Investigating Intracranial Pressure Elevation and the Cerebrospinal Fluid Proteome Post-Stroke

Rebecca Joan Hood B. Biomed Sci. (Hons)

March, 2019 Thesis submitted in fulfilment of the requirements for the degree of Doctorate of Philosophy in Human Physiology.

This research was supported by an Australian Government Research Training Program (RTP) Scholarship

I hereby certify that the work embodied in the thesis is my own work, conducted under normal supervision. The thesis contains no material which has been accepted, or is being examined, for the award of any other degree or diploma in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made. I give consent to the final version of my thesis being made available worldwide when deposited in the University’s Digital Repository, subject to the provisions of the Copyright Act 1968 and any approved embargo.

Rebecca J. Hood

Acknowledgements

Firstly, I would like to thank my wonderful supervisors Neil Spratt, Mark Baker, Damian McLeod and my unofficial supervisor Peter Dunkley. We did it!! Words cannot express my gratitude. No matter how busy any of you were, you always made the time to provide assistance and guidance whenever I needed it. I have learned so many important lessons from each of you and will be lucky if one day I am able to help junior researchers in the way that you have helped me.

Next, I would like to thank the Spratt Lab (past and present). I have grown so much as a researcher and as a person since I began with you all in 2010 (mostly for the better) and have felt so lucky to have worked alongside so many wonderful people and made some life-long friends. You kept me laughing, positive (even when things seemed bleak) and kept me company in the MSB dungeon where I spent the majority of my days. I would like to give special shout outs to my mentor (Lucy) and mentee (Kirb) – I cannot thank either of you enough for your support and encouragement throughout my PhD (including keeping me sane at the end). Additional shout outs to Daniel, Jackie (my protégé), Kah Ni and Cait (Mart). I’d also like to extend my gratitude to unofficial members of the Spratt Lab Nishani, Murielle and Rohan for your constant encouragement. Lastly, I would be remiss if I did not extend my warmest thanks to Deb! You have made every day that I have worked in the Spratt Lab brighter just by being yourself.

I would also like to thank my adoptive lab – the Bakers! Thank you for making me feel so welcome in your lab! Rachel, Louise and Jacob your patience and unwavering support as I attempted to navigate the confusing world of mass spectrometry could not be more appreciated. You are wonderful scientists and I have enjoyed every second of working with you.

I would like to thank those who have supported me financially. I have been incredibly lucky to receive financial support from the University of Newcastle as well as Neil Spratt through a UoN 50:50 scholarship. In addition, I would also like to acknowledge the wonderful Anne Greaves who generously provided a top up scholarship allowing me to focus more time on my studies.

To all of my wonderful friends (including the Spratts and Bakers) who have supported me throughout my terrible life choice joyful post graduate studies. You have kept me fed, housed (when I was too tired to drive home) and supported in more ways than I can count. Special shout out to Coles #0885, Rachelle, Brodie, the Druitts, the Hedleys2 and the Grennalls, and especially my twin – Claire! Lastly, and most importantly, I would like to thank my wonderful family. You have supported my every crazy idea over the past 28 (and a bit) years and this has been no exception. You have fostered my love of all things science since childhood and encouraged me every step of the way whether it was buying me a science kit when I was in primary school, encouraging me to ask questions or buying me the same forensic book every Christmas. Thank you for all that you have done and continue to do for me. (Also Josh, thanks for screening all of my calls and then mostly calling me back. Could not have done this without you!)

Table of Contents

Abstract ...... 1

Chapter 1 Introduction ...... 3

1.1 Stroke ...... 3 1.2 Intracranial Pressure ...... 4 1.3 Investigating the trigger for ICP elevation after mild-moderate stroke ...... 15 1.4 CSF composition ...... 19 1.5 Aims and rationale ...... 23

Chapter 2: Methods ...... 24

2.1 Introduction ...... 24 2.2 Substances investigated for effect on ICP ...... 24 2.3 Human CSF ...... 25 2.4 Animals ...... 26 2.5 Surgical Procedures ...... 27 2.6 Proteomic Analysis of Human CSF Samples ...... 33 2.7 Statistical Analysis ...... 35

Chapter 3 Validation of epidural pressure measurement in the rat ...... 36

3.1 Introduction ...... 36 3.2 Part 1: Intracranial catheter systems ...... 36 3.3 Part 2: The influence of dural contact on ICP waveform and accuracy ...... 56 3.4 Part 3: Is a recovery period required between surgery and ICP measurement? ... 61 3.5 Part 4: Accuracy of epidural pressure measurement using a fibre optic pressure sensor ...... 66 3.6 Part 5: Accuracy of serial measurement of epidural and subdural pressure when ICP Is high ……………………………………………………………………………………………...74 3.7 Summary ...... 76

Chapter 4 The effect of arginine vasopressin on intracranial pressure ..... 77

4.1 Introduction ...... 77 4.2 Experimental Design ...... 78 4.3 Results ...... 79 4.4 Discussion ...... 85

Chapter 5 Human CSF Infusion ...... 88

5.1 Introduction ...... 88 5.2 Part 1: Human CSF transfusion ...... 88 5.3 Part 2: The CSF proteome and ICP elevation ...... 105 5.5 Summary ...... 121

Chapter 6 General Discussion and Conclusions ...... 122

6.1 Epidural pressure measurement is accurate and reliable in rats using fibreoptic technology ...... 122 6.2 Evidence of a potential molecular trigger for ICP elevation post-stroke ...... 124 6.3 Investigation of potential triggering molecules ...... 126 6.4 Could this ICP trigger be conserved across other neurological conditions involving raised ICP? ...... 127 6.5 Conclusions ...... 129

Chapter 7 References ...... 130

Chapter 8 Appendix ...... 144

Appendix A: Publications not included in this thesis ...... 144 Appendix B: Copyright Permissions ...... 178

Abstract

Background: Our group have recently made a number of important findings regarding intracranial pressure (ICP) elevation post-stroke. Until recently, it was assumed that only patients with large, malignant infarction experienced ICP elevation. However, recent experimental and clinical findings from our laboratory suggest the existence of ICP elevation at 24 hours after mild-moderate stroke. The assumption was that this patient population did not experience ICP elevation due to small volumes of cerebral oedema (the assumed cause of ICP elevation in all disorders of neurological injury). When investigating the underlying mechanism behind the dramatic, yet transient ICP elevation after mild-moderate stroke in rats we discovered that the rise occurred independently of oedema, suggesting a novel mechanism of ICP elevation. Further investigation into the mechanisms led to the discovery that the 24 hour ICP rise can be triggered in stroke-free animals, by administration of cerebrospinal fluid (CSF) from stroke animals. Preliminary evidence, from studies transfusing human stroke-CSF into rats, suggests that the ICP trigger may also exist in human CSF. Stroke is known to alter CSF composition and there have been a handful of studies showing correlation between CSF proteins e.g. arginine vasopressin (AVP) and ICP elevation.

Aims: The aims of this PhD were to 1) Confirm the accuracy and reliability of epidural pressure measurement in rats using a fibreoptic pressure sensor. 2) Determine whether intraventricular infusion of AVP causes delayed ICP elevation in naïve rats. 3) Determine whether intraventricular infusion of human stroke-CSF causes ICP elevation in naïve rats. 4) Identify potential protein candidates in human stroke-CSF that may be responsible for delayed ICP elevation.

Methods: Accuracy and reliability of epidural pressure measurement was assessed over a range of pressures against both subdural and intraventricular pressure (the gold standard). Epidural pressure measurement was then used to monitor changes in ICP following infusion of AVP (0.02–5 ng) or human CSF (from patients diagnosed with stroke, chronic hydrocephalus, subarachnoid haemorrhage, or control patients) into naïve (disease free) outbred, Wistar rats. Human CSF samples were then analysed using a Q-Exactive Plus mass spectrometer to determine relative protein expression. Stroke- and chronic hydrocephalus- CSF samples that caused ICP elevation ³5 mmHg in the recipient animals were compared against samples that did not cause a rise in ICP.

Results: Epidural pressure measurement was shown to demonstrate a strong correlation across a range of pressures, with pressure measured in both the subdural (rc = 0.89) and intra-ventricular spaces (rc = 0.11). Preliminary experiments found that infusion of AVP did

1 not cause significant ICP elevation above baseline (n = 4/ group; p>.05). Infusion of both stroke- and chronic hydrocephalus-CSF caused significant delayed ICP elevation in recipient animals (p=.02 and .002 respectively). Investigation of the stroke and hydrocephalus CSF proteomes resulted in the identification of 17 and 32 proteins respectively showing significantly altered expression when ICP elevating samples were compared against samples that did not affect ICP. Galectin -1 and Calcium/calmodulin dependent protein kinase II beta were identified as being significantly more abundant in the samples that caused ICP elevation in both stroke- and chronic hydrocephalus-CSF samples.

Conclusions: In this thesis, I have presented data that expands upon the working hypothesis that CSF composition is important in the delayed ICP elevation seen in humans and animals 24 hours after mild-moderate stroke. My findings confirm the validity of epidural pressure measurement in rats and expand on previous findings by showing this using fibreoptic pressure sensing technology. Secondly, my data indicates that AVP is unlikely to be the primary driver behind the delayed ICP elevation seen at 24 hours post-stroke in rats. Thirdly, my data suggests the presence of a molecular trigger for ICP elevation in the CSF of human stroke patients. Furthermore, by showing a similar rise in ICP following infusion of human stroke-CSF I have confirmed the clinical relevance of our animal work. Finally, I have identified two candidates of interest that may be important for ICP regulation not just post-stroke but also for patients with chronic hydrocephalus. These proteins may be potentially important for the diagnosis of stroke patients at risk of developing damaging ICP elevation. They may also be relevant to other neurological conditions involving ICP elevation. These preliminary findings are strongly suggestive of a novel mechanism of ICP elevation after stroke and justify further investigation into other neurological diseases of which ICP elevation is a harmful consequence.

2 Chapter 1 Introduction

1.1 Stroke

Stroke is a leading cause of death and the leading cause of adult disability worldwide (1). In Australia alone approximately 56,000 people suffer a new or recurrent stroke each year (2), with treatment and ongoing care costing approximately 5 billion dollars (3). The biggest risk factor for stroke is increasing age, with the risk of stroke reportedly doubling at every decade of life after the age of 55 (4). With Australia’s aging population it is predicted that by 2050 stroke incidence will more than double, increasing the already significant health and economic burden (2).

A stroke occurs when there is a disruption in the blood supply to the brain caused by either vessel occlusion (ischaemic stroke) or vessel rupture (haemorrhagic stroke) (5). The resulting neurological injury caused by the ischaemia, inflammation or direct compressive injury from the haemorrhage results in a loss of cellular function and has the potential to cause cell death. Ischaemic stroke accounts for approximately 80% of strokes in Australia (6) and will be the main focus of this thesis. ‘Stroke’ will therefore refer to the ischaemic pathology, unless otherwise stated.

1.1.1 Pathophysiology

Following vessel occlusion, ischaemia causes brain injury at a cellular level, by initiating the ischaemic cascade (7). This includes loss of function of membrane ion pumps (leading to cytotoxic oedema), release of excitatory neurotransmitters, mitochondrial failure, production of reactive oxygen species, and finally cell death (8, 9).

The affected tissue can be divided into two significant regions, the infarct core and the penumbra (10, 11). The infarct core is made up of irreversibly damaged brain tissue. This is the area of the brain suffering the greatest reduction in blood flow. Within minutes of occlusion the cells in this region suffer severe hypoxia and ATP depletion leading to metabolic failure and cell death. The tissue surrounding the core is a region of severely hypoperfused, yet potentially salvageable tissue, known as the ischaemic penumbra (10, 11). This tissue is functionally impaired yet remains viable due to residual blood supply networks (collateral blood vessels) (12, 13). Despite this residual flow, without reperfusion the cells in this region will die and become incorporated into the infarct core; this is known as infarct expansion (14). The extent of the damage from ischaemic stroke usually depends on the duration, severity and

3 location of ischaemia. The aim of current acute therapies is to restore blood flow to the penumbra as fast as possible.

1.1.2 Early infarct expansion

Approximately 10-40% of stroke patients experience early infarct expansion (15-17). These patients, who typically present to hospital with mild-moderate stroke symptoms, deteriorate within 1 to 2 days post-stroke (15, 18). Until recently it was thought that early infarct expansion occurred due to rethrombosis of reperfused vessels. However, there is a lack of clinical evidence to support this hypothesis (15). In fact, advanced imaging studies have shown stable arterial occlusion between acute and follow-up imaging (16, 17, 19). Interestingly, a key feature of these patients is deterioration of leptomeningeal collateral blood supply between acute and follow up imaging (collateral failure) (17). Many possible mechanisms have been proposed as to the cause of collateral failure including thrombus growth, collateral vessel thrombosis (20), venous steal (21), Reversed Robin Hood syndrome (22), and blood pressure fluctuations after autonomic dysfunction (23). However the exact cause remains to be elucidated.

Under physiological conditions cerebral blood flow is maintained across a wide range of cerebral perfusion pressures (CPP) by autoregulation (24). This involves vasoconstrictive and dilatory responses of cerebral arterioles in response to changes in perfusion pressure. Autoregulation of cerebral blood flow is lost during stroke, therefore perfusion of the collateral circulation (supplying the ischaemic penumbra) becomes dependent upon CPP (24). CPP is the difference between mean arterial pressure (MAP) and intracranial pressure (ICP) (25). A decrease in MAP or an increase in ICP has the potential to reduce collateral blood flow by reducing the driving pressure of flow across collateral vessels. We have recently confirmed this experimentally by quantifying blood flow through individual collateral vessels during artificial ICP elevation in animals following middle cerebral artery occlusion (MCAo) (26). Collateral flow decreased linearly with increasing ICP, to a mean 55% reduction in flow at 30 mmHg. Such a large reduction in flow would be expected to have important implications for penumbral survival. We hypothesise that post-stroke ICP elevation may be a major contributing factor to collateral failure and early infarct expansion.

1.2 Intracranial Pressure

Elevated ICP can complicate several forms of acute neurological injury including stroke (27- 29), intracerebral haemorrhage (30), subarachnoid haemorrhage (SAH) (31) and traumatic brain injury (32). It can also contribute to secondary neurological injury and potentially lead to death.

4 1.2.1 Invasive measurement of ICP

Pressure sensors

Clinically, ICP is measured invasively (33, 34). Most commonly ICP Is measured via fluid filled catheters. Fluid filled pressure catheters are comprised of a fluid filled tube connected to a pressure manometer often via a three-way stop-cock. Moreover, fluid filled catheters permit direct access to the cerebrospinal fluid (CSF) (if placed within the lateral ventricles) (34, 35). This can be useful for infusion of therapeutic agents as well as the potential to remove CSF for testing and/ or to lower ICP. However, they are prone to blockages and air bubbles (36) and sensitivity to hydrostatic artefacts (37), potentially leading to false ICP measurements. This is more of an issue in animal studies due to the very fine gauge catheters required.

Another option for ICP measurement is to use solid state technology. This includes strain gauge (piezoelectric) as well as fibre optic pressure sensors. Solid state pressure sensors are more expensive than fluid filled catheters however they provide higher signal quality and are less likely to fail during use (36, 38, 39). Comparative studies have shown no statistically significant differences between ICPs recorded by either piezoelectric or fibre optic sensors (40). However, during use, piezoelectric pressure sensors are sensitive to electromagnetic interference as well as temperature fluctuations (41). Therefore, there are a number of variables that need to be controlled when using piezoelectric technology (41). In contrast, fibreoptic pressure technology is less susceptible to noise/ electromagnetic interference (42). Taken together with the higher sensitivity over fluid filled technology, fibreoptic pressure sensors represent the more ideal means of ICP measurement.

Location of pressure measurement

Several locations for measurement of ICP exist. Cannulation of the cerebroventricular system is considered the clinical “gold standard” for measuring ICP (34, 43, 44). This involves drilling through the skull and inserting a pressure catheter through the parenchyma and into one of the lateral ventricles. Despite being recognised as the most accurate site for ICP measurement, intraventricular pressure measurement is incredibly invasive. There is also an increased risk of infection (45, 46), haemorrhage (47), hydrocephalus (36) and potential difficulty of placement (especially if there is midline shift) (34, 48, 49). Therefore other sites such as intraparenchymal (50-52), subarachnoid (46), subdural (46) epidural (51, 52) and lumbar spine (53) have also been used in humans (Figure 1.1).

The epidural space is an appealing location for pressure measurement as placement of pressure sensors is comparatively simple and it does not require the dura mater to be pierced.

5 This reduces the chance of infection, haemorrhage and hydrocephalus (36) as compared to the more invasive locations. However, in humans, the dura mater is extremely thick (~270 to 1000 µm (54, 55)) and it is hypothesised that the thickness and inelasticity of the dura prevent reliable pressure signals being transmitted between the parenchyma and pressure sensor. Therefore, epidural pressure measurement is rarely used in humans due to questions over the accuracy of measurements (56, 57).

Figure 1.1 Schematic showing locations for intracranial pressure (ICP) measurement. This diagram shows placement of pressure sensors in the a. epidural, b. subdural, c. subarachnoid, d. intraparenchymal, e. intraventricular spaces.

6 Experimentally, there have been numerous techniques developed for measuring ICP in rat studies. Locations for measurement include the intraventricular (58, 59), intraparenchymal (38, 60), subarachnoid (61-63), cisterna magna (38, 64, 65), lumbar spine (65) and epidural spaces (36, 38, 39, 66, 67) (Figure 1.1).

Epidural pressure measurement is more common in experimental studies compared with clinical studies. The justification for its use in rats is that the dura mater is far thinner (~80 µm) (68) than in humans and is unlikely to impede pressure readings. Despite the widespread use of epidural pressure monitoring in rats there are only three studies that investigate the accuracy and reliability of epidural pressure against other locations of pressure measurement and one that compares it against the gold standard of intraventricular pressure (36, 38, 60). These studies present conflicting results. Verlooy et al. (1990) raised questions about the accuracy of epidural pressure measurement in rats. The authors anecdotally reported artificially high ICP values when epidural pressure was measured using a fibreoptic pressure sensor compared with pressure measured at the cisterna magna using a fluid filled catheter (38). In contrast, two recent studies have shown excellent correlation between epidural pressure and both intraparenchymal (60) and intraventricular (36) pressures using fluid filled catheters. A potential, yet unlikely factor that may have contributed to the conflicting findings are the methodological differences in pressure sensing technology. Indeed, the studies that showed correlations of ICP measurement were performed using fluid filled catheters, whilst the other study used a fibreoptic sensor. However, a more plausible explanation for the conflicting findings is that these studies differed in the extent to which contact was made between the sensor and the dura. Both of the papers showing good correlation between pressure measurement locations specify that the tip of the sensor was flush with the inside of the skull (36, 60). By contrast Verlooy et al. (1990) did not ensure that the tip of the probe was not in contact with the dura (38). They anecdotally reported the discrepancy between the measuring locations but did not provide any specific data. Therefore, the exact cause of this discrepancy requires investigation to determine the efficacy of epidural pressure measurement. It is highly likely that epidural pressure measured with fibreoptic technology is as accurate (but more sensitive) than fluid filled. However, further investigation is required to confirm the accuracy and reliability of epidural pressure measurement using fibreoptic pressure sensors.

1.2.2 Non-invasive ICP measurement

Currently all techniques of ICP measurement require the removal of part of the skull and the accuracy of most measurements decline over time. Therefore, clinical ICP measurement is reserved for patients where ICP is likely to become critically elevated e.g. large, malignant

7 stroke (complete infarction of the MCA territory, accompanied by space occupying mass effect (33, 69)). To circumvent the complications that can arise from clinical ICP measurement, many attempts have been made to develop accurate and reliable non-invasive methods for estimating ICP. Methods include measuring the diameter of the optic nerve sheath, equilibrating intraocular pressure with ICP and imaging to estimate intracranial volume (70, 71). Despite promising results, to date none of the methods have been proven to be sufficiently accurate for clinical use. The most extensively studied method involves measuring intracranial blood flow velocity using transcranial Doppler (72). A method developed by Schmidt et al. (1999) has demonstrated significant correlation with invasive pressure measurement, when changes in ICP over time were assessed in seven patients with traumatic brain injury (73). In a larger study of 137 traumatic brain injury patients, the same algorithm accurately depicted long-term trends in ICP compared with invasive measurement (74). This indicates that non-invasive ICP may provide a suitable estimation of ICP in patients at risk of ICP elevation but who do not qualify for probe implantation.

1.2.3 ICP elevation after stroke

In humans, ICP elevation is known to occur after large, malignant infarcts becoming elevated between 2 and 5 days post-stroke (29, 75, 76). ICP is currently not monitored in patients with mild-moderate stroke. This is in part due to the invasive nature of monitoring and because of the assumption that this patient population do not experience raised ICP. However, how do we know that ICP elevation does not increase after mild-moderate stroke if we have not looked at it?

ICP elevation has been shown to occur in numerous animal models of stroke (38, 67, 77-79). In 2012, our group recognised a link between minor stroke and dramatic, yet transient (over hours) ICP elevation at 24 hours post-stroke in rats (80). Previous studies by Kotwica, and Silasi showed a similar ICP elevation temporal profile in rats however they failed to identify the importance of stroke volume in their findings (67, 77). Follow up studies from our group confirmed these findings in three different strains of rats (80-83), using young and aged animals (82) and, using transient (80, 82, 83) and permanent MCAo (81) and, photothrombotic stroke (81). The ICP elevation observed in these studies (20–30 mmHg) is similar to that observed following large, hemispheric strokes.

Evidence of a similar rise in higher order animals is scant and conflicting. Bell et al. (1991) reported post-stroke ICP elevation at 18 hours in cats with small stroke (7% of total brain volume) (78), but a study by Wells et al. (2015) did not find significant ICP elevation at 24 hours in an ovine model of mild-stroke (79). Reasons for the conflicting results are unclear

8 however, the animal studies raise an important question – could a similar rise in ICP occur in patients with mild-moderate stroke? To answer this question our lab used non-invasive ICP monitoring in patients with mild-moderate stroke (84). ICP was assessed at 6 hours post- stroke and again at 24 hours post-stroke revealing a significant increase from baseline at 24 hours when compared to healthy control patients (Figure 1.2) (84). Although the rise in ICP was modest compared with our animal studies it must be noted that ‘baseline’ ICP was measured at 6 hours post-stroke. We have preliminary animal evidence showing that ICP has already begun to rise at this time point in animals following experimental stroke (Unpublished observations, S. Azarpeykan, 2018). Therefore, if the same is true for humans our change in ICP from baseline values may be underestimated. Furthermore, it may be possible that even a rise such as this would be enough to reduce flow to the tissue immediately surrounding the penumbra (suffering reduced blood flow but not enough to affect function of the cells – oligaemic tissue) and expand the area of the penumbra however this requires further investigation. Importantly, this is the first evidence of ICP elevation in this patient population and challenges our current understanding of ICP elevation post-stroke. This begs the question – are the underlying mechanisms that cause ICP elevation after mild-moderate stroke the same as after large, malignant infarcts?

Figure 1.2: The change in intracranial pressure between two timepoints (ΔICP) in stroke patients compared with healthy controls. Box and whisker plot showing the median (horizontal line), interquartile (box) and absolute (whisker) ranges of data from each cohort with outliers (solid circles). The difference in pressure measured non-invasively in healthy volunteers (blue; n = 75) and mild-moderate stroke patients (red; n = 10) from the first to second time point (0-18 hours control, 6-24 hours post-stroke). ** p=.00012. (84)

9 1.2.4 Mechanisms of ICP elevation post-stroke

ICP elevation can be attributed to an increase in the volume of one or more of the intracranial contents – brain tissue, cranial blood, and CSF, as suggested by the Monroe-Kellie doctrine. In short, the Monroe-Kellie doctrine states that the intracranial compartment is a closed system within the non-expandable, rigid skull made up of three non-compressible elements: tissue, blood and CSF (85-87). This system conserves mass (i.e. remains constant) due to equal inflow and outflow parameters. Under different physiological conditions, an increase in one of the parameters will cause a reduction in one or both of the remaining two. The significance of these observations is that the skull cannot easily accommodate any additional volume. An increase in any one of the components (tissue (oedema), blood or CSF) leading to an increase in volume, must be met by a decrease in one or both of the other two components. ICP is not only related to the volume within the intracranial cavity but also the ability of the cavity to accommodate this volume. The intracranial cavity only has a small reserve for accommodating additional volume, which, once exhausted, will result in a dramatic rise in ICP.

Tissue Swelling (Cerebral Oedema)

Cerebral oedema is defined as an abnormal accumulation of water in the intracellular or interstitial space. There are 3 categories of oedema: cytotoxic, ionic and vasogenic. Cytotoxic oedema occurs as a result of the failure of ion pumps during the ischaemic cascade. Within minutes to hours after stroke affected cells begin to swell as a result of sodium accumulation leading to intracellular water influx (88). Cytotoxic oedema tends to correspond with the severity and duration of ischaemia and is potentially reversible if energy is restored. However, excessive water accumulation has the potential to cause cellular rupture. It is unlikely that cytotoxic oedema influences ICP as it is merely a redistribution of water (and ions) from the extracellular to the intracellular space. Ionic oedema occurs secondary to cytotoxic oedema. As the water and ions are drawn from the extracellular space to the intracellular space an osmotic gradient is formed between the tissue and blood which draws more intravascular water and ions into the tissue causing tissue swelling (89). Importantly, for ionic oedema to occur there needs to be a source of fluid i.e. blood flow (88). Therefore, ionic oedema is more likely to occur in the penumbra than the infarct core. Vasogenic oedema occurs following the breakdown of tight junctions between endothelial cells that make up the blood-brain-barrier. Although the exact mechanisms leading to this are unclear (90). This allows the passage of plasma proteins and fluid into cerebral tissue leading to increased extracellular fluid volume. The resulting swelling has the potential to displace brain structures and even hemispheres, leading to the compression of neurons and cerebral blood vessels.

10 The current understanding of the mechanisms underlying post-stroke ICP elevation in humans comes from patients with malignant infarction. In this patient population ICP elevation is associated with cerebral oedema (27, 29, 75, 76). However, there are some clinical studies that provide evidence to challenge this understanding. A study by Frank (1996) reported that out of 19 patients with malignant stroke and oedema only 5 patients had ICP elevation (76). Furthermore, Schwab et al. (1996) and Poca et al (2010) showed that ICP elevation was either preceded by cerebral herniation (caused by significant swelling) (75) or not present at all (91). Despite these findings cerebral oedema is still widely considered to be the sole cause of ICP elevation post-stroke. Small strokes result in little oedema. Therefore, there is a general assumption that patients with mild-moderate stroke do not experience ICP elevation. However, as explained above, we have found clinical evidence that ICP elevation does in fact increase in patients with small stroke which challenges this assumption (84).

Further evidence to challenge the current understanding of post-stroke ICP elevation comes from preclinical models of acute neurological injury. When investigating the mechanism behind the dramatic, yet transient ICP elevation observed at 24 hours post-stroke in rats with mild-moderate infarcts we noticed that ICP elevation occurred independently of cerebral oedema (80, 82, 83). This was confirmed using histology, wet weight-dry weight and in vivo magnetic resonance imaging to measure oedema (83). Furthermore, in one cohort of Sprague-Dawley rats, treatment with short duration hypothermia was found to completely prevent ICP elevation despite the presence of large oedema volumes (83). Combined with the human data this suggests that cerebral oedema may not be the sole cause of ICP elevation post-stroke. Interestingly, a study by John and Colbourne (2016) supports this hypothesis in another animal model of neurological injury. When investigating ICP elevation after experimental intracerebral haemorrhage they found that hypothermia treatment reduced ICP elevation but not oedema volumes (92). It could be hypothesised that there may be similarities in the underlying pathophysiology of the ICP rise in the two conditions. Many biological processes are recruited in response to a range of different insults therefore it can be further speculated that this may also apply to other disorders associated with ICP elevation. If it is not cerebral oedema causing the transient ICP elevation at 24 hours post stroke, it must be asked – what is the mechanism behind ICP elevation after mild-moderate stroke?

Cerebral Blood Volume

Cerebral blood volume (CBV) is the amount of blood contained within the cranial cavity. In order to maintain stable ICP there needs to be an equilibrium between arterial inflow and venous outflow. The majority of this volume lies within the cerebral venous circulation (93). ICP is known to increase in conditions that involve a reduction in cerebral venous outflow

11 including venous thrombosis and stenosis (94). In some patients with idiopathic intracranial hypertension (chronically raised ICP with no apparent cause) slight venous stenosis causes ICP elevation, which causes further compression of the vessel, thereby further increasing ICP (95, 96). It is not unreasonable to hypothesise that CBV may play a role in the delayed ICP rise we see at 24 hours post-stroke. Changes in CBV and its role in ICP elevation post-stroke have not been investigated clinically. Preliminary evidence from our laboratory using perfusion computed tomography maps indicated a decrease in cerebral blood volume in both the ipsilateral and contralateral hemisphere 24 hours post-stroke in rats. As this is the time we have shown peak ICP to occur CBV unlikely to be the primary cause of ICP elevation (Figure 1.3).

a. b.

Figure 1.3. Total cerebral blood volume (CBV) post-stroke in rats a. Absolute intracranial blood volumes measured in the ipsilateral (ipsi) and contralateral (contra) hemispheres with perfusion computed tomography pre- and 24-hours post stroke b. Example slice showing a CBV map in the centre of the stroke territory. Total CBV was calculated in slices across the whole of the brain. Blood volume = mean volume of blood in whole forebrain (mean ± SD; n = 4). ** p<.1). McLeod et. al 2011, Unpublished

Cerebrospinal fluid volume

CSF is the fluid that bathes the brain and spinal cord. Alterations in CSF volume leading to ICP elevation are thought to play a role in a number of neuropathological conditions including congenital and obstructive hydrocephalus (97), aqueductal stenosis (98) and idiopathic intracranial hypertension (96). Computed tomography and magnetic resonance imaging techniques have been used to determine CSF volume, however investigations often focus on

12 cranial CSF and fail to include the spinal component (99, 100). Quantifying CSF volume is technically challenging, especially in small animals given small volumes distributed throughout the brain and spinal cord (~200 to 300 µL of cranial CSF in rats (99)). Combined with the fact that oedema is thought to be the sole cause of ICP elevation, it is unsurprising that there are no studies investigating CSF volume post-stroke.

CSF volume is directionally proportional to the production and outflow of CSF. Therefore, increased CSF production and/ or decreased outflow, or a blockage to flow will cause ICP to rise. Due to the difficulties of measuring CSF volume, CSF production (101-104) and outflow resistance (105-107) are often used as surrogate measures for CSF volume however the accuracy and reliability of these methods are debated (106, 108-111).

There is no consensus as to the exact mechanisms underlying CSF production, flow and drainage, however it is clear that alterations to normal physiology can contribute to elevated ICP. Our current understanding of CSF production suggests that the majority of CSF (80- 90%) is produced by the cells of the choroid plexus (CP) (97, 112-116). CP CSF is produced in 2 stages. Stage one involves filtration of plasma across the choroidal capillary endothelium, driven by the hydrostatic pressure gradient between choroidal capillary blood and interstitial fluid (97). Stage two involves the active secretion of water (via transporters such as the Na+/K+/2Cl− cotransporter (117) and aquaporin(AQP)1 (118)), ions and macromolecules into the CSF (119, 120). Additional CSF is generated by the bulk flow of interstitial fluid from the parenchyma (115, 121). CP production of CSF has been shown to be influenced by neural stimulation (122), ion transporters (123), altered AQP expression and activity (118), neuropeptides e.g. arginine vasopressin (AVP) (124) and fibroblast growth factor (125).

Expression of AQP1 has been shown to affect CP production of CSF (118, 126). AQPs are water channels that facilitate water transport between cells and fluid compartments. AQP1 is highly expressed in CP epithelial cells (127). The precise contribution of AQP1 to CSF production is yet to be established, however AQP1 knockout mice exhibit reduced CSF production by as much as 25%, consequently causing a two-fold reduction in ICP compared with wild-type control mice (118, 126). Using a global model of cerebral ischaemia, Akdemir et al. (2016) found expression of AQP1 significantly increased in the CP between 24 and 48 hours post-stroke in rats, despite initial damage to the CP at the time of stroke (128). Neither CSF production nor ICP was measured in this study. However, it can be hypothesised that an increase in expression may contribute to increased CSF production.

There is also mounting evidence to suggest that mechano- and osmoreceptors in the CP may also be involved in regulation of CSF production e.g. transient receptor potential vanilloid 4

13 (TRPV4). Preston et al (2018) identified that activating TRPV4 increased ion flux and conductance (markers of permeability) in a porcine CP cell line (129). Expression of TRPV4 was increased following ischaemia (130) whilst antagonism using a specific TRPV4 antagonist (HC-067047) has been shown to reduce brain water accumulation and blood-brain-barrier breakdown in the ipsilateral hemisphere (131).

There are relatively few conditions that are known to increase CSF production. Those that do include CP hyperplasia (132), tumours (choroid papilloma) (133, 134) and infections such as meningitis (which can also affect CSF outflow) (135). From its site of production, CSF flows from the ventricles into the subarachnoid space. It was previously assumed that this transport is driven by both CSF production, as well as arterial vessel pulsation in the CP (136). Recent findings suggest that flow is also influenced by both respiration rate, and to a lesser extent, heart rate (137). However, as demonstrated by Ridgway et al. (1987), pulsation will result in bidirectional flow (138). Studies in rodents have shown that from the subarachnoid space CSF enters the brain along periarterial channels permitting CSF entry into the parenchyma and then exiting through perivenous circulation (139). These perivascular (glymphatic) channels are important sites of CSF/ interstitial fluid exchange and waste clearance (139). The mechanisms underlying the glymphatic circulation are unclear, however this constant exchange makes CSF composition an excellent indicator of the neuronal environment.

