Extracellular fluid, cerebrospinal fluid and plasma biomarkers of axonal and neuronal injury following intracerebral hemorrhage

Spontaneous intracerebral hemorrhage (ICH) is the most devastating form of stroke. To refine treatments, improved understanding of the secondary injury processes is needed. We compared energy metabolic, amyloid and neuroaxonal injury biomarkers in extracellular fluid (ECF) from the perihemorrhagic zone (PHZ) and non-injured (NCX) brain tissue, cerebrospinal fluid (CSF) and plasma. Patients (n = 11; age 61 ± 10 years) undergoing ICH surgery received two microdialysis (MD) catheters, one in PHZ, and one in NCX. ECF was analysed at three time intervals within the first 60 h post- surgery, as were CSF and plasma samples. Amyloid-beta (Aβ) 40 and 42, microtubule associated protein tau (tau), and neurofilament-light (NF-L) were analysed using Single molecule array (Simoa) technology. Median biomarker concentrations were lowest in plasma, higher in ECF and highest in CSF. Biomarker levels varied over time, with different dynamics in the three fluid compartments. In the PHZ, ECF levels of Aβ40 were lower, and tau higher when compared to the NCX. Altered levels of Aβ peptides, NF-L and tau may reflect brain tissue injury following ICH surgery. However, the dynamics of biomarker levels in the different fluid compartments should be considered in the study of pathophysiology or biomarkers in ICH patients.


Patients.
Adult patients admitted to the Neurosurgical Department of University Hospital of Linköping, Sweden, requiring emergent surgery with craniotomy for spontaneous intracerebral hemorrhage (ICH) were prospectively included in the study 2 . The ICH was surgically evacuation by routine microneurosurgical technique 2 and patients received an external ventricular drain (EVD; DePuy Synthes, Raynham, USA) for monitoring intracranial pressure. One microdialysis (MD) catheter (CMA-71 Brain Catheter, M-Dialysis, Solna, Sweden) was inserted, at a 45-degree angle, via the craniotomy into the perihemorrhagic zone (PHZ) defined as < 1 cm of the evacuated ICH, and one catheter was inserted in non-eloquent and non-injured cortex (NCX) remote to the ICH; either ipsilateral to the ICH or in the contralateral hemisphere via a separate frontal burr hole. MD catheter locations were verified with a post-operative CT-scan.
Seven of the patients were included in a previous publication from our group evaluating energy metabolic disturbances following ICH surgery 2 .
Pre-and post-operatively the patients were treated by a standardized neurocritical care protocol as previously described 2 . None of the patients had any known neurodegenerative disorder, dementia or Alzheimer's disease at time of ICH onset.
Plasma and CSF sampling. Blood plasma was drawn daily through an arterial line catheter. CSF samples were drawn from the EVD daily following discarding of the initial 2 mL of CSF. Plasma and CSF samples were then centrifuged at 1800 G for 10 min at 4 degrees Celsius and the supernatant was decanted into 1 mL aliquots and stored at -80 degrees C until further analysis.
Microdialysis. Microdialysis catheters of 10 mm length with a molecular weight cut-off of 100 kDa (CMA-71, M-dialysis AB, Solna, Sweden) were used in accordance with department routines. Catheters were perfused with 5% human albumin in a solution containing the excipients sodium chloride, N-acetyl-DL-tryptophan and caprylic acid (Albunorm, 50 g/l, Octapharma AB, Stockholm, Sweden), at a rate of 0.3 µL/min using the CMA 106 perfusion pump (M-Dialysis AB, Solna, Sweden) 4 . The first 2 h of sampling were discarded according to consensus praxis 22,23 . Microdialysis samples were collected every 2 h for routine analysis of small molecular metabolites (glucose, lactate, pyruvate, glycerol and glutamate) [22][23][24][25] . Following this analysis, the remaining MD sample (approximately 30 µL/vial) was frozen and stored at − 20 °C, and typically within 2-8 weeks transferred to Eppendorf vials and stored at − 86 °C until further analysis.
Analytical methods. Analysis of energy-metabolic markers. The ISCUS Flex® analyser (M Dialysis AB, Solna, Sweden) was used bedside in the neurointensive care unit (NICU) to determine extracellular levels of glucose, lactate, pyruvate, glycerol and glutamate immediately after sample collection. As per the manufacturer's instructions the lower limit of detection (LLOD) was 1.0 μmol/L for glutamate, 0.1 mmol/L for glucose and lactate, 10 μmol/L for pyruvate and 0.22 mg/mL for glycerol. Metabolite concentrations in the MD samples were analysed by the enzymatic method. Sample volume required was 0.5 µL for glucose, 0.2 µL for lactate, 0.5 µL for pyruvate, 0.5 µL for glycerol, 1 µL for glutamate and 0.5 µL for urea, leaving approximately 30 µL for the biomarker analysis.
Analysis of neuroaxonal injury biomarkers. All fluid samples were analysed for Aβ40, Aβ42, tau and NfL using commercially available assays on the Single molecule array (Simoa) HD-1 Analyser (Quanterix, Billerica, MA). Specifically, NfL concentration was measured using the NF-Light kit, whilst Aβ40, Aβ42, and tau concentrations were measured using the Neurology 3-Plex panel (Quanterix Billerica, MA). Plasma samples were diluted fourfold, according to the kit inserts. CSF and microdialysis samples were diluted 200-or 400-fold (depending on analyte concentration). All measurements were performed in one round of experiments by board-certified laboratory technicians who were blinded to clinical data. Intra-assay coefficients of variation were below 10%.

