Spinal cerebrospinal uid ow is increased in rats with elevated intracranial pressure 18 hours after cortical ischaemic stroke

Background A dramatic oedema-independent intracranial pressure (ICP) rise occurs 24 hours post-stroke in rats and may explain infarct expansion. Underlying mechanisms of this rise are unknown but evidence suggests cerebrospinal uid (CSF) dynamics are involved. Methods We investigated how CSF ow changes post-stroke and how this relates to ICP by infusing CSF tracer into the lateral ventricles of rats and assessing transport time and total tracer transport to the spinal subarachnoid space over a 90 minute period. Results Stroke animals with ICP rise had faster tracer transit when compared with stroke animals without ICP rise (27.6 ± 4, n = 6, vs 48.6 ± 4.5 mins, n = 6) or animals subjected to a sham procedure (47.9 ± 4 mins, n = 8), F(2,17) = 0.1, p ≤ 0.01. There was a correlation between tracer transit time and ΔICP (R = -0.52, p=0.02) and infarct volume (R = -0.6, p=0.04). There was no difference in total tracer observed. Conclusions Faster tracer transit in stroke animals may be explained by impairment of other CSF outow pathways, whereby, spinal drainage acts as a compensatory mechanism. Investigation into the disruption of other CSF drainage routes post-stroke may offer insight into the underlying mechanisms of infarct expansion post-stroke.


Abstract
Background A dramatic oedema-independent intracranial pressure (ICP) rise occurs 24 hours post-stroke in rats and may explain infarct expansion. Underlying mechanisms of this rise are unknown but evidence suggests cerebrospinal uid (CSF) dynamics are involved. Methods We investigated how CSF ow changes post-stroke and how this relates to ICP by infusing CSF tracer into the lateral ventricles of rats and assessing transport time and total tracer transport to the spinal subarachnoid space over a 90 minute period. Results Stroke animals with ICP rise had faster tracer transit when compared with stroke animals without ICP rise (27.6 ± 4, n = 6, vs 48.6 ± 4.5 mins, n = 6) or animals subjected to a sham procedure (47.9 ± 4 mins, n = 8), F(2,17) = 0.1, p≤0.01. There was a correlation between tracer transit time and ΔICP (R = -0.52, p=0.02) and infarct volume (R = -0.6, p=0.04). There was no difference in total tracer observed. Conclusions Faster tracer transit in stroke animals may be explained by impairment of other CSF out ow pathways, whereby, spinal drainage acts as a compensatory mechanism. Investigation into the disruption of other CSF drainage routes post-stroke may offer insight into the underlying mechanisms of infarct expansion post-stroke.

Background
Ischaemic stroke is a leading cause of death and disability worldwide [1]. Stroke severity is variable and some patients present with relatively minor stroke. While occlusion of small branch vessels accounts for many of those with minor symptoms, it is now recognised that some patients with minor symptoms may have large vessel occlusion [2]. Around 10-20% of patients with minor stroke or transient ischaemic attack will experience early recurrent stroke, and the vast majority of these occur in the initially ischaemic vascular territory [3,4].
The most likely cause of infarct expansion is failure of leptomeningeal collateral vessels. Failure of initially good collateral blood ow is associated with infarct growth following ischaemic stroke [5]. The cause of collateral vessel failure has not been de nitively established. However, after stroke, blood ow in these vessels is largely driven by cerebral perfusion pressure (CPP) which is sensitive to changes in intracranial pressure (ICP) [6]. In support of this, our previous work showed that elevation of ICP during middle cerebral artery occlusion (MCAo) in rats caused a linear reduction of collateral blood ow [7]. We previously identi ed a transient but dramatic rise in ICP 24 hours after minor ischemic stroke in rats [8]. For the rst time, this study reported ICP rise after minor ischemic stroke. The time point of this ICP rise, taken with our understanding of how ICP in uences collateral blood ow, suggests a possible mechanism of collateral failure and infarct expansion.
Our understanding of the underlying mechanisms of this ICP rise is limited. We know that oedema is not the primary cause [9] and pilot data from our lab found no contribution of cerebral blood volume [unpublished data]. Resistance to CSF out ow is increased at 18 hours post-stroke in a rat model of cortical ischaemia and at 24 hours post-stroke in a model of striatal ischaemia [10,11]. This suggests that CSF volume is increased at these time points.
The relative importance and contribution of CSF drainage pathways are not well de ned [12]. CSF drains from the cerebral ventricular system to the spinal subarachnoid space and canal before moving into lymphatic vessels of the sacral spine [13,14]. This offers an opportunity to image CSF dynamics in vivo by focusing on spinal CSF ow. In this study, we evaluated how spinal CSF ow is altered post-stroke and whether this is related to ICP rise. We aimed to determine whether CSF ow from the cranium to the spinal subarachnoid space is impaired after stroke and whether this relates to ICP rise.