More commonly, CSF volume pathologies are caused by blockages of CSF flow and/ or absorption. Obstructive hydrocephalus is a condition arising from a blockage in CSF flow within the ventricular system due to a tumour or aqueductal stenosis. Blockages in CSF drainage may also result in communicating hydrocephalus and ICP elevation as seen in some patients following subarachnoid haemorrhage (140). There has been much debate concerning the routes of CSF drainage. Despite a lack of evidence, it was commonly accepted that the drainage of CSF primarily occurs through the arachnoid villi and granulations of the superior sagittal sinus (97, 141, 142). More recent studies suggest that this may only occur at high ICPs (143). This suggests an alternate route of CSF drainage at normal or slightly raised ICPs. Studies by Johnston et al. have shown the cribriform plate and nasal submucosa to be a potential CSF drainage pathway in sheep, mice, rabbits, pigs, rats (144, 145) and a non-human primate (145). This has been supported by in vivo imaging studies by our group that show these pathways to be the primary drainage site of CSF in rats (99). Additionally, in post-mortem examinations Johnston has found similarities of these extracranial lymphatic pathways in humans (145). It must be noted that CSF pathways degenerate rapidly after death so the significance of these findings in humans is yet to be confirmed in vivo. Since the tracers are toxic to humans it is difficult to confirm the sites of CSF drainage in clinical studies. Further evidence to support the importance of lymphatic CSF outflow pathways has been 14 shown using tracers injected into the CSF entering the lymphatics of the head, neck and spinal region (146, 147). Additional outflow pathways have been shown around spinal nerve roots in sheep (148, 149) and through lymphatic vessels in mouse dural sinuses and along dural vessels (150, 151). However, there is much debate within this field and the contribution of different routes of CSF outflow is yet to be determined.

Since there is so much uncertainty regarding CSF outflow, very little information exists about how it is regulated. Increased outflow resistance has been induced experimentally by Nagra et al (2010) using a kaolin model of communicating hydrocephalus (105). The authors found decreased lymphatic CSF absorption and ventriculomegaly in hydrocephalic mice. However they were not able to identify the precise mechanism underlying the outflow changes. Experimental blockage of the cribriform plate has also been shown to substantially reduce CSF outflow leading to ICP elevation (152, 153). Morphological changes to the cribriform plate e.g. ossification in ageing have been found post-mortem in patients with Alzheimer’s Disease (154) (a condition shown to feature reduced CSF outflow (155, 156)). Other conditions that may involve changes to CSF outflow include meningitis which is suspected to cause long term changes in resistance due to scarring and in rare cases hydrocephalus can be caused by impaired CSF absorption due to increased venous pressure in the dural sinuses or at the jugular veins (157).

1.3 Investigating the trigger for ICP elevation after mild-moderate stroke

We suspect that an increase in CSF volume is a possible mechanism underlying the transient, oedema-independent ICP elevation observed at 24 hours post-stroke in rats. If this is indeed the case, it must be asked, what is the cause of this rise?

1.3.1 The choroid plexus

One potential cause of increased CSF volume after stroke is CP damage. Experimental studies have shown that stroke can result in an increase in CP permeability, CP epithelial swelling and apoptosis (158-160). Ennis and Keep (2006) showed a reduction in CP blood flow caused by permanent MCAo, and two vessel occlusion with hypotension (159). With as little as 10 minutes of occlusion authors observed a doubling in the permeability of the blood- CSF-barrier. By increasing the occlusion time to 30 minutes permeability was three times higher than in control rats. It is important to note that the authors only investigated permeability after two vessel occlusion with hypotension. Blood flow reduction to the CP was greater during permanent MCAo (62%) than in the two vessel occlusion model (13%) therefore it is not unreasonable to speculate that the former would cause more damage to the CP. Similar effects on CP permeability were reported by Nagahiro (1994) after MCA and internal carotid

15 occlusion using magnetic resonance imaging contrast enhancement in rats (158). The effects of increased permeability on CSF production and/or ICP were not investigated in either study.

The original studies from our lab showing ICP elevation at 24 hours post-stroke were performed in rats following temporary MCAo using the intraluminal thread occlusion model. As discussed above, MCAo has the potential to reduce blood flow to the CP (159) likely as a direct result of blocking the anterior choroidal artery (161). Whilst there is no direct evidence linking CP damage and ICP elevation, we hypothesised that damage to the CP could lead to altered CSF production and composition and influence ICP. However, when investigating this hypothesis in two separate studies we found that CP damage is not required for ICP elevation. First, using permanent MCAo with a short (1.5mm) or long (4mm) thread we found that only animals with the short thread, designed to keep the anterior choroidal artery patent had ICP elevation (17.3 ± 7.5 mmHg at 24 hours post-stroke) (81). Animals with the long thread, designed to block the anterior choroidal artery did not experience ICP elevation at 24 hours post-stroke. Moreover, we found no correlation between 24 hour ICP elevation and the degree of CP damage after permanent MCAo. It must be noted that CP damage was assessed at 3 days post stroke, not at the time of ICP elevation, which may have allowed time for CP recovery (162). We have also observed ICP elevation in a photothrombotic model of stroke which causes cortical infarcts with no CP involvement (23.1 ± 5.4 mmHg at peak) (81). Together, these results indicate that CP damage is not required for ICP elevation.

1.3.2 The ischaemic penumbra

During development of the photothrombotic stroke model mentioned above, it was observed that ICP only increased when the injury was submaximal. This led us to hypothesise that submaximal injury (ischaemic penumbra) is required for ICP elevation at 24 hours post-stroke. We investigated this hypothesis by randomising rats to photothrombosis with either low or standard light exposure over the same area (0.13 W/ cm2 for 2 minutes or 0.3 W/ cm2 for 20 minutes respectively). At 24 hours post-stroke ICP was significantly higher in animals with submaximal injury (23.1 ± 5.4 mmHg) compared with maximal infarcts (11.8 ±3.7 mmHg) (81). There is controversy surrounding the existence of penumbral tissue with the photothrombotic model, however low light intensity for a short time (as was done in our study) has been shown to induce a penumbra like ‘region at risk’ (163). Supporting evidence for this hypothesis comes from our permanent MCAo study where ICP elevation was only observed in animals with patent anterior choroidal arteries which supply both the CP and act as collateral vessels to the penumbra (81). Patency of the anterior choroidal artery results in significantly more penumbral tissue and a trend toward less infarct core, compared with an occluded anterior choroidal artery (81). Furthermore, this hypothesis is also supported by our clinical data where patients

16 (n = 10) were selected based on their high penumbra to core ratio (mean baseline penumbra and core volumes were 83.73 ± 46.31 mL and 18.29 ± 17.27 mL respectively) (84). However, this will need to be confirmed by measuring ICP in patients with a wide range of penumbral volumes.

The 24 hour time delay for ICP elevation suggests the possibility of some form of active cellular processes. Despite electrical silence, protein synthesis in the penumbra is preserved (albeit reduced) (164, 165). Select proteins including heat shock proteins and hypoxia induced factor are upregulated in this tissue despite a general reduction in protein synthesis (166). Expression of heat shock protein-70 (167) for example, increases early after ischaemia and aids cell survival by refolding denatured proteins (168) and inhibiting apoptosis (169). Given that ICP elevation was observed in the presence of larger volumes of penumbral tissue, it is possible the penumbra is releasing factors that contribute to the 24 hour ICP rise.

1.3.3 Neurohumoral factors in post-stroke CSF

To test this hypothesis we designed a set of experiments to determine whether there was a molecular trigger for ICP elevation in post-stroke CSF. Transfusion of CSF from rats with mild- moderate stroke (collected 6 hours post-3 hour MCAo) into the lateral ventricles of naïve (disease free) recipient animals triggered ICP elevation in recipient rats. This rise showed a similar time delay to that seen post-stroke in rats, and importantly no rise was seen in recipients of control (non-stroke) CSF (Figure 1.4). Therefore we concluded that a molecule (or molecules) in post-stroke CSF cause ICP to rise.

To determine the clinical relevance of our findings we performed the same biological ICP assay whereby CSF from a patient with mild-moderate stroke (collected 5 hours post-stroke) was infused into the lateral ventricles of naïve recipient rats. Excitingly, we observed a similar ICP response in animals that had received infusion of human stroke-CSF. In contrast, this response was not observed in animals that received CSF from patients with chronic hydrocephalus (n = 2/ group; Figure 1.5). Our findings suggest the existence of a completely novel mechanism of ICP elevation after stroke. However, further experimentation is required to confirm these preliminary results.

17

30 * * * * SE ** diff 20 ** * Stroke Control 10 ICP (mmHg)

0 0 1 16 18 20 22 24 Hours

Figure 1.4. Intracranial pressure (ICP) following cerebrospinal fluid (CSF) transfusion. Data (mean ± SE) for ICP changes over time in male, outbred Wistar rats following CSF transfusion (n = 6/ group). Stroke-CSF was collected from male, outbred Wistar rats 6 hours post-stroke (3 hour MCAo and 3 hour reperfusion) whilst control CSF was collected after 6 hours of anaesthesia. (* = p<.05; ** = p<.001; *** = p<.0001). Unpublished data, R. Hood, D.D. McLeod (2012)

CSF infusion Stroke CSF 1 Stroke CSF 2 40 Control CSF 1 Control CSF 2 30

20

ICP (mmHg) 10

0 0 16 18 20 22 24 Time (Hours)

Figure 1.5. Intracranial pressure (ICP) in rats following human cerebrospinal fluid (CSF) transfusions. Data for ICP changes over time in male, outbred Wistar rats following CSF transfusion (n = 2/ group). Stroke CSF samples were collected from patients at 5 hours post- stroke, whilst control samples were collected from patients with chronic hydrocephalus. Unpublished data, C. Logan, D.D. McLeod, D. Beard, R. Li (2013).

18 1.4 CSF composition

CSF composition is complex and dynamic, serving as an essential medium for transporting molecules from one area to another. The major component of CSF is water (99%) and the remaining 1% is made up of glucose (approx. 600 ng/ µL), various ions (Na+, K+, Ca2+, Mg2+ and Cl-) and proteins (approx. 350 ng/ µL) (97, 170). Although representing <1% of the total content, CSF contains a vast array of proteins (171-173). However, the functional significance of the vast majority of proteins has yet to be identified in the CNS, particularly in the context of stroke.

CSF composition can be used as an important tool for monitoring changes in brain metabolism, barrier function and permeability and aid in the diagnosis of neurological disorders including meningitis (174) and multiple sclerosis (175). This is because the composition of CSF provides clinicians and researchers a molecular snapshot of the CNS environment. Part of the appeal of investigating CSF is that it bathes the brain and so provides an idea of the molecular changes occurring in the brain without being overly invasive.

There are several factors that can contribute to changes in CSF composition, including choroid plexus secretions to maintain homeostasis and assist in repair, as well as disease or damage to the blood-brain-barrier (97, 162, 176-179). CSF composition is also heavily influenced by the neuronal environment due to its role in protein and metabolite clearance from the extracellular fluid of the parenchyma. Many neurological diseases result in changes in CSF composition including traumatic brain injury (180), stroke (181) and hydrocephalus (182). Perhaps the most drastic change in composition occurs in SAH in which blood is directly released into CSF (183).

The delayed nature of the ICP elevation seen in our studies is suggestive of active cellular processes and protein synthesis. Whether or not a change in the protein composition has an effect on the largest component of CSF (water), has not yet been characterised in neurological disease. It could be hypothesised that a change in the CSF proteome may have effects that result in increased intracranial water volume and thus, increase ICP.

1.4.1 Stroke and the CSF proteome

Whilst we understand many of the physiological processes that occur during stroke, there is little understanding of how stroke influences CSF composition. The major reason for this is that stroke is diagnosed by advanced imaging, therefore there is no need to investigate CSF composition for diagnosis. Furthermore, CSF collection is invasive, and is a potential contraindication for intravenous thrombolytic therapy and anticoagulants (184).

19 As discussed, we have already identified the ischaemic penumbra as a potential source of proteins that may be released into the CSF to cause delayed ICP elevation (81). Furthermore, we have shown that ICP elevation is not associated with CP damage. If presence of an ICP elevating molecule is not related to CP damage, is a functioning CP required for production of the triggering molecule? The CP is a central source of fluid regulating peptides in CSF and has been shown to alter CSF composition in response to disease. During forebrain ischaemia the CP secretes growth factors into CSF to aid in repair (179). It is not unreasonable to speculate that the CP might secrete fluid regulating peptides in response to stroke which may influence water balance in the brain leading to ICP elevation.

Alternatively, it is also possible that the CSF proteome may be influenced by foreign factors being released into the CSF via a breakdown of the blood-brain-barrier. There is debate over the exact time course of blood-brain-barrier breakdown in the first few hours of stroke however it is clear that the barrier opens within the first 6 hours of stroke (185-188). The timing of this opening fits within the 6 hour time window of CSF collection in the donor rats in our previous study. Upon breakdown of the blood-brain-barrier, blood constituents including plasma proteins, immune cells or inflammatory mediators may enter the parenchyma. Whether these factors themselves influence ICP elevation, or whether these factors elicit further CSF compositional change resulting in ICP elevation is cause for speculation.

A search of the literature returned only one study that has investigated global protein expression changes in CSF post-stroke. Simats et al. (2018) found that at 2 hours post-MCAo there were 716 differentially expressed proteins within rat CSF compared with sham animals (181). Perhaps unsurprisingly, they found that proteins related to the inflammatory response were the most over represented in CSF at 2 hours post-stroke. As the study was investigating potential stroke biomarkers, any potential links between the proteins identified and ICP elevation were not investigated. With the exception of the above study that investigated total protein expression changes, the remainder of studies that investigate post-stroke CSF compositional changes do so by investigating expression of specific factors (189-195). These investigations have revealed only a handful of factors whose expression within CSF has been shown to correlate with (192-195) or directly cause ICP elevation via infusion into the CSF (193, 196-198). Interestingly, of these factors, bradykinin, angiotensin II and arginine vasopressin have also been shown to be upregulated after ischaemic stroke.

Bradykinin

CSF bradykinin has been shown to significantly correlate with ICP in a number of acute neurological diseases (193). Kunz et al. (2013) showed significantly elevated levels of CSF bradykinin, peaking between 24-48 hours, in patients with ischaemic stroke, traumatic brain 20 injury, and both intracerebral and SAH (193). Unfortunately, the authors did not measure the time course of ICP changes in any of the disease states. Interestingly the authors also showed a significant correlation between bradykinin levels and cerebral oedema suggesting that this was the underlying mechanism behind the rise. Importantly CSF in this study was collected from patients with malignant infarction therefore the mechanism might differ from what occurs after mild-moderate stroke (193). In experimental stroke, upregulation of bradykinin B2 receptors was seen in affected tissue observed within 2 hours of stroke induction with upregulation persisting in the ischaemic penumbra out to 24 hours (199).

Angiotensin II

Expression of angiotensin II has been shown to be significantly upregulated in CSF within 2 hours of experimental stroke (181). Intraventricular infusion of angiotensin II has been shown to increase ICP in rats (196). However, somewhat paradoxically central administration of angiotensin II was shown to reduce blood flow to the CP, which would theoretically decrease CSF production. Such a finding has prevented a clear understanding of the exact mechanism behind the elevation (200). As receptors for angiotensin II have been shown on rabbit CP epithelial cells (200) therefore, it is not unreasonable to speculate that CSF volume may be contributing to the observed elevation in pressure seen in recipient animals (196).

Arginine vasopressin

Perhaps the most promising candidate for a triggering molecule for ICP elevation is AVP. AVP is a nonapeptide that is involved in water homeostasis. In the kidneys AVP is known to upregulate AQP2 expression to increase reabsorption of water back into the blood (201, 202). In the brain, it functions as a neurotransmitter and neuromodulator and has been shown to increase cerebral water permeability (203, 204), increase brain water volume and oedema formation post-stroke (204-208), and influence CSF dynamics (124) and ICP (197, 198, 209, 210). AVP is present in the CSF of both humans (195, 211-213) and animals (211, 214, 215).

AVP concentration in healthy patients has been investigated in numerous studies with values ranging from <1 to 11.5 pg/mL (211). Expression can be influenced by the CP as well as by diffusion of catabolites from the brain. Increased CSF AVP concentrations have been found in numerous neurological disorders involving increased ICP including stroke (194), SAH (216), intracerebral haemorrhage (194, 195), traumatic brain injury (194), idiopathic intracranial hypertension (194, 195) and hydrocephalus (194, 195). Furthermore, AVP levels have been found to correlate with ICP in intracerebral haemorrhage, idiopathic intracranial hypertension and hydrocephalus (195). However, it remains unknown whether AVP is a cause of elevated ICP or is upregulated in response to raised ICP.

21 Intracerebral infusion performed on rabbits (198, 210), rats (196, 209) and goats (197) have produced conflicting results of AVPs effect on ICP. Differences in dose, time of ICP measurement and AVP delivery make the results difficult to interpret. However, mode of administration into the CSF is thought to influence the ICP response. Continuous infusion has typically been shown to increase ICP (197, 198) whilst bolus injection causes the inverse response (209, 210). Interestingly however, both have also been found in some studies to exert no effect over ICP (196, 198). It has been hypothesised that these differences are a result of vasoconstriction caused by local effects of the bolus infusion of AVP on surrounding blood vessels. However, the underlying mechanism behind why continuous infusion of AVP causes ICP elevation is yet to be elucidated. Importantly, these studies all measure short term changes in ICP (either during or immediately post-AVP infusion). By contrast, delayed effects of AVP administration on ICP elevation, such as the delayed rise in ICP seen after minor stroke in our studies, have yet to be investigated. Could intracerebral infusion of AVP exert delayed effects on ICP?

Indirect evidence from AVP receptor blockade studies (using V1a receptor antagonists administered intraventricularly at the time of stroke) have shown attenuated oedema volumes, increased AQP4 expression and a reduction in blood-brain-barrier permeability out to 24 hours post-infusion (204, 217). Therefore, it is plausible that AVP infusion may exert delayed effects over ICP.

22 1.5 Aims and rationale

Until recently ICP elevation was assumed only to occur in patients with malignant infarction as a result of cerebral oedema. Our lab has found evidence of ICP elevation at 24 hours after mild-moderate stroke in animals and humans. When investigating the underlying mechanism behind this previously unrecognised ICP elevation in rats, we discovered that it occurs in the presence of minimal oedema. Upon further investigation, it was determined that the 24 hour ICP rise can be triggered in stroke-free rats, by administration of CSF collected from donor rats within 6 hours of stroke induction. This indicated that there is a molecule(s) released into CSF during stroke that causes ICP elevation in rats. The central hypothesis on which this PhD was based, is that factors within CSF contribute to the ICP rise at 24 hours after mild- moderate stroke in humans. Identification of this molecule(s) may have important implications for the diagnosis of patients with mild-moderate stroke who are at risk of developing raised ICP. As these patients do not currently undergo ICP monitoring, their risk of ICP elevation is largely unrecognised. Therefore, successful identification of a triggering molecule(s) may create an entirely new avenue for therapy in stroke and potentially other neurological disorders for which elevated ICP contributes to disability and/ or death.

The aims of this PhD were to:

• Confirm the accuracy and reliability of epidural pressure measurement in rats using a fibreoptic pressure sensor. • Determine whether intraventricular infusion of AVP causes delayed ICP elevation in naïve rats. • Expand on our pilot data to determine whether infusion of human stroke-CSF causes ICP elevation in naïve rats. • Identify potential protein candidates in human stroke-CSF that may be responsible for delayed ICP elevation.

23 Chapter 2: Methods

2.1 Introduction

Our research into ICP elevation post-stroke has raised many important questions regarding the mechanisms underlying the oedema independent rise that occurs at 24 hours post-stroke in rats. Human studies were considered unsuitable to answer these questions. Therefore an animal model was deemed to be the most appropriate model for investigation of CSF factors. In order to measure epidural pressure in the rat brain, extensive troubleshooting was undertaken in order to identify a pressure sensor suitable for use with our published method and is presented in Chapter 3. Also presented there, are improvements that were made to the method during the troubleshooting process and the validation of the probe and improved methods. The final method, as well as those methods not requiring modification for use in this project are presented below.

2.2 Substances investigated for effect on ICP

Substances that were administered intraventricularly in these studies are listed below:

• Artificial cerebrospinal fluid (aCSF) (Harvard Apparatus, USA). • [Arg8]-Vasopressin solution (AVP; V0377; Sigma-Aldrich, USA), 0.02 ng, 0.1 ng and 5 ng in 200 µL aCSF injection volume. • Human CSF (200 µL injection volume) collected via lumbar puncture from patients diagnosed with: o Ischaemic stroke o Chronic hydrocephalus o SAH o Normal CSF (collected for diagnostic testing but subsequently found negative)

24 2.3 Human CSF

2.3.1 Patient Selection

This study was approved by the University of Newcastle Human Research Ethics Committee (reference numbers H-2014-0040 (Chapter 5: Parts 1 and 2), H-2014-0376 (Chapter 5: Part 1)), the Harbin University Ethics Committee (reference number 2011DFA31470 (Chapter 5: Parts 1 and 2)) and the Hunter New England Human Research Ethics Committee (reference number 14/02/19/4.07 (Chapter 5: Part 1)). Informed consent was obtained from either the patient or person responsible if the patient was unable to consent prior to inclusion in this study, with the option to opt out at any time.

2.3.2 Ischaemic stroke and hydrocephalus CSF

Clinical CSF samples were collected from 35 patients being treated between February and December, 2012 at the Second University Hospital, Harbin University, China with the diagnosis of either ischaemic stroke or chronic hydrocephalus. Ischaemic stroke patients included in this study had computed tomography scans/ magnetic resonance imaging and were clinically assessed for stroke severity prior to CSF collection. Inclusion criteria for stroke patients included the following: 18-years-old or above; not pregnant; clinical evidence of acute ischaemic stroke; small vessel occlusion as identified on acute imaging; baseline National Institute of Health Stroke Severity (NIHSS) score of less than 20. Hydrocephalus patients included in this study were undergoing routine CSF drainage. Patients had computed tomography scans/ magnetic resonance imaging and were clinically assessed for symptoms prior to CSF collection. Inclusion criteria included: 18-years-old or above; not pregnant; clinical evidence of chronic hydrocephalus.

2.3.2.1 Sample Collection

CSF samples were collected via lumbar puncture by a sterile sampling technique. The first millilitre of CSF was discarded to remove debris/blood contaminants, before collection of clear CSF into a sterile tube. The samples were frozen at -80 °C within 30 minutes of collection until lyophilisation and transfer to Australia.

2.3.2.1 Sample Rehydration

Samples were rehydrated to their original volume using sterile water. To eliminate cells and other insoluble material the CSF samples were centrifuged at 14 G at 4 °C for 5 minutes. The samples were then stored at -80 °C in 500 µL aliquots in 2 mL glass vials (Waters).

25 2.3.3 SAH, IIH and Control CSF

Clinical CSF samples were collected from 6 patients being treated at John Hunter Hospital, New Lambton Heights, Australia. Selected patients were diagnosed with SAH (n = 4) or suspected of having seizure disorder (n = 1) or normal pressure hydrocephalus (n = 1) and later found negative (control samples). Inclusion criteria included: aged 18-years-old or above; patients that were already having CSF being collected as part of ongoing, therapeutic care; not pregnant.

CSF from SAH and control patients was collected via lumbar puncture using the same technique as described for stroke and chronic hydrocephalus CSF. Samples were stored at - 80 ˚C until use.

2.4 Animals

Over the course of my PhD the lab experienced significant issues with technology and reliability of results. During extensive troubleshooting we investigated obtaining animals from different locations (Perth, Melbourne and Newcastle) as well as different strains of rats (outbred Wistar, inbred Wistar, Wistar Kyoto and Sprague Dawley). The final studies were performed using male, outbred Wistar and Sprague Dawley rats (body weight: 255-520 g; Monash Animal Research Platform, Vic; Animal Services Unit, University of Newcastle, NSW, Australia). Animals were fed standard laboratory rat chow ad libitum and housed in an area of constant temperature (22 ± 1 °C) and humidity on a 12 hour light-dark cycle. Experimental protocols were approved by the Animal Care and Ethics Committee of the University of Newcastle, Australia (reference number A-2013-343) and conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes.

2.4.1 Instrumentation and techniques

Preparation of rats for experimentation involved implantation of:

I. Femoral arterial line for MAP and heartrate

II. 2-4 screws for ICP measurement and intraventricular infusion

III. Fibreoptic pressure sensor(s) for ICP measurement in the epidural, subdural and intraventricular space

IV. Intraventricular catheter (IVC) for intraventricular infusion

The preparation is described below.

26 2.5 Surgical Procedures

2.5.1 Anaesthesia and intra-operative monitoring

Rats were anaesthetised with 5% isoflurane (Bomac, Australia) and maintained throughout the procedure using isoflurane (1.5-3%) in a mixture of nitrogen (N2) and oxygen (O2) (30-50% vol O2) through a nose cone attached to a stereotaxic frame at a total flow rate of 2 L/ minute.

Both the anaesthetic and N2:O2 levels were adjusted as required to maintain respiratory rate and blood oxygen saturation (SpO2) monitored using an oximeter pod connected to a foot probe (ADInstruments, Australia). Temperature was maintained at 37 °C throughout the procedures using a rectal probe thermometer connected to a homeothermic heat mat (Faculty of Health Workshop, University of Newcastle, Australia (Chapters 3 and 4); DC Temperature Control System (FHC,USA) (Chapters 5 and 6)). All monitors were connected to a PowerLab 8/30 data acquisition system (ADInstruments, Australia). Physiological parameters were recorded using LabChart Pro software, Version 7 (ADInstruments, Australia).

Rats were monitored from anaesthesia induction to approximately 80 minutes post infusion (day 1) (or until ICP had reached levels similar to pre-infusion) and from anaesthesia induction (approximately 16 hours post infusion) to 24 hours post-infusion (day 2).

2.5.2 Surgical Preparation

Surgical sites were prepared by shaving the skin and sterilising with chlorhexidine in 70% EtOH. A Tegaderm dressing (3M, USA) and sterile drape were used to prevent any unwanted debris from entering the incision sites. Prior to any surgical procedures, long-acting, local anaesthetic Bupivacaine (2 mg/kg 0.05%; Pfizer, Australia) was administered subcutaneously to incision sites. Atropine (0.5 mg/kg; Apex Laboratories, Australia) was also administered subcutaneously to reduce airway mucosal secretions during the lengthy experimental procedures.

2.5.3 Mean Arterial Pressure Measurement

Femoral artery catheter construction

All catheters were constructed using common laboratory materials. A 10 cm length of Silastic tubing (inner diameter 0.3 mm, outer diameter 0.64 mm; SAI Infusion Technologies, USA) was connected to a 3 cm length of polyurethane tubing (inner diameter 0.2 mm, outer diameter 0.4 mm; SAI). The polyurethane tubing was inserted 0.5 mm into the Silastic tubing on one end, and a 26 G blunt-end needle (SAI) was inserted into the other end. Before use the catheter was sterilised and tested for leaks by flushing with 70% ethanol, sterile saline and then air

27 (from a sterile syringe). Before insertion into the femoral artery, the femoral line catheter was filled with heparinised saline (10 IU/ mL; Pfizer, Australia) to prevent any air from entering the catheter.

Catheter Insertion

A 3 cm incision was made along the inner thigh above the femoral vessels on the right leg (left leg for day 2), with care taken to avoid damage to the vessels. The femoral vessels (saphenous nerve, femoral artery and femoral vein) were then located and exposed and the femoral artery was gently blunt dissected away from both the nerve and vein. Three pieces of 5-0 silk suture (braided silk; Ethicon, USA) were placed underneath the artery. The distal piece of suture was tied, the middle and proximal pieces were loosely tied and Lignocaine (20 mg/mL; Troy Laboratories, Australia) was applied topically (approx. 0.1 - 0.2 mL) to dilate the vessel. A vascular clamp was placed in between the two loose ties to temporarily occlude the vessel and the artery was pierced adjacent to the distal occlusion suture. The catheter tip was inserted into the vessel until it reached the clamp and the proximal and middle ties were then secured around the vessel and indwelling catheter. The distal tie was also secured around the catheter to provide further stability. The skin was then closed with simple discontinuous silk sutures and the catheter was exteriorised between the suture ties. The animal was then flipped to prone in preparation for the ICP surgery. To measure MAP the femoral catheter was connected to a fluid filled pressure transducer (CODAN, Germany) and pressure was recorded electronically on the previously mentioned LabChart software.

Catheter Removal

Following completion of monitoring, and prior to recovering the rat from anaesthesia (day 1 only), the incision site was re-opened for withdrawal of the femoral catheter. The ties holding the catheter in position were loosened and the catheter was slowly and carefully retracted until the tip of the catheter was seen through the arterial wall. The proximal tie was then tightened and the catheter fully removed and the middle tie also tightened to ensure the vessel was completely ligated. The skin was closed with simple discontinuous sutures using 5-0 silk thread.

2.5.4 Intracranial Pressure Measurement

The method chosen for ICP measurement was developed by Murtha et al. (2012) and modified for this investigation (39) (see Chapter 3 for details). The rat’s skull was positioned in a stereotaxic frame (Faculty of Health Workshop, University of Newcastle, Australia) to restrict any movement and the skin was prepared for incision (as described in section 2.5.2). A 2 cm

28 incision was made above the frontal bone and blunt dissection was used to clear away the soft tissue and the periosteum. The skin and muscle were then retracted and landmarks Lambda and Bregma were located. Depending on the experiment, one to three holes (2 mm in diameter) were burred into the parietal bones to expose the dura mater without piercing it (Chapter 3: Parts 1 and 2 and Chapters 4 and 5, n = 1 hole (2 mm lateral and 2 mm caudal to Bregma); Chapter 3: Parts 3-5, n = 3 holes (1.3 mm lateral and 0.3 mm caudal to Bregma over the right lateral ventricle; and 2 x 2 mm lateral and 2 mm caudal to Bregma in the left and right parietal bones). The coordinates of the screws were chosen so that the screws were not located over any major blood vessels. Sterile saline solution (0.9%; Pfizer, Australia) was gently flushed into the burr holes to clear away debris, before 2mm diameter x 5 mm long hollow poly-ether ether ketone (PEEK) screws (internal diameter 0.8 mm; Solid Spot LLC, USA) were carefully screwed into the skull. Prior to insertion, screws were checked to make sure the shafts hadn’t been blocked by debris. Screws were secured to the skull with ethyl 2- cyanoacrylate (Super Glue Ultra Fast liquid, UHU, Australia) and dental cement (Vertex- Dental B.V., The Netherlands). Once the dental cement had dried, a fibre optic pressure transducer (OPP-M250-X-80SC-1.5PTFE-XN-100PIT3-P1-SILICONE SEAL; Opsens, Canada) was inserted into the ICP screw so that the tip was positioned in the screw just above the dura (Figure 2.1). Additional Opsens pressure sensors were placed in the subdural as well as intraventricular spaces (Chapter 3 only). For subdural pressure measurement a sterile 27 G Sprotte spinal needle was inserted approximately 0.5 mm past the bottom of the screw – far enough to pierce the dura and create a continuum between the inside of the screw and the subdural space but not far enough to cause substantial damage to the brain. Similarly, for the intraventricular pressure measurement a sterile 27 G Sprotte needle was inserted approximately 4 mm through the parenchyma to pierce the right ventricle. Upon access to the right lateral ventricle the Sprotte was removed and a pressure sensor was then passed through the needle tract and into the ventricle for ICP monitoring. Two layers of caulking material (Silagum, DMG Dental, Germany) were then applied to the opening of the screw and around the ICP probe to create an airtight seal and secure the probe in place. Once the Silagum had set, the probe location was confirmed by ensuring a good signal, and by making sure the trace responded to abdominal compressions (Figure 2.2). Abdominal compression reduces venous return of blood to the heart and causes a transient increase in ICP for the duration of the compression. Pressure was transmitted to the PowerLab data acquisition system and digitised for recording using LabChart software. Once recordings were completed, the ICP probe was simply pulled back through the Silagum and placed in a 0.1% Terg-A-Zyme (Sigma Aldrich, Australia) and distilled water solution. Silagum was then reapplied over the opening created by removal of the probe, to create an airtight seal. On day 2 the Silagum cap

29 covering the ICP screw was removed and the ICP probe was re-inserted using the method described above.

Figure 2.1. Intracranial pressure (ICP) probe location schematic. This diagram shows the placement of the ICP probe in the epidural (ED), subdural (SD) and intraventricular (IVP) spaces. It also shows the intraventricular catheter (IVC) used in cerebrospinal fluid infusion positioned in the left lateral ventricle. Caulking material (Silagum, Germany) was applied to the top of each screw to create an air tight seal. Figure modified from Murtha et al. (2012).

30

Figure 2.2. Intracranial pressure (ICP) and blood pressure traces responding to abdominal compression. ICP measured simultaneously in the epidural, subdural and intraventricular spaces. The abdomen was temporarily compressed (1-2 sec) to reduce cerebral venous return. This increases the intracranial volume and causes ICP elevation (*). Also shown is arterial blood pressure which drops after the initial ICP rise. Respiratory rate can be calculated based on each major depression in the signal (#) and heart rate can be assessed based in the minor depression (+).

31 2.5.5 Intraventricular Infusion

Intraventricular catheter construction

The rounded tip from a 27 G Sprotte spinal needle was broken off from the needle shaft (10 mm) and inserted 1 mm into one end of a 16 cm length of PE-10 tubing (internal diameter 0.381 mm, external diameter 0.061 mm; SAI) so that the round tip was facing away from the tubing. A 30.5 G blunt end needle (BD Microlance, USA) was inserted 0.5 cm into the other end of the PE-10 tubing for syringe attachment. Before use the catheter was sterilised and tested for leaks by flushing with 70% ethanol, sterile saline and then air (from a sterile syringe). Before insertion, the IVC was pre-loaded with the infusate to prevent the infusion of air bubbles.

Intraventricular infusion

In all studies, an additional screw was inserted superior to the left lateral ventricle during ICP surgery. Positioning of the screw was based on stereotaxic coordinates (1.3 mm lateral and 0.3 mm caudal to Bregma). This screw location was initially validated by infusing Evans Blue dye (960 kDa) and aCSF (Harvard Apparatus, USA) into the lateral ventricles of rats (n = 24) and observing dye distribution within the ventricular system after euthanasia.