Statistical methods.
As biomarker levels were non-normally distributed, non-parametric Kruskal-Wallis test was employed for multiple group comparisons, and Friedman's test for repeated measures when biomarker levels were compared over time. To correct for multiple comparisons the Bonferroni method was used. Wilcoxon signed rank test was employed to compare two related groups. Correlations were investigated using Spearman's rho (ρ). All statistical tests were 2-sided and the significance threshold was set at p < 0.05. Normally distributed data are presented as mean and standard deviation (SD). Non-normally distributed data are pre- Energy-metabolic markers. The initial 68 h of monitoring of the energy metabolites revealed significantly lower extracellular fluid (ECF) level of glucose in the PHZ compared to the NCX ( Fig. 1a; p = 0.004), however, levels were above critical in both locations 23 . ECF level of lactate was significantly higher in the PHZ when compared to the NCX (data not shown; p < 0.0001), as was the pyruvate (data not shown; p < 0.0001), lactate/pyruvate ratio (LPR; Fig. 1b; p = 0.001), glycerol ( Fig. 1c; p < 0.0001) and glutamate ( Fig. 1d; p < 0.0001). Aβ, tau and NF-L levels in the different bodily fluid compartments. Median Fig. 2a-d). Levels of all biomarkers were significantly higher in CSF and ECF than in plasma (p < 0.001; Fig. 2a-d).