Animals
Procedures were carried out on male outbred Wistar rats aged between 8 and 12 weeks (n = 20) weighing between 280-320 g. All experimental animal procedures used in this project were in accordance with the Australian Code of Practice for the Care and Use of Animals for Scienti c Purposes and were approved by the Animal Care and Ethics Committee of the University of Newcastle (A-2013-343). The studies were conducted and the manuscript prepared in accordance with the ARRIVE guidelines [15].
Animals were excluded from experiments if they presented with congenital deformities that obstructed surgical intervention. Animals were only involved in the analysis following con rmation of stroke and ventricle penetration of infusion catheter after histology.
Animals were assigned to stroke or sham groups on day 0, prior to intervention. Blinding was not carried out during experimental procedures; however, the investigator was blinded at the time of data extraction and analysis: each animal was assigned an experiment number and analysis was carried out on all data in one sitting with no indication of treatment to the investigator. Five animals per group were required to detect a 30% change in CSF tracer transit between stroke and sham animals with 80% power and α error probability of 0.05 (SD = 6.2). Power calculations were conducted on pilot data using power calculation software (G*Power 3.1.9.2). Pilot data was included in the nal analysis.

Anaesthesia and monitoring
Rats were anaesthetised with iso urane (5% induction, 2-2.5% maintenance) in 50:50% N 2 :O 2 . Incision sites were injected subcutaneously (s.c.) with 2 mg/kg 0.05% Bupivacaine (P zer, Sydney, Australia). Core body temperature was regulated via a thermocouple rectal probe (RET-2, Physitemp Instruments Inc, Clifton, New Jersey, USA) and heat mat. Blood gases were monitored on day 0 prior to stroke and day 1 prior to laminectomy from 0.1 ml blood samples from a femoral arterial line. This line was also used for arterial blood pressure monitoring. Prior to recovery, an additional Bupivacaine injection (0.3 ml, 0.05%, s.c.) and rectal paracetamol (250 mg/kg; GlaxoSmithKline, Brentford, UK) were administered for overnight pain relief. Saline was administered intraperitoneally (2 × 1.5 mL) to replace uid loses. Following surgery, animals were returned to their cages with free access to food and water. Cages were placed half over a heat mat to allow animals to thermoregulate during recovery. ICP measurement ICP was measured using a bre-optic pressure sensor (Opsens Solutions, Quebec, Canada). The wire was sealed 5 mm inside a hollow saline-lled screw in the left parietal bone (-1.8 mm lateral and − 2 mm posterior to Bregma). Correct positioning was con rmed when a trace of pulse and respiratory oscillations were observed after the wire was sealed in place. Baseline ICP was recorded for 30 minutes prior to photothrombotic stroke, and day 1 ICP was recorded from 18 to 21 hours.

Photothrombotic stroke
A light source (100,000 LUX, 5 mm diameter, Olympus Corporation LG-PS2, Tokyo, Japan) was placed over the right parietal bone 0.5 mm from the skull surface, 1.2 mm posterior to Bregma. While the light source was concentrated on the skull, Rose Bengal in saline (0.01 g/kg; Sigma, St Louis, Missouri, USA) was infused into the right femoral vein followed by 1 mL saline. The light source remained constant for 20 minutes after Rose Bengal infusion. Sham animals were administered saline alone.

Laminectomy
Laminectomy was carried out as previously described [16]. Brie y, an incision was made from the back of the head caudally to the mid-thoracic region. The connective tissue was blunt dissected away and a cut was made through the midline of each muscle layer. The paraspinous muscles were retracted to expose the spinal column and connective tissue was blunt dissected. A dental drill (Saeshin Precision, Pahodong, Korea) was used to thin the laminae of the C7 and T1 vertebrae. The laminae were then cut and removed at these points to expose the spinal cord.

Evans blue infusion
Evans blue was infused bilaterally into the lateral ventricles (2% w/v in aCSF, 20 µL total volume, 2 µL/min). Intraventricular catheters were sealed in head screws positioned 0.8 mm posterior and 1.8 mm lateral to Bregma on both parietal bones. The catheters were connected to two syringes containing Evans blue and infusion was controlled by a syringe driver (Harvard Apparatus, Holliston, MA, USA). Penetration of the ventricles was later con rmed by histology.

Image acquisition and ImageJ analysis
Images were acquired using a standard microscopic eyepiece camera connected to a PC running basic image acquisition software. An image was captured every minute for 90 minutes. These images were then loaded into ImageJ as a sequence. The images were converted to an 8-bit format with the colour channels split; only the red channel was used for analysis as this minimised contrast from blood vessels.
Two regions of interest were selected: the spinal cord and an area out-with the spinal cord to correct for background noise. Difference between each image and baseline were calculated to show contrast development.