The IVC was inserted approximately 4 mm through the parenchyma with the tip placed in the left lateral ventricle. The needle was then sealed in place using Silagum. ICP was continuously monitored during insertion to ensure that pressure did not rise significantly as a result of needle insertion. A syringe driver (Pump 11 Elite, Harvard Apparatus, USA) was used for the intraventricular infusions. Infusion rates varied between studies (see individual chapters for information). The catheter was left in place until day 1 monitoring was completed (approximately 80 minutes post-infusion or when ICP returned to baseline levels) at which point it was slowly removed and the screws were re-sealed with Silagum. For the probe location studies the IVC was left in place and the aCSF infusion changed to an infusion containing aCSF + 3% Evans Blue to confirm catheter location (terminal experiments).

2.5.6 Post-operative monitoring and care

After anaesthesia had been terminated (on day 1), the animals were placed in a cage and monitored continuously for 30 minutes and then half hourly until 2 hours. Due to the animal’s limited thermoregulatory capacity after anaesthesia half the cage was placed on a heat mat.

32 2.5.7 Confirmation of injection location

Following cessation of surgery, the isoflurane was increased to 5% and the animals were transported to a corkboard and placed supine with their limbs taped down. To ensure maintenance of deep anaesthesia a falcon tube containing an isoflurane soaked gauze pad was placed over the animal’s nose and taped to the corkboard. Transcardial perfusion was performed by infusion of 50 mL of saline (0.9%) followed by 400 mL of ice cold paraformaldehyde (4%) (Sigma Aldrich, Australia) into the left ventricle. Following perfusion the animals were decapitated and their heads were placed in 10% neutral buffered formalin at 4 °C for 2-3 days to allow for in-skull fixation. The brains were then removed and checked for haemorrhages as well as to confirm location of infusion.

2.6 Proteomic Analysis of Human CSF Samples

2.6.1 Protein preparation for mass spectrometry

Proteins were quantified (Pierce BCATM Protein Assay Kit, ThermoFischer Scientific, Germany) and 100 µg of protein/ sample was diluted to 360 µL with Milli Q water. Disulfide bonds were reduced with 40 µL of 100 mM 1,4-dithiothreitol (Sigma-Aldrich, USA) for 30 minutes at room temperature. To alkylate the cystines, 40 µL of 200 mM iodoacetamide (Sigma-Aldrich, USA) was then added and the samples incubated in the dark for 30 minutes at room temperature. Proteins were precipitated for trypsin digestion by methanol-chloroform precipitation. The sample were combined with 440 µL MeOH and 200 µL of CHCl3 (2:2:1 Sample:Methanol:Chloroform) and vortexed until an emulsion was formed. Samples were then spun for 2 minutes at 17000 x G in a centrifuge (Heraeus Pico 17 Microcentrifuge, ThermoFisher Scientific, Germany) followed by removal of the top fraction ~2 mm above the protein interface. Another 300 µL MeOH was added and the sample gently mixed before centrifuging at 14000 x G for 20 minutes. The supernatant was discarded and the protein pellet air-dried. Proteins were digested overnight at 37 °C and 200 rpm (Eppendorf Thermomixer Compact, Eppendorf, Germany) in 50 µL trypsin solution consisting of 2 µg (1:50 ratio trypsin:protein) trypsin (Promega, USA) diluted in 2% Acetonitrile (ACN), 40 mM Ammonium Bicarbonate. Samples were then spun at 14000 x G for 25 minutes and 40 µL of supernatant transferred into glass vials (Waters, Australia). Samples were acidified (pH 2-3) using 4 µL of 10% Trifluroacetic acid.

33 2.6.2 High Pressure Liquid Chromatography

For all experiments, chromatography was performed using an Ultimate 3000 Ultra High Pressure Liquid Chromatography system (Dionex, Australia). In each run, digested CSF peptides (1 µL for data independent experiments, 2 µL for data independent experiments) were loaded onto a C18 column (Nano-column C18 PepMap 100, 75 µm ID x 150 mm, 3 µm, 100 A). The sample was then eluted using a gradient of 2-40% buffer B over 65 minutes, 90% buffer B for 2 minutes, 2% buffer B for 13 minutes (buffer A consisting of HPLC grade water with 0.1% formic acid (v/v) and buffer B consisting of HPLC grade water with 80% acetonitrile (v/v) + 0.1% formic acid) at a flow rate of 10 µL/ minute.

2.6.3 Mass Spectrometry

The eluted peptides were analysed in either data dependent or data independent acquisition mode in a Q-Exactive Plus Orbitrap (Thermo Fisher Scientific, Denmark; Acquisition software version: 2.5-204201/ 2.5.0.2042).

For the data dependant acquisition experiments peptides were analysed in full MS mode using a resolution of 70,000 from m/z 370-1400 (automatic gain control target of 1x106 or 50 ms injection time), with the 20 most intense peaks selected for fragmentation. MS/MS scans were performed following collision induced dissociation, with a scan range of 200-2000 m/z and a dynamic exclusion of 20s (AGC target of 2e5, NCE of 28, MS/MS intensity threshold of 1.8e4).

For data independent experiments an initial MS scan of 50 ms with a mass range of 390 to 810 m/z, followed 20 ms scans of 5 m/z isolation windows covering the 390-810 m/z scan range (MSX count of 5, AGC target of 2e5, NCE of 28).

2.6.4 Spectral library

The data-dependent fragmentation data were searched using SEQUEST HT within Proteome DiscovererTM (software version 2.0.0.802; ThermoFisher Scientific) against all human proteins (all isoforms) in the Swissprot database (downloaded 8/6/2017) with: a strict false discovery rate of 1%, a precursor mass tolerance of 10 ppm, a fragment mass tolerance of 0.02 Da, a maximum of 2 missed cleavages, and variable modifications of Oxidation (M), phosphorylation (S,T,Y), Deamidation (N,Q), methyl (K), carbamidomethyl (C), N-terminal acytlation. All CSF samples were used to build a spectral library in Skyline (218) in order to accommodate for variation between individual patients as well as between different conditions.

34 2.6.5 Protein Quantification

2.6.5.1 Skyline

Search results were imported into Skyline (218) to generate a library file. DIA files were imported into the document, de-multiplexed to reduce chemical noise as well as improve peak selectivity and matched to the spectral library. Low quality peptides were filtered and peak areas exported for further analysis.

2.6.5.2 Perseus

Further analysis was performed using the freely available software Perseus (219). Peak areas imported and Log2 transformed (85) to reduce sample spread. Areas were normalised by sample median subtraction, and missing values were replaced from a shifted normal distribution. A t-test was performed to determine significant values, and a protein expression was considered different between groups if p <.05 and fold change >2. Principal component analysis (PCA) was also conducted using Perseus.

2.7 Statistical Analysis

The majority of statistical analyses were performed using GraphPad Prism version 7.0 for Mac (La Jolla, CA, USA). All parametric data was analysed using two tailed Student’s t-tests for between group comparisons and one way analysis of variance (ANOVA) for multiple group comparisons. Non-parametric data was analysed using Mann-Whitney tests for between group comparisons and Kruskal-Wallis ANOVA for multiple comparisons. Paired tests were used for comparison between baseline and peak ICP and unpaired tests were used when comparing between treatment and control at peak ICP. Linear regression analysis was performed to assess the correlation to determine the relationship between probe use and baseline ICP (Chapter 3) and CSF donor characteristics and peak ICP above baseline.

Concordance correlation coefficients (rc) were calculated using Stata 11.1 (StataCorp. TX, USA) to determine the agreement between pressure measured in the epidural and subdural spaces against the “gold standard” of intraventricular pressure. Assistance with statistical analysis was provided by CReDITSS.

35 Chapter 3 Validation of epidural pressure measurement in the rat

3.1 Introduction

In order to conduct the desired experiments presented in Chapters 4 and 5 our method of ICP monitoring required validation to ensure the accuracy and reliability of the pressure sensor, as well as the location of pressure measurement. Each aspect of the method validation will be presented in the following sections.

3.2 Part 1: Intracranial catheter systems

SAMBA probes (420 LP, SAMBA Sensors, Sweden) had been routinely used in our laboratory with excellent consistency across numerous studies (26, 80-83). The impetus to use a new system came when the last of our SAMBA probes reached the end of its lifespan (SAMBA Sensors had filed for bankruptcy in 2012 and we were unable to source any replacement probes). Therefore, I aimed to identify a suitable replacement pressure sensor for use with our published method of epidural pressure measurement (39). Due to the simplicity of our ICP monitoring method I originally planned to simply substitute the SAMBA probe with a similar fibreoptic pressure catheter made by Opsens (Canada). However, performance of the probes progressively declined with repeated use. This led to the sequential testing of probes made by Fiso and Millar and eventually a modified version of our original Opsens probe. Unexpected difficulties caused by differences in probe design resulted in substantial troubleshooting and methods were adapted to suit the individual probes. This is a retrospective analysis of my findings.

3.2.1 Experimental design

The expectation that SAMBA probes could easily be replaced by any fibreoptic probe of similar dimensions was challenged by differences in probe design, resulting in unexpected troubleshooting both in vivo and on the bench. A number of pressure systems were tested in our laboratory throughout the course of my PhD studies (see Figure 3.1 for a schematic of the probes; see Table 3.1 for probe specifications), including:

1. Opsens a. Unsheathed Opsens b. Sheathed Opsens c. Sheathed Opsens (Silicone tip; Opsens(Si)) 2. Fiso 3. Millar

36 Analysis was retrospectively performed on data collected from a number of experimental cohorts conducted in our laboratory, the majority from experiments associated with this thesis. ICP data from ninety six male, outbred Wistar rats (weight range 255-519 g) and seven Sprague Dawley rats (weight range 338-423 g) was analysed.

All animals underwent cranial surgery and had baseline ICP monitored in the epidural space prior to administration of any treatment. In experiments where epidural ICP was measured simultaneously at different sites, an average of the ICP values was taken. In the majority of experiments, ICP was also measured 16-24 hours after baseline ICP had been established and the animal had undergone intraventricular infusion of a specific substance (Chapter 2- Methods). However, since these animals had undergone treatment (intraventricular infusion of aCSF (Chapters 3 and 4), AVP (Chapter 4) or human CSF (Chapter 5)), the day 2 ICP values were not analysed for this study except where non-physiological/ pathophysiological effects were seen on the traces. ICP effects from negative drift and interferences were obtained from both baseline ICP measurement as well as longer term (8> hours) ICP monitoring. Negative drift was assessed as ICP values <0 mmHg. Interference was assessed as non-physiological effects on ICP caused by outside influences i.e. temperature, light, heat and electrical interference. See Table 3.2 for a comparative overview of findings.

Unless otherwise specified, probes were prepared according to our published method (39). In brief: probes were zeroed in air at the same height as the screw and then inserted through the saline filled screw shaft into the epidural space and sealed in place with Silagum after a signal had been established. Upon completion of monitoring, probes were removed from Silagum and cleaned according to manufacturer recommendations.

3.2.2 Contribution of investigators

This study involved analysis of data collected by members of the laboratory. The literature review and experimental design was performed by Rebecca Hood (RH). The majority of animals used in this analysis were performed by RH (n = 93). Dr Lucy Murtha (LM), Dr Daniel Beard (DB), Caitlin Logan (CL) and Kirby Warren also assisted with some of the initial Opsens(NS) surgeries (n = 10). Modifications of the published method were conceptualised by RH, Prof. Neil Spratt, Dr Damian McLeod, CL, DB and LM. All the experiments on the modifications were performed by RH. All data were analysed by RH.

37

Figure 3.1: Schematic showing differences in probe tip design. This diagram illustrates difference in the design of 5 different pressure sensors (manufactured by SAMBA Sensors, Opsens, Fiso Technologies and Millar) that were assessed for their suitability for epidural intracranial pressure monitoring in the rat. Note, this figure is simplified and is for illustrative purposes only.

38 Manufacturer Model No. Probe Type Probe Sheath Pressure range Outer diameter Port No. of (Coating) (according to at probe tip traces manufacturer analysed specifications) SAMBA Sensors 420LP Fibreoptic No sheath -50 to +300 mmHg 0.42 mm End NA (radio-opaque coated)

Opsens OPP-M250-X- 80SC-1.5PTFE- Fibreoptic No sheath -50 to +300 mmHg 0.1 mm End 1 (bare fibre) XN-50PI-P1

Opsens OPP-M250-X-805C-1.SPTFE- Fibreoptic Polyimide -50 to +300 mmHg 0.42 mm End 22 (no silicone at tip) XN-100PIT4-P1

Opsens OPP-M250-X-80SC-1.5PTFE- Fibreoptic Polyimide -50 to +300 mmHg 0.304 mm End 53 (silicone seal at tip) XN-100PIT3-P1-SILICONE SEAL

Fiso FISO-LS-2FR-10 (FOP-MIV-20 Fibreoptic Polyimide ±300 mmHg; 0.64 mm End 17 (gel at tip) sensor)

Millar SPR407 Piezoelectric No sheath -50 to +300 mmHg 0.66 mm Side 10 (nylon coated)

Table 3.1: Specifications of probes tested in Chapter 3 – Part 1.

39 Property Sensor Reliability Durability High fidelity signal Drift Accuracy Interference SAMBA Sensors 100+ uses 100+ uses Y None Accurate None Opsens NA 1 use - Y N/A (broke 1st use) Accurate None Unsheathed Fragile and brittle Opsens Sheathed 5-10 uses 10+ uses Declined with use Increased with use Accurate None Opsens (Si) 100+ uses 100+ uses Y None Accurate None Fiso Drifted over 10+ uses Y Occurred with changes in Influenced by Hydration status time hydration status of probe hydration status of tip. At time of insertion and probe tip during long term monitoring. Resulted in negative baseline values. Millar Declined 1 use – Declined over time None Pressure output Heat, light, room with use coating easily easily influenced pressure, damaged by interference electromagnetism, hydration status Increased with use Table 3.2: Comparative overview of findings of the probes tested in Chapter 3 – Part 1

40 3.2.3 Opsens

The first probes tested were Opsens OPP-M250 (Opsens, Canada) fibre optic pressure sensors (referred to as Opsens). We tested two different Opsens probes - unsheathed and sheathed (Opsens(Si) was not available at the time). Similarities between the Opsens and SAMBA probes include the size and location of the pressure sensor (at the tip of the probe) and pressure range (according to manufacturer specifications; Table 3.1 for details). The major difference between probes was that the SAMBA optical fibre was protected by a radio- opaque coating, whilst the unsheathed Opsens fibre had no coating and the sheathed Opsens fibre was protected by a polyimide sheath. Despite these differences it was expected that either probe would be a suitable replacement for the SAMBA system.

3.2.3.1 Results

Unsheathed Opsens

This probe provided a high fidelity ICP signal in one animal, however due to its fragility the fibre snapped when it was bumped by the operator before completion of the first recording.

Sheathed Opsens

Sheathed Opsens probes provided a signal that was comparable to what was obtained with SAMBA (Figure 3.2). However, signal quality declined with repeated probe use (5-10 uses; see Figure 3.4 in section 3.2.4 Fiso of this chapter). Presence of a signal was observed prior to sealing in 4/22 animals (only assessed in baseline ICP recordings). In the remaining animals a signal was found to develop as the probe was sealed into place (see Figure 3.10 from Part 2 of this chapter). We were able to obtain a stable baseline ICP trace in 20/22 animals, ranging from 2.51–13.59 mmHg. In the remaining 2 animals stable baseline ICP was not achieved due to experimental error (n=1) and signal drift (n=1). Signal stability when the probe tip was in the air (for zeroing) also declined with use and signal drift was observed when the tip was in saline as well as when it was sealed within the screw (max range -50 mmHg to +300 mmHg). When investigating the cause of signal drift, there were no effects of either air flow or room pressure. Visual inspection of the probe tip under the operating microscope showed evidence of a build-up of organic material in the recess between the end of the sheath and the probe tip.

41

Figure 3.2: Representative epidural intracranial pressure (ICP) waveforms. ICP waveforms recorded from male, outbred Wistar rats using five different pressure sensors. Probe type (from top to bottom): SAMBA Sensors, sheathed Opsens, Fiso, Millar and silicon tipped sheathed Opesns (Opsens(Si)). All traces show clear respiratory and pulse pressure waveforms. Nb. Recordings made in different animals.

42 3.2.3.2 Discussion

While unsheathed Opsens probes provided a high fidelity signal, the fibre was extremely fragile and snapped before the experiment was completed. Despite many similarities in the design of the unsheathed Opsens and the SAMBA probes, the lack of a coating to protect the fibre in the Opsens design resulted in it being brittle and very susceptible to damage. It is highly likely that the radio-opaque coating of the SAMBA probe provided protection to the brittle fibre. Many attempts were made to add our own coating to the unsheathed fibre however we were unable to coat the probe without damaging the probe tip. Moreover, attempts to sheath the probe (up to the tip) with heat wrap, glass capillaries, silastic and polyurethane tubing were unsuccessful due to the fragility of the optical fibre. It was decided that without a protective coating, probes would likely be too fragile for our use.

Sheathed Opsens probes were found to be far more robust than their unsheathed counterparts. However, probes became unreliable within 10 uses, displaying reduced signal fidelity and signal drift, despite initially producing high fidelity ICP waveforms similar to that produced with SAMBA. The number of reliable ICP recordings varied between probes (5-10 uses) and probe failure could not be immediately identified or predicted. Therefore, knowing when to change probes was challenging. The loss of fidelity did not affect the absolute values and thus, was not of major concern. The signal drift did however, become a major obstacle of using the sheathed Opsens probes because without a stable pressure in air we were unable to reliably zero the probes. It is likely that both issues occurred as a direct result of the build- up of biological material within the recessed tip. I hypothesise that the organic material influenced the flexible membrane at the tip of the probe thus resulting in reduced signal fidelity and drift. Attempts at removing the build-up of organic material including brushing with soft paintbrushes and isopropyl alcohol wipes, or soaking the tip overnight in 0.1% Terg-A-Zyme were unsuccessful. We also attempted to reduce the recess by trimming the protective sheath, however this resulted in damage to the delicate membrane,

Difficulties were encountered when obtaining a stable ICP waveform prior to Silagum application. In our original experiments, SAMBA probes were sealed into place upon visualisation of a stable ICP waveform between 0-15 mmHg. This was not always possible using the sheathed Opsens probes despite repeated attempts, often by multiple operators. In experiments where a stable ICP waveform was not visualised pre-seal, a waveform would become visible within minutes of Silagum application. The difference between SAMBA and the sheathed Opsens pre-seal stability is likely caused by the proximity of the sensor tip to the dura mater. The unsheathed, bare fibre design of the SAMBA probe permitted direct contact between the sensor and the dura. I hypothesise that the close proximity of the sensor with the

43 dura permitted stable ICP recordings prior to the probe being sealed into place. The sheathed Opsens sensor, however, is recessed within the protective sheath at the tip of the probe preventing the sensor from coming into contact with the dura. Potential effects of dural contact on absolute values will be investigated in Part 2 of this chapter.

A limitation of this investigation is that probe accuracy could not be assessed via direct comparison with SAMBA probes. Prior to testing the Opsens probes the last of our SAMBA probes had broken. However, baseline values were within the physiological range previously reported by us and others using both solid state and fluid filled systems (26, 36, 60, 80-83).

In conclusion neither unsheathed nor sheathed Opsens probes were found to be suitable for measuring ICP using our published method. Without protection from a sheath, unsheathed Opsens probes were extremely fragile and not able to be used repeatedly. The addition of a protective sheath improved the longevity of Opsens probes which were successfully used to measure epidural pressure. However, sheathed probes were found to be unreliable due to the accumulation of organic material within the recess at the tip of the probe leading to a reduction in signal fidelity and signal drift. Therefore, I decided to test a sheathed fibreoptic catheter with no recess at the tip.

44 3.2.4 Fiso

Fiso probes (FISO-S-2FR-10; FISO Technologies, Canada) were next investigated. These are fibreoptic catheters with a similar design to the sheathed Opsens probes, in terms of a protective sheath covering the tip of the probe. However, Fiso probes are designed with protective gel over the sensor which fills the gap between the sensor and is flush with the end of the sheath, thus preventing any recess. I hypothesised that the protective gel would limit build-up at the tip of the probe and prevent the signal drift and reduction in signal quality previously observed with the sheathed Opsens probes.

3.2.4.1 Results

Signal drift (range -20 mmHg to +12 mmHg) was observed prior to the probes being inserted into the screws for ICP monitoring. The signal would be stable in saline and then drift would occur immediately as probe tips were exposed to air. This could be minimised by drying out the probe tip (soaking the probe tip in EtOH and allowing it to air dry; Figure 3.3a) and zeroing when the tip was protected from the air by an additional sheath (Figure 3.3b). There was a recurrent pressure drop of up to 20 mmHg (the lower limit of the prespecified range in some experiments) upon insertion of the probe tip into the saline filled screw. A signal was visualised prior to sealing in all experiments. Fiso probes produced high quality signals that did not lose fidelity with repeated use (Figures 3.2 and 3.4). Negative drift (as low as -20 mmHg) occurred in 10/34 recordings (including both day 1 and day 2 recordings) after Silagum application, however quickly returned to a stable signal (Figure 3.5). Baseline ICPs ranged from -4.2 to 9 mmHg. With 7/17 baselines £0 mmHg. Negative drift also occurred in 2 experiments after approximately 4 hours of continuous ICP monitoring (on day 2 of experimental procedures) (Figure 3.6).

45

Figure 3.3: Representative pressure recordings showing preparation of Fiso probe tip before zeroing. Fiso pressure trace as probe is moved from liquid to air. In an effort to stabilise pressure signal probe tips were soaked in EtOH and dried whilst being either a. directly exposed to air or b. protected within an additional sheath.

Figure 3.4: Simultaneous recording of intracranial pressure (ICP) using two different pressure sensors. ICP was recorded simultaneously via a Fiso catheter (top) and a sheathed Opsens catheter (bottom). The Fiso trace shows a lower mean ICP, however a clearer ICP signal. Both traces show respiratory and pulse pressure waveforms.

46

Figure 3.5: Negative depression in absolute intracranial pressure (ICP). Representative epidural pressure trace showing a transient negative depression in absolute ICP post-seal as measured using a Fiso pressure sensor. Within minutes of the depression ICP returns to baseline values. *Abdominal compression

Figure 3.6: Negative drift over several hours. Representative epidural pressure trace showing negative drift of the pressure signal after approximately 4 hours of stable ICP recording using a Fiso pressure sensor.

47 3.2.4.2 Discussion

Fiso probes produced high fidelity pressure waveforms that did not diminish in quality with use. However, accuracy of the probes was questionable due to the observed signal drift both prior to- and during-use. Signal drift was a major issue when troubleshooting the Fiso probes. The original probe preparation for our published method involved zeroing the probe in the air at the height of the screw. This was not possible with Fiso probes without first drying the tip and then zeroing the signal when the probe tip was inside a sheath prior to inserting it into the screw. This limited drift in the air, however it failed to prevent drift from occurring when the probe was inserted into the saline filled screw. It also had no effect on the negative drift observed at later time points throughout monitoring. After much investigation it was noticed that the signal drift predominantly occurred at transition points where the probe tip would either be removed from or placed into liquid. This was not a problem with either the SAMBA or Opsens probes. The major difference between probe designs was the protective gel at the tip of the Fiso catheters. This suggested that the hydration status of the gel at the probe tip greatly influenced the absolute values measured by the probes and therefore the accuracy of the measurements.

In this study, almost half of all baseline values were below 0 mmHg. Previous investigations of ICP in rats (including from this laboratory) have shown baseline values to be ³0 mmHg (26, 36, 60, 80Uldall, 2014 #711Hiploylee, 2014 #732, 81-83). One potential cause of negative ICP is CSF leak (220). It is possible that meninges were damaged in these animals during screw insertion. However, this is unlikely to be the cause as firstly, CSF would have been visible within the screw and secondly, the system was sealed. This would mean that fluid would have simply been shifting between compartments as opposed to leaking out from within the craniospinal system. Furthermore, from experience, even if the dura is torn during surgery causing a CSF leak, measured pressure is still positive (unpublished observation). It is therefore unlikely that the baseline ICP measurements obtained in this study reflected the true, physiological pressure. The specific reason for the negative baseline ICP values observed in this study requires further investigation, however, the most likely cause is an incorrect “zero point”. If the probe had a falsely high pressure reading when it was zeroed in air, it is entirely possible that when sealed within the skull the measured pressure output would be lower than that of the actual pressure. The likely cause of an inaccurate zero point in this study is the hydration status of the gel protecting the probe tip. Evidence to support this comes from the negative drift observed as the dry probe tips were inserted within the saline filled screw. It is likely this signal fluctuation occurred as a direct result of the re-hydration of the probe tip by the saline. If the zero point when the probe tip was dry differed from the zero point when the probe tip was wet, then zeroing the dehydrated probe tip in air likely confounded the absolute

48 values obtained in situ when the probe tip was rehydrated. This may not be a significant problem if measuring DICP measured continuously, however it presents a significant problem if absolute values are important. Furthermore, this may be a potential confounder of results in a repeated measures study, such as the experiments in this thesis where probes have to be removed and reinserted on each monitoring day. Experiments were initially planned to assess differences between absolute values when the probe was hydrated vs dehydrated. However, we did not ultimately proceed with these experiments due to the uncertainty of probe reliability over longer monitoring periods.

Substantial negative drift was observed in two of the experiments (during day 2 monitoring). This drift was unexpected and occurred despite stable baseline ICP recordings (for 4+ hours) with no change in the ICP waveform. In these animals visual inspection was performed on the screws when probes were removed and the screws appeared to be dry. The exact cause of the delayed negative drift is unclear however, based on the observed influence of hydration status over absolute values it is reasonable to hypothesise that the saline in the screws may have been absorbed by extradural vessels permitting the probe tip to dry out. As the probe is sealed within the screw, it is impossible to ascertain how much saline is remaining within the screw shaft without removing the probe and disturbing the system. Negative drift was only noticed in two animals, however negative drift could potentially mask ICP rises and therefore may go unnoticed in animals that experience ICP elevation. Based on this unreliability it was determined that Fiso probes were unsuitable for longer term ICP monitoring in vivo due to the sensitivity of the gel at the probe tip to hydration.

49 3.2.5 Millar

With no suitable fibreoptic probes available at that time I decided to investigate Millar pressure catheters (SPR407; Millar, USA). Millar catheters are piezoelectric and measure pressure via changes in electrical resistance across the sensor. Other than the way it measures pressure there are numerous differences between the Millar catheters and the previously tested fibreoptic pressure sensors including location of pressure sensor (side port), probe tip size (2 F outer diameter) and tip coating (nylon), requiring modification of our published method prior to testing. Firstly, the diameter of the hole within the screw shaft was increased from 0.7 mm to 1 mm to accommodate the increased size of the probe tip. Secondly, to minimise known effects from temperature and hydration status (according to the manufacturer) on absolute values it was recommended that the probes be soaked (for at least 30 minutes) and zeroed in 37 °C saline before in vivo use. Lastly, the method of probe removal from the Silagum at the end of the experiment was altered. Previously, simply applying upward pressure on the fibreoptic probes was all that was required to remove the probes from Silagum (with the exception of the unsheathed Opsens that broke before removal). This was not a problem previously as the probe tip diameter was either uniform with the coating or within a sheath. However, the Millar catheters tip was 0.5 F bigger than the body, resulting in a theoretical weak point at the junction between the catheter body and sensor tip. To ensure that this connection was not damaged due to the upward pressure placed on this junction, probes were carefully removed from the screw by dissecting the Silagum from around the catheter body and removing the probe tip from the larger opening. However, during initial testing of the probe (prior to in vivo use) visual inspection of the probe tip showed a bend at the connection between the tip and body of the probe as well as scratches to the nylon coating from our tools. Therefore, in an effort to protect the delicate probe sheath and tip from being damaged by our sharp tools, Silagum was pried off the dental cement whilst the probe was still sealed in place and then simply pushed up along the length of the probe.

3.2.5.1 Results

One animal was excluded from analysis as a stable baseline ICP could not be established due to electromagnetic interference. A stable signal developed within minutes of applying the Silagum in all experiments (Figure 3.2). Baseline ICP ranged from 3.49–15.24 mmHg. Baseline ICP increased with repeated probe use (R2 = 0.71; p=.005; Figure 3.7). Signal quality declined and interference increased with repeated use (Figure 3.8). Interference was observed during the third experiment and increased with each subsequent use of the probe until there was so much interference that a stable baseline ICP could not be obtained. Electromagnetic interference was caused by external sources such as the homeothermic heat

50 mat under the rat, static electricity generated from brushing the animal’s fur and even from proximity of researchers. There was also interference from light, heat, and room pressure changes despite the probe being sealed within the screw. Attempts to ground the probe, rat and electrical equipment were not successful and did not minimise interference. Visual inspection of a probe that had approximately 10 uses (compared with a brand new probe) found damage to the nylon coating of the probe potentially caused by shearing of the nylon coating as the Silagum was moved along the body of the probe. Additionally, there appeared to be liquid and air bubbles inside the nylon coating.

20 R2 = 0.7049, p = 0.0046

15

10

ICP (mmHg) ICP 5

0 2 4 6 8 10 Experiment No.

Figure 3.7: Intracranial pressure (ICP) at baseline vs. probe use. Baseline epidural pressure (mmHg) was measured in male, outbred Wistar rats using piezoelectric Millar pressure sensors. Linear regression analysis of baseline ICP versus the experiment number. Absolute values recorded at baseline increased with repeated use of the same Millar sensor. Data points are baseline ICP measurements recorded in n = 9 animals.

51

Figure 3.8: Decrease in performance of Millar pressure sensors with repeated use. Representative pressure waveforms of epidural pressure measurement in male, outbred Wistar rats using Millar pressure sensors. Damage to the probe accumulated during use resulting in a. a reduction of signal fidelity (from top to bottom: 1 x use, 8 x use, 10 x use). Signal interference also increased with repeated use of the sensor. Representative mean arterial pressure (MAP; top) and intracranial pressure (ICP; bottom) traces showing interference to ICP waveform from b. light and c. electromagnetism confounding the absolute pressure readings. The lack of change in the blood pressure trace indicates no physiological changes occurred.

52 3.2.5.2 Discussion

Millar catheters were found to be suitable for single use measurement of epidural ICP using our published method. However, the accuracy and reliability of the probes became increasingly questionable due to interference from electromagnetism, light, heat and changes in room pressure. Furthermore, the cost of individual probes (more than 2x that of Fiso and 10x that of Opsens probes) precluded consideration of using the probes once only.

There were no observed negative effects of using a probe with a side port within a screw to measure ICP. In our original fibreoptic studies an ICP signal was established prior to the probe being sealed into place to ensure the sensor was in the correct location within the screw. Due to the tip’s side port it was impossible to have the sensor directly in contact with the dura using our preparation, therefore visual confirmation of probe location was not possible prior to Silagum application. However, this did not appear to influence the appearance or fidelity of the ICP waveform generated by the Millar catheter as probes were sealed into place. This was a similar finding to what was observed using the sheathed Opsens probes if waveform visualisation was not achieved prior to Silagum application. Unfortunately, potential effects this may have had on baseline ICP values are hard to determine due to the likely confounding of measurements by electrical interference. Although, absolute values are within the published range (26, 36, 60, 80-83) the correlation between experiment number and increasing baseline ICP value (R2 = 0.71; p=.005) casts doubt over the accuracy of these measurements.

The major obstacle encountered with Millar is the interference that affected both signal quality and absolute values. Piezoelectric pressure sensors are known to be susceptible to interference (41). In this study, the observed increasing interference is most likely a direct result of the accumulating damage to the body of the catheter. The delicate nylon coating of the probe body made probes particularly susceptible to damage caused by removal from the Silagum. Damage increased with each use of the probe, permitting water infiltration within the body of the probe, resulting in the probes increased sensitivity to interference. Attempts to limit damage to the probe during removal from Silagum were unsuccessful. Therefore, Millar catheters were deemed unsuitable replacements for SAMBA probes using our published method of epidural pressure measurement.

53 3.2.6 Opsens(Si)

The final probes tested were sheathed Opsens probes with silicone at the tip (Opsens(Si); OPP-M250; Opsens, Canada). Since the original experiments using Opsens probes (both sheathed and unsheathed) the company had modified their design (in part based on our feedback) and added a protective layer of silicone over the tip to fill the recess and prevent the accumulation of organic material. Despite the initial trouble with the original sheathed Opsens probes, fibreoptic probes were still considered the most suitable replacements for SAMBA and we hoped that the new design would eliminate the problems that were faced because of the recessed tip on the sheathed Opsens catheters. To maximise the chances of success the method of probe preparation was modified based on findings from the Fiso and Millar probe troubleshooting. This included: 1) soaking probes in saline for at least 30 minutes prior to use and 2) zeroing the probe with the sensor tip in saline at the height of the screw. In the sheathed Opsens experiments there were difficulties obtaining a stable signal prior to sealing the probe with Silagum. Furthermore, the Millar experiments showed that prior visualisation of the ICP waveform was unnecessary for stable baseline ICP recording. Therefore, probes were intentionally sealed into place without prior visualisation of the ICP waveform. Lastly, the absolute value of the pressure recording was measured in saline upon removal from the screw (after recording was finished in the animal) to ensure that there were no effects from drift (either positive or negative) over time. If no drift had occurred the pressure should read ~0 mmHg when returned to saline. Previously probes had been moved straight into cleaning fluid.

3.2.6.1 Results

No signal drift was experienced when zeroing and inserting the Opsens(Si) probes. Signals would develop within minutes of the probes being sealed into place (Figure 3.2). ICP baselines ranged from 3.03-9.24 mmHg and there was no relationship between absolute values and probe use (R2 = 0.2, p>.05). During longer term monitoring (up to 8 hours) we did not observe any obvious negative drift. After removal from the animal, the mean difference from pre- insertion zeroing (as measured in saline) was 0.43 ± 1.2 mmHg.

