Ethics, consent, and permissions

The study was performed in accordance with the Declaration of Helsinki and was approved by the Poznan University of Medical Sciences' bioethics committee (approval no. 983/13). All participants or their families provided written informed consent. Two physicians (anaesthesiologist and neurosurgeon) independently assessed level of patient's consciousness by Glasgow coma scale (GCS) at admission and daily. Patients with GCS score of 14 and lower were qualified as incapable of giving informed consent, therefore written consent was obtained from the next of kin on the behalf of participants. Patients with GCS score of 15 (and 14 in case of eye opening score of 3) underwent decision-making capacity assessment with probing questions [10]. In case of failure, the next of kin provided surrogate consent.

Study population

Between December 2013 and June 2015, 152 patients with aneurysmal SAH, confirmed by computerised tomography (CT) and digital subtraction angiography, were admitted to the neurosurgical ward. Inclusion criteria for the study were early (< 24 hour post rupture) endovascular occlusion of the aneurysm and acute hydrocephalus (HCP) treated with external ventricular drainage (EVD) before 48 hours post rupture. HCP was diagnosed on the basis of the original CT findings (relative bicaudate index of more than 1 or focal dilation of ventricular system due to obstruction). Patients with decreased consciousness and diagnosed HCP or high likelihood of HCP development (thick intraventricular blood clot) underwent cerebrospinal fluid (CSF) diversion. Due to high risk of brain tissue shift (13/18 patients had intraparenchymal haematoma) and suspected obstructive aetiology of HCP (18/18 patients had intraventricular blood) EVD was preferred. Exclusion criteria were age below 18, pregnancy, history of central nervous system (CNS) disease (meningitis, stroke), active infection of CNS and active systemic disease (diabetes mellitus, rheumatoid arthritis, malignancy, cirrhosis, renal failure). To rule out potential impact of a surgery-related trauma on a CSF homoeostasis, patients who underwent clipping were also excluded. This left a total of eighteen patients in the study group. The control group consisted of eight patients with minor neurological disorders (e.g. tension headache), where CSF was obtained during diagnostic lumbar puncture (LP). The patients with abnormal CSF results, elevated C-reactive protein (CRP) level and white blood cell (WBC) count on the day of LP were not enrolled. The exclusion criteria were the same as for the study group.

Management protocol

On admission the clinical status was assessed using the Hunt & Hess scale (HH), World Federation of Neurosurgical Societies scale (WFNS), and GCS. After aneurysm occlusion and EVD insertion a CT scan of the head was carried out on post-SAH day 2–3 to assess any procedure related brain injury. All patients received continuous intravenous infusion of nimodipine for at least 10 days, hypotension was avoided using vasopressors and euvolemia was maintained. DCI in our study was defined as neurological impairment (such as hemiparesis, aphasia, apraxia, hemianopsia, or neglect) or a decrease of at least 2 points on the GCS [either on the total score or one of its individual components (eye, motor on either side, verbal)]. Any such deterioration should last for at least one hour, not be apparent immediately after aneurysmal occlusion, and not attributable to non-neurological causes [11]. Every patient diagnosed with DCI was treated with induced hypertension (20–30% rise of mean arterial pressure above baseline). Cerebral infarction due to DCI was defined as presence of cerebral infarction on CT scan of the brain within 6 weeks after SAH, not present on the post coiling/EVD CT scan in patients diagnosed with DCI. Treatment outcome was assessed at three months using the Glasgow outcome scale (GOS).

Sample collection and assays

Haemoglobin (Hgb), CRP level and WBC count were assessed daily. Automatic analysers XT 2000i (Sysmex, Japan), Cobas 6000 (Roche diagnostic, USA) and ACL TOP 500 (Instrumentation Laboratory, Italy) were used for measurements. CSF samples were collected from the EVD at three time points, on post-SAH day 0–3, 5, and 10–12. The final sample was not obtained in five cases, on account of early EVD removal. CSF cell count was checked at least twice per patient and CSF culture was ordered at least once on post-SAH day 10–12. Each CSF sample was centrifugated, and stored at -80°C until assayed. ELISA commercial kits SEA753Hu and SEA663Hu (USCN Life Science, Wuhan, China) with a lower detection limit of 0.117 ng/ml and 0.133 ng/ml respectively, were used to assess sTLR2 and 4 concentrations. Values below the detection limit were extrapolated using ELISA kit standard curves.