Histological Analysis
The animal was euthanised at 3 hours post-infusion and was transcardially perfused with 0.9% saline. The brains were then xed in neutral-buffered formalin before being processed and para n embedded.
Coronal sections of 5 and 10 µm were cut and stained with hematoxylin and eosin. Images were scanned using a digital slide scanner (Aperio Technologies, Vista, CA, USA) and image analysis was carried out within the Aperio programme. Image analysis was con rmed by an independent investigator and any cases with > 10% discrepancy were agged for review.

Experimental design
The experimental timeline is outlined in Fig. 1. Physiological variables were recorded at baseline on day 0 prior to photothrombotic stroke and at 18 hours post stroke, prior to surgical intervention. Screws were placed over the lateral ventricles and a dental cap tted on day 0, ready for infusion the following day. A C7-T1 laminectomy was performed on day 1 to expose this area of the spinal cord. Bilateral infusion of Evans blue into the lateral ventricles took place at 18 hours post-stroke. White light images were acquired of the C7-T1 spinal cord every 60 seconds from zero to 90 minutes post-infusion and images were analysed using ImageJ software (NIH). The animals were sacri ced at 3 hours post-infusion and histological analysis was carried out as described.
Although we previously reported a transient ICP rise at 24 hours post-stroke [9], we chose 18 hours poststroke as our time point for this investigation. This is because we would expect any changes to CSF ow to be present at this earlier time point, which would contribute to the ICP rise we observe later. If we investigate CSF changes at 22 to 24 hours, there is a high likelihood of missing these changes, as we would expect the system to start to return to baseline levels at this point.

Statistical analysis
Statistical analyses were carried out using GraphPad Prism 7 (GraphPad Software, La Jolla, CA, USA). Data was tested for normal distribution using a D'Agostino and Pearson normality test. Comparisons between two groups were analysed by unpaired Student's t-test for normally distributed data and by Mann-Whitney U-test for non-normally distributed data. Comparisons between three or more groups were carried out using one-way ANOVA. All normally distributed data values are presented graphically as average ± standard deviation and non-normally distributed data as median and interquartile range (IQR). Relationships were determined by Pearson correlation unless otherwise stated. Table 1 shows the physiological parameters of stroke and sham animals on day 0 and day 1.

Exclusions
Three animals were excluded from experimentation as they presented with congenital deformities that prevented surgical intervention. Four animals were excluded due to technical errors during procedures and one animal died before the procedure could be completed.
ICP rise occurred in six out of 12 stroke animals ICP rise (≥ 5 mmHg) occurred in stroke animals but not sham (Fig. 2). ΔICP was signi cantly higher in stroke animals (3.53 mmHg, IQR = 0.06 to 9.41) than in sham animals (-0.71 mmHg, IQR = -2.47 to 1.09 n = 8), U = 19, p = 0.025. We found that 6/12 stroke animals had an ICP rise between baseline (day 0) and 18 hours post-stroke, and there was a clear bimodal distribution from those that did not have a rise. For subsequent analyses to determine whether there was an association between ICP rise and CSF ow, we separated stroke animals into two groups depending on whether they had an ICP rise or not.

Increased infarct volume correlates with increased CSF ow
We found that there was a strong trend to a correlation between infarct volume and ΔICP from baseline to 18 hours post-stroke, although not statistically signi cant (Fig. 4A), R = 0.55, p = 0.07, Spearman's correlation. However, we identi ed a signi cant correlation between infarct volume and T50%max, which demonstrates faster transit of Evans blue to the C7-T1 spinal subarachnoid space in animals with larger infarct volumes (Fig. 4B), R = -0.6, p = 0.04.

Maximum contrast did not differ between groups
To estimate the contribution of CSF secretion to changes in CSF ow, we determined the maximum change in contrast from 0 to 90 minutes post-infusion for each animal (Fig. 5A). The maximum change in contrast did not signi cantly differ between stroke animals with ICP rise (175.2 ± 15.4 Arbitrary Units (AU), n = 6), stroke animals without ICP rise (152 ± 18.1 AU, n = 6), and sham animals (125.1 ± 16.7 AU, n = 8), F(2, 17) = 2.28, p = 0.13.