3.2.6.2 Discussion

Opsens(Si) was found to be reliable for epidural pressure measurement using our published method. Stable ICP waveforms were observed during both short and long durations of measurement (>8 hours) as well as for repeated measures analysis. Therefore, it was determined that Opsens(Si) is a suitable replacement pressure catheter for epidural pressure measurement using our established model.

54 3.2.7 Summary

Various technical difficulties were encountered whilst identifying a suitable replacement for SAMBA probes due to differences in probe design. Out of the five probe models, from 3 manufacturers that were investigated, I concluded that Opsens(Si) probes are the most suitable pressure sensors for epidural ICP measurement using our established method. Opsens(Si) probes were easy to prepare and were robust enough to withstand multiple uses without showing any signs of probe deterioration either physically or through changes in output. Furthermore, probes were found to be reliable for both short and long term monitoring periods without a loss of fidelity of signal drift occurring. As a direct result of my findings ICP measurement with these probes has become standard practice in our laboratory.

55 3.3 Part 2: The influence of dural contact on ICP waveform and accuracy

In our original ICP measurement studies, appropriate probe location within the screw was confirmed by visualisation of a pressure waveform prior to sealing the catheter in place (39). Whilst investigating replacement probes for SAMBA in Part 1 of this chapter I noticed that visualisation of a pressure waveform prior to sealing was not always possible. In these cases, waveforms would develop within minutes of Silagum application, as an airtight seal was formed around the probe tip within the screw. Understanding the differences between probe designs led me to hypothesise that waveform visualisation prior to creation of an airtight seal was likely only possible when the probe sensor was in direct contact with the dura mater. Anecdotal evidence from Verlooy et al. (1990) suggests that dural contact may confound ICP recordings (38). Therefore, the aim of this study was to investigate whether dural contact influences baseline ICP values using our published method of epidural pressure measurement.

3.3.1 Experimental Design

To determine whether probe sensors placed in contact with the dura produced higher absolute ICP values than those placed above the dura, retrospective analysis was performed on baseline ICP data from SAMBA, sheathed Opsens and Opsens(Si) experiments conducted as part of this thesis. All animals underwent ICP surgery and had baseline values recorded prior to any treatment. Baseline ICP recordings were categorised as either being in contact with the dura (signal before seal; n = 37) or not in contact with the dura (no signal before seal; n= 69) (Figure 3.10).

3.3.2 Contribution of investigators

This study involved analysis of data from multiple members of the laboratory. The experimental design and majority of experiments used in this analysis (n = 86) were performed by Rebecca Hood (RH). Dr Lucy Murtha, Dr Daniel Beard, Caitlin Logan and Kirby Warren also assisted with some of the initial SAMBA/ sheathed Opsens surgeries (n=20). All data were analysed by RH.

56

Figure 3.10: Pressure waveform visualisation. Intracranial pressure (ICP) waveforms generated when the pressure senor was a) pressed against the dura (signal present before seal) or b) superior but not in contact with the dura (no signal present before seal). Grey shading represents Silagum application. Nb. Recorded baseline ICP measurement not shown as it is assessed after at least 30 minutes of stable pressure recording.

57 3.3.3 Results

In total, two SAMBA baseline recordings were excluded from this analysis. In one animal we were unable to obtain a stable baseline ICP as the animal had a haemorrhage during probe insertion. The other recording was excluded as the ear bars (used to keep the rats skull steady during ICP surgery; see Chapter 2) were not removed from the animal prior to baseline ICP recording. We have previously shown that the ear bars cause ICP elevation while in place (39). Therefore, baseline ICP in this animal is likely to have been artificially high. Physiological parameters are listed in Table 3.3. Baseline ICP was significantly lower in the no contact group versus the contact group (5.7 ± 1.9 mmHg v 7.6 ± 3.8 mmHg, respectively; p=.02; Figure 3.11).

Contact (n = 37) No Contact (n = 69) P value SpO2 (%) 98.5 ± 1.4 96.3 ± 2.8 p<.0001 Respiratory Rate (BPM) 55 ± 7 61 ± 7 p<.0001 Heart Rate (BPM) 383 ± 49 424 ± 31 p<.0001 MAP (mmHg) 92.6 ± 13.4 101.2 ± 8.5 p=.0003 Table 3.3: Mean physiological parameters at baseline in animals undergoing epidural ICP measurement. Abbreviations: SpO2, oxygen saturation; BPM, breaths (respiratory rate) or beats (heart rate) per minute; MAP, mean arterial pressure

20 *

15

10

ICP (mmHg) ICP 5

0 Contact No Contact (n=37) (n=69)

Figure 3.11: Influence of contact between the pressure sensor and dura on baseline intracranial pressure (ICP). Box and whisker plot showing the median (horizontal line), interquartile (box) and absolute (whisker) ranges of data from each experimental cohort, with outliers (triangles). Baseline ICP was assessed when pressure catheters were either in contact with the dura (n = 37) or superior to the dura (no contact, n = 69). Absolute ICP values were significantly higher when probes were in contact with the dura, than when there was no contact between the probe and dura (p=.02).

58 3.3.4 Discussion

The data show that ICP values were higher when the sensor was in contact with the dura. This suggests that dural contact causes artificially high absolute ICP values. Importantly, whilst statistically significantly different, the mean absolute difference observed was biologically minor (1.9 mmHg). There were significant differences in all of the measured physiological variables between groups. Respiratory rate, heart rate and MAP were lower in the contact group. Interpretation of these results are difficult due to the method of anaesthesia. At the time of measurement animals had not yet experienced any experimental intervention i.e stroke or intraventricular infusion, therefore they should have had intact cerebral autoregulation to maintain a constant CPP. As discussed in Section 1.1.2 (Chapter 1) CPP = MAP – ICP, therefore any decrease in MAP should be met with a similar decrease in ICP in order to maintain adequate cerebral perfusion. It could be hypothesised then that if MAP was equal between groups ICP may have actually been even higher in the contact group than what was recorded. However, isoflurane has been shown to affect autoregulation (221), therefore any confounding of absolute pressures by the physiological variables is difficult to interpret.

Interestingly, when placing the probe against the dura there was often a transient increase in pressure that quickly returned to baseline values (Figure 3.10 a). Suggesting that even if the probe is in contact with the dura, the effects on ICP are minor. A similar transient rise at the time of probe insertion has been reported by Verlooy et al. (1990) (38). The authors speculate that the transient nature of the rise is caused by dural elasticity i.e. the dura becomes stretched due to tension from the probe and thus pressures return to baseline values.

There have been no published studies to date directly comparing epidural pressure values measured simultaneously (or otherwise) from sensors placed in contact with or not in contact with the dura. However, the present findings are supported by evidence from published animal studies showing excellent correlation between epidural pressure and both intraventricular and intraparenchymal pressures when the method of implantation involves the probe being placed above/not in contact with the dura (36, 60). In contrast, anecdotal reports from Verlooy (1990) suggest that epidural pressure is not accurate when there is contact between the sensor tip and dura (38). However, they did not report any specific values therefore it is difficult to interpret their findings.

There are some limitations of this study. This was a retrospective analysis of data obtained from multiple cohorts of animals over a four year period. The data was available due to the troubleshooting of multiple pressure sensors which required original methods to be modified

59 (Part 1 of this Chapter). Therefore, animals were not randomised to dural contact versus no contact and blinding was not possible as presence or absence of an ICP waveform was visually confirmed prior to application of Silagum. This may have introduced bias into the results. Furthermore, I was not able to investigate the relationship between the degree of force applied to the probe and absolute ICP.

The results of this study show a statistically significant difference between groups suggesting that avoiding contact is methodologically desirable. For this reason, all subsequent ICP measurements in this thesis were made with probes located above and not in contact with the dura. However, despite the statistical significance, the observed differences were biologically minor and are likely to be of minimal importance to results.

60 3.4 Part 3: Is a recovery period required between surgery and ICP measurement?

Numerous published methods for epidural pressure measurement assess pressure the same day as surgery (36, 39, 60), but in practice researchers often include a recovery period between the initial surgery and ICP measurement (197, 198, 222). The rationale behind these recovery periods is to permit any acute or subacute changes from surgery to subside before pressure is assessed. Interestingly however, there is a lack of evidence to support the existence of such changes occurring. Furthermore, there is evidence to suggest that it may actually be harmful depending on the location of the pressure catheter (36). Measurements of ICP within our laboratory are taken at baseline (pre-intervention, on the same day as surgery) and again at 16-24 hours post-surgery/ intervention (e.g. CSF transfusion). Therefore, it was important to 1. confirm if any changes in ICP between these two time points occur as a result of the intervention and 2. ensure that results are not confounded by the ICP surgery itself. The aim of this study was to investigate whether ICP measured at baseline the same day or 24 hours-post surgery is influenced by the surgery itself.

3.4.1 Experimental Design

Male, outbred Wistar rats (n = 11; weight 285-349 g) underwent ICP measurement surgery to provide access to the epidural and subdural spaces as well as both lateral ventricles. To establish whether baseline ICP measurement was affected by the ICP surgery itself, ICP was measured either immediately post-ICP surgery (group 1; n = 6) or 24 hours post-surgery (group 2; n = 5) (Figure 3.12). Two separate groups of animals were used due to the likelihood that piercing the dura on day 1 could itself affect ICP on day 2 due to CSF leak. Baseline pressure was assessed at each location via the sequential introduction of pressure sensors into the epidural, subdural and intraventricular spaces. Responsiveness to changes in pressure was confirmed in all traces by abdominal compression at various times throughout the procedures. Baseline ICP data measured simultaneously at each location was compared between groups.

61

Figure 3.12: Experimental design – Comparison of intracranial pressure (ICP) measured in the epidural, subdural and intraventricular space. Group 1 animals (n = 6) underwent surgery (S) to measure mean arterial pressure (MAP) and ICP before sequential introduction of epidural (red arrow), subdural (blue arrow) and intraventricular (black arrow) pressure catheters. After baseline (B) ICP was established at each location, ICP was artificially elevated to ³20 mmHg via intraventricular infusion of artificial cerebrospinal fluid at increasing flow rates. When ICP was sufficiently high, both epidural and subdural probes were removed (upward arrows) and reinserted (at the elevated ICP) before infusion was stopped and animals were euthanized. Group 2 animals (n = 5) underwent the same procedures, however, there was a 24 hour recovery period between the initial surgery and ICP measurement/elevation. In addition, group 2 animals had MAP surgery in the opposite leg (S2) prior to ICP assessment on day 2.

62 3.4.2 Results

Physiological parameters measured at baseline are listed in Table 3.4. There were no differences in mean ICP between group 1 and group 2 at any location (epidural, p=.8; subdural, p=.1 and intraventricular, p=.9; Figure 3.13; see Table 3.5 for mean ICP at different locations). Mean ICP across locations was 4.7 ± 0.6 mmHg when measured on the same day as surgery (group 1) and 5.4 ± 0.8 mmHg when measured at 24 hours post-surgery (group 2). ICP was not significantly different between groups (p=.9).

Group 1 Group 2 Combined SpO2 (%) 95.5 ± 2.4 97.6 ± 2.1 96.5 ± 2.4 RR (BPM) 63 ± 4 69 ± 9 66 ± 7 HR (BPM) 415 ± 19 416 ± 23 416 ± 20 MAP (mmHg) 108.7 ± 7.7 98.6 ± 7.3 104.1 ± 8.9

Table 3.4. Mean (±SD) physiological parameters at baseline for each group. Abbreviations: SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute; MAP, mean arterial pressure.

Probe location Mean baseline ICP (mmHg) P value Group 1 Group 2 Combined (G1 v G2) ED 4.1 ± 1.3 4.5 ± 2.3 4.3 ± 1.8 p=.8 SD 4.7 ± 1.4 6.1 ± 1.2 5.3 ± 1.4 p=.1 IVP 5.4 ± 2.8 5.6 ± 1.5 5.5 ± 2.2 p=.9

Table 3.5: Mean (±SD) baseline intracranial pressure (ICP) at each location of pressure measurement for group 1 and 2. Abbreviations: ED, epidural pressure; SD, subdural pressure; IVP, intraventricular pressure.

63

10

8

6

4

ICP (mmHg) ICP 2

0 Day 1 Day 2 Day 1 Day 2 Day 1 Day 2

Figure 3.13. Baseline intracranial pressure (ICP) measured the same day or 1 day post- ICP surgery, at different locations within the intracranial compartment. Baseline ICP (mean ±SD) as measured simultaneously in the epidural (red), subdural (blue) and intraventricular (black) space of all animals on the day of ICP surgery (group 1; solid bars) or the day after surgery (group 2; open bars). There were no significant differences between pressures measured on either day (p>.1)

64 3.4.3 Discussion

The results indicate that there is no confounding effect of surgery on absolute ICP on the same day or the subsequent day after surgery at any of the measured locations. There have been no published studies that specifically investigate the effects of surgery on ICP measurement. Furthermore, effects of surgery have not been investigated in previous epidural ICP validation studies. Indirect evidence comes from epidural validation studies using both telemetered (60) and implanted (36) pressure sensors that show stable ICP across numerous days post-surgery (in one case out to 60 days post-surgery). This indicates that a recovery period is not required for accurate measurement after ICP measurement surgery. As there were no differences between cohorts, both groups have been combined for the analysis outlined in the remainder of this chapter of determining the accuracy of epidural pressure measurement.

65 3.5 Part 4: Accuracy of epidural pressure measurement using a fibre optic pressure sensor

Despite being widely used to measure ICP, there have been very few published studies validating the accuracy of epidural pressure measurement in rats. Two recent publications have shown epidural pressure to be accurate and reliable when compared against intraventricular (36) and intraparenchymal pressures (60). Both of these studies used fluid filled catheters. Importantly, the former study only measured resting (baseline) ICP whilst the latter only measured ICP after injury (no baseline). The one study that has investigated epidural pressure measurement using a fibreoptic device reported that epidural pressure did not correlate well with pressure measured in the cisterna magna (38). Despite the conflicting findings it is plausible that the presence of the dura influences the reliability of epidural pressure measurements (as is the case with humans). This could be investigated by comparing epidural and subdural pressure measurement. The most likely reason for this discrepancy is more to do with probe placement than the probe type itself (as discussed in Part 2 of this chapter), therefore I hypothesise that epidural pressure measured using a fibreoptic sensor will correlate extremely well with both subdural and the “gold standard” of intraventricular pressure. The aim of this study was to assess the accuracy and reliability of epidural against subdural and intraventricular pressure measurement using fibreoptic pressure sensors over a range of pressures.

3.5.1 Experimental Design

These experiments were designed to assess the accuracy of epidural pressure measurement against subdural and intraventricular pressure over a range of pressures. To do this I artificially elevated ICP using intraventricular infusion of aCSF in a stepwise manner via continuous infusion (see Table 3.6 for flow rates). Pressure was measured simultaneously in the epidural, subdural and intraventricular spaces using fibreoptic pressure sensors. Flow rates began at 4 µL/ minute and increased as necessary to rates that would sustain elevated ICP of approximately 20 mmHg or higher. Each flow rate was maintained for 10 minutes to allow for pressures to plateau Figure 3.14. ICP as well as MAP, SpO2, RR, and HR were assessed every 2.5 minutes during infusion.

66

aCSF Infusion Rates (µL/ min) ∆IVP mmHg (#no. animals) Mean (± SD) Range 0.1 - 5 (n = 11) 5.4 ± 2.7 0 - 12 5 - 10 (n = 11) 15.1 ± 8 4 - 36 10 - 15 (n = 11) 24.6 ± 13.1 8 - 68 15 - 20 (n = 10) 33.8 ± 13.9 12 - 68 20 - 25 (n = 8) 37.3 ± 13.9 16 - 64 25 - 30 (n = 6) 36.8 ± 12 20 - 56 >30 ( n = 1) 24 24

Table 3.6: Rates of artificial cerebrospinal fluid (aCSF) infusion required to produce intraventricular pressure (IVP) elevation.

67

Figure 3.14: Increase in intracranial pressure (ICP) in response to artificial cerebrospinal fluid (aCSF) infusion. Representative ICP traces showing epidural, subdural and intracerebroventricular pressure during artificial elevation of pressure. ICP was increased by infusion of aCSF into the left lateral ventricle of male, outbred Wistar rats (n = 11). ICP was measured simultaneously by probes implanted in the epidural (red trace), subdural (blue trace) and intraventricular space (black signal). The infusion rate was increased stepwise from 4 µL/ minute to 24 µL/ minute.

68 3.5.2 Results

Exclusions

The epidural pressure measurements of two animals were excluded due to technical failures. The subdural and intraventricular pressure measurements of the animals were not affected, therefore they were included in this analysis.

Baseline ICP

Mean ICP (± SD) was similar across all of the different probe locations; epidural 4.3 ± 1.8 mmHg, subdural 5.3 ± 1.4 mmHg, intraventricular 5.5 ± 2.2 mmHg (p=.3; Figure 3.15, Table 3.5). Physiological variables at baseline for all animals are listed in Table 3.4.

Correlation between epidural, subdural and intraventricular ICP

Absolute epidural ICP correlated well with both subdural rc = 0.89 (95% CI, 0.85-0.93) and intraventricular pressures rc = 0.81 (95% CI, 0.75-0.87). Subdural pressure also correlated well with intraventricular pressure rc = 0.84 (95% CI, 0.8-0.9) (Figure 3.16). Correlations improved when they were calculated based on relative changes (D baseline) in ICP. Relative epidural pressure correlated well with both subdural rc = 0.9 (95% CI, 0.86-0.94) and intraventricular pressure rc = 0.85 (95% CI, 0.8-0.9). Subdural pressure also correlated well with intraventricular pressure rc = 0.87 (95% CI, 0.83-0.91) (figure 3.16).

10

8

6

4 ICP (mmHg) ICP 2

0 ED SD IVP Probe Location

Figure 3.15. Baseline intracranial pressure (ICP) as measured at different locations within the intracranial compartment. a. Mean (±SD) baseline ICP as measured simultaneously in the epidural (ED; red), subdural (SD; blue) and intraventricular (IVP, black) space of all animals in both groups. There were no significant differences in baseline ICP based on probe location (p=.3).

69

a. b. 40 30 ρc = 0.81, p<.0001 ρc = 0.85, p<.0001

30 20

20

10 10 Epidural ICP (mmHg) Epidural ICP

0 Epidural Pressure (mmHg) 0 Δ 0 10 20 30 40 0 10 20 30 40 Intraventricular ICP (mmHg) Δ Intraventricular Pressure (mmHg)

c. d. 40 30 ρc = 0.89, p<.0001 ρc = 0.9, p<.0001

30 20

20

10 10 Epidural ICP (mmHg) Epidural ICP

0 Epidural Pressure (mmHg) 0 Δ 0 10 20 30 40 0 10 20 30 40 Subdural ICP (mmHg) Δ Subdural Pressure (mmHg)

e. f.

40 ρc = 0.84, p<.0001 30 ρc = 0.87, p<.0001

30 20

20

10 10 Subdural ICP (mmHg) Subdural ICP

0 Subdural Pressure (mmHg) 0 0 10 20 30 40 Δ 0 10 20 30 40 Intraventricular ICP (mmHg) Δ Intraventricular Pressure (mmHg)

Figure 3.16: Concordance correlation coefficient (rc) of absolute and relative (D) changes in intracranial pressure (ICP) between different locations within the rat brain. ICP was simultaneously recorded in the epidural, subdural and intraventricular space of male, outbred, Wistar rats (n = 11). ICP was artificially elevated by the infusion of artificial cerebrospinal fluid. These graphs show the concordance between: epidural and intraventricular a) absolute and b) relative pressures; epidural and subdural c) absolute and d) relative pressures; and subdural and intraventricular e) absolute and f) relative pressures. Graph made with GraphPad Prism 7.

70 3.5.3 Discussion

There was no difference in baseline ICP when comparing absolute values between locations. Pressure correlated well between all locations both in terms of absolute and relative pressure measurement across a wide range of pressures. Intraventricular infusion of aCSF resulted in intraventricular and subdural pressure elevation in all animals, but surprisingly, there were 2 outlier epidural pressure responses to infusion. In one animal, epidural ICP was not affected at all by aCSF infusion. In another animal, epidural ICP initially responded very similarly to intraventricular and subdural pressures. However, when the flow rate was increased to 12 µL/minute epidural pressure plateaued, and despite increasing the flow rate epidural pressure did not increase further. Subdural and intraventricular pressure continued to increase with increasing flow rate in this animal. In the case of former it is likely that an ‘artificial seal’ was created during screw insertion, by incomplete clearance of the skull in the burr hole, or adhesive or dental cement entering the hole whilst screws were being sealed. This would have compromised the ability of the pressure sensor to detect changes in pressure. As for the latter outlier, the most likely cause of the epidural ICP plateau mid-way through infusion, despite initially increasing with both subdural and intraventricular ICPs is that there was a leak that occurred during the experimental process. From previous experience, when there is a leak in the system, the ICP trace will tend to drift negatively (unpublished observations). The most common cause of this is the improper application of Silagum. Usually this will be noticed at baseline and rectified by applying an additional layer of Silagum. As there was no negative drift detected at baseline, it is likely that the leak occurred during infusion as a result of the artificially increased pressure in the system. The increasing pressure caused by the infusion would have masked any negative drift that may have occurred due to the formation of the leak and thus, resulted in the observed plateau. This is also supported by the data from Part 5 of this chapter – a reliable ICP trace could not be obtained when probes were reinserted and the data point had to be excluded. Importantly, both probes displayed an ICP signal despite no change in epidural pressure in response to aCSF infusion. This suggests that assessing the suitability of a method for ICP measurement at baseline does not necessarily equate with accuracy and reliability of measurements at higher pressures. Furthermore, the outliers highlight the importance of adequate preparation of the skull and an air tight seal.

Interestingly, divergence from intraventricular pressure was noted with both epidural and subdural pressure at higher pressures. The regression slopes comparing epidural and subdural ICP with intraventricular ICP were less than 1, suggesting that both epidural and subdural ICP slightly underestimated the pressure at higher pressures (as measured intraventricularly). The exact cause of this is unknown, however I hypothesise that the

71 intraventricular infusion itself caused a transcerebral/ transparenchymal pressure gradient, whereby pressure was higher in the ventricles than at the brain periphery. Pressure gradients have been proposed in various models of disease particularly those involving ventriculomegaly (223). There is little evidence supporting the existence of pressure gradients in healthy rats. Furthermore, there was no evidence of a pressure gradient in the baseline ICP values. ICP was increased by infusing aCSF into the ventricular system to create a range of ICP values. The flow rate of aCSF infused into the ventricular system was up to 440% higher than the normal flow rate of CSF through the cerebral aqueduct (3.0 ± 1.5 µL/ minute (99)). It is theoretically possible then, that as the flow rate increased so too did the pressure within the ventricle causing an artificial pressure gradient. As our measure of ‘true ICP’ was intraventricular pressure it is therefore likely that this is what caused the divergence of absolute and relative values at the higher pressures/ flow rates. The exact contribution of flow rate is difficult to determine as individual animals reacted very differently to aCSF infusion (max flow rates 20 µL/ minute to 68 µL/ minute). Flow rate was simply used as a tool to reach a target intraventricular ICP. A concurrent study from our group, found excellent correlation between intraventricular pressure and both epidural and subdural pressures over a range of pressures after stroke (unpublished results. A. Patabendige, 2017). Therefore, it is likely that the method of ICP elevation resulted in the creation of an artificial pressure gradient. A method to assess this hypothesis would be to use intrathecal infusion of aCSF to avoid directly increasing pressure or volume within the brain by intraventricular infusion.

There are many advantages of measuring epidural pressure over other locations including ease of placement, it is the least invasive location and is associated with lower rates of complication than other, more invasive locations of pressure measurement including the intraparenchymal and intraventricular spaces. The results of this study contribute to existing findings showing good correlation between epidural pressure and intraventricular (36) or intraparenchymal pressure (60) using fluid-filled catheters in rats. This study expands on the previous findings by showing good correlation using fibreoptic pressure sensors which as previously discussed, have many advantages over fluid filled technology (Chapter 1). Furthermore, I have shown good correlation over a range of pressures. There have been no studies validating subdural pressure measurement in rats, however the similarities observed with pressure measured epidurally are not surprising considering the thickness of the dura mater in rats (~80 µm (68)). In this study, epidural ICP was not observed to be systematically higher than pressure measured either subdurally or intraventricularly. Clinically, epidural ICP has been found to overestimate pressure when compared with intraventricular (224, 225), intraparenchymal (57, 224) and even subdural ICP (226). In humans, this is likely due to the

72 thickness and mechanical qualities of the dura. This is less likely to be a problem in rats because the dura is far thinner (68).

The method of ICP elevation is both a strength and a weakness of this study. By infusing aCSF at increasing flow rates we were able to create a range of controllable pressure values to test the accuracy of epidural and subdural ICP. Previous investigations have assessed pressure at baseline/ resting ICP and used either short bursts of increased pressure from abdominal compressions (36) or induced neurological disease (36, 60) to show a responsiveness to changes in pressure which are more difficult to control. It is likely that our method itself slightly overestimated the differences between measurement locations seen at high ICPs by causing an artificial pressure gradient between intraventricular and both epidural and subdural measurement locations. Furthermore, inter-animal variability in responses to ICP infusion caused variability between the maximum ICP that we were able to reach and the maximum flow rate required. It is likely that this was influenced by the CSF outflow resistance of the individual animals. Ideally, our stepwise infusion approach would have permitted the calculation of CSF outflow resistance, however the presence of stable plateaus in each animals ICP trace was not often easy to determine. Such plateaus are required to calculate outflow resistance. In fact, in a number of animals I was unable to generate change in pressure despite increasingly high flow rates, which contributed to the variability in maximum flow rate.

In the current study I have shown that epidural and subdural pressure measurement in rats is both accurate and reliable over a range of pressures using fibre optic pressure sensing technology. My results validate the use of epidural pressure measurement in future studies, given that it correlates well with the gold standard of intraventricular pressure measurement.

73 3.6 Part 5: Accuracy of serial measurement of epidural and subdural pressure when ICP Is high

Measurements of ICP within our laboratory are taken at baseline and again at 16-24 hours post-surgery/ intervention. The 16-24 hour time points are used in order to capture any changes in ICP when we expect ICP to be elevated. However, we cannot be sure if ICP will already be elevated when we begin monitoring during this period. Therefore, I wanted to ascertain whether our epidural ICP measurements remained accurate when ICP is already high. In this study I aimed to determine the accuracy of repeated measures analysis when ICP is already high.

3.6.1 Experimental Design

To assess accuracy of serial pressure measurement when ICP is already high, ICP was artificially elevated via infusion of aCSF (in the same cohort of animals from Parts 3 and 4 of this chapter). Epidural, subdural and intraventricular pressure was recorded simultaneously during artificial ICP elevation. Flow rate was adjusted to achieve a plateau close to 20 mmHg (as measured intraventricularly). Once ICP had plateaued, both the epidural and subdural probes were removed whilst the intraventricular probe remained in place and continued to record ICP, in part to ensure that the process of removing the other catheters did not itself alter the ICP. The epidural and subdural probes were sequentially re-introduced (see Figure 3.13 for timeline). Originally, I planned to compare both epidural and subdural pressure values with intraventricular pressure. However, as noted in Part 4 of this chapter, as intraventricular pressure increased so too did divergence of both epidural and subdural pressures from intraventricular pressure. Therefore, intraventricular pressures at reinsertion were matched with the same value recorded during continuous infusion, and then the corresponding epidural/ subdural pressures at that time point were compared with the reinsertion value. Values that could not be matched where the difference between continuous and reinsertion intraventricular pressure were <-2 or >2 were not included in this analysis.

3.6.2 Results

Two animals were excluded from this analysis as they did not have matched intraventricular pressure values. A subdural pressure data point is absent because the probe became damaged during removal after the continuous infusion and so could not measure a reinsertion value. Physiological parameters measured at matched intraventricular pressures are presented in Table 3.4.

74 Absolute values correlated exceptionally well between epidural pressure measured continuously and at reinsertion (rc 0.96 (95%CI, 0.91-1.01); Figure 3.17a). Similarly, there was good correlation between the continuous and reinsertion subdural pressures (rc 0.87 (95%CI, 0.67–1.06); Figure 3.17b).

Mean physiological parameters (± SD) Continuous Reinsertion SpO2 (%) 95.7 ± 3.7 94.9 ± 3.3 RR (BPM) 65.7 ± 7.8 58.6 ± 13 HR (BPM) 378 ± 24 398 ± 22.5 MAP (mmHg) 99.7 ± 10.8 108.3 ± 12.5

Table 3.4. Mean physiological parameters at matched intraventricular pressures when intracranial pressure (ICP) was measured continuously vs serially. Abbreviations: SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute; MAP, mean arterial pressure.

a. b. 40 40 ρc = 0.96, p<0.0001 ρc = 0.87, p<0.0001

30 30

20 20 (Reinsertion) (Reinsertion) 10 10 Epidural ICP (mmHg) Epidural ICP Subdural ICP (mmHg) Subdural ICP 0 0 0 10 20 30 40 0 10 20 30 40 Epidural ICP (mmHg) Subdural ICP (mmHg) Continuous Continuous

Figure 3.17: Concordance correlation coefficient (rc) of matched pairs of pressure measurements. ICP was recorded simultaneously in the intraventricular (IVP), a) epidural and b) subdural spaces of male, outbred Wistar rats (n = 9) whilst pressure was artificially elevated by infusion of artificial cerebrospinal fluid. Pressure was measured continuously in all locations from baseline to ~20 mmHg (as measured at IVP). Once IVP had plateaued at ~20 mmHg, epidural and subdural probes were removed and reinserted to assess their accuracy whilst intraventricular infusion was continued at the same rate. Each symbol represents a matched pair. Graph made with GraphPad Prism 7.

75 3.6.3 Discussion

When investigating the accuracy of epidural and subdural pressure measurement across serial measures, I found that both locations of measurement showed excellent correlation between continuous and reinsertion pressure measurements at known elevated intraventricular pressures. This finding confirms the reliability of serial measurement of epidural and subdural pressure at high ICP. There is little previous literature that is directly comparable with these studies. In the few investigations that have confirmed the accuracy of ICP, investigators have assumed the accuracy of epidural ICP at higher values based on the correlation between epidural pressure and pressure measured either intraparenchymally or intraventricularly (36, 60). However, in both of these cases ICP was high as a result of disease prior to pressure assessment therefore the accuracy of absolute measurements was just assumed. In this study, continuous infusion of aCSF was used to achieve a known ICP elevation at levels similar to that shown post-stroke whilst measuring ICP simultaneously in the epidural, subdural and intraventricular spaces. I ensured ICP was maintained at this level by leaving the intraventricular catheter in place as the epidural and subdural catheters were removed and reinserted. These findings demonstrate that epidural ICP is a reliable location for pressure measurement when measured serially. The results further validate the use of epidural ICP measurement.

3.7 Summary

In summary, the results presented in this chapter have confirmed the suitability of epidural pressure measurement in rats. Substantial troubleshooting highlighted the importance of probe design for accurate and reliable epidural ICP measurement. From this work, I identified a robust and reliable fibreoptic pressure sensor for ICP monitoring in rats (Chapter 3: Part 1). I then assessed the accuracy and reliability of epidural pressure measurement in rats using this technology. Dural contact was identified as potential source of variability in baseline ICP (Chapter 3: Part 2). I also found there was no need for a recovery period between surgery and pressure measurement (Chapter 3: Part 3). I showed epidural pressure measurement using fibre optic pressure sensors to be as accurate as both subdural and intraventricular pressure over a range of pressures (Chapter 3: Part 4) and serial measurement at these locations is reliable at ‘pathological’ ICPs (Chapter 3: Part 5). The technologies and methods presented in this chapter are now standard in our laboratory and were used to assess ICP throughout the experiments presented in the remainder of this thesis.

76 Chapter 4 The effect of arginine vasopressin on intracranial pressure

4.1 Introduction

Previous studies from our laboratory provided evidence to suggest that a molecule in CSF causes delayed ICP elevation post-stroke in rats. Whilst, there are many possible candidates, one of the most promising is AVP. AVP is present in the CSF of both humans (195, 211-213, 227) and animals (211, 214, 215). CSF AVP concentration is increased in numerous neurological conditions involving raised ICP, including stroke, hydrocephalus, intracranial haemorrhage, idiopathic intracranial hypertension and intracranial tumours (194, 195). There have been no direct studies investigating the relationship between post-stroke CSF AVP concentration and ICP. However, indirect evidence from effects of AVP on post-ischaemic cerebral oedema and brain water permeability supports the hypothesis that AVP may influence ICP post-stroke (217).

Sorensen et al. (1984) showed a direct correlation between AVP concentration and ICP (195). Currently, it is unclear whether AVP is a cause or consequence of raised ICP. There have been a number of attempts to elucidate this cause and effect relationship via infusion of AVP into CSF, however results have been inconsistent. AVP infusion has been found to increase (197, 198), decrease (209, 210) or have no effect on ICP (196, 198), both during and immediately after intraventricular infusion. Methodological differences between studies including species, dose and mode of administration, make results difficult to interpret. Mode of administration i.e. bolus vs continuous infusion of AVP is thought to influence the ICP response. Bolus injection is typically associated with a reduction in ICP whilst continuously infused AVP is typically associated with ICP elevation. However, both delivery methods have also shown no effect on ICP. Therefore, further investigation is required to determine the effects of AVP infusion on ICP.

The previous investigations have examined the short term effects of AVP on ICP (max time 150 minutes post infusion). There have been no studies investigating delayed effects of AVP infusion on ICP elevation. Therefore, it is currently unknown whether AVP exerts any delayed effect on ICP similar to that previously observed in our laboratory following stroke (80-83). Indirect evidence from AVP receptor blockade studies (using V1a receptor antagonists administered intraventricularly at the time of stroke) have shown attenuated oedema volumes, increased aquaporin 4 (AQP4) expression and a reduction in blood-brain-barrier permeability out to 24 hours post-infusion (204, 217). Therefore, I hypothesised that AVP infusion would exert delayed effects over ICP. The aim of this pilot experiment was to investigate whether

77 intracerebroventricular infusion of AVP causes acute (0-80 minutes) and/ or delayed (16-24 hours) ICP elevation.

