD insertion, for cases of ICH and intracranial neoplasms, their sensitivity and specificity lie close to the sensitivity and specificity of CI over-all. However, both their AUC-ROC also fail to adequately diagnose HCAVM.
In addition to a single CI determination, changes in its trend during subsequent CSF sampling may also potentially help diagnose HCAVM. Available studies looking into serial CI showed trends of slowly increasing CI values until the time of conventional diagnosis of HCAVM. To further analyze these trends, the CI escalation ratio was a parameter proposed by dividing the highest CI derived by the baseline CI at the time of EVD insertion [6, 12]. However, because of the inconsistent CSF sampling schedules in our sample, this study only focused on the utility of CI at the point of initial HCAVM suspicion. The significance of looking into the trend and/or the CI escalation ratio may be better studied in a prospective study which will ensure a predefined CSF sampling schedule from EVD insertion until its removal.
As of writing, this had been the largest study evaluating the diagnostic utility of CI as it reviewed the records of a total of 161 patients who underwent EVD insertion and analyzed the parameter’s utility in 71 patients who were suspected with HCAVM. This is also the first study to include the pediatric population. In addition, as compared to previous studies which predominantly included patients with various forms of ICH, our sample included a significant proportion with intracranial neoplasms.
Additional limitations of the study, however, must also be acknowledged. First, its retrospective design may have posed unrecognized confounders. Second, among those with intracranial neoplasms, the interpretation of CI may be confounded by potential CSF alterations owing to leptomeningeal spread or prior treatments received. The utility of CI in this particular subgroup may be further explored in another study.
link