Discussion
In this study, we demonstrated that CSF ow from the cranial compartment to the spinal subarachnoid space is faster after stroke. We observed that oedema-independent ICP rise is present 18 hours after cortical photothrombotic stroke in rats and that this ICP rise correlates to faster movement of CSF tracer to the spinal subarachnoid space.
Resistance to CSF out ow is increased at 18 and 24 hours post-stroke in cortical and striatal ischaemia models, respectively [10,11]. We expected stroke animals to have decreased and delayed movement of CSF tracer to the spinal subarachnoid space, indicative of impaired and slowed CSF movement which would possibly explain ICP rise at 24 hours post-stroke [9]. Our data show that animals experiencing an ICP rise post-stroke had faster transit of CSF tracer to the spinal subarachnoid space, while animals that did not experience ICP rise post-stroke were similar to sham animals. This observation was further con rmed when we identi ed a correlation between ICP rise at 18 hours post-stroke and Evans blue transit time to the C7-T1 spinal subarachnoid space. These ndings show that changes to CSF ow poststroke are linked to ICP rise; however, it remains unclear if ICP rise is the direct result of these changes or vice versa. Previous studies have observed increased transport of CSF tracer into the extracranial lymphatics system and through arachnoid projections with increased ICP [17,18]. It is possible that increased CSF transit to the spinal subarachnoid space occurs in response to ICP rise following impaired CSF drainage via another cranial drainage pathway post-stroke. This is a possible compensatory mechanism given that CSF drains along spinal nerve routes extending from intervertebral spaces and into the peripheral lymphatics located in the sacral spine [14].
Despite faster transit to the C7-T1 subarachnoid space, the maximum amount of tracer present at this region between 0 to 90 minutes post-infusion was not signi cantly different. The lack of signi cant difference between the groups suggests that increased CSF secretion may not be involved in the ICP rise we observe. If CSF secretion was increased in animals experiencing ICP rise, then we would expect maximum observed contrast to be lower as the tracer becomes more dilute.
We investigated whether oedema could be the underlying cause of ICP elevation at 18 hours post-stroke, perhaps contributing to faster Evans blue transport to the spinal subarachnoid space. We found that oedema volume did not correlate with either ΔICP rise or T50%max. This nding is in line with our previous report that oedema does not correlate with ICP rise at 24 hours after transient MCAo [9] and with work showing similar observations after permanent MCAo [11]. Oedema volumes were assessed early and would likely be greater at later time points. However, we still observe ICP rise from 18 to 21 hours, which suggests that mechanisms other than oedema contribute to ICP rise and increased tracer movement after photothrombotic stroke.
In this study, 6 out of 12 stroke animals had ICP rise 18 hours post-stroke. This ICP rise was smaller than we previously reported with MCAo at 24 hours and photothrombotic stroke at 22 to 24 hours post-stroke [8,9,19]. Here, we used a photothrombotic technique to produce a cortical ischaemia instead of a striatal ischaemia. This was the preferred model for our study of CSF ow as we aimed to maintain choroid plexus integrity throughout our experiments. MCAo, a striatal stroke model, reduces blood ow to the choroid plexus by around 62% and causes choroidal oedema leading to reduced blood-CSF barrier integrity and increased CSF secretion [20]. Further, the earlier time point of our investigation may explain the smaller ICP rise, as we would expect ICP to continue to rise and peak at 22 hours post-stroke as previously observed [19]. We chose this time point to observe any changes that may contribute to ICP rise as we expect changes to reduce later as the system returns to normal.
Experiments to elucidate physiological CSF dynamics are challenging and often involve some sort of intervention. In this study, we infused the CSF tracer, Evans blue, directly into the lateral ventricles of the animals [18,21,22]. This type of intervention has the potential to disturb normal CSF dynamics; however, this technique is widely used and we took care to limit the infusion rate to lower than that of CSF production (2.66-2.84 µl/min) to maintain the integrity of the system as much as possible [23].
Several CSF drainage hypotheses have been presented in recent years and we are yet to meet a consensus on the topic [24][25][26]. In recent years, lymphatic drainage of CSF from both the cranium and the spinal space have gained merit in the eld, with several groups demonstrating the presence of CSF tracers in the peripheral lymphatic system, primarily the cervical lymphatics [21,[27][28][29][30]. This may be an interesting direction of investigation to understand how CSF drainage is altered post-stroke.

Conclusions
Overall, we demonstrate that CSF ow from the cranial compartment to the spinal subarachnoid space is increased 18 hours post-stroke. Taken with our observation of increased resistance to CSF out ow at 18 hours post-stroke [10] and that oedema does not correlate to ICP rise or Evans blue transit, we believe that altered CSF movement and drainage from the CNS contributes to ICP rise post-stroke. This hypothesis must be further explored to elucidate the exact changes to CSF dynamics that occur. These changes may offer insight into the underlying mechanism of ICP elevation and collateral failure and could be used as a therapeutic target to improve patient outcome after minor ischaemic stroke.