4.2 Experimental Design

This experiment was performed to establish the effects of AVP infusion on ICP. Male, outbred, Wistar (n = 21) and Sprague Dawley (n = 9) rats (weight 255-338 g) underwent baseline recording of epidural ICP before continuous intracerebroventricular infusion of 0.02 ng, (n = 12), 0.1 ng (n = 5), or 5 ng (n = 4) AVP in 200 µL aCSF, or vehicle control (n = 9) at 4 µL/ minute for 50 minutes (total volume 200 µL). Continuous infusion of AVP was chosen as the mode of delivery as it has previously been shown to cause ICP elevation (197, 198). In all experiments, acute (during infusion- to 30 minutes post-infusion) and delayed (16-24 hours post-infusion) effects of continuous AVP infusion on ICP were measured. In addition to ICP,

SpO2, RR, HR and MAP were monitored simultaneously (see Figure 4.1 for timeline). To minimise the effects of regression to the mean, caused by within animal variability of ICP over time, peak DICP was determined in individual animals and analysed along with the corresponding SpO2, RR, HR and MAP values.

Figure 4.1: Experimental timeline. Animals underwent surgery for mean arterial pressure (MAP) and intracranial pressure (ICP) measurement. Baseline ICP (B) was established before intraventricular infusion of AVP or vehicle control (4µL/ minute for 50 minutes). At 2 hours post-infusion animals were recovered overnight and at 16 hours post-infusion animals were re-anaesthetised and both ICP and MAP were measured between 16-24 hours post-infusion. At 24 hours post-infusion animals were sacrificed.

78 4.3 Results

4.3.1 Exclusions

In total, 14 animals were excluded from the results of this study. Thirteen of these exclusions were made due to unreliable pressure measurement with Fiso catheters (see Chapter 3 for more information) (n = 4, outbred Wistar (0.02 ng AVP), n = 9, Sprague Dawley (n = 4, 0.02 ng AVP; n = 5, vehicle control)). The remaining Wistar rat was excluded after suffering a pulmonary embolism during recovery (0.1 ng AVP recipient).

Of the included animals, a 5 ng recipient was euthanised at approx. 20 hours post-infusion due to difficulties keeping physiological variables stable under anaesthetic. However, there was nothing to suggest that results up until this point had been affected therefore the animal’s data was included in the results.

4.3.2 Physiological Parameters

Physiological parameters measured at baseline, during and post-infusion are presented in Table 4.1. Physiological parameters at peak DICP (0-80 minutes, and 16-24 hours post- infusion) are presented in Table 4.2.

4.3.3 Acute effects of AVP infusion on ICP (0-80 minutes post-infusion)

Mean baseline ICP across all groups was 5.5 ± 1.4 mmHg (Table 4.1). Mean peak DICP was not significantly higher than control in any AVP group between 0-80 minutes post infusion (Table 4.3; p>.2).

Area under the curve (AUC) (DICP (mmHg) vs time (hours)) was calculated from the start of infusion, during infusion (50 minutes), to 30 minutes post-infusion (80 minutes) (see Figure 4.2a-d for individual ICP profiles). Mean AUC was not significantly different between any dosage and control (0.02 ng: 1.9 ± 0.7 mmHg/ hour; 0.1 ng: 7.4 ± 5.5 mmHg/ hour; 5 ng: 3.1 ± 0.5 mmHg/ hour; control: 3.2 ± 3.2 mmHg/ hour; p=.2; Figure 4.2e).

79 Time post-start of infusion (mean ± SD) Baseline 0.5 h 1 h 16 h 20 h 24 h Absolute ICP (mmHg) 0.02 ng 4.9 ± 1.2 6.9 ± 1.8 5.9 ± 1.2 8 ± 2.3 6.7 ± 1.4 7.7 ± 3.2 0.1 ng 5.2 ± 1.3 13.3 ± 5.5 9.8 ± 4.1 7.6 ± 4.2 5.9 ± 2.3 4.4 5 ng 5.8 ± 2.4 9.2 ± 3.2 7.1 ± 2.1 9.4 9.3 ± 4.8 Control 5.9 ± 0.7 8.8 ± 2.6 7.5 ± 2.8 6.1 ± 1.5 5.7 ± 1.3 6.0 ± 2.4 MAP (mmHg) 0.02 ng 100.5 ± 7.3 100.3 ± 9.3 101.6 ± 8.4 88.3 ± 2.1 95.4 ± 6.4 100.1 ± 6.9 0.1 ng 103.9 ± 9.9 105.7 ± 16.3 106.3 ± 14.6 84.9 ± 4.2 95.4 ± 12 92 5 ng 96.2 ± 5.9 99.2 ± 3.7 100.2 ± 7.6 93 103 ± 18.5 Control 105.6 ± 3.2 114.9 ± 8.5 117.8 ± 11.9 93.5 ± 9.1 99.5 ± 4.4 106.3 ± 6.7 SpO2 (%) 0.02 ng 96.3 ± 2.8 96.5 ± 3.1 96.8 ± 2.6 99 ± 1.7 98.3 ± 2.4 98.3 ± 1.7 0.1 ng 97 ± 2 96 ± 2.4 95.3 ± 2.2 98.5 ± 2.1 97.5 ± 3.1 100 5 ng 97.2 ± 1.7 96.4 ± 2.2 96 ± 1.3 100 100 Control 94.8 ± 2.2 96.5 ± 2.4 97 ± 2 100 99 ± 1.2 99 ± 1.4 RR (BPM) 0.02 ng 64 ± 1.7 62 ± 1.9 57 ± 2.8 59 ± 3.5 64 ± 6.9 62 ± 3.4 0.1 ng 67 ± 7.8 65 ± 1.7 67 ± 4.1 63 ± 3.5 64 ± 5.4 53 5 ng 65 ± 4.1 66 ± 7.2 61 ± 7.3 69 66 ± 7.8 Control 61 ± 6.7 63 ± 6.3 62 ± 11.8 64 ± 15.5 65 ± 6.2 56 ± 3.5 HR (BPM) 0.02 ng 426.5 ± 17 418.1 ± 25.3 397.6 ± 19.9 425.5 ± 18.1 418.2 ± 18.7 417.2 ± 11.5 0.1 ng 446.3 ± 31.1 440.3 ± 35.6 430.9 ± 39.2 395.8 ± 8.2 400 ± 36.3 384.2 5 ng 431 ±10.2 425.1 ± 7.2 404.9 ± 15.4 418.4 406.3 ± 20.3 Control 429.3 ± 33.7 432 ± 42.1 434.8 ± 55.6 414.5 ± 65.5 418.4 ± 11.2 408.3 ± 1.6

Table 4.1: Mean physiological parameters before, during and after intracerebroventricular infusion of arginine vasopressin (0.02, 0.1 or 5 ng) or vehicle control (n = 4/ group). Abbreviations: ICP, intracranial pressure; MAP, mean arterial pressure; SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute.

80

Physiological parameters at peak D ICP (mean ± SD) 0-80 minutes 16-24 hours Absolute ICP (mmHg) 0.02 ng 7.4 ± 2.4 9.9 ± 2.8 0.1 ng 14.6 ± 6.1 7.3 ± 2.4 5 ng 10.22 ± 3.5 10.9 ± 5.1 Control 10.5 ± 4.7 8.1 ± 0.9 BP (mmHg) 0.02 ng 100.4 ± 9.1 95.4 ± 11.1 0.1 ng 109.5 ± 22.2 91.3 ± 6.2 5 ng 99.3 ± 9.4 100.8 ± 6.1 Control 117.4 ± 12.2 96.8 ± 10.8

SpO2 (%) 0.02 ng 96.8 ± 3.2 99.3 ± 1.5 0.1 ng 96.5 ± 2.7 98.3 ± 2.4 5 ng 96.7 ± 2.1 99.8 ± 1.1 Control 95.8 ± 1.5 99.8 ± 0.5 RR (BPM) 0.02 ng 60 ± 1.6 64 ± 12.1 0.1 ng 67.5 ± 8.5 66 ± 9.7 5 ng 64.3 ± 4.8 70 ± 7.3 Control 62.3 ± 5.1 58 ± 7.8 HR (BPM) 0.02 ng 417 ± 20.4 429.3 ± 17.3 0.1 ng 442 ± 37.6 413 ± 50.9 5 ng 421.7 ± 14.4 421.6 ± 17.1 Control 436.1 ± 41.8 396.3 ± 31.6

Table 4.2: Mean physiological parameters at peak intracranial pressure (ICP) above baseline (DICP) between 0-80 minutes and 16-24 hours post-infusion of arginine vasopressin (AVP). Animals underwent intracerebroventricular infusion of AVP (0.02, 0.1 or 5 ng) or vehicle control (n = 4/ group). Abbreviations: ICP, intracranial pressure; MAP, mean arterial pressure; SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute.

Mean Peak D ICP (± SD) 0-80 minutes 16-24 hours 0.02 ng 2.5 ± 1.3 5.1 ± 3.1 0.1 ng 9.4 ± 6.4 2.1 ± 3.1 5 ng 4.4 ± 1.9 5.1 ± 4.8 Control 4.5 ± 4.6 2.2 ± 1.2

Table 4.3: Mean peak intracranial pressure above baseline (D ICP) between 0-80 minutes and 16-24 hours post-infusion of arginine vasopressin (AVP). Animals underwent intracerebroventricular infusion of AVP (0.02, 0.1 or 5 ng) or vehicle control (n = 4/ group).

81 a. b. 20 20

15 15

10 10

5 5 ICP (mmHg) ICP (mmHg) ICP Δ Δ 0 0

-5 -5 B 0 0.5 1.0 1.5 B 0 0.5 1.0 1.5 Time (h) Time (h)

c. d. 20 20

15 15

10 10

5 5 ICP (mmHg) ICP ICP (mmHg) ICP Δ Δ 0 0

-5 -5 B 0 0.5 1.0 1.5 B 0 0.5 1.0 1.5 Time (h) Time (h)

e. 15

10

5 AUC (mmHg/ h)

0 Control 0.02 ng 0.1 ng 5 ng (aCSF)

Figure 4.2: Effects of intraventricular infusion of arginine vasopressin (AVP) on intracranial pressure (ICP). Graphs showing acute changes in ICP over time (baseline-80 minutes post-infusion) in all included animals. ICP was measured in animals infused with a. artificial cerebrospinal fluid (aCSF; control), b. 0.02 ng, c. 0.1 ng, or d. 5 ng AVP (shaded area = infusion). e. Bar graph showing the mean area under the curve (AUC) of the change in pressure (DICP mmHg) over time (0-80 minutes post-start of infusion) for all groups shown in a-d. No significant differences were found between any dose and control (p>.05).

82 4.3.4 Delayed effects of AVP infusion on ICP (16-24 h post-infusion)

Mean peak DICP was variable between individual animals of the three AVP dose groups, however was not significantly higher than control in any AVP group between 16-24 hours post infusion (Table 4.3; p>.3). The timing of peak ICP varied between animals within each group (range 0.02 ng: 16-23.5 hours; 0.1 ng: 16.5-21 hours, 5 ng: 17-23.5 hours; control: 16-23.5 hours post-infusion).

AUC was variable between animals at each dose (see Figure 4.3a-d for individual ICP profiles). Due to an incomplete data set (early euthanasia of an animal in the 5 ng group) AUC was calculated between 17 and 20.5 hours. Mean AUC was not significantly different between any AVP dosage and control (0.02 ng: 20.7 ± 24.8 mmHg/ hour; 0.1 ng: 22.7 ± 24.9 mmHg/ hour; 5 ng 41 ± 48 mmHg/ hour; control: 14.5 ± 11.8 mmHg/ hour; p=.8; figure 4.3e). The results did not change when mean AUC was calculated between 16.5 and 23.5 hours in the 0.02 ng, 0.1 ng and control groups (p=.7).

83 a. b. 15 15

10 10

5 5 ICP (mmHg) ICP (mmHg) ICP

Δ 0 Δ 0

-5 -5 B 16 18 20 22 24 B 16 18 20 22 24 Time (h) Time (h)

c. d. 15 15

10 10

5 5 ICP (mmHg) ICP ICP (mmHg) ICP Δ 0 Δ 0

-5 -5 B 16 18 20 22 24 B 16 18 20 22 24 Time (h) Time (h)

e. 100

80

60

40

AUC (mmHg/ h) 20

0 Control 0.02 ng 0.1 ng 5 ng (aCSF)

Figure 4.3: Effects of intraventricular infusion of arginine vasopressin (AVP) on intracranial pressure (ICP). Graphs showing delayed changes in ICP over time (baseline and between 16-24 hours post-infusion) in all included animals. ICP was measured in animals infused with a. artificial cerebrospinal fluid (aCSF; control), b. 0.02 ng, c. 0.1 ng, or d. 5 ng AVP. e. Bar graph showing the mean area under the curve (AUC) of the change in pressure (DICP mmHg) over time (17-20.5 hours post-start of infusion) for all groups shown in a-d. No significant differences were found between any dose and control (p>.05).

84 4.4 Discussion

The aim of this pilot study was to assess acute (0-80 minutes) and delayed (16-24 hours) effects of intraventricular AVP infusion on ICP elevation. Baseline ICP across all groups was within the range reported by us and others (36, 39, 80-83). No significant differences in mean peak DICP (Table 4.3), nor AUC (Figures 4.2e and 4.3e) were observed between control and any AVP treatment group between 0-80 minutes or 16-24 hours post-infusion. Furthermore, there was no evidence of a dose-response relationship between AVP and ICP elevation. The results from within this dataset (n = 4/ group) indicate that it is unlikely that AVP infusion causes dramatic acute (0-80 minutes) or delayed (16-24 hours) ICP elevation in rats.

In this study, AVP infusion did not cause mean peak ICP elevation significantly different to controls either during or immediately post-infusion, or between 16-24 hours post infusion. The results of this study are supported by two previous studies reporting that AVP does not exert acute effects over ICP (196, 198). However, the lack of any consistent effects of AVP infusion on ICP contrasts with previous studies reporting ICP lowering (209, 210) or elevating effects (197, 198) either during- or immediately post-AVP infusion. Methodological differences between studies make the results different to interpret. The most obvious difference between this study and those reporting decreased ICP post-infusion is the mode of delivery of AVP. In this study, AVP was infused slowly into the ventricular system via continuous infusion. In contrast, the two studies reporting a decrease in ICP post-infusion delivered AVP via a bolus injection (209, 210). Both studies speculate the potential mechanisms underlying the observed ICP reductions including increased CSF outflow (210) and reduced cerebral blood volume (209). However, neither hypothesis accounts for the lack of a response to bolus infusion reported by Sorensen et al. (1990) (198). Furthermore, these studies require careful interpretation as they both used saline as their control and for their vehicle (as opposed to aCSF), showed arguably biologically insignificant ICP reductions of <2 mmHg and neither showed a dose response relationship.

Importantly, the results in this study also differed from previous continuous infusion studies showing ICP elevation post-infusion. However, similarly to papers observing decreased ICP in response to AVP infusion, there are methodological differences to consider, one major difference being the length of time between surgical preparation and AVP administration. In the current study AVP was infused on the same day as surgery, whilst the earliest infusion time post-intracerebroventricular cannulation in the previous studies was 48 hours (the maximum time was >2 weeks post-cannula implantation). The ICP response to AVP infusion in these studies increased as time post-cannulation increased. As previously discussed, permanent ventricular cannulation can cause disturbances in the ventricular system, which

85 have been shown to induce hydrocephalus and influence ICP (36). Additionally, AVP expression has been shown to be significantly higher in patients with hydrocephalus (194, 213). Although the significance of these findings is yet to be understood, if hydrocephalus had been induced in the cannulated animals and hydrocephalus increases AVP expression, it is possible that this may alter the sensitivity of the intraventricular system to AVP thereby confounding the results.

There have been no studies to date that investigate delayed effects of intracerebroventricular AVP infusion on ICP. The current study was conceived around the idea that increased AVP expression in post-stroke CSF may contribute to the ICP rise observed in post-stroke CSF recipient animals between 16-24 hours post-transfusion. Even at the supramaximal dose (5 ng) mean peak ICP elevation in this study was much lower than what was observed in the previous CSF transfusion study (mean peak ICP 26.3 ± 12.3 mmHg (unpublished results, R. Hood, D. D. McLeod)). This suggests that there are other factors within post-stroke CSF that contribute to the observed ICP elevation in the CSF recipients. Whilst the current results do not rule out the possibility that AVP is involved in post-stroke ICP elevation, they do suggest that AVP is not solely responsible for the rise.

This study has several strengths. Firstly, we investigated a large range of doses of AVP up to and over levels likely to occur after stroke. The dose range of AVP appeared to be safe and tolerable when infused intracerebroventricularly in the current study. The supramaximal dose of 5 ng AVP equates to an approximate CSF concentration of 12.5 ng/ mL (assuming CSF volume is roughly 200 µL in rats (99)) which is substantially higher than the reported physiological levels within CSF (3.8 to 18.5 pg /mL (211, 214, 215, 228)). There are no studies to date measuring post-stroke AVP concentration in CSF of rats however in human studies post-stroke CSF AVP levels are approximately double that of control values (194). If this is true for rats, then the estimated final concentration would similarly be substantially higher than levels that would be expected post-stroke. Therefore, if post-stroke CSF AVP levels were solely responsible for ICP elevation, we should have been able to demonstrate this in the current study. Secondly, AVP was infused on the same day as temporary ventricular cannulation and ICP measured within 24 hours. It is therefore unlikely, that our method of infusion nor our method of ICP measurement would have significantly affected our results. Thirdly, although it was not measured in this study, continuous infusion of AVP permits greater distribution within the CSF system than bolus infusion (198). This would likely permit AVP to reach any site(s) of action. Finally, analysing peak ICP (regardless of the time of peak) prevented individual responses to AVP infusion from regressing to the mean at different time points.

86 A limitation of this study is that it is underpowered to exclude small to moderate effects of AVP on ICP. Therefore, the results require careful interpretation. There was no statistically significant evidence to suggest a relationship between AVP infusion and ICP, however, this does not mean that one does not exist. It was hypothesised that AVP may be one of the molecules involved in the delayed ICP elevation previously observed in stroke rats and in naïve recipients of stroke rat CSF. Even at the supramaximal AVP dose the effect size observed in this study (peak ICP 2.8 ± 2.6 mmHg above controls) was very modest compared with that observed in the original CSF transfusion study (peak ICP ~12 mmHg above controls). We were looking for a dramatic effect of AVP infusion on ICP elevation. Due to the moderate effect size within this data, coupled with the lack of a dose response, it was decided the study would not be continued past 4 animals/ group. Another limitation of this study is that ICP was not continuously monitored as we wanted to limit the effects of prolonged anaesthesia on both days of the experiment. Therefore, any ICP changes that occurred between monitoring periods were not recorded. It is possible, yet unlikely that ICP elevation occurred between the two monitoring periods as the literature suggests that it either shows immediate or delayed effects on pressure. Furthermore, in our original CSF transfusion assay the ICP changes we observed occurred between 16-24 hours post-infusion. Therefore, if AVP was solely responsible for the ICP rise observed in the stroke CSF recipients it would be expected that the AVP recipient animals in this study would show evidence of ICP elevation within this time window. Future studies could confirm this using implanted/telemetered ICP sensors which can be used to measure pressure in awake animals. Unfortunately, there are currently no implantable solid state sensors that are small enough to measure epidural pressure in rats. It should also be noted that there were two strains of rat used in this study. As explained in Chapter 2: Methods (Section 2.4), the different strains were chosen during troubleshooting. Unfortunately, I had to exclude the Sprague Dawley AVP trials, therefore any influence that the different strains may have had on the results is unknown.

In conclusion, within the dataset (n = 4/ group) it was found that AVP infusion did not significantly increase ICP between 0-80 minutes or 16-24 hours post-infusion. Coupled with previous findings, this suggests that AVP is not likely to be solely responsible for the delayed ICP elevation observed after CSF transfusion from stroke rats to naïve recipients or in stroke animals.

87 Chapter 5 Human CSF Infusion

5.1 Introduction

Our group have discovered a delayed ICP rise following mild-moderate stroke in rats (80-83). Using non-invasive ICP measurement, we found that ICP is also elevated at 24 hours post- stroke in humans with mild-moderate stroke (84). This is the first evidence of ICP elevation in this patient population. When investigating potential mechanisms behind the rise in rats, we showed that CSF transfusion from post-stroke rats into naïve (disease free) rats causes delayed ICP elevation in CSF recipients (Figure 1.4, Chapter 1). This indicates that there is a factor(s) released in to CSF during stroke that causes ICP elevation in rats. We have preliminary evidence suggesting a similar response in recipient rats after CSF transfusion from human stroke patients (Figure 1.5, Chapter 1). Based on the animal data, combined with the human pilot, I hypothesised that there is a molecular trigger in post-stroke human CSF that causes ICP elevation. This study was designed to confirm the results of the initial pilot study and to expand on those findings by identifying potential proteomic triggers of the rise.

5.2 Part 1: Human CSF transfusion

5.2.1 Aims and hypotheses

This study was designed with three specific aims.

1. Determine whether human post-stroke CSF causes ICP to rise between 16-24 hours post-infusion in naïve rats. 2. Investigate sample characteristics that may influence the likelihood of CSF to cause ICP elevation. 3. Investigate whether there was a similar triggering molecule in the CSF of chronic hydrocephalus and SAH patients.

I hypothesised that:

1. Infusion of CSF from stroke patients will cause ICP elevation in recipient rats. 2. Stroke CSF collected at earlier time points will cause higher ICP elevation compared with samples collected at later time points. 3. Infusion of SAH, but not chronic hydrocephalus CSF will cause ICP elevation in recipient rats.

88 5.2.2 Experimental Design

The first experiments were performed to determine whether CSF from stroke patients causes delayed ICP elevation in recipient rats. Human lumbar CSF was collected from four different patient groups. The first CSF donor group consisted of 9 patients with mild-moderate stroke with CSF collected within the first 24 hours post-stroke. The next CSF donor group consisted of 9 patients with chronic hydrocephalus who were undergoing routine CSF drainage. The rationale behind using chronic hydrocephalus was that the ICP elevating molecule was likely to be an acute factor released within the first hours of stroke and therefore unlikely to be present in a chronic condition. Furthermore, preliminary evidence from a pilot study (n = 2) suggested a dichotomy in the ICP response of naïve animals after infusion of CSF into the brain where animals that received stroke-CSF had a distinct ICP response whilst there was an absence of ICP response in animals that received hydrocephalus-CSF (see Figure 1.5 in Chapter 1). The third CSF donor group consisted of 4 patients with SAH. And the last group consisted of 2 patients who had CSF collected for testing of neurological disorders but found to be negative (control CSF).

It is not known if all patients with mild-moderate stroke experience raised ICP, or if the ICP elevating molecule will be present at later timepoints (we have only shown presence of the molecule(s) in CSF collected at or within 6 hours of stroke). Therefore, to maximise the chances of identifying CSF samples that caused ICP elevation I chose to test the samples using biological replicates (testing one sample per rat), as opposed to pooling the samples.

To maintain consistency with the previous experiments CSF was infused intraventricularly into naïve recipient rats (n = 24; weight 309 – 344 g; outbred, Wistar) at 4 µL/ minute for 50 minutes (total volume 200 µL). Similarly, these experiments replicated the timing of previous infusion experiments. Epidural ICP was compared between pre-infusion baseline and 16-24 hours post-infusion (see Figure 5.1 for timeline). In addition to ICP SpO2, RR, HR and MAP were monitored simultaneously. To minimise the effects of regression to the mean, caused by within animal variability of ICP over time, peak DICP was determined in individual animals and analysed along with the corresponding SpO2, RR, HR and MAP values.

89

Figure 5.1: Experimental timeline. Animals underwent surgery for mean arterial pressure (MAP) and intracranial pressure (ICP) measurement. Baseline ICP was established before intraventricular infusion of human cerebrospinal fluid (4 µL/ minute for 50 minutes). At 2 hours post-infusion animals were recovered overnight and at 16 hours post-infusion animals were re-anaesthetised and both ICP and MAP were measured between 16-24 hours post-infusion. At 24 hours post-infusion animals were euthanised.

90 5.2.3 Results

5.2.3.1 CSF donor characteristics

Stroke-CSF donors

Of the 9 stroke-CSF samples tested, 2 (22%) were from women. Mean patient age was 52 years (median age 58 years; range: 27 to 65 years). Mean NIHSS score was observed to be 5.6 (median score 5; range: 1 to 14). Mean opening pressure was 11 mmHg (median opening pressure 9.5 mmHg; range: 8.8 to 13.2 mmHg). Median CSF collection time was 20 hours post-stroke (range: 3 to 23 hours). See Table 5.1 for individual sample characteristics and the specific causes of cerebral ischaemia and Figure 5.2 for representative scans.

Chronic hydrocephalus-CSF donors

Of the 9 chronic hydrocephalus-CSF samples chosen, 7 (78%) were from women. Mean patient age was 47 years (median age 47 years; range: 25 to 69 years). Median CSF collection time was 2 months post-symptom onset (range: 1 to 36 months).

SAH-CSF donors

Of the 4 SAH-CSF samples used in this study 3 (75%) were from women. Mean patient age was 55 years (median age was 52.5 years; range: 46 to 69 years). Median CSF collection time was 19 hours post-SAH (range: 3 to 96 hours).

Patients negative for neurological disorders (control-CSF donors)

The control-CSF samples came from an 18 year old female tested but found negative for seizure disorder and an 88 year old female tested but found negative for normal pressure hydrocephalus.

91

# Gender Age CSF collection Clinical Findings NIHSS CT Findings Opening time post-stroke Score pressure (mmHg) 1 M 65 5 h Receptive aphasia and ataxia 5 Left MCA syndrome PACI 9.5 2 M 39 20 h Dysphasia 1 Left MCA syndrome PACI 13.2 3 F 45 22 h Right hemiparesis and hemianesthesia 3 Left MCA stenosis 13.2 Capsulostriatal infarct 4 M 62 23 h Left hemiparesis 8 Right hemisphere 13.2 Watershed infarct 5 M 59 23 h Right hemiparesis and dysarthria 7 Left hemisphere lacunar 9.5 syndrome 6 M 59 4 h Left hemiparesis and dysarthria 4 Right hemisphere lacunar 9.5 stroke 7 M 58 4 h Right hemiparesis and hemianesthesia 3 Left MCA and ACA PACI 9.5 8 F 27 23 h Left hemiparesis, hemianesthesia and 6 Right MCA PACI 8.8 dysphasia 9 M 54 3 h Right hemiparesis, hemianesthesia and 14 Left MCA TACI 12.5 dysphasia Table 5.1: Clinical parameters of stroke patients. Samples found to cause intracranial pressure elevation in recipient animals are bolded. Abbreviations; MCA, middle cerebral artery; ACA, anterior cerebral artery; PACI, partial anterior circulation infarct; TACI, total anterior circulation infarct.

92

Figure 5.2: Representative images of selected stroke-CSF donor patients. a. Diffusion- weighted magnetic resonance imaging (DW-MRI) of right middle cerebral artery (MCA) partial anterior circulation infarct (PACI); b. Computer tomography (CT) image of left MCA and anterior cerebral artery PACI; c. T2-weighted MR image of left MCA syndrome PACI; d. DW- MRI of left MCA stenosis, capsulostriatal infarct; e. CT image of left MCA total anterior circulation infarct.

93 5.2.3.2 CSF recipient animals

Physiological parameters

Physiological parameters at baseline and peak ICP (16-24 hours post infusion) are listed in Table 5.2. Physiological parameters measured between 18-24 hours post-infusion are presented in Table 5.3.

Physiological parameters at baseline and peak D ICP (mean ± SD) Baseline Peak ICP ICP (mmHg) Stroke CSF 6 ± 2.2 12.3 ± 7.2 Hydrocephalus CSF 5.5 ± 1.1 9.7 ± 2.4 SAH CSF 6.1 ± 2.5 8.4 ± 0.8 Control CSF 4.6 ± 1 3.4 ± 0.1 MAP (mmHg) Stroke CSF 98.3 ± 5.6 94 ± 4.8 Hydrocephalus CSF 100.3 ± 6.1 93.2 ± 5.7 SAH CSF 86.7 ± 5.8 94.2 ± 13.3 Control CSF 91.3 ± 0.5 92.1 ± 2

SpO2 (%) Stroke CSF 97.2 ± 2.1 97.3 ± 2.3 Hydrocephalus CSF 95.6 ± 2.2 97.7 ± 1.8 SAH CSF 93.8± 3.6 97.9 ± 0.6 Control CSF 89 ± 1.4 96.5 ± 0.7 RR (BPM) Stroke CSF 62.4 ± 3.2 64 ± 4.4 Hydrocephalus CSF 66.9 ± 3.2 72.1 ± 10.3 SAH CSF 67.3 ± 3.8 74.5 ± 3 Control CSF 69 ± 4.2 68 ± 14.4 HR (BPM) Stroke CSF 453.2 ± 17.2 409 ± 23 Hydrocephalus CSF 434.8 ± 23.6 421.4 ± 21.1 SAH CSF 421.4 ± 22 410.2 ± 31.7 Control CSF 433.2 ± 9.7 374.8 ± 11

Table 5.2: Mean physiological parameters at baseline and peak intracranial pressure above baseline (DICP) between 16-24 hours post-infusion of human cerebrospinal fluid (CSF). Abbreviations: ICP, intracranial pressure; MAP, mean arterial pressure; SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute. Stroke n = 9, hydrocephalus n = 9, subarachnoid haemorrhage (SAH) n = 4 and control n = 2.

94 Time post-start of infusion (mean ± SD) Baseline 18 h 20 h 22 h 24 h ICP (mmHg) Stroke CSF 6 ± 2.2 7.4 ± 2.5 10.2 ± 5.4 7.7 ± 2.4 7.6 ± 2.6 Hydrocephalus CSF 5.5 ± 1.1 8.5 ± 1.9 6.4 ± 2.6 6.7 ± 2.2 6.9 ± 2.9 SAH CSF 6.1 ± 2.5 5.2 ± 1.4 5.5 ± 1.5 5.8 ± 2.3 5.1 ± 2 Control CSF 4.6 ± 1 1.8 ± 0.8 1.6 ± 0.9 2.9 ± 0.6 3.4 MAP (mmHg) Stroke CSF 98.3 ± 5.6 92.4 ± 3.6 95.5 ± 5.7 93.2 ± 6.3 95.4 ± 7.5 Hydrocephalus CSF 100.3 ± 6.1 90.6 ± 5 95 ± 5.1 99.6 ± 7.2 103.8 ± 9.9 SAH CSF 96.7 ± 5.8 95.5 ± 6.2 98.4 ± 7.9 102.4 ± 10 103.6 ± 11.4 Control CSF 91.3 ± 0.5 89.7 ± 1.2 90.2 ± 1.3 93 ± 4.7 93.6 SpO2 (%) Stroke CSF 97.2 ± 2.1 98.4 ± 1.4 97.4 ± 2.1 97.3 ± 1.7 97.2 ± 2.37 Hydrocephalus CSF 95.6 ± 2.2 97.2 ± 1.6 97.8 ± 1.7 97.4 ± 1.9 97.9 ± 2.3 SAH CSF 93.8 ± 3.6 97.8 ± 0.5 96.4 ± 1.1 97.9 ± 1.3 95.4 ± 3.3 Control CSF 89 ± 1.4 96.5 ± 0.7 97 ± 1.4 98.5 ± 0.7 97 RR (BPM) Stroke CSF 62.4 ± 3.2 63.8 ± 12.3 61.1 ± 10.8 59.9 ± 11.4 63.4 ± 5.8 Hydrocephalus CSF 66.9 ± 3.2 69.4 ± 4.8 67.8 ± 2.2 63.1 ± 4.3 62.9 ± 6.6 SAH CSF 67 ± 3.4 73.3 ± 4.1 68.5 ± 7.2 66.3 ± 3.3 63 ± 9 Control CSF 69 ± 4.2 75 ± 18.4 70 ± 8.5 67 ± 1.4 78 HR (BPM) Stroke CSF 453.2 ± 17.2 423.2 ± 16.1 409.3 ± 21.7 412 ± 21 419.8 ± 13.1 Hydrocephalus CSF 434.8 ± 23.6 414.5 ± 26.2 422.4 ± 24.3 421.5 ± 24.4 419.2 ± 17.6 SAH CSF 4214 ± 22 390.2 ± 31.7 375.1 ± 30.2 398.2 ± 36.3 387.5 ± 27.1 Control CSF 433.2 ± 9.7 384.7 ± 27.8 384 393.1 ± 25.3 382.5

Table 5.3: Mean physiological parameters before, during and after intracerebroventricular infusion of human cerebrospinal fluid (CSF) from stroke, hydrocephalus, subarachnoid haemorrhage (SAH) and control patients (samples negative for neurological disease). Abbreviations: ICP, intracranial pressure; MAP, mean arterial pressure; SpO2, oxygen saturation; RR, respiratory rate; HR, heart rate; BPM, breaths (RR) or beats (HR) per minute. Stroke n = 9, hydrocephalus n = 9, SAH n = 4, control n = 2.