Statistical analysis

Values for normally distributed numerical data have been expressed as mean and standard deviations; for ordinal or non-normally distributed numerical data as median and interquartile range, and for categorical data as counts and percentages. Since the sTLR2 and 4 levels do not show normal distribution they are presented as median and interquartile range. The normality of data distribution was assessed using the Shapiro-Wilk test. The correlations were assessed by Spearman’s exact test and degree of correlation was described 'according to a rule of thumb': -0.3 ≤ correlation coefficient (cc) ≤ 0.3 as “negligible correlation”; 0.3 < cc ≤ 0.5 (-0.5 ≤ cc < -0.3) as “low correlation”; 0.5 < cc ≤ 0.7 (-0.7 ≤ cc < -0.5) as “moderate correlation”; 0.7 < cc ≤ 0.9 (-0.9 ≤ cc < -0.7) as “high correlation”; cc < 0.9 ≤ 1.0 (-1.0 ≤ cc < -0.9) as “very high correlation” [12]. A value of P < 0.05 was considered statistically significant when comparing. Data were analyzed using STATISTICA 10 (Stat Soft Inc., Tulsa, OK, USA) and StatXact (Cytel Inc., Cambridge, MA, USA).

Origin of sTLR2 and 4

ELISA kits used in our study lack the ability to differentiate between strictly soluble and membrane forms of TLRs. We believe that the sTLR values presented consist of (I) soluble TLRs released by immune cells as a TLR signalling blockade, and (II) membrane TLRs released incidentally as a result of necrosis and cell membrane fragmentation. This raises the question as to which form dominates in CSF. In studies using identical ELISA kits, membrane forms of TLRs have been ignored in discussions about the origins of the assayed sTLR [3,4,13,14]. TLRs are abundantly expressed in CNS and their mRNA is upregulated in the brain in SAH and ischaemic models [15–17]. Most animal studies show that both necrosis and apoptosis are modes of cell death following SAH, but necrosis is not considered to be a major factor [18,19]. On this basis, and in line with other authors, we have assumed that the membrane TLR form makes only a minor contribution to the measured sTLRs values.

Lack of sTLR2 and 4 in CSF from the control group is consistent with the inflammatory-dependent nature of these receptors. Physiologically, sTLR2 and 4 have been shown to be present in blood plasma, saliva, breast milk and amniotic fluid [3,20–22]. As mentioned above, presence of sTLR2 and 4 in CSF may be a consequence of active production (intrathecal or peripheral) and incidental release (intrathecal necrosis) (Fig 3). The only established source of soluble TLR forms are blood mononuclear cells (BMCs) [3]. Their intrathecal appearance may be due to some SAH-related migration through the blood-brain barrier (BBB), or that BMCs are simply released from the subarachnoid blood clot into the CSF [23]. An alternative local sTLR2 and 4 source may be the microglia. These resident CNS macrophages express both TLR2 and 4 on their surfaces, and are activated during SAH [24,25]. BBB breakdown observed in SAH allows migration of blood-derived proteins into the CNS. Potentially, sTLR2 and 4 circulating in blood may reach the CSF by this route [26]. Although it is suspected to be a minor contribution, incidental release of membrane TLRs due to cell necrosis within the CNS should not be neglected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Possible sources of sTLR2 and 4 in CSF.