Discussion
In this study, the use of paired microdialysis (MD) enabled sampling of extracellular fluid (ECF) of the perihemorrhagic zone (PHZ) as well as of non-injured, control cortex (NCX) at a distance from the ICH. In addition, we also sampled cerebrospinal fluid (CSF) and plasma, and we measured markers of neuroaxonal injury in www.nature.com/scientificreports/ these three compartments over the first 60 h following surgery, which represents the time period when MD was clinically indicated as part of the multimodal monitoring in the neurocritical care setting which, for most ICH patients, is 2-4 days post-onset 27 . We observed that levels of Aβ40 were lower, and levels of tau higher, in the PHZ when compared to NCX. All biomarkers were measureable in the three compartments, with the highest levels observed in CSF, and lowest in plasma. Over time, the dynamics were different in the three compartments without any correlations across ECF, CSF and plasma, emphasising that for adequate interpretation of cerebral events leading to the release of a biomarker, knowledge of the relationship of biomarker levels in different compartments is crucial. www.nature.com/scientificreports/ Similarly to previous results published by our group 2 , increased ECF lactate-pyruvate ratio (LPR) indicated a metabolic crisis in the PHZ. In addition, we found lower ECF Aβ-40, and higher ECF tau concentrations, in the PHZ without any correlations between biomarker levels and metabolic markers of metabolic distress. This could reflect PHZ pathophysiology, including edema formation, mitochondrial dysfunction, and inflammatory responses 28,29 . Similar to our findings of lower Aβ40 in the PHZ, a previous study of 18 traumatic brain injury (TBI) and SAH patients found decreased ECF levels of Aβ40 and Aβ42 10 plausibly reflecting a decreased neuronal synaptic activity. In another TBI study, levels of Aβ40 and Aβ42 were higher in diffuse axonal injury (DAI) compared to focal TBI 15 . In these studies, the MD catheters were placed predominately in non-injured tissue, and not in brain tissue close to a focal lesion. Thus, these data are sampled from a less injured brain region than the PHZ region evaluated here.
We observed higher levels of the axonal injury biomarker tau in the PHZ compared to NCX, similar to findings of previous studies of TBI 9,14 , and SAH 7 patients, whereas there was no difference in levels of NF-L in the PHZ compared to NCX.
Levels of Aβ-40, Aβ42, tau and NF-L were significantly higher in CSF compared to ECF and plasma, although all were measureable in all three compartments. Higher biomarker, in CSF than in ECF or plasma has also been found in previous studies in TBI and SAH patients 7,9,10 possibly due to reduced relative recovery in ECF 10 . Following release of biomarkers produced by the brain injury into the ECF, monitoring by MD is preferable. However, the invasive nature, the small focal area that can be measured, and relatively poor time resolution of the method enables its use only in highly selected patients 30 . In TBI, biomarkers of neuroaxonal injury can be detected in ECF, CSF and plasma often in falling concentrations 31,32 . Our findings of lower levels of biomarkers in the ECF than in the CSF may reflect a reduced relative recovery across the MD membrane 30 , reduced recovery due to biomarker adsorption to surfaces 5,33 , physiological factors including tissue clearance 34 and variability in www.nature.com/scientificreports/ blood-brain-barrier (BBB) integrity causing contamination by serum levels [35][36][37] . Peripheral plasma levels on the other hand, present the least invasive method although challenges include dilution effect of CNS derived biomarkers, and contamination by peripheral non-CNS production, reflected in our present study by significantly lower levels in plasma of all biomarkers. ECF concentrations of Aβ did not fluctuate over time, implying a consistent production or release can a stable relative recovery be assumed 10 . We cannot exclude a decreased relative recovery masking any increased cerebral release of Aβ peptides, however 38 . In contrast, plasma and CSF levels of Aβ increased over the monitoring period, possibly reflecting an increased permeability of Aβ through the blood brain barrier (BBB) 39 , or increased peripheral production 40,41 .
Levels of tau decreased with time in ECF and plasma, and to some extent also in CSF, which could be reflective of a high level of axonal damage initially following ICH, which then decreases over the first 60 h following surgery.
Levels of neurofilament light (NF-L) were stable over time in ECF and CSF but showed significant increase in plasma over the initial monitored time period. Such an increase in plasma NF-L following acute brain injury has also been shown in several studies of a variety of neurological disorders 18,[42][43][44][45][46][47] , with higher levels associated with more severe injury and poorer functional outcome.
There was a strong correlation between the levels of Aβ40 and Aβ42 within each compartment, which was expected since they share the same precursor protein (amyloid precursor protein; APP) and are typically secreted in a stable ratio 48 . Similarly, there was a correlation between ECF tau and NF-L-both considered markers of axonal injury-a finding in line with previous studies of SAH and TBI patients 9,49-52 . However, rather surprisingly, there was no correlation between individual levels of biomarkers in the ECF when compared to levels in CSF or plasma. This lack of correlation suggests caution when interpreting plasma or CSF biomarkers levels as indicators of ECF levels. Thus, more work is needed to understand the dynamics of evolving tissue using plasma or CSF biomarkers.
No previous study has to our knowledge compared levels and dynamics of Aβ, tau and NF-L in ECF, CSF and plasma 53 . In a previous study of six TBI patients, the levels of the F 2 -isoprostane 8-iso-prostaglandin F 2α, a biomarker of oxidative stress, were higher in ECF when compared to both plasma and ventricular CSF 13 , contrary to our present findings. A study of IL-6 levels in SAH patients showed higher levels in CSF than in ECF and lowest in plasma. Furthermore, IL-6 levels in CSF and ECF could predict neurologic deterioration which plasma levels could not 54 . As we have not explored relative recovery of each biomarker in this study, we cannot determine their true extracellular concentration. It is, however, plausible that the ECF concentration sampled by MD represents only a fraction of the true extracellular concentration which may be even higher than that observed in CSF. The sampling of Aβ from ECF is challenged by a tendency for Aβ to adsorb to microdialysis membrane, tubing, and vials which was avoided in our study by the use of Albumin in the perfusate 10 . Our set-up with paired catheters also allowed for a comparison of ECF biomarker levels between MD catheters, and presumably the relative recovery is similar between the NCX and PHZ.

Conclusion
We found lower levels of extracellular Aβ40 and higher levels of extracellular tau in the perihemorrhagic zone (PHZ) when compared to non-injured cortex (NCX), during the first 60 h following surgical evacuation of intracerebral hemorrhage. These data suggest ongoing neuroaxonal injury in the PHZ, allowing for monitoring of the secondary injury process. Furthermore, median levels of Aβ40, Aβ42, tau and NF-L were higher in the cerebrospinal fluid than in ECF, which in turn was much higher than in plasma. We found poor correlation between levels of Aβ, tau and NF-L in the ECF when compared to CSF and plasma. Since the development of e.g. edema may be prolonged and monitoring with microdialysis not be feasible, studies of CSF and blood can be used. However, our results emphasize that for adequate interpretation of cerebral events leading to the release of a biomarker, knowledge of biomarker levels, dynamics and correlations in different compartments is crucial.

Data availability
The data supporting the findings in this study are available from the corresponding author, upon reasonable request.