95 The effect of CSF infusion on ICP in recipient animals

Temporal profiles for the individual animals are presented in Figure 5.3a-d. ICP significantly increased from 6 ± 2.2 mmHg at baseline to 12.3 ± 7.2 mmHg at peak ICP in the stroke-CSF recipient group (p=.02) (Figure 5.3e). The 95th percentile for baseline ICP amongst stroke- CSF recipients was 8.96 mmHg. Five stroke-CSF recipients demonstrated an increase in ICP at peak ICP that was greater than this value. Of those five stroke-CSF recipients, the mean increase in ICP at peak was 16.9 ± 6.5 mmHg. Median time to peak was 20.5 hours post- infusion (range 18-23.5 hours). In the hydrocephalus group ICP also significantly increased from 5.7 ± 1.2 mmHg at baseline to 9.7 ± 2.4 mmHg at peak ICP (p=.002) (Figure 5.3e). Median time to peak was 17.5 hours post-infusion (range 16.5 to 19.5 hours). The 95th percentile for baseline ICP amongst hydrocephalus-CSF recipients was 7.63 mmHg. Eight out of the nine hydrocephalus-CSF recipients demonstrated an increase in ICP at peak that was greater than this value. Of those CSF recipients, the mean increase in ICP at peak was 10.2 ± 2.3 mmHg. When groups were compared there was no difference in peak DICP (stroke, 6.3 ± 6.2 mmHg vs hydrocephalus, 4.3 ± 2.8 mmHg; p=.3). AUC (DICP (mmHg) vs time (hours)) was calculated between 18 and 23 hours post-infusion. There was no difference in mean AUC between groups (stroke, 26.3 ± 18.1 mmHg/ hour; hydrocephalus, 35.7 ± 28.1 mmHg/ hour; p=.4) (Figure 5.3f).

In the SAH-CSF recipient group ICP did not significantly increase from baseline to peak (6.1 ± 2.5 mmHg vs 8.4 ± 0.8 mmHg; p=.3). Time of peak ICP occurred between 16.5 hours and 20.5 hours post-infusion. ICP did not increase in the two control CSF recipients, ICP was 3.9 mmHg and 5.3 mmHg at baseline and 3.3 mmHg and 3.4 mmHg at peak (during 16-24 hour recording), peaking at 19.5 hours and 24 hours post-infusion, respectively. Mean peak DICP was 2.3 ± 2.6 mmHg in SAH-CSF recipients and peak DICP was -0.6 mmHg and -1.9 mmHg in the control CSF recipients. Mean AUC was 8.5 ± 17 mmHg/ hour for SAH CSF and 0 mmHg/ hour in the control CSF recipients (Figure 5.3f). Due to unequal sample numbers neither control, nor SAH-CSF recipient groups were compared against the stroke and/ or hydrocephalus-CSF recipient groups.

96

a. b. Stroke Hydrocephalus 30 30

20 20

10 10 ICP (mmHg) ICP ICP (mmHg) ICP

0 0 B 16 18 20 22 24 B 16 18 20 22 24 Time (h) Time (h)

c. d. SAH Normal (Control) 30 30

20 20

10 10 ICP (mmHg) ICP (mmHg) ICP

0 0 B 16 18 20 22 24 B 16 18 20 22 24 Time (h) Time (h)

e. f. ** 30 100 * Stroke Hydroceph. 80 SAH 20 Control 60

40 10 ICP (mmHg) ICP 20 AUC (mmHg/ hour) 0 0 Baseline Peak Stroke Hydroceph. SAH Control (n=9) (n=9) (n=4) (n=2)

Figure 5.3: Effects of intraventricular infusion of human cerebrospinal fluid (CSF) on intracranial pressure (ICP) in recipient rats. ICP of individual animals at pre-infusion baseline and between 16-24 hours post-infusion of CSF from a. stroke, b. hydrocephalus, c. subarachnoid haemorrhage (SAH) or d. control patients. e. ICP of stroke (red circles), hydrocephalus (blue triangles), SAH (green squares) and control (black crosses) CSF recipients at baseline and peak (16-24 hours). f. Area under the curve (DICP (mmHg)/ time (hour)).

97 Correlations between CSF donor characteristics and peak DICP in recipient animals

Regression analysis indicated no significant relationship between peak DICP of stroke-CSF recipients and donor NIHSS score (r2 = 0.02; p=.7), CSF collection time post-stroke (r2 <0.01; p=.99) or CSF opening pressure (r2 = 0.13; p=.3) (Figure 5.4a-c). There was no correlation between CSF collection time and time of peak ICP (r2 = 0.1; p=.8; Figure 5.4d).

a. b. 25 25 r2 = 0.02; p = 0.7 r2 <0.01; p = 0.99 20 20

15 15

10 10 ICP (mmHg) ICP ICP (mmHg) ICP Δ 5 Δ 5

Peak 0 Peak 0

-5 -5 0 5 10 15 0 5 10 15 20 25 NIHSS Score CSF collection time post-stroke (h)

c. d. 25 24 r2 = 0.13; p = 0.3 r2 = 0.01; p = 0.8 20 22 15

10 20 ICP (mmHg) ICP

Δ 5 18 post-infusion (h) Time of peak ICP

Peak 0

-5 16 100 120 140 160 180 200 0 5 10 15 20 25

Opening pressure (cmH2O) CSF collection time post-stroke (h)

Figure 5.4. Illustrative comparison of the peak change in intracranial pressure (peak DICP). Stroke CSF recipients versus a. patient NIHSS score b. CSF collection time post- stroke c. opening pressure at CSF collection from stroke-CSF donor patients. No significant relationships were found between DICP and NIHSS score, CSF collection time, or opening pressure) (p>.05). d. Illustrative comparison of the CSF collection time post-stroke (in patients) and the time of peak ICP in the recipient animals. No relationship was found (p>.05).

98 5.2.4 Discussion

Mean peak ICP elevation was significantly higher than baseline in both stroke and hydrocephalus CSF recipients. Furthermore, no differences were detected in peak ICP or AUC between stroke- and hydrocephalus-CSF recipient groups. In contrast, infusion of SAH- CSF did not cause mean peak ICP elevation above baseline between 16-24 hours. Similarly, infusion of control-CSF did not cause ICP elevation above baseline at any time between 16- 24 hours post-infusion. The results suggest the presence of an ICP elevating factor(s) within both stroke- and hydrocephalus-CSF, but not in SAH- or control-CSF.

My findings are consistent with our preclinical data showing a similar ICP response profile in rats that received an infusion of stroke-CSF from donor rats (Unpublished, R. Hood, D. D McLeod). The results of this study provide the first direct evidence of the presence of an ICP elevating molecule(s) in the CSF of stroke patients. This suggests that a molecule(s) is released into the CSF during the first 24 hours post-stroke that trigger a delayed ICP elevation. While this study investigated the ICP response in rats, it is entirely possible that the same ICP response occurred in the patients from whom the CSF was collected. However, this is difficult to ascertain in these patients as we do not have baseline ICP values, therefore DICP cannot be calculated. CSF composition has been shown to be altered as a consequence of stroke in both human (189-191) and animal (181) studies. However, the effects of the vast majority of compositional changes have yet to be understood. A handful of investigations have shown correlations between CSF proteins and post-stroke ICP elevation e.g. AVP, bradykinin (193) in humans yet there is a lack of supporting evidence to show causation. Furthermore, investigations into how these molecules affect ICP have focused on acute as opposed to delayed effects. The composition of CSF will be investigated in Part 2 of this chapter.

Importantly, stroke-CSF was collected from patients with mild-moderate stroke with NIHSS scores below that considered to predict ICP elevation (≥20 (229)), who up until recently (84) were not thought to experience any ICP elevation at all due to their small oedema volumes. Interestingly, mean peak DICP elevation in the stroke CSF recipient rats (6.3 ± 6 mmHg) was similar to what was observed in our non-invasive ICP pilot study in patients (approximately 5 mmHg above baseline) (84). I hypothesise that altered CSF composition may play a role in this newly discovered ICP elevation at 24 hours in patients with mild-moderate stroke (84). CSF was infused into the existing CSF of recipient animals, therefore the molecules within the infused CSF would be diluted by existing CSF and we may not have observed the full extent of the ICP response in recipient animals. Consequently, I speculate that the ICP response caused by a triggering molecule would be greater in the CSF donor. Moreover, the observed ICP elevation occurred in stroke-free animals. The fact that any ICP elevation was observed

99 certainly warrants further investigation. With this being said, the observed ICP elevation may be sufficient to contribute to worse outcomes in the context of stroke. There is no defined threshold for pathological ICP elevation however, under normal circumstances, ICP is assumed to become pathological only once values exceed 20 mmHg. However, what about in the damaged brain? We have shown that even a modest rise in ICP reduces flow through the leptomeningeal collateral vessels after stroke (81). This may be particularly important to the vulnerable ischaemic penumbra where collateral circulation is crucial for tissue survival and any reduction in flow may lead to expansion of the infarct core.

Interestingly, not every stroke-CSF sample caused ICP elevation in the recipient rats (peak DICP ranged from -0.38 mmHg to 19.39 mmHg). Five CSF samples caused increased peak ICP that was greater than the 95th percentile of baseline ICP values, whilst the increase in ICP was negligible for the remaining four samples. This finding is in keeping with our human non- invasive ICP findings (84) where ICP was observed to be greater than the 95th percentile of control patients in 7/10 patients with mild-moderate stroke. It is likely that expression of the ICP elevating factor(s) may differ between patients and therefore the tested CSF samples. Stroke is an extremely heterogenous disease which makes it difficult to control variables between patients in a small clinical study such as this. Furthermore, stroke has a progressive pathophysiology. The extent of tissue damage as well as the nature of tissue damage e.g. ischaemic cascade (minutes), vasogenic oedema (hours) changes over time. Simats et al. (2018) showed differential expression of 716 CSF proteins within the first 2 hours of experimental stroke compared with sham animals (181). This number is likely to be influenced by a plethora of factors including the location of stroke, comorbidities and time post stroke to name a few. In this study, CSF collection time ranged from 3 to 23 hours post-stroke. Our original pilot study using controlled CSF collection times suggested that the triggering molecule was present at 6 hours post-stroke and caused ICP elevation in recipient animals at what would have been 24 hours post-stroke in the donor animals. However, no relationship between CSF collection time and peak DICP was observed, nor time of peak ICP. This suggests that expression of this ICP elevating molecule(s) is maintained for at least the first 24 hours post-stroke in humans. Furthermore, it suggests that the molecule(s)’ mechanism of action in the recipient is not related to the time of stroke. Similarly, there was no correlation between NIHSS score and peak ICP, suggesting that stroke severity does not influence expression of the ICP triggering molecule in post-stroke CSF.

In the present study ICP elevation was observed in the majority (8/9) of hydrocephalus-CSF recipients suggesting the presence of a molecular trigger contributing to ICP elevation. This was an unexpected finding based on the lack of an ICP response in hydrocephalus-CSF

100 recipients seen in our pilot study (n = 2) (Figure 1.5, Chapter 1) and the chronic nature of the disease. Chronic hydrocephalus is caused by an accumulation of CSF within the craniospinal space leading to increased ICP (230). However the mechanisms underlying the pathophysiology are complex and poorly understood (231, 232). CSF composition has been shown to be altered in chronic hydrocephalus (233) therefore, it is plausible that there may be molecular contributors to disease progression. There have been numerous investigations into CSF biomarkers of chronic hydrocephalus (see review by Tarnaris et al. (2006) (234))(233] therefore, it is plausible that there may be molecular contributors to disease progression. There have been numerous investigations into CSF biomarkers of chronic hydrocephalus (see review by Tarnaris et al. (2006) {Tarnaris, 2006 #829). However, as with stroke, none have been shown to directly elevate ICP. The results in this study suggest that further investigation of hydrocephalus CSF composition might provide insight to the complex mechanisms underlying this disease.

Peak ICP elevation above baseline was not significant in animals that received CSF from control or SAH patients. The results suggest the absence of a molecular trigger for ICP elevation in the SAH-CSF samples that were tested however, interpretation of these findings are difficult due to the small number of samples that we were able to test. An SAH occurs due to the rupture of a blood vessel over the surface of the brain, in the subarachnoid space releasing blood directly into the CSF. The absence of an ICP response in SAH-CSF recipient animals is particularly surprising considering that a hallmark of SAH is ICP elevation concurrent with blood contamination of the CSF (235-238). Furthermore, a study by Conzen et al (2018) found using an experimental model of SAH that ICP was almost 3 times higher when blood was injected into the subarachnoid space versus when the equivalent volume of saline was injected at the same rate (238). Suggesting that presence of blood constituents results in higher ICP than just the total volume of infusate alone. This may have potentially been influenced by our sample collection method where samples were initially spun down to remove. It is possible that blood breakdown products were still present in the SAH CSF, however the samples did not undergo further molecular investigation. It would be interesting to observe the ICP elevation in recipient animals of SAH CSF that still contained blood cells. While ICP is initially elevated after SAH, it does not persist beyond the first few hours post- SAH. However, some SAH patients experience prolonged ICP elevation or ICP spikes with high pressures >20 mmHg measured up to 14 days post SAH (31). The samples tested in this study were collected at various times including out to 96 hours post-SAH. Therefore, it is theoretically possible that if there was a molecular component to ICP elevation post-SAH that our samples would have covered an appropriate time window for either of the ICP rises.

101 Unfortunately, a deeper investigation into SAH CSF was limited by the number of samples that I was able to obtain.

The exact mechanism behind the ICP rise was not investigated in this study however based on the Monroe-Kellie doctrine, the target of the triggering molecule is likely related to water balance (CSF/ oedema) or cerebral blood volume regulation, which when disturbed causes ICP to become elevated. We have previously shown that the ICP rise seen at 24 hours post- stroke in rats is primarily caused by a mechanism other than cerebral oedema (80, 82, 83). Although oedema was not measured, I hypothesise that ICP elevation in this study occurred via a similar oedema independent increase in pressure in this study due to similarities in the timing of the ICP profile between this study and our previous studies as well as the absence of tissue damage (which would result in oedema). Furthermore, chronic hydrocephalus is characterised by an excessive amount of CSF in the craniospinal cavity. Therefore, I speculate that the mechanism may be related to CSF volume. However as this was an exploratory study to determine if CSF from stroke patients causes ICP to rise, the exact mechanism behind ICP elevation was not investigated. A subsequent study measuring CSF production and outflow post-infusion of stroke- or hydrocephalus-CSF in rats may help to elucidate the mechanisms underlying the current ICP rise.

The similarity between ICP responses of the CSF recipient animals of both stroke- and hydrocephalus-CSF begs the question, could it be a similar molecular trigger causing ICP elevation in both CSF recipient groups? Despite differences in the initial cause of the diseases, there are some similarities in the pathology of stroke and hydrocephalus. Ischaemic stroke is caused by reduced blood flow leading to hypoxic and ischaemic damage. Braun et al. found upregulation of lactate (a sensitive marker of ischaemia and hypoxia) in an experimental model of hydrocephalus (239-241). Chronic hydrocephalus can cause a reduction in cerebral blood flow which can affect perfusion of periventricular tissue leading to neuronal damage (242). Could this blood flow reduction also cause submaximal tissue injury similar to what we have associated with ICP elevation after stroke? Similarities in the CSF proteome that may be relevant for ICP elevation will be investigated in Part 2 of this chapter.

This study has many strengths. The biological assay used was chosen because it allows for the screening of samples to test their biological relevance to ICP elevation. It also permits the testing of CSF composition as a whole rather than just selecting individual molecules of interest. Furthermore, by using human CSF I was able to strengthen the clinical relevance of our animal findings. Another advantage of using human CSF is that the volume able to be obtained for testing is several orders of magnitude higher than that available from rats (up to 15 mL vs ~200 µL respectively). This permits additional testing of each sample. This study

102 was undertaken to confirm the results of our pilot study and determine whether further investigation into CSF composition and the mechanisms behind ICP elevation were warranted. I chose to do biological replicates (using a different sample/ animal) over technical replicates (using the same sample across multiple animals). This approach enabled efficient use of animals and maximised the chances of identifying CSF samples that caused ICP elevation. The CSF transfusion assay was a successful screening tool in this regard. This approach can also be considered a limitation due to the lack of technical replicates of each sample. Performing technical replicates would ensure that each sample consistently causes ICP elevation in recipient animals and strengthen the results of this study. Another limitation of this study was the accessibility of CSF samples, particularly SAH and control CSF samples. Lumbar puncture is an invasive procedure, therefore we were restricted to collecting CSF from patients who were already undergoing CSF collection. Unfortunately, this resulted in uneven sample sizes between groups. As a result, the findings of this study must be interpreted accordingly. Finally, although the location of the injection was confirmed after sacrifice, neither histological analysis, nor oedema assessment was performed on the recipient animals brains due to institutional safety requirements. This may have provided some mechanistic insight into why some animals showed ICP elevation whilst others did not e.g global oedema or ventriculomegaly. The safety requirements also precluded molecular analysis on the brain tissue that may have provided insight into protein expression in the animals that had ICP rise versus those that did not. As discussed above, the results support the hypothesis that there is a triggering molecule in the CSF of stroke and hydrocephalus patients causing ICP elevation. It is unlikely that the results are due to the interspecies CSF transfusion i.e. human CSF (irrespective of disease status) causes ICP elevation in rats. CSF is largely acellular and therefore is not likely to illicit an immune response in recipient animals and furthermore there was no response to infusion in control and especially SAH-CSF recipients. Thus, the results suggest that human to rat CSF transfusion alone does not cause ICP elevation in recipient animals. This is in keeping with previous CSF transfusion studies which reported no adverse effects of interspecies CSF transfusion (243-245).

My findings contribute to our group’s growing body of evidence pointing to a novel mechanism of delayed ICP elevation after mild-moderate stroke in both humans and animals (80-84). Furthermore, the results of the current study support the clinical relevance of our animal transfusion data and suggest a similar molecular trigger is present in the CSF of human stroke patients. Additionally, I have found potential evidence of a similar novel mechanism for ICP elevation in hydrocephalus. Future studies will confirm and expand upon these findings by increasing the number of biological and technical replicates. This will help us to characterise which samples cause ICP elevation. Importantly, ICP elevation occurred in the absence of

103 neurological injury and this is strongly suggestive of an entirely new mechanism of ICP elevation. If confirmed this will have major implications for diagnosis and treatment of raised ICP in stroke and hydrocephalus patients.

104 5.3 Part 2: The CSF proteome and ICP elevation

After confirming the potential presence of ICP elevating molecules in human CSF the next logical step was to investigate CSF composition. It was originally planned that I would compare stroke and hydrocephalus CSF samples to identify factors involved in ICP elevation post-stroke based on the dichotomous response seen in the original human CSF pilot study (Figure 1.5, Chapter 1). However, the results of the experiments, presented in Part 1 of this chapter, suggested the presence of the ICP elevating factor(s) in both stroke and hydrocephalus-CSF. Therefore, the experiments presented in this section were re-designed to identify differentially expressed CSF proteins within each disease based on the ICP response of recipient animals.

5.3.1 Aim and hypothesis

The aim of this study was to identify proteins within human CSF samples from stroke and hydrocephalus patients that may be involved in the delayed ICP elevation that was observed in recipient rats in Part 1 of this chapter. I hypothesised that proteins would be differentially expressed between ICP elevating and non-elevating CSF samples.

5.3.2 Experimental Design

To maximise the chance of identifying relevant molecules of interest, I chose to compare CSF samples within each disease group and differentiate them based on the ICP response profile of the recipient animals. Samples within stroke and hydrocephalus groups (from Part 1 of this chapter) were differentiated into those that caused an ICP rise versus those that did not. The control CSF data (Chapter 5 - Part 1) and aCSF infusion data (Chapter 4) indicated that control infusion did not result in DICP rise ³5 mmHg, thus 5 mmHg (DICP rise) was determined to be the threshold for this study. Therefore, CSF samples that caused a rise in recipient animals above 5 mmHg were considered ‘ICP elevating’ and those that did not cause a rise above 5 mmHg were considered ‘non ICP elevating’ for proteomic analysis. Analysis of the individual ICP profiles demonstrated a subset of stroke (44%) and hydrocephalus (33%) CSF recipients had ICP elevation ³5 mmHg above baseline.

Quantitative proteomic analysis was performed on all CSF samples that were tested in Part 1 of this chapter (n = 9/ group). Briefly, each sample was reduced and alkylated before proteins were precipitated using methanol-chloroform two phase system. The protein was then digested with trypsin and the peptides acidified. Label free data dependent and data independent analysis (DIA) was performed on each sample (see Chapter 2 for details). In order to accommodate the variability in CSF protein content between individual patients and

105 diseases all samples were used to build the spectral library. The DIA data was then matched against proteins within the library and relative expression of proteins was investigated between sample groups. In order to determine if key pathways were involved in the observed ICP elevation (Part 1 of this chapter), pathway enrichment analysis was performed using the freely available software - DAVID bioinformatics (https://david.ncifcrf.gov). In order to determine if CSF samples could be independently separated based on ICP elevation in rats, a PCA was performed. PCA takes all the variables that move in the one (same) direction and groups them into a component (e.g. proteins more abundant in CSF samples that cause ICP elevation). A second set of variables that move in the same direction (e.g. proteins more abundant in CSF samples that do not cause ICP elevation) is then taken and PCA determines which samples group together.

5.3.3 Results

Sample exclusion

Of the 18 samples that were investigated for this study one of the ICP elevating stroke-CSF samples and two of the hydrocephalus-CSF samples (ICP elevating: n = 1; non-ICP elevating: n = 1) were excluded due to processing errors.

DIA Library

In total, 1758 proteins were identified within the cohort of samples. These proteins were then quantified across all samples.

Proteomic profiling of stroke-CSF

A total of 17 proteins were found to be differentially expressed (>2-fold, p<.05) between ICP elevating (peak ICP ³5 mmHg above baseline) and non-elevating samples. Expression of 15 of these proteins was found to be significantly more abundant in the samples that caused ICP elevation (Table 5.4; Figure 5.5), the remaining two proteins were significantly more abundant in the non-elevating samples (Table 5.5; Figure 5.6). Galectin-1 (LEG1) was found to be 10.2 fold more abundant in the samples that caused ICP elevation (p=.03). Dishevelled-associated activator of morphogenesis 2 (DAAM2) and netrin receptor DCC (DCC) were found to be the most statistically significant proteins in the ICP elevating samples (fold change 6.8 and 2.07 respectively; p=.002 for both). Tubulin alpha-4A chain and matrix metalloprotease 2 were found to be more abundant in the non-ICP elevating samples.

106 Swiss-Prot ID Acronym Protein Name P Value Fold Increase P09382 LEG1 Galectin-1 0.03 10.2 P01591 IGJ Immunoglobulin J chain 0.04 8.1 Q15185 TEBP Prostaglandin E synthase 3 0.03 7.2 Q86T65 DAAM2 Dishevelled-associated activator of 0.002 6.8 morphogenesis 2 P62937 PPIA Peptidyl-prolyl cis-trans isomerase 0.02 5.3 P80108 PHLD Phosphatidylinsitol-glycan-specific <0.05 4.7 phospholipase D P14618-3 KPYM Isoform 3 of Pyruvate kinase PKM 0.02 4.4 P23435 CBLN1 Cerebellin-1 0.02 4.1 Q96PZ0 PUS7 Pseudouridylate synthase 7 homolog 0.03 3.8 Q13554 KCC2B Calcium/calmodulin-dependent protein 0.02 3.4 kinase type II subunit beta P62937-2 PPIA-2 Isoform 2 of Peptidyl-prolyl cis-trans 0.03 3.2 isomerase A P15559 NQO1 NAD(P)H dehydrogenase [quinone] 1 0.01 2.6 P01210 PENK Proenkephalin-A <0.05 2.1 P43146 DCC Netrin receptor DCC 0.002 2.07 P02766 TTHY Transthyretin 0.03 2.02 Table 5.4: Proteins found to be higher in abundance in ICP elevating (³ 5 mmHg) stroke CSF samples.

Swiss-Prot ID Acronym Protein Name P Value Fold Increase P68366 TBA4A Tubulin alpha-4A chain 0.02 35.3 P08253 MMP2 Matrix metalloprotease 2 0.0008 7.3 Table 5.5: Proteins found to be higher in abundance in the non-ICP elevating stroke CSF samples.

107 a. 10 ICP Elevating CSF Non-ICP Elevating CSF *

5 * * ** expression * 2 0 b. DAAM2 -5 2 ICP Elevating CSF

-10 Non-ICP Elevating CSF Normalised Log 0 LEG1 IGJ TEBP DAAM2 PPIA expression 2 -2

-4 10 * * * -6 5 * * Normalised Log S1 S2 S3 S4 S5 S6 S7 S8 expression

2 0

-5

-10

-15 Normalised Log c. DCC PHLD KPYM CBLN1 PUS7 KCC2B 7.5 ICP Elevating CSF

5.0 15 * Non-ICP Elevating CSF expression

2 2.5 10 0 * **

expression * * 2 5 -2.5

0 -5.0 Normalised Log S1 S2 S3 S4 S5 S6 S7 S8 -5

-10 Normalised Log PPIA-2 NQO1 PENK DCC TTHY

Figure 5.5: Normalised Log2 expression data comparing intracranial pressure (ICP) elevating and non-elevating stroke cerebrospinal fluid (CSF) samples. a. Mean (±SD) normalised Log2 expression of proteins found to be higher in abundance (> 2-fold, p<.05) in CSF samples that caused ICP elevation in rats (peak ICP ³5 mmHg above baseline) in Part 1 of this chapter. Proteins with the highest significant differences are highlighted by boxes. b. Expression profile of dishevelled-associated activator of morphogenesis 2 (DAAM2) in individual CSF samples. c. Expression profile of netrin receptor DCC (DCC) in individual CSF samples. Abbreviations: LEG1: Galectin-1, IGJ: Immunoglobulin J chain, TEBP: Prostaglandin E synthase 3, PPIA: Peptidyl-prolyl cis-trans isomerase, PHLD: Phosphatidylinsitol-glycan-specific phospholipase D, KPYM: Isoform 3 of Pyruvate kinase PKM, CBLN1: Cerebellin-1, PUS7: Pseudouridylate synthase 7 homolog, KCC2B: Calcium/calmodulin-dependent protein kinase type II subunit beta, PPIA-2: Isoform 2 of Peptidyl-prolyl cis-trans isomerase A, NQO1: NAD(P)H dehydrogenase [quinone] 1, PENK: Proenkephalin-A, TTHY: Transthyretin.

108 a. b. 6 4 Non-ICP Elevating CSF * ICP Elevating CSF 4 *** Non-ICP Elevating CSF 2 expression expression 2 2 2 ICP Elevating CSF

0 0

-2 -2 Normalised Log Normalised Log TBA4A MMP2 S1 S2 S3 S4 S5 S6 S7 S8

Figure 5.6: Normalised Log2 expression data comparing intracranial pressure (ICP) elevating and non-elevating stroke cerebrospinal fluid (CSF) samples. a. Mean (±SD) normalised Log2 expression of proteins found to be higher in abundance (>2-fold, p<.05) in CSF samples that did not cause ICP elevation in rats (peak ICP <5 mmHg above baseline) in Part 1 of this chapter. b. Expression profile of matrix metalloprotease 2 (MMP2) in individual CSF samples. Abbreviations: TBA4A: Tubulin alpha-4A chain.

109 Pathway analysis of the 15 proteins that were more abundant in the ICP elevating CSF samples showed seven significantly enriched pathways (p<.05; with at least 2 proteins identified as being involved in the pathway). The top 5 enriched pathways identified are shown in Figure 5.7.

Over Expressed in ICP Elevating Stroke CSF

Protein homodimerisation activity Extracellular Space RNA-dependent DNA biosynthetic process Poly(A) RNA binding Extracellular exosome

0 5 10 15 -Log p-value 2 Figure 5.7: Gene ontology analysis. Gene ontology analysis of proteins that were >2-fold more abundant in intracranial pressure (ICP) elevating cerebrospinal fluid (CSF) samples. Only those proteins that were found to be significantly more abundant were analysed using DAVID bioinformatics. Shown are the top five significant enrichment terms for biological processes. a. Over Expressed in ICP Elevating Hydrocephalus CSF

Actin filament bundle assembly In order to determine if CSF samples could be independently separated based on ICP Actin crosslink formation elevation, a PCA wasActin performed. filament network ICP formation elevating samples could not easily be distinguished from samples that did not causeActin ICP filament elevation. bundle Interestingly, using this approach, two types Extracellular exosome of populations clustered (Figure 5.8). However, within the cluster, there are samples that do, 0 5 10 15 and do not cause ICP elevation. -Log2 p-value b. Over Expressed in non-ICP Elevating Hydrocephalus CSF

Extracellular Exosome Extracellular Space Endoplasmic reticulum lumen Extracellular region Response to estradiol

0 5 10

-Log2 p-value

Figure 5.8: Principle component analysis of stroke-CSF samples. The normalised Log2 data for each sample were uploaded into the statistical software package Perseus. There was no obvious separation between stroke-CSF samples that caused ICP elevation (≥5 mmHg, closed circles) from those that did not (open circles) when infused into naïve rats.

110 Proteomic profiling of hydrocephalus-CSF

In total, 32 proteins were found to be differentially expressed (>2-fold, p<.05) between ICP elevating and non-elevating hydrocephalus-CSF samples. Expression of 19 of these proteins were found to be significantly more abundant in the samples that caused ICP elevation (Table 5.6; Figure 5.9). Calcium/calmodulin-dependent protein kinase type II subunit beta (KCC2B) was 51.3 fold more abundant in ICP elevating CSF samples (p=.004). Whilst, transalodolase (TALDO) and lumican (LUM) showed the most statistically different expression between sample groups (fold change 19.5, p=.0001 and fold change 8.2, p=.0008 respectively). The remaining 13 proteins were found to be more abundant in the non-ICP elevating samples (Table 5.7, Figure 5.10). Trypsin-3 (TRY3) (fold change 76, p=.03) and V-set and transmembrane domain-containing protein 2B (VTM2B) (fold change 7.2, p=.005) showed the highest change in abundance and significance respectively.

Swiss-Prot ID Acronym Protein Name P Value Fold Increase Q13554 KCC2B Calcium/calmodulin-dependent protein 0.004 51.3 kinase type II subunit beta Q2TV78 MST1L Putative macrophage stimulating 1-like 0.002 37.7 protein Q92932 PTPR2 Receptor-type tyrosine-protein 0.04 19.9 phosphatase N2 P37837 TALDO Transaldolase 0.0001 19.5 P15814 IGLL1 Immunoglobulin lambda-like polypeptide <0.05 17.4 1 Q6UX71-2 PXDC2-2 Isoform 2 of Plexin domain-containing 0.04 8.7 protein 2 P51884 LUM Lumican 0.0008 8.2 P01011-2 AACT Isoform 2 of Alpha-1-antichymotrypsin 0.007 7.2 P09382 LEG1 Galectin-1 0.04 6 Q6UX71 PXDC2 Plexin domain-containing protein 2 0.03 4.4 Q6UXB8 PI16 Peptidase inhibitor 16 0.03 4.2 O60568 PLOD3 Procollagen-lysine,2-oxoglutarate 5- 0.02 3.4 dioxygenase 3 Q14651 PLSI Plastin-1 0.008 3.3 P13797 PLST Plastin-3 0.008 3.3 P01624 KV315 Immunoglobulin kappa variable 3-15 <0.05 2.3 A0A0C4DH55 KVDO7 Immunoglobulin kappa variable 3D-7 <0.05 2.3 P14136 GFAP Glial fibrillary acidic protein 0.04 2.2 P00918 CAH2 Carbonic anhydrase 2 0.04 2.2 P20742-2 PZP-2 Isoform 2 of Pregnancy zone protein <0.05 2.1 Table 5.6: Proteins found to be higher in abundance in ICP elevating (³ 5 mmHg) hydrocephalus CSF samples.

111

Swiss-Prot ID Acronym Protein Name P Value Fold Increase P35030 TRY3 Trypsin-3 0.03 76 P15151 PVR Poliovirus receptor <0.05 59.2 P02652 APOA2 Apolipoprotien A-II <0.05 50.5 Q9UNW1 MINP1 Multiple inositol polyphosphate 0.03 45.6 phosphatase 1 Q08174 PCDH1 Protocadherin-1 0.02 32.5 P02511 CRYAB Alpha-crystallin B chain 0.01 14.9 P24821 TENA Tenascin 0.04 10.4 Q06481-5 APLP2 Isoform 5 of Amyloid-like protein 2 0.03 7.6 A6NLU5 VTM2B V-set and transmembrane domain- 0.005 7.2 containing protein 2B P01602 KV105 Immunoglobulin kappa variable 1-5 <0.05 5 P29279 CTGF Connective tissue growth factor 0.03 4.1 Q86UX2-3 ITIH5-3 Isoform 3 of Inter-alpha-trypsin inhibitor 0.01 2.9 heavy chain H5 P02452 CO1A1 Collagen alpha-1(I) chain 0.03 2.8 Table 5.7: Proteins found to be higher in abundance in the non-ICP elevating hydrocephalus CSF samples.

112 a. 15 ICP Elevating CSF Non-ICP Elevating CSF *** 10 * * b. expression ** ** *** * TALDO

2 5 2 ICP Elevating CSF 0 Non-ICP Elevating CSF 0 -5 expression

2 -2 -10 Normalised Log -4 KCC2B MST1L PTPR2 TALDO IGGL1 PXDC2-2 LUM -6

-8 Normalised Log 10 H1 H2 H3 H4 H5 H6 H7 * 5 ** * * * ** expression

2 0

-5

-10 c. LUM

15 -15 Normalised Log AACT LEG1 PXCD2 PI16 PLOD3 PLSI ICP Elevating CSF 10 expression 2

15 * 5 Non-ICP Elevating CSF * 10 * * * 5 ** expression

2 0 Normalised Log 0 H1 H2 H3 H4 H5 H6 H7 -5

-10

-15 Normalised Log PLST KV315 KVDO7 GFAP CAH2 PZP-1

Figure 5.9: Normalised Log2 expression data comparing intracranial pressure (ICP) elevating and non-elevating hydrocephalus cerebrospinal fluid (CSF) samples. a. Mean (±SD) normalised Log2 expression of proteins found to be higher in abundance (>2-fold, p<.05) in CSF samples that caused ICP elevation in rats (peak ICP ³5 mmHg above baseline) in Part 1 of this chapter. Proteins with the highest significant differences are highlighted by boxes. b. Expression profile of transaldolase (TALDO) in individual CSF samples. c. Expression profile of lumican (LUM) in individual CSF samples. Abbreviations: KCC2B: Calcium/calmodulin- dependent protein kinase type II subunit beta, MST1L: Putative macrophage stimulating 1-like protein, PTPR2: Receptor-type tyrosine-protein phosphatase N2, IGLL1: Immunoglobulin lambda-like polypeptide 1, PXDC2-2: Isoform 2 of Plexin domain-containing protein 2, AACT: Isoform 2 of Alpha-1-antichymotrypsin, LEG1: Galectin-1, PXDC2: Plexin domain-containing protein 2, PI16: Peptidase inhibitor 16, PLOD3: Procollagen-lysine,2-oxoglutarate 5- dioxygenase 3, PLSI: Plastin-1, PLST: Plastin-3, KV315: Immunoglobulin kappa variable 3- 15, KVDO7: Immunoglobulin kappa variable 3D-7, GFAP: Glial fibrillary acidic protein, CAH2: Carbonic anhydrase 2, PZP-2: Isoform 2 of Pregnancy zone protein.