(A) Migration of activated BMCs across the BBB during SAH and active release of sTLRs into CSF. (B) BMCs trapped in subarachnoid clot actively release sTLRs into CSF. (C) DAMPs bind to TLRs on microglia surface and trigger TLR signalling followed by active release of sTLRs into CSF. (D) Membrane TLRs released incidentally into CSF as a result of necrosis and cell membrane fragmentation. (E) BMCs release sTLRs peripherally into the blood plasma. sTLRs leak through the disrupted by SAH BBB. Abbreviations: CSF = cerebrospinal fluid; sTLR = soluble Toll-like receptor; BMC = blood mononuclear cells; DAMP = danger-associated molecular patterns; BBB = blood-brain barrier.

https://doi.org/10.1371/journal.pone.0156171.g003

Release kinetics of sTLR2 and 4

In our study, sTLR2 and 4 level changes followed a specific pattern; neither can be detected on post-SAH day 0–3, but then increase with time in the majority of cases. By contrast, plasma concentrations of sTLR2, sTLR4, and IL-1Ra peak in 2–4 hours following lipopolysaccharide (LPS) administration in all patients [3]. Delayed sTLR2 and 4 appearance in CSF denies necrotic origin of sTLR2 and 4, since some cell death might be expected at an early stage following aneurysm rupture, and this would lead to the presence of sTLR in the early samples. Conversely, delayed kinetics may reflect the time required for the migration of activated BMCs to the CNS before local sTLRs production can be initiated. Alternatively, unlike LPS, DAMPs may trigger a prolonged sTLR2 and 4 production pathway in migrated BMCs or microglia. As mentioned above, current reports describe only a rapid (peak in 2–4 hours) release of sTLR2 and 4 to blood plasma after stimulation with LPS (an example of a PAMP).

Potential significance

Soluble forms of TLRs function as the first line of blockade in TLR signalling [2]. Abundant expression of TLRs in CNS (especially on microglia) can lead to a massive inflammatory reaction [24]. Reported observations of (I) early release of DAMPs during SAH, (II) upregulation of downstream mediators and TLRs in brain tissue and (III) increase of final protein products in CSF and brain tissue, indicate both activation of TLR signalling and failure of anti-inflammatory mechanisms [16,27–31]. Attenuation of SAH associated neuronal apoptosis in TLR4 depleted mice and alleviation of EBI in rats administered exogenous TLR signalling blocker, prove both the substantial role of TLRs in experimental SAH, and the therapeutic potential of TLR signalling blockers [32,33]. Revealed in our study negligible to moderate correlations between sTLRs and patients’ condition seem to lessen significance of sTLRs in human SAH. We found this observation unexpected, thus it is discussed below. Firstly, CSF concentrations of soluble TLRs are low in comparison with other body fluids. Blood plasma from healthy volunteers has 3–8 times the concentration of sTLR2, while experimental endotoxemia leads to plasma concentrations of sTLR2 and 4 which are 10–100 times greater [3,4]. Secondly, sTLR2 and 4 were absent from CSF at 0–3 days and thus cannot influence TLRs signalling, which peaks 2–6 hours after aneurysm rupture [16]. Thirdly, in vast majority we have found negligible to low correlation between sTLR2 and 4 levels and the severity of inflammation (defined by WBC, CRP level), clinical status on admission and treatment outcome. Furthermore, complete absence of soluble TLRs in CSF of 5 patients did not affect their treatment outcome, or lead to any differences in the monitored parameters. The results of our study suggest that delayed release kinetics and low concentrations of soluble TLRs (while DAMPs levels are elevated), might be responsible for low overall correlations with clinical status.

Limitations of the study

EVD infection is a potential bias. Microbial CNS inflammation would be likely to lead to profound alterations in sTLRs levels. Since all of these patients had an intraventricular drain, infection was a distinct possibility. However, no drain was removed early as a result of infection, and none showed evidence of infection on removal. Nearly all patients received antibiotics for other unrelated infections while the drains were in situ what could be a bias. It is recognised that this group of patients does not represent the full spectrum of aneurysmal SAH, which limits the range of possible conclusions. The study protocol restricted number of enrolled patients and determined their poor admission status. However the basic correlation between presenting status and clinical outcomes does not suggest any other unusual feature for this group. On the basis of available literature, we assumed robust activation of TLR signalling in our patients, and did not measure levels of TLR pathway products to confirm this [16,32,34].

Future directions

We consider that confirmation of sTLR release into CSF following SAH should encourage further studies in view of the possible role in delayed brain injury. In order to investigate this further, we would suggest a larger series of all grade SAH patients, with stricter monitoring of potential CNS infection and antibiotic intake. sTLR levels should be analysed together with other elements of the TLR signalling pathway in both blood and CSF.