113 a. ICP Elevating CSF Non-ICP Elevating CSF 6 * * * * * * * 2 expression 2 -2

b. VTM2B -6 2 Non-ICP Elevating CSF -10 Normalised Log 1 ICP Elevating CSF TRY3 PVR APOA2 MINP1 PCDH1 CRYAB TENA expression

2 0

-1 10 * -2

5 * -3 * Normalised Log

expression **

2 * * H1 H2 H3 H4 H5 H6 H7 0

-5

-10 Normalised Log APLP2 VTM2B KV105 CTGF ITH5-3 CO1A1

Figure 5.10: Normalised Log2 expression data comparing intracranial pressure (ICP) elevating and non-elevating hydrocephalus cerebrospinal fluid (CSF) samples. a. Mean (±SD) normalised Log2 expression of proteins found to be higher in abundance (>2-fold, p<.05) in non-ICP elevating CSF samples (peak ICP <5 mmHg above baseline) in Part 1 of this chapter. Proteins with the highest significant differences are highlighted by boxes. b. Expression profile of V-set and transmembrane domain-containing protein 2B (VTM2B) in individual CSF samples. Abbreviations: TRY3: Trypsin-3, PVR: Poliovirus receptor, APOA2: Apolipoprotien A-II, MINP1: Multiple inositol polyphosphate phosphatase 1, PCDH1: Protocadherin-1, CRYAB: Alpha-crystallin B chain, TENA: Tenascin, APLP2: Isoform 5 of Amyloid-like protein 2, KV105: Immunoglobulin kappa variable 1-5, CTGF: Connective tissue growth factor, ITIH5-3: Isoform 3 of Inter-alpha-trypsin inhibitor heavy chain H5, CO1A1: Collagen alpha-1(I) chain.

114 Over Expressed in ICP Elevating Stroke CSF

Protein homodimerisation activity Extracellular Space RNA-dependent DNA biosynthetic process Poly(A) RNA binding Extracellular exosome

0 5 10 15 -Log p-value Bioinformatic analysis of enriched pathways showed 8 signif2icantly enriched gene ontology terms (p<.05) in the ICP elevating samples and 12 pathways in samples that did not elevate ICP. The top 5 enriched pathways identified in both groups are shown in Figure 5.11.

a. Over Expressed in ICP Elevating Hydrocephalus CSF

Actin filament bundle assembly Actin crosslink formation Actin filament network formation Actin filament bundle Extracellular exosome

0 5 10 15

-Log2 p-value b. Over Expressed in non-ICP Elevating Hydrocephalus CSF

Extracellular Exosome Extracellular Space Endoplasmic reticulum lumen Extracellular region Response to estradiol

0 5 10

-Log2 p-value

Figure 5.11: Gene ontology analysis. Gene ontology analysis of proteins that were >2-fold more abundant in cerebrospinal fluid samples that a. caused intracranial pressure (ICP) elevation or b. did not cause ICP elevation in Part 1 of this chapter. Only those proteins that were found to be significantly more abundant in each group were analysed using DAVID bioinformatics. Shown are the top five significant enrichment terms for biological processes.

115 There was minimal clustering observed when samples were separated using PCA analysis (Figure 5.12). ICP elevating samples could not easily be distinguished from samples that did not cause ICP elevation following a PCA analysis.

Figure 5.12: Principle component analysis of hydrocephalus CSF samples. The normalised Log2 data for each sample were uploaded into the statistical software package Perseus. There was no obvious separation between hydrocephalus CSF samples that caused ICP elevation (above 5 mmHg, closed triangles) from those that did not (open triangles) when infused into naïve rats.

116 Proteins common to Stroke- and Hydrocephalus-CSF

Expression of LEG1 and KCC2B were expressed in higher abundance in both stroke and hydrocephalus ICP elevating CSF samples. Figure 5.13 shows the relationship between normalised log2 expression of LEG 1 and KCC2B with peak DICP from Part 1 of this chapter. There was no association between expression of LEG1 (R2 = 0.2, p=.1) and peak DICP. A moderate yet significant association was found between expression of KCC2B and peak DICP (R2 = 0.5, p=.002).

LEG1 KCC2B 4 5 R2 = 0.2, p=.1 R2 = 0.5, p=.002 Stroke 3 Hydrocephalus 0 2 expression expression 2 2 1 -5

0 -10 -1

-15

-2 Normalised Log Normalised Log -5 0 5 10 15 20 25 -5 0 5 10 15 20 25 Peak Δ ICP (mmHg) Peak Δ ICP (mmHg)

Figure 5.13: Relationship between peak change in intracranial pressure (DICP from Part 1 of this chapter) and the normalised Log2 expression of proteins within cerebrospinal fluid (CSF). CSF collected from humans suffering from stroke or hydrocephalus was injected intraventricularly into naïve recipient rats. ICP was measured before-, during- and after infusion. A second aliquot of CSF was then analysed using mass spectrometry and the data was normalised and transformed before being compared between samples. The relationship between the relative abundance of a. Galectin-1 (LEG1) and b. Calcium/calmodulin-dependent protein kinase type II subunit beta (KCC2B) and the peak D ICP is depicted in this figure.

117 5.3.4 Discussion

The aim of this study was to identify proteins in the CSF of stroke and hydrocephalus patients that may have contributed to the observed ICP elevation in recipient rats (Part 1 of this chapter). Proteomic analysis was used to compare the CSF proteome of samples that caused ICP elevation (DICP ³5 mmHg) and those that did not (DICP <5 mmHg). A total of 50 proteins (17 stroke and 33 hydrocephalus) were identified as being differentially expressed between sample groups. It is likely that CSF transfusion introduced the ICP elevating factor into the rat CSF or resulted in increased expression of an ICP regulating molecule in the rat brain. Therefore, I will discuss proteins that were found to be higher in abundance in samples that caused ICP elevation.

Within the stroke CSF samples, 15 proteins were found to be higher in abundance in ICP elevating CSF samples. DCC was shown to be one of the most significantly differentially expressed proteins between groups. DCC mediates netrin-1 responses and is implicated in axonal guidance and neural migration (246). Furthermore, DCC has been associated with apoptotic cell death in the absence of netrin-1 (247, 248) Both netrin-1 and netrin receptor DCC expression have been shown to be upregulated following experimental stroke peaking at 14 days in rats (249). In this study netrin-1 was not identified using our proteomic approach. Whether there was an absence of netrin-1 in the samples or the levels were below detection is unclear, but it is suggestive of the expression of DCC being linked with apoptosis. This is of particular interest as apoptosis is more likely to be present in reperfused, penumbral tissue (250) which, as we have shown is associated with delayed ICP elevation (81).

Three of the most statistically significant enriched gene ontology functions identified in ICP elevating CSF involve cell-cell communication via exosomes and gene regulation. Exosomes transport proteins, lipids and genetic material between cells in both physiological and pathophysiological conditions. Experimental studies have shown exosomal transport of miRNAs involved in regulating repair processes in the brain post-stroke (251). The ICP response observed in Part 1 of this thesis and our previous studies is delayed in nature. This indicates that the observed ICP elevation occurs in response to active cellular processes. Could this transport lead to synthetic processes in recipient animals that are involved in ICP regulation?

PCA analysis could not distinguish between the two populations of stroke CSF samples. This data suggest that whatever is causing the rise in ICP is subtle. Interestingly, the PCA did identify two distinct clusters, however the reason behind the separation needs further investigation.

118 When investigating the more abundant proteins in the ICP elevating hydrocephalus CSF two out of the 19 over expressed proteins (LEG1 and KCC2B) stood out as they also appeared as more abundant in the stroke ICP elevating CSF samples. LEG1 was found to be 10.2 and 6 fold higher in ICP elevating stroke- and hydrocephalus-CSF samples respectively. In the CNS, LEG1 is expressed in both neurons and glia (mainly astrocytes) (252). There have been no investigations into LEG1 expression in hydrocephalus. However, expression has been shown to increase in astrocytes surrounding photothrombotic lesions in the rat brain (253) and LEG1 has been shown to be neuroprotective against ischaemic cell damage (253, 254). Interestingly, mRNA for LEG1 has been detected in CP tissue (255), however the authors did not correlate gene expression with protein expression. Furthermore, the phenotype was not investigated. There was no relationship between normalised log2 expression of LEG1 and peak DICP (from Part 1 of this study). KCC2B expression was found to be 3.4 and 51.3 fold higher in ICP elevating stroke and hydrocephalus CSF samples respectively. KCC2B belongs to the serine/ threonine protein kinase family (256) and as with LEG1, KCC2B has not been investigated in the context of hydrocephalus. However, KCC2B has been well studied in the context of ischaemic stroke and was recently found to be upregulated in CSF at 2 hours post- stroke (181). The relationship between KCC2B expression and infarct volume has been investigated and whilst inhibition has been shown to be neuroprotective (257), knockdown has been shown to promote cell death (258). In the present study, normalised log2 expression of KCC2B was found to correlate with peak DICP (from Part 1 of this study). This suggests a potential role for KCC2B in delayed ICP elevation in rats. There is very little information regarding potential mechanisms of action of KCC2B on ICP. Evidence of a potential relationship between KCC2B and cerebral oedema comes from a study by Gunnarson et al. (2005) showing a binding site for KCC2B on AQP4, suggesting that the protein may be involved in AQP4 permeability or regulation (259). This may have important consequences for ICP as expression of AQP4 has been linked with the development and also the resolution of cerebral oedema (260). Unfortunately, amongst the numerous investigations into KCC2B expression and infarct volume none of the authors report specific oedema volumes, despite the infarct volume calculations being corrected for oedema, (257, 258). Further investigation is required to identify potential mechanisms of action of KCC2B on ICP elevation however it is a very promising protein.

As was the case with stroke, cell-cell communication via exosomes was the most enriched gene ontology term in ICP elevating hydrocephalus CSF. The remaining four terms were found to involve actin filament formation and assembly. A search of the literature did not return any studies that provide insight into these gene ontology terms and either hydrocephalus or ICP. However, plastin-1 and -3 expression (upregulated in hydrocephalus-CSF) was found to

119 be associated with these terms. A study by Narita et al. (2012) identified both proteins as being part of the proteome for cilia in juvenile swine CP epithelial cells (261). Upregulation of both proteins in CSF may be indicative of damage to the cilia which has been associated with altered CP function and the formation of hydrocephalus in mice (233). However, this is speculative and will need to be investigated further.

PCA analysis of the seven hydrocephalus CSF samples could not distinguish between ICP elevating and non-elevating CSF samples. Again, this indicates a subtle cause behind the observed ICP elevation in Part 1 of this chapter.

In this study I used state of the art proteomic technology to identify numerous protein candidates that may be involved in ICP elevation. The majority of these proteins have not previously been associated with ICP elevation. A strength of this study lies within its design. Although CSF is 99% water the remaining 1% is extremely complex. Under physiological conditions proteomic studies of the CSF have revealed expression of over 4000 different proteins (172) and CSF composition is known to change during disease e.g. stroke (181, 189- 195). In this study I directly linked ICP elevation in rats with protein expression. By comparing CSF samples based on their ICP elevating properties, I was able to minimise the identification of proteins related to patient differences, changes in the CSF proteome caused by stroke and hydrocephalus unrelated to ICP and I identified numerous proteomic candidates within each group that may be involved in the delayed ICP rise observed in rats. Additionally, the samples were run individually as opposed to being ‘pooled’, which would have masked individual differences. This allowed not only the comparison of CSF profiles between groups but permitted the comparison of expression between individual samples and to determine any common protein changes. Furthermore, by not pooling the samples expression of less abundant proteins were less likely to be masked by samples with more abundant proteins.

There are some limitations of these findings that must be taken into account before future studies are undertaken. Firstly, it must be acknowledged that this is a discovery study with a small sample size. Replication in a larger population and comparison to control CSF samples will be important to confirm the results of this study. Secondly, validation of protein expression by other techniques such as ELISA or immunoblotting is important to validate the mass spectrometry results. Thirdly, the samples were not immunodepleted. If the ICP elevating protein is expressed at low abundance then there is the possibility that its expression will be masked by proteins that are highly expressed in CSF such as albumin. Additionally, a longer liquid chromatography gradient might be useful to increase the number of identified proteins. Lastly, the data did not undergo any correction for multiple comparisons and therefore requires careful interpretation.

120 In conclusion, I have identified several proteins in the CSF of stroke and hydrocephalus patients whose expression may be related to ICP elevation in rats. This study should be regarded as a pilot discovery study to identify possible candidates involved in ICP elevation. However, if the validity of my results is confirmed then this may provide new targets for diagnosis and treatment of ICP elevation in stroke, hydrocephalus and potentially other neurological disorders in which raised ICP is a potential complication.

5.5 Summary

The first important result presented in this chapter was that infusion of CSF from stroke patients causes ICP elevation in recipient rats. This indicates the presence of ICP elevating factors within human stroke-CSF and suggests the existence of a novel mechanism of ICP elevation. The second important finding was that a similar ICP response to CSF infusion was also observed in chronic hydrocephalus-CSF recipient rats. If the newly identified ICP ‘trigger’ also occurs in chronic hydrocephalus it suggests the possibility that this potential mechanism may be important in other forms of neurological disease. Lastly, I have identified numerous potential molecular candidates that may be involved in the ICP response that was observed in recipient rats.

121 Chapter 6 General Discussion and Conclusions

The primary objective of this thesis was to determine whether there is a molecular component contributing to ICP elevation after ischemic stroke. The experiments performed in the preceding chapters were based upon preliminary data suggesting the presence of an ICP triggering molecule in post-stroke CSF. Before investigating this potential mechanism of ICP elevation the first step was to ensure that our method of ICP measurement was accurate and reliable for serial measurement of epidural pressure. The next step taken was to investigate AVP – an exciting candidate molecule that I had hypothesised may be involved in triggering ICP elevation in the recipient rats. I then aimed to translate the rat CSF-transfusion findings by testing whole CSF samples from stroke patients to determine if they induce ICP elevation in recipient rats, indicating presence of a triggering molecule. Lastly, I sought to investigate the CSF proteome to identify proteins that may be involved in the observed ICP elevation.

6.1 Epidural pressure measurement is accurate and reliable in rats using fibreoptic technology

Over the course of my PhD studies several solid state pressure catheters (fibreoptic and piezoelectric) were thoroughly investigated for their suitability in epidural pressure measurement. Substantial troubleshooting highlighted the importance of probe design for accurate and reliable epidural ICP measurement (Chapter 3: Part 1). From this work, I identified a robust and reliable fibreoptic pressure sensor for ICP monitoring in rats. Advantages of fibreoptic technology over fluid filled and piezoelectric pressure sensors include a greater sensitivity to the ICP pressure waveform (39) as well as less susceptibility to interference (36, 37, 41).

Using this pressure sensing technology, I made a number of important findings regarding epidural pressure measurement in rats. Firstly, I found that the surgical procedure itself does not influence baseline ICP as I found no difference in baseline ICP measured immediately- or 24 hours post-surgery (Chapter 3: Part 3). This was critical to ascertain in this study as ICP was compared between baseline ICP (measured on the same day as surgery) and 16-24 hours post-infusion. Many investigations into ICP test interventions several days to weeks after surgery, often with an indwelling catheter. A study by Uldall et al. (2014) showed that chronic implantation of an epidural catheter did not affect ICP. In contrast, chronic implantation of an intraventricular catheter resulted in the development of hydrocephalus (36). My findings show that a recovery period is not required for accurate and reliable ICP measurement. Secondly, I demonstrated that epidural pressure measurement is as accurate 122 as subdural and intraventricular pressure over a wide range of pressures (Chapter 3: Part 4). My findings expand on the existing literature showing the validity of epidural pressure measurement in rats using fluid filled technology (36, 60). Furthermore, to my knowledge this is the first time subdural ICP has been validated against any other monitoring location in the rat. Finally, I showed that both epidural and subdural pressure measurement are reliable over repeated measures at ‘pathological’ ICPs (Chapter 3: Part 5). This is of particular importance when investigating serial ICP measurements that may change over time as a result of an ICP elevating intervention.

However, this technique may also be confounded by improper placement of the probe tip in contact with the dura. In this thesis, I showed that pressure recorded in the epidural space is statistically significantly higher at baseline when the pressure sensor is in contact with the dura compared with those placed above, and not in contact with the dura (Chapter 3: Part 2). This highlights the importance of probe placement when measuring ICP. This may be an explanation for the conflicting reports of epidural accuracy in the literature (36, 38, 60). Importantly, despite the statistical significance, the biological significance of the higher baseline values were minor. While this may not be a major confounder for studies with huge effect sizes, it may affect interpretation of findings when the effect size is small. Thus, location of the tip should be an important technical consideration when monitoring ICP and methodological consideration in the development of new methods for ICP measurement.

In the clinical setting, intraventricular pressure measurement is considered the gold standard. Epidural pressure is not reliable in humans as the thick and inelastic dura impedes accurate ICP measurement. My results show that this is not the case in rats, likely due to the substantially thinner dura (68). There are many advantages of epidural pressure measurement including minimal tissue damage and reduced risks of infection and hydrocephalus (36). Therefore, the ability to measure epidural pressure is advantageous in experimental studies since unlike the situation in patients, we can measure pressure accurately without interfering noticeably with the system. My results expand on the efficacy of using epidural pressure measurement in rats and factors that may confound accurate pressure measurement in the epidural space. Coupled with the advantages of epidural pressure monitoring I argue that the epidural space is the best location for ICP monitoring in rats.

123 6.2 Evidence of a potential molecular trigger for ICP elevation post- stroke

I have previously shown ICP elevation in naïve rats following infusion of stroke CSF collected from donor rats (Figure 1.4, Chapter 1). This suggests that there is a molecule released into the CSF during stroke that triggers ICP elevation. Using the same biological assay, I have shown a similar ICP response in naive rats who received an infusion of human stroke-CSF (Chapter 5: Part 1). This suggests the presence of a similar molecular trigger in post-stroke human CSF and strengthens the clinical relevance of our experimental findings. Until now, the relationship between post-stroke CSF compositional changes and ICP elevation has not been studied. One reason for this is that although CSF collection is still a routine part of admission to a neurology ward in some parts of the world, CSF is not routinely collected from stroke patients in Australia, as it is invasive and is a potential contraindication for intravenous thrombolytic therapy and anticoagulants (184). Additionally, until the recent recognition by our group of an oedema independent mechanism of ICP elevation, oedema was assumed to be the sole cause of post-stroke ICP elevation, therefore other mechanisms had not been considered.

My findings contribute to the growing body of evidence challenging the long held assumption that post-stroke ICP elevation is solely caused by cerebral oedema after large, malignant infarction. Importantly, CSF was collected in this study from donors with mild-moderate stroke in both CSF transfusion studies (animal and human). This patient population do not experience significant oedema volumes and until recently were not thought to experience ICP elevation (84). Evidence from our animal studies suggests that the ICP elevation observed at 24 hours after small stroke occurs via a mechanism other than cerebral oedema (80-83). Although the specific mechanism behind the ICP elevation in this study was not directly investigated, my findings provide the first evidence of a potential trigger for ICP elevation in these patients.

6.2.1 Future directions

Collection of CSF coupled with non-invasive ICP monitoring

The results presented in Chapter 5: Part 1 show that ICP is elevated in rats following infusion of human stroke-CSF. One way to confirm the clinical relevance of these findings would be to measure ICP at the time of CSF collection and then again at 24 hours post-stroke. Identification of proteins within CSF could then be performed on samples associated with ICP rise in the CSF donors. When investigating patients with smaller strokes, patient selection would be extremely important due to the contraindications of lumbar puncture. Therefore, 124 patients who have small stroke with NIHSS <5 might be good candidates, as the use of tPA is not routine practice in many centres. As these patients present with minor symptoms it is assumed that the risk of thrombolytic treatment outweighs the benefit (262). Recent studies from our laboratory, including the results presented in this thesis, have suggested that mild- moderate stroke increases the risk of delayed ICP rise (80-84), which we have shown in a separate study, can reduce collateral blood flow. Reduction in collateral blood flow may further reduce the crucial blood supply to the penumbra potentially worsening outcome. It therefore seems imperative that this population is further investigated. ICP monitoring is not currently performed in patients with small strokes due to the invasiveness of the monitoring equipment and currently perceived risk-to-benefit ratio. Therefore, non-invasive ICP monitoring could be utilised.

In this study, only 5/9 stroke-CSF samples were shown to cause ICP elevation, suggesting that the concentration of the triggering molecule was different between samples. Isolating the CSF from only those patients with higher ICPs would theoretically provide an enriched sample and comparing them to CSF from non-ICP elevated patients might provide greater insight into the underlying molecular mechanisms.

Is there a similar molecular trigger in patients with large, malignant infarction?

It is unknown whether a similar molecular trigger exists in the CSF following malignant infarction. These patients typically experience ICP elevation at later time points than seen after small stroke (48-72 hours). At present, the explanation for ICP elevation in these patients is cerebral oedema, however, a small number of clinical studies suggest this may not be the complete picture (75, 76, 91). It is not unreasonable to hypothesise that there may be an alternate mechanism of ICP elevation in some patients with malignant infarction. However, mass effect is a contraindication of lumbar puncture due to the risk of herniation (184) so collection of these samples would have to be made in patients already undergoing decompressive hemicraniectomy, where CSF is readily accessible/ being removed to lower pressure.

Investigation of the mechanism behind the triggered ICP elevation

Similarities between the current study and our previous experiments suggest that this rise in ICP may occur independently of cerebral oedema. However, the exact mechanism of the rise was not investigated as part of this study. Therefore, experiments will be undertaken to elucidate the mechanism that is being activated by the CSF triggering molecule(s). According to the Monroe-Kellie doctrine, ICP elevation occurs via an uncompensated increase in tissue (cerebral oedema), cerebral blood or CSF volume. Serial CT imaging could be used for

125 assessment of global oedema (263) pre and post-infusion of human CSF although quantification may be difficult especially in rats due to poor spatial resolution. Similarly, CTP imaging would provide insight into changes in cerebral blood volume post-infusion. As discussed in Chapter 1, surrogate markers are often used to assess changes in CSF volume. Therefore, CSF production (101) and outflow resistance (105) could be compared between stroke- and control-CSF recipient groups to infer whether CSF volume might be increasing to contribute to the observed ICP elevation in these animals.

6.3 Investigation of potential triggering molecules

Having discovered that transfusion of post-stroke CSF between rats caused ICP elevation, the next step was identifying candidate molecules that may be involved in triggering ICP elevation in the recipient rats. The experiments performed in Chapter 4 were designed to directly investigate one molecule of interest – AVP. AVP was considered to be a molecule of interest based on the correlation between AVP concentration and ICP elevation in numerous neurological conditions including stroke (194, 213) as well as physiological evidence showing an association between intraventricular infusion of AVP and subsequent ICP elevation (197, 198). Even at the supramaximal dose (5 ng), AVP did not consistently cause ICP elevation in this study, suggesting that it is unlikely to be the sole molecule responsible for triggering ICP elevation in CSF recipient animals. However, as elevated AVP concentration is associated with ICP elevation in humans, my results do not rule out the possibility that AVP concentration still might be used as a diagnostic marker for raised ICP. Due to the complexity of the CSF proteome it was not feasible, nor ethical to investigate every protein of interest in this way. Therefore, I investigated changes in the proteome between samples based on their ICP elevating effects in rats. The combination of physiology with the molecular studies prevented the unnecessary investigation of proteins that may not be important to ICP elevation. Through this method, I identified 17 potential protein candidates for the triggering molecule(s) (Chapter 5: Part 2). Confirmation of a triggering molecule would have important implications for diagnosis and treatment of elevated ICP after stroke.

6.3.1 Future directions

Confirmation of the ICP elevation triggering molecule(s)

In this study, I identified a number of proteins that may be involved in the observed ICP elevation (Chapter 5). Future studies will confirm potential causative relationships between these proteins and ICP in experiments similar to that conducted in Chapter 4 with AVP, in which the protein will be infused directly into the CSF of a naïve rat. Any molecule(s) shown

126 to elevate ICP, will then be inhibited in the CSF immediately following stroke in a further set of experiments as a proof of concept ICP preventative treatment.

6.4 Could this ICP trigger be conserved across other neurological conditions involving raised ICP?

The observation of ICP elevation in animals that received infusion of hydrocephalus-CSF was a surprising finding (Chapter 5: Part 1). One plausible interpretation of this result could be that there is a similar molecular trigger causing ICP elevation in hydrocephalus-CSF. Interestingly, when investigating the CSF proteome, LEG-1 and KCC2B were found to be higher in abundance in both the stroke- and hydrocephalus-CSF samples that caused ICP elevation. Further investigation is required to determine the relevance of these proteins to ICP elevation, however the significance of this finding is that this mechanism may be conserved across these two diseases. It must be asked then, could this previously unrecognised, but potentially fundamental mechanism be important in other disorders of ICP elevation where, as is the case in stroke, mechanisms other than oedema have not generally been considered?

Most injury responses in biology are recruited in response to a range of different insults. Therefore, there may be some conservation in the mechanisms of neurological diseases that involve raised ICP. ICP is a known contributor to worsened outcomes in other acute neurological diseases including traumatic brain injury, intracerebral haemorrhage and SAH all of which have pathophysiological parallels with ischaemic stroke (31, 264, 265). All of these disorders involve blood-brain-barrier disruption, cerebral oedema, necrotic and apoptotic cell death and disturbed cerebral blood flow autoregulation. Furthermore, John and Colbourne (2016) have found evidence of an oedema independent rise in ICP after collagenase induced intracerebral haemorrhage in rats (92). Surprisingly, despite the dramatic acute ICP elevation seen in experimental SAH at the time of stroke (266), ICP elevation was not observed in recipients of SAH-CSF. Whether this is due to differing mechanisms (acute versus delayed ICP rise), or just due to small sample size (perhaps a similar ICP response dichotomy was simply missed), requires further investigation. However, it is possible that other ICP related disorders may also involve a CSF molecular triggering response.

One other neurological disorder of particular interest is idiopathic intracranial hypertension (chronically raised ICP with no apparent cause) (267, 268). In this population, ICP elevation is not associated with cerebral oedema (269). CSF composition has been suspected to be important in disease progression however, no specific molecule(s) has been determined as causing ICP elevation as yet. It is a plausible hypothesis that CSF composition may influence ICP elevation in these patients given traditional factors do not. Furthermore, since CSF

127 drainage is a therapeutic manoeuvre in these patients, obtaining samples for testing should be far less complicated than in stroke and controls.

6.4.1 Future directions

Assessing the relevance of the current findings in similar disorders

Given the small sample sizes of SAH- and control-CSF used in this study, a larger study is needed to confirm our current findings. In addition, investigating CSF samples from other neurological disorders, including traumatic brain injury, intracerebral haemorrhage and idiopathic intracranial hypertension will elucidate the conservation of this mechanism across diseases. Both animal studies and proteomic analysis would be performed.

128 6.5 Conclusions

The findings presented in this thesis are both exciting and have the potential to alter our understanding of ICP elevation after mild-moderate stroke. Firstly, I confirmed the accuracy and reliability of epidural pressure measurement in rats at both baseline and ‘pathological’ pressures. Moreover, I expanded on previous studies by validating the use of fibreoptic pressure sensing technology. Furthermore, insight gained during extensive troubleshooting ultimately led to the improvement of our original method. This was crucial for the success of my PhD and the ongoing experiments of the laboratory. Secondly, I showed that AVP is unlikely to be involved in the delayed ICP elevation observed in rats post-stroke or post- infusion of stroke-CSF. Thirdly, I found evidence of a potential ‘trigger’ of ICP elevation in human stroke-CSF. Excitingly, presence of a similar ICP trigger was observed in CSF from patients with chronic hydrocephalus. My findings contribute our labs growing body of evidence supporting the existence of ICP elevation after small stroke and present evidence of a novel mechanism of ICP elevation. Moreover, they raise questions over our current understanding of ICP elevation post-stroke and may extend to other disorders where raised ICP contributes to disability and/or death for which current treatment options are limited. Finally, I identified numerous potential triggering molecules that may contribute to the observed ICP elevation. Identification of a triggering molecule in CSF may one day aid in the diagnosis of patients at risk of ICP elevation. If confirmed, this will potentially provide a whole new avenue for therapy for these prevalent disorders.

129 Chapter 7 References

1. Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Med. 2006;3(11):e442. 2. Deloitte Access Economics. No postcode untouched - Stroke in Australia, 20172017. 3. Deloitte Access Economics. The economic impact of stroke in Australia, 2013. 2013. 4. Kalaria RN, Akinyemi R, Ihara M. Stroke injury, cognitive impairment and vascular dementia. Biochimica et biophysica acta. 2016;1862(5):915-25. 5. Investigators WMPP. The World Health Organisation MONICA Project (monitoring trends and determinants in cardiovascular disease). Journal of Clinical Epidemiology. 1988;41:105-14. 6. Australian Institute of Health and Welfare. Cardiovascular Disease Mortality: Trends at Different Ages. 2010. 7. Deb P, Sharma S, Hassan KM. Pathophysiologic mechanisms of acute ischemic stroke: An overview with emphasis on therapeutic significance beyond thrombolysis. Pathophysiology. 2010;17(3):197-218. 8. Xing C, Arai K, Lo EH, Hommel M. Pathophysiologic cascades in ischemic stroke. International journal of stroke : official journal of the International Stroke Society. 2012;7(5):378-85. 9. Dirnagl U, Iadecola C, Moskowitz MA. Pathobiology of ischaemic stroke: an integrated view. Trends Neurosci. 1999;22(9):391-7. 10. Astrup J, Siesjo BK, Symon L. Thresholds in cerebral ischemia - the ischemic penumbra. Stroke. 1981;12(6):723-5. 11. Baron JC. Perfusion thresholds in human cerebral ischemia: historical perspective and therapeutic implications. Cerebrovasc Dis. 2001;11 Suppl 1:2-8. 12. Brozici M, van der Zwan A, Hillen B. Anatomy and functionality of leptomeningeal anastomoses: a review. Stroke. 2003;34(11):2750-62. 13. Miteff F, Levi CR, Bateman GA, Spratt N, McElduff P, Parsons MW. The independent predictive utility of computed tomography angiographic collateral status in acute ischaemic stroke. Brain. 2009;132(Pt 8):2231-8. 14. Beard DJ, Murtha LA, McLeod DD, Spratt NJ. Intracranial Pressure and Collateral Blood Flow. Stroke. 2016;47(6):1695-700. 15. Ali LK, Saver JL. The ischemic stroke patient who worsens: new assessment and management approaches. Rev Neurol Dis. 2007;4(2):85-91. 16. Asdaghi N, Hameed B, Saini M, Jeerakathil T, Emery D, Butcher K. Acute perfusion and diffusion abnormalities predict early new MRI lesions 1 week after minor stroke and transient ischemic attack. Stroke. 2011;42(8):2191-5. 17. Campbell BC, Christensen S, Tress BM, Churilov L, Desmond PM, Parsons MW, et al. Failure of collateral blood flow is associated with infarct growth in ischemic stroke. J Cereb Blood Flow Metab. 2013;33(8):1168-72. 18. Georgiadis D, Engelter S, Tettenborn B, Hungerbuhler H, Luethy R, Muller F, et al. Early recurrent ischemic stroke in stroke patients undergoing intravenous thrombolysis. Circulation. 2006;114(3):237-41. 19. Coutts SB, Modi J, Patel SK, Demchuk AM, Goyal M, Hill MD, et al. CT/CT angiography and MRI findings predict recurrent stroke after transient ischemic attack and minor stroke: results of the prospective CATCH study. Stroke. 2012;43(4):1013-7. 20. Liebeskind DS. Collateral circulation. Stroke. 2003;34(9):2279-84. 21. Pranevicius O, Pranevicius M, Pranevicius H, Liebeskind DS. Transition to collateral flow after arterial occlusion predisposes to cerebral venous steal. Stroke. 2012;43(2):575-9. 22. Alexandrov AV, Sharma VK, Lao AY, Tsivgoulis G, Malkoff MD, Alexandrov AW. Reversed Robin Hood syndrome in acute ischemic stroke patients. Stroke. 2007;38(11):3045- 8.

130 23. Palamarchuk I, Kimpinski K, Lippert C, Hachinski V. Nocturnal deterioration after ischemic stroke and autonomic dysfunction: hypothesis and implications. Cerebrovasc Dis. 2013;36(5-6):454-61. 24. Urrutia VC, Wityk RJ. Blood pressure management in acute stroke. Neurol Clin. 2008;26(2):565-83, x-xi. 25. Czosnyka M, Pickard JD. Monitoring and interpretation of intracranial pressure. Journal of neurology, neurosurgery, and psychiatry. 2004;75(6):813-21. 26. Beard DJ, McLeod DD, Logan CL, Murtha LA, Imtiaz MS, van Helden DF, et al. Intracranial pressure elevation reduces flow through collateral vessels and the penetrating arterioles they supply. A possible explanation for 'collateral failure' and infarct expansion after ischemic stroke. J Cereb Blood Flow Metab. 2015. 27. Hacke W, Schwab S, Horn M, Spranger M, De Georgia M, von Kummer R. 'Malignant' middle cerebral artery territory infarction: clinical course and prognostic signs. Arch Neurol. 1996;53(4):309-15. 28. Jeon SB, Park JC, Kwon SU, Kim YJ, Lee S, Kang DW, et al. Intracranial Pressure Soon After Hemicraniectomy in Malignant Middle Cerebral Artery Infarction. J Intensive Care Med. 2018;33(5):310-6. 29. Ropper AH, Shafran B. Brain edema after stroke. Clinical syndrome and intracranial pressure. Arch Neurol. 1984;41(1):26-9. 30. Papo I, Janny P, Caruselli G, Colnet G, Luongo A. Intracranial pressure time course in primary intracerebral hemorrhage. Neurosurgery. 1979;4(6):504-11. 31. Zoerle T, Lombardo A, Colombo A, Longhi L, Zanier ER, Rampini P, et al. Intracranial pressure after subarachnoid hemorrhage. Critical care medicine. 2015;43(1):168-76. 32. Balestreri M, Czosnyka M, Hutchinson P, Steiner LA, Hiler M, Smielewski P, et al. Impact of intracranial pressure and cerebral perfusion pressure on severe disability and mortality after head injury. Neurocrit Care. 2006;4(1):8-13. 33. Brain Trauma F, American Association of Neurological S, Congress of Neurological S, Joint Section on N, Critical Care AC, Bratton SL, et al. Guidelines for the management of severe traumatic brain injury. VII. Intracranial pressure monitoring technology. J Neurotrauma. 2007;24 Suppl 1:S45-54. 34. Zhong J, Dujovny M, Park HK, Perez E, Perlin AR, Diaz FG. Advances in ICP monitoring techniques. Neurol Res. 2003;25(4):339-50. 35. Muralidharan R. External ventricular drains: Management and complications. Surg Neurol Int. 2015;6(Suppl 6):S271-4. 36. Uldall M, Juhler M, Skjolding AD, Kruuse C, Jansen-Olesen I, Jensen R. A novel method for long-term monitoring of intracranial pressure in rats. J Neurosci Methods. 2014;227:1-9. 37. Crenshaw AG, Styf JR, Mubarak SJ, Hargens AR. A new "transducer-tipped" fiber optic catheter for measuring intramuscular pressures. J Orthop Res. 1990;8(3):464-8. 38. Verlooy J, Selosse P, Van Reempts J, Haseldonckx M, Borgers M. Fiberoptic intracranial pressure monitoring in rats. J Neurosci Methods. 1990;31(1):1-6. 39. Murtha L, McLeod D, Spratt N. Epidural intracranial pressure measurement in rats using a fiber-optic pressure transducer. J Vis Exp. 2012(62). 40. Zacchetti L, Magnoni S, Di Corte F, Zanier ER, Stocchetti N. Accuracy of intracranial pressure monitoring: systematic review and meta-analysis. Crit Care. 2015;19:420. 41. Zhao X, Ho D, Gao S, Hong C, Vatner DE, Vatner SF. Arterial Pressure Monitoring in Mice. Curr Protoc Mouse Biol. 2011;1:105-22. 42. Kaufman KR, Wavering T, Morrow D, Davis J, Lieber RL. Performance characteristics of a pressure microsensor. J Biomech. 2003;36(2):283-7. 43. Lundberg N, Troupp H, Lorin H. Continuous recording of the ventricular-fluid pressure in patients with severe acute traumatic brain injury. A preliminary report. J Neurosurg. 1965;22(6):581-90. 44. Guillaume J, Janny P. [Continuous intracranial manometry; importance of the method and first results]. Rev Neurol (Paris). 1951;84(2):131-42.

131 45. Beer R, Lackner P, Pfausler B, Schmutzhard E. Nosocomial ventriculitis and meningitis in neurocritical care patients. J Neurol. 2008;255(11):1617-24. 46. Aucoin PJ, Kotilainen HR, Gantz NM, Davidson R, Kellogg P, Stone B. Intracranial pressure monitors. Epidemiologic study of risk factors and infections. Am J Med. 1986;80(3):369-76. 47. Binz DD, Toussaint LG, 3rd, Friedman JA. Hemorrhagic complications of ventriculostomy placement: a meta-analysis. Neurocrit Care. 2009;10(2):253-6. 48. Saladino A, White JB, Wijdicks EF, Lanzino G. Malplacement of ventricular catheters by neurosurgeons: a single institution experience. Neurocrit Care. 2009;10(2):248-52. 49. Bekar A, Dogan S, Abas F, Caner B, Korfali G, Kocaeli H, et al. Risk factors and complications of intracranial pressure monitoring with a fiberoptic device. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2009;16(2):236-40. 50. Berlin T, Murray-Krezan C, Yonas H. Comparison of parenchymal and ventricular intracranial pressure readings utilizing a novel multi-parameter intracranial access system. Springerplus. 2015;4:10. 51. Eide PK. Comparison of simultaneous continuous intracranial pressure (ICP) signals from ICP sensors placed within the brain parenchyma and the epidural space. Med Eng Phys. 2008;30(1):34-40. 52. Eide PK, Sorteberg W. Simultaneous measurements of intracranial pressure parameters in the epidural space and in brain parenchyma in patients with hydrocephalus. J Neurosurg. 2010;113(6):1317-25. 53. Lenfeldt N, Koskinen LO, Bergenheim AT, Malm J, Eklund A. CSF pressure assessed by lumbar puncture agrees with intracranial pressure. Neurology. 2007;68(2):155-8. 54. Reina MA, Lopez-Garcia A, Dittmann M, de Andres JA. [Structural analysis of the thickness of human dura mater with scanning electron microscopy]. Rev Esp Anestesiol Reanim. 1996;43(4):135-7. 55. Kuchiwaki H, Inao S, Ishii N, Ogura Y, Gu SP. Human dural thickness measured by ultrasonographic method: reflection of intracranial pressure. J Ultrasound Med. 1997;16(11):725-30. 56. Poca MA, Sahuquillo J, Topczewski T, Penarrubia MJ, Muns A. Is intracranial pressure monitoring in the epidural space reliable? Fact and fiction. J Neurosurg. 2007;106(4):548-56. 57. Bruder N, N'Zoghe P, Graziani N, Pelissier D, Grisoli F, Francois G. A comparison of extradural and intraparenchymatous intracranial pressures in head injured patients. Intensive Care Med. 1995;21(10):850-2. 58. Jiang J, Tyssebotn I. Measurement of cerebrospinal fluid pressure in conscious rats. Undersea Hyperb Med. 1997;24(1):39-43. 59. Mandell EC, Zimmermann E. Continuous measurement of cerebrospinal fluid pressure in unrestrained rats. Physiology & behavior. 1980;24(2):399-402. 60. Hiploylee C, Colbourne F. Intracranial pressure measured in freely moving rats for days after intracerebral hemorrhage. Experimental neurology. 2014;255:49-55. 61. McLeod D, Murtha L, Pitsillides K. A new method of attaching a solid-state pressure sensor for measurement of chronic intracranial pressure in freely moving rats using biotelemetry. The FASEB Journal 2012;26:1098.14. 62. McLeod D, Murtha L, Burke J, Pitsillides K. Biotelemetry of intracranial pressure in freely moving rats: a detailed description of surgical implantation for chronic studies. The FASEB Journal 2013;21(1):700.11. 63. Guild SJ, McBryde FD, Malpas SC. Recording of intracranial pressure in conscious rats via telemetry. J Appl Physiol (1985). 2015;119(5):576-81. 64. Zwienenberg M, Gong QZ, Lee LL, Berman RF, Lyeth BG. ICP monitoring in the rat: comparison of monitoring in the ventricle, brain parenchyma, and cisterna magna. J Neurotrauma. 1999;16(11):1095-102. 65. Kusaka G, Calvert JW, Smelley C, Nanda A, Zhang JH. New lumbar method for monitoring cerebrospinal fluid pressure in rats. J Neurosci Methods. 2004;135(1-2):121-7.

132 66. Jallo J, Saetzler R, Mishke C, Young WF, Vasthare U, Tuma RF. A chronic model to simultaneously measure intracranial pressure, cerebral blood flow, and study the pial microvasculature. J Neurosci Methods. 1997;75(2):155-60. 67. Silasi G, MacLellan CL, Colbourne F. Use of telemetry blood pressure transmitters to measure intracranial pressure (ICP) in freely moving rats. Curr Neurovasc Res. 2009;6(1):62- 9. 68. Maikos JT, Elias RA, Shreiber DI. Mechanical properties of dura mater from the rat brain and spinal cord. J Neurotrauma. 2008;25(1):38-51. 69. Simard JM, Sahuquillo J, Sheth KN, Kahle KT, Walcott BP. Managing malignant cerebral infarction. Curr Treat Options Neurol. 2011;13(2):217-29. 70. Robba C, Bacigaluppi S, Cardim D, Donnelly J, Bertuccio A, Czosnyka M. Non- invasive assessment of intracranial pressure. Acta neurologica Scandinavica. 2016;134(1):4- 21. 71. Raboel PH, Bartek J, Jr., Andresen M, Bellander BM, Romner B. Intracranial Pressure Monitoring: Invasive versus Non-Invasive Methods-A Review. Crit Care Res Pract. 2012;2012:950393. 72. Naqvi J, Yap KH, Ahmad G, Ghosh J. Transcranial Doppler ultrasound: a review of the physical principles and major applications in critical care. Int J Vasc Med. 2013;2013:629378. 73. Schmidt B, Czosnyka M, Schwarze JJ, Sander D, Gerstner W, Lumenta CB, et al. Cerebral vasodilatation causing acute intracranial hypertension: a method for noninvasive assessment. J Cereb Blood Flow Metab. 1999;19(9):990-6. 74. Schmidt B, Czosnyka M, Smielewski P, Plontke R, Schwarze JJ, Klingelhofer J, et al. Noninvasive Assessment of ICP: Evaluation of New TBI Data. Acta Neurochir Suppl. 2016;122:69-73. 75. Schwab S, Aschoff, A., Spranger, M., Albert, F., Hacke, W. The value of intracranial pressure monitoring in acute hemispheric stroke. Neurology. 1996;47(2):393-8. 76. Frank JI. Large hemispheric infarction, deterioration, and intracranial pressure. Neurology. 1995;45(7):1286-90. 77. Kotwica Z, Hårdemark H, Persson L. Intracranial pressure changes following middle cerebral artery occlusion in rats. Res Exp Med. 1991;191(2):99-104. 78. Bell RD, Frazer GD, Osterholm JL, Duckett SW. A novel treatment for ischemic intracranial hypertension in cats. Stroke. 1991;22(1):80-3. 79. Wells AJ, Vink R, Helps SC, Knox SJ, Blumbergs PC, Turner RJ. Elevated Intracranial Pressure and Cerebral Edema following Permanent MCA Occlusion in an Ovine Model. PLoS One. 2015;10(6):e0130512. 80. Murtha LA, McLeod DD, McCann SK, Pepperall D, Chung S, Levi CR, et al. Short- duration hypothermia after ischemic stroke prevents delayed intracranial pressure rise. International journal of stroke : official journal of the International Stroke Society. 2014;9(5):553-9. 81. Beard DJ, Logan CL, McLeod DD, Hood RJ, Pepperall D, Murtha LA, et al. Ischemic penumbra as a trigger for intracranial pressure rise - A potential cause for collateral failure and infarct progression? J Cereb Blood Flow Metab. 2016;36(5):917-27. 82. Murtha LA, Beard DJ, Bourke JT, Pepperall D, McLeod DD, Spratt NJ. Intracranial Pressure Elevation 24 h after Ischemic Stroke in Aged Rats Is Prevented by Early, Short Hypothermia Treatment. Front Aging Neurosci. 2016;8:124. 83. Murtha LA, McLeod DD, Pepperall D, McCann SK, Beard DJ, Tomkins AJ, et al. Intracranial pressure elevation after ischemic stroke in rats: cerebral edema is not the only cause, and short-duration mild hypothermia is a highly effective preventive therapy. J Cereb Blood Flow Metab. 2015;35(12):2109. 84. Kovacs T, Murtha L, Beard D, McLoeod D, Hood R, García-Esperón C, et al. Intracranial Pressure Rises 24 Hours After Mild-Moderare Ischaemic Stroke - A Potential Trigger For Early Neurological Deterioration? European Stroke Journal. 2017;2(15):98-478. 85. Monroe J. Observations on the structures and functions of the nervours system. Edinburgh. 1783.

133 86. Kellie G. An account of the appearances observed in the dissection of two of the three individuals presumed to have perished in the storm of the 3rd, and whose bodies were discovered in the vicinity of Leith on the morning of the 4th November 1821 with some reflections on the pathology of the brain. Trans Med Chir Sci, Edinburgh. 1824;1:84-169. 87. Burrows G. On Disorders of the Cerebral Circulation, and on the Connection between Affections of the Brain and Diseases of the Heart. The Medico-Chirurgical Review. 1846;4(7):34-48. 88. Simard JM, Kent TA, Chen M, Tarasov KV, Gerzanich V. Brain oedema in focal ischaemia: molecular pathophysiology and theoretical implications. Lancet Neurol. 2007;6(3):258-68. 89. Mori K, Miyazaki M, Iwase H, Maeda M. Temporal profile of changes in brain tissue extracellular space and extracellular ion (Na(+), K(+)) concentrations after cerebral ischemia and the effects of mild cerebral hypothermia. J Neurotrauma. 2002;19(10):1261-70. 90. Ayata C, Ropper AH. Ischaemic brain oedema. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2002;9(2):113-24. 91. Poca MA, Benejam B, Sahuquillo J, Riveiro M, Frascheri L, Merino MA, et al. Monitoring intracranial pressure in patients with malignant middle cerebral artery infarction: is it useful? J Neurosurg. 2010;112(3):648-57. 92. John RF, Colbourne F. Delayed localized hypothermia reduces intracranial pressure following collagenase-induced intracerebral hemorrhage in rat. Brain Res. 2016;1633:27-36. 93. Mellander S, Johansson B. Control of resistance, exchange, and capacitance functions in the peripheral circulation. Pharmacol Rev. 1968;20(3):117-96. 94. Ridha MA, Saindane AM, Bruce BB, Riggeal BD, Kelly LP, Newman NJ, et al. MRI findings of elevated intracranial pressure in cerebral venous thrombosis versus idiopathic intracranial hypertension with transverse sinus stenosis. Neuroophthalmology. 2013;37(1):1- 6. 95. Yada K, Nakagawa Y, Tsuru M. Circulatory disturbance of the venous system during experimental intracranial hypertension. J Neurosurg. 1973;39(6):723-9. 96. Preuss M, Hoffmann KT, Reiss-Zimmermann M, Hirsch W, Merkenschlager A, Meixensberger J, et al. Updated physiology and pathophysiology of CSF circulation--the pulsatile vector theory. Childs Nerv Syst. 2013;29(10):1811-25. 97. Johanson CE, Duncan JA, 3rd, Klinge PM, Brinker T, Stopa EG, Silverberg GD. Multiplicity of cerebrospinal fluid functions: New challenges in health and disease. Cerebrospinal Fluid Res. 2008;5:10. 98. Jellinger G. Anatomopathology of non-tumoral aqueductal stenosis. J Neurosurg Sci. 1986;30(1-2):1-16. 99. Murtha LA, Yang Q, Parsons MW, Levi CR, Beard DJ, Spratt NJ, et al. Cerebrospinal fluid is drained primarily via the spinal canal and olfactory route in young and aged spontaneously hypertensive rats. Fluids Barriers CNS. 2014;11:12. 100. Chiu C, Miller MC, Caralopoulos IN, Worden MS, Brinker T, Gordon ZN, et al. Temporal course of cerebrospinal fluid dynamics and amyloid accumulation in the aging rat brain from three to thirty months. Fluids Barriers CNS. 2012;9(1):3. 101. Karimy JK, Kahle KT, Kurland DB, Yu E, Gerzanich V, Simard JM. A novel method to study cerebrospinal fluid dynamics in rats. J Neurosci Methods. 2015;241:78-84. 102. Pappenheimer JR, Heisey SR, Jordan EF, Downer J. Perfusion of the cerebral ventricular system in unanesthetized goats. American Journal of Physiology. 1962;230(5):763-44. 103. Heisey SR, Pappenheimer JR, Held D. Bulk Flow and Diffusion in Cerebrospinal Fluid System of Goat. American Journal of Physiology. 1962;203(5):775-&. 104. Yasuda T, Tomita T, McLone DG, Donovan M. Measurement of cerebrospinal fluid output through external ventricular drainage in one hundred infants and children: correlation with cerebrospinal fluid production. Pediatr Neurosurg. 2002;36(1):22-8. 105. Nagra G, Wagshul ME, Rashid S, Li J, McAllister JP, 2nd, Johnston M. Elevated CSF outflow resistance associated with impaired lymphatic CSF absorption in a rat model of kaolin- induced communicating hydrocephalus. Cerebrospinal Fluid Res. 2010;7(1):4.

134 106. Sullivan HG, Miller JD, Griffith RL, 3rd, Carter W, Jr., Rucker S. Bolous versus steady- state infusion for determination of CSF outflow resistance. Ann Neurol. 1979;5(3):228-38. 107. Marmarou A, Shulman K, Rosende RM. A nonlinear analysis of the cerebrospinal fluid system and intracranial pressure dynamics. J Neurosurg. 1978;48(3):332-44. 108. Oreskovic D, Klarica M. Measurement of cerebrospinal fluid formation and absorption by ventriculo-cisternal perfusion: what is really measured? Croat Med J. 2014;55(4):317-27. 109. Martins AN, Newby N, Doyle TF. Sources of error in measuring cerebrospinal fluid formation by ventriculocisternal perfusion. Journal of neurology, neurosurgery, and psychiatry. 1977;40(7):645-50. 110. Marakovic J, Oreskovic D, Jurjevic I, Rados M, Chudy D, Klarica M. Potential error in ventriculocisternal perfusion method for determination of cerebrospinal fluid formation rate in cats. Coll Antropol. 2011;35 Suppl 1:73-7. 111. Oreskovic D, Klarica M, Vukic M, Marakovic J. Evaluation of ventriculo-cisternal perfusion model as a method to study cerebrospinal fluid formation. Croat Med J. 2003;44(2):161-4. 112. Pollay M, Curl F. Secretion of cerebrospinal fluid by the ventricular ependyma of the rabbit. Am J Physiol. 1967;213(4):1031-8. 113. Welch K. Secretion of Cerebrospinal Fluid by Choroid Plexus of the Rabbit. Am J Physiol. 1963;205:617-24. 114. Dandy WE. Experimental Hydrocephalus. Ann Surg. 1919;70(2):129-42. 115. Oreskovic D, Klarica M. The formation of cerebrospinal fluid: nearly a hundred years of interpretations and misinterpretations. Brain Res Rev. 2010;64(2):241-62. 116. Davson H. The Cerebrospinal Fluid. Ergeb Physiol. 1963;52:20-72. 117. Steffensen AB, Oernbo EK, Stoica A, Gerkau NJ, Barbuskaite D, Tritsaris K, et al. Cotransporter-mediated water transport underlying cerebrospinal fluid formation. Nat Commun. 2018;9(1):2167. 118. Oshio K, Watanabe H, Song Y, Verkman AS, Manley GT. Reduced cerebrospinal fluid production and intracranial pressure in mice lacking choroid plexus water channel Aquaporin- 1. FASEB J. 2005;19(1):76-8. 119. Mobasheri A, Marples D. Expression of the AQP-1 water channel in normal human tissues: a semiquantitative study using tissue microarray technology. American journal of physiology Cell physiology. 2004;286(3):C529-37. 120. Murphy VA, Johanson CE. Na(+)-H+ exchange in choroid plexus and CSF in acute metabolic acidosis or alkalosis. Am J Physiol. 1990;258(6 Pt 2):F1528-37. 121. Redzic Z. Molecular biology of the blood-brain and the blood-cerebrospinal fluid barriers: similarities and differences. Fluids Barriers CNS. 2011;8(1):3. 122. Lindvall M, Owman C. Autonomic nerves in the mammalian choroid plexus and their influence on the formation of cerebrospinal fluid. J Cereb Blood Flow Metab. 1981;1(3):245- 66. 123. Javaheri S, Wagner KR. Bumetanide decreases canine cerebrospinal fluid production. In vivo evidence for NaCl cotransport in the central nervous system. J Clin Invest. 1993;92(5):2257-61. 124. Johanson CE, Preston JE, Chodobski A, Stopa EG, Szmydynger-Chodobska J, McMillan PN. AVP V1 receptor-mediated decrease in Cl- efflux and increase in dark cell number in choroid plexus epithelium. Am J Physiol. 1999;276(1 Pt 1):C82-90. 125. Johanson CE, Szmydynger-Chodobska J, Chodobski A, Baird A, McMillan P, Stopa EG. Altered formation and bulk absorption of cerebrospinal fluid in FGF-2-induced hydrocephalus. Am J Physiol. 1999;277(1 Pt 2):R263-71. 126. Oshio K, Song Y, Verkman AS, Manley GT. Aquaporin-1 deletion reduces osmotic water permeability and cerebrospinal fluid production. Acta Neurochir Suppl. 2003;86:525-8. 127. Johansson PA, Dziegielewska KM, Ek CJ, Habgood MD, Mollgard K, Potter A, et al. Aquaporin-1 in the choroid plexuses of developing mammalian brain. Cell Tissue Res. 2005;322(3):353-64.

135 128. Akdemir G, Kaymaz F, Gursoy-Ozdemir Y, Akalan N, Akdemir ES. The time course changes in expression of aquaporin 4 and aquaporin 1 following global cerebral ischemic edema in rat. Surg Neurol Int. 2016;7:4. 129. Preston D, Simpson S, Halm D, Hochstetler A, Schwerk C, Schroten H, et al. Activation of TRPV4 stimulates transepithelial ion flux in a porcine choroid plexus cell line. American journal of physiology Cell physiology. 2018;315(3):C357-C66. 130. Jie P, Lu Z, Hong Z, Li L, Zhou L, Li Y, et al. Activation of Transient Receptor Potential Vanilloid 4 is Involved in Neuronal Injury in Middle Cerebral Artery Occlusion in Mice. Mol Neurobiol. 2016;53(1):8-17. 131. Jie P, Tian Y, Hong Z, Li L, Zhou L, Chen L, et al. Blockage of transient receptor potential vanilloid 4 inhibits brain edema in middle cerebral artery occlusion mice. Front Cell Neurosci. 2015;9:141. 132. Aziz AA, Coleman L, Morokoff A, Maixner W. Diffuse choroid plexus hyperplasia: an under-diagnosed cause of hydrocephalus in children? Pediatr Radiol. 2005;35(8):815-8. 133. Milhorat TH, Hammock MK, Davis DA, Fenstermacher JD. Choroid plexus papilloma. I. Proof of cerebrospinal fluid overproduction. Childs Brain. 1976;2(5):273-89. 134. Rickert CH, Paulus W. Tumors of the choroid plexus. Microsc Res Tech. 2001;52(1):104-11. 135. Bauer DF, Tubbs RS, Acakpo-Satchivi L. Mycoplasma meningitis resulting in increased production of cerebrospinal fluid: case report and review of the literature. Childs Nerv Syst. 2008;24(7):859-62. 136. Mestre H, Tithof J, Du T, Song W, Peng W, Sweeney AM, et al. Flow of cerebrospinal fluid is driven by arterial pulsations and is reduced in hypertension. Nat Commun. 2018;9(1):4878. 137. Yamada S, Miyazaki M, Yamashita Y, Ouyang C, Yui M, Nakahashi M, et al. Influence of respiration on cerebrospinal fluid movement using magnetic resonance spin labeling. Fluids Barriers CNS. 2013;10(1):36. 138. Ridgway JP, Turnbull LW, Smith MA. Demonstration of pulsatile cerebrospinal-fluid flow using magnetic resonance phase imaging. Br J Radiol. 1987;60(713):423-7. 139. Iliff JJ, Wang M, Liao Y, Plogg BA, Peng W, Gundersen GA, et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid beta. Sci Transl Med. 2012;4(147):147ra11. 140. Gjerris F, Borgesen SE, Sorensen PS, Boesen F, Schmidt K, Harmsen A, et al. Resistance to cerebrospinal fluid outflow and intracranial pressure in patients with hydrocephalus after subarachnoid haemorrhage. Acta Neurochir (Wien). 1987;88(3-4):79-86. 141. Tripathi R. Tracing the bulk outflow route of cerebrospinal fluid by transmission and scanning electron microscopy. Brain Res. 1974;80(3):503-6. 142. Welch K, Friedman V. The cerebrospinal fluid valves. Brain. 1960;83:454-69. 143. Silverberg GD, Mayo M, Saul T, Rubenstein E, McGuire D. Alzheimer's disease, normal-pressure hydrocephalus, and senescent changes in CSF circulatory physiology: a hypothesis. Lancet Neurol. 2003;2(8):506-11. 144. Nagra G, Koh L, Zakharov A, Armstrong D, Johnston M. Quantification of cerebrospinal fluid transport across the cribriform plate into lymphatics in rats. Am J Physiol Regul Integr Comp Physiol. 2006;291(5):R1383-9. 145. Johnston M, Zakharov A, Papaiconomou C, Salmasi G, Armstrong D. Evidence of connections between cerebrospinal fluid and nasal lymphatic vessels in humans, non-human primates and other mammalian species. Cerebrospinal Fluid Res. 2004;1(1):13. 146. Kida S, Pantazis A, Weller RO. CSF drains directly from the subarachnoid space into nasal lymphatics in the rat. Anatomy, histology and immunological significance. Neuropathol Appl Neurobiol. 1993;19(6):480-8. 147. Casley-Smith JR, Foldi-Borsok E, Foldi M. The prelymphatic pathways of the brain as revealed by cervical lymphatic obstruction and the passage of particles. Br J Exp Pathol. 1976;57(2):179-88. 148. Brierley JB, Field EJ. The connexions of the spinal sub-arachnoid space with the lymphatic system. J Anat. 1948;82(3):153-66.

136 149. Zakharov A, Papaiconomou C, Djenic J, Midha R, Johnston M. Lymphatic cerebrospinal fluid absorption pathways in neonatal sheep revealed by subarachnoid injection of Microfil. Neuropathol Appl Neurobiol. 2003;29(6):563-73. 150. Aspelund A, Antila S, Proulx ST, Karlsen TV, Karaman S, Detmar M, et al. A dural lymphatic vascular system that drains brain interstitial fluid and macromolecules. J Exp Med. 2015;212(7):991-9. 151. Louveau A, Smirnov I, Keyes TJ, Eccles JD, Rouhani SJ, Peske JD, et al. Structural and functional features of central nervous system lymphatic vessels. Nature. 2015;523(7560):337-41. 152. Mollanji R, Bozanovic-Sosic R, Zakharov A, Makarian L, Johnston MG. Blocking cerebrospinal fluid absorption through the cribriform plate increases resting intracranial pressure. Am J Physiol Regul Integr Comp Physiol. 2002;282(6):R1593-9. 153. Mollanji R, Bozanovic-Sosic R, Silver I, Li B, Kim C, Midha R, et al. Intracranial pressure accommodation is impaired by blocking pathways leading to extracranial lymphatics. Am J Physiol Regul Integr Comp Physiol. 2001;280(5):R1573-81. 154. Ethell D, Woltjer R. Changes in Cribriform Plate Morphology are Associated with Alzheimer’s Disease. Alzheimer's & Dementia: The Journal of the Alzheimer's Association. 2017;13(7):1282. 155. de Leon MJ, Li Y, Okamura N, Tsui WH, Saint-Louis LA, Glodzik L, et al. Cerebrospinal Fluid Clearance in Alzheimer Disease Measured with Dynamic PET. J Nucl Med. 2017;58(9):1471-6. 156. Peng W, Achariyar TM, Li B, Liao Y, Mestre H, Hitomi E, et al. Suppression of glymphatic fluid transport in a mouse model of Alzheimer's disease. Neurobiol Dis. 2016;93:215-25. 157. Trevisi G, Frassanito P, Di Rocco C. Idiopathic cerebrospinal fluid overproduction: case-based review of the pathophysiological mechanism implied in the cerebrospinal fluid production. Croat Med J. 2014;55(4):377-87. 158. Nagahiro S, Goto S, Korematsu K, Sumi M, Takahashi M, Ushio Y. Disruption of the blood-cerebrospinal fluid barrier by transient cerebral ischemia. Brain Res. 1994;633(1- 2):305-11. 159. Ennis SR, Keep RF. The effects of cerebral ischemia on the rat choroid plexus. J Cereb Blood Flow Metab. 2006;26(5):675-83. 160. Gillardon F, Lenz C, Kuschinsky W, Zimmermann M. Evidence for apoptotic cell death in the choroid plexus following focal cerebral ischemia. Neuroscience letters. 1996;207(2):113-6. 161. McLeod DD, Beard DJ, Parsons MW, Levi CR, Calford MB, Spratt NJ. Inadvertent occlusion of the anterior choroidal artery explains infarct variability in the middle cerebral artery thread occlusion stroke model. PLoS One. 2013;8(9):e75779. 162. Johanson CE, Palm DE, Primiano MJ, McMillan PN, Chan P, Knuckey NW, et al. Choroid plexus recovery after transient forebrain ischemia: role of growth factors and other repair mechanisms. Cell Mol Neurobiol. 2000;20(2):197-216. 163. Hu X, Brannstrom T, Gu W, Wester P. A photothrombotic ring stroke model in rats with or without late spontaneous reperfusion in the region at risk. Brain Res. 1999;849(1-2):175- 86. 164. Hossmann KA. Disturbances of cerebral protein synthesis and ischemic cell death. Prog Brain Res. 1993;96:161-77. 165. Hossmann KA. Viability thresholds and the penumbra of focal ischemia. Ann Neurol. 1994;36(4):557-65. 166. Sharp FR, Lu A, Tang Y, Millhorn DE. Multiple molecular penumbras after focal cerebral ischemia. J Cereb Blood Flow Metab. 2000;20(7):1011-32. 167. Kinouchi H, Sharp FR, Koistinaho J, Hicks K, Kamii H, Chan PH. Induction of heat shock hsp70 mRNA and HSP70 kDa protein in neurons in the 'penumbra' following focal cerebral ischemia in the rat. Brain Res. 1993;619(1-2):334-8. 168. Hohfeld J, Cyr DM, Patterson C. From the cradle to the grave: molecular chaperones that may choose between folding and degradation. EMBO Rep. 2001;2(10):885-90.

137 169. Ravagnan L, Gurbuxani S, Susin SA, Maisse C, Daugas E, Zamzami N, et al. Heat- shock protein 70 antagonizes apoptosis-inducing factor. Nat Cell Biol. 2001;3(9):839-43. 170. Fishman RA. Cerebrospinal fluid in disease of the nervous system. Philadelphia: Saunders; 1980. 171. Zhang Y, Guo Z, Zou L, Yang Y, Zhang L, Ji N, et al. A comprehensive map and functional annotation of the normal human cerebrospinal fluid proteome. J Proteomics. 2015;119:90-9. 172. Pan S, Zhu D, Quinn JF, Peskind ER, Montine TJ, Lin B, et al. A combined dataset of human cerebrospinal fluid proteins identified by multi-dimensional chromatography and tandem mass spectrometry. Proteomics. 2007;7(3):469-73. 173. Maurer MH. Proteomics of brain extracellular fluid (ECF) and cerebrospinal fluid (CSF). Mass Spectrom Rev. 2010;29(1):17-28. 174. Kleine TO, Zwerenz P, Zofel P, Shiratori K. New and old diagnostic markers of meningitis in cerebrospinal fluid (CSF). Brain Res Bull. 2003;61(3):287-97. 175. Polman CH, Reingold SC, Banwell B, Clanet M, Cohen JA, Filippi M, et al. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol. 2011;69(2):292-302. 176. Hakvoort A, Johanson CE. Growth factor modulation of CSF formation by isolated choroid plexus: FGF-2 vs. TGF-beta1. European journal of pediatric surgery : official journal of Austrian Association of Pediatric Surgery [et al] = Zeitschrift fur Kinderchirurgie. 2000;10 Suppl 1:44-6. 177. Johanson C, Stopa E, Baird A, Sharma H. Traumatic brain injury and recovery mechanisms: peptide modulation of periventricular neurogenic regions by the choroid plexus- CSF nexus. J Neural Transm. 2011;118(1):115-33. 178. Johanson CE, Gonzalez AM, Stopa EG. Water-imbalance-induced expression of FGF- 2 in fluid-regulatory centers: choroid plexus and neurohypophysis. European journal of pediatric surgery : official journal of Austrian Association of Pediatric Surgery [et al] = Zeitschrift fur Kinderchirurgie. 2001;11 Suppl 1:S37-8. 179. Johanson CE, McMillan PN, Palm DE, Stopa EG, Doberstein CE, Duncan JA. Volume Transmission-Mediated Protective Impact of Choroid Plexus - Cerebrospinal Fluid Growth Factors on Forebrain Ischemic Injury. In: Westman HSSaJ, editor. Blood-spinal cord and brian barriers in health and disease. Amsterdam ; Boston: Elsevier Academic Press; 2004. 180. Zetterberg H, Smith DH, Blennow K. Biomarkers of mild traumatic brain injury in cerebrospinal fluid and blood. Nat Rev Neurol. 2013;9(4):201-10. 181. Simats A, Garcia-Berrocoso T, Ramiro L, Giralt D, Gill N, Penalba A, et al. Characterization of the rat cerebrospinal fluid proteome following acute cerebral ischemia using an aptamer-based proteomic technology. Sci Rep. 2018;8(1):7899. 182. Del Bigio MR. Hydrocephalus-induced changes in the composition of cerebrospinal fluid. Neurosurgery. 1989;25(3):416-23. 183. Wu Q, Wang XL, Yu Q, Pan H, Zhang XS, Zhang QR, et al. Inflammasome Proteins in Cerebrospinal Fluid of Patients with Subarachnoid Hemorrhage are Biomarkers of Early Brain Injury and Functional Outcome. World Neurosurg. 2016;94:472-9. 184. Engelborghs S, Niemantsverdriet E, Struyfs H, Blennow K, Brouns R, Comabella M, et al. Consensus guidelines for lumbar puncture in patients with neurological diseases. Alzheimers Dement (Amst). 2017;8:111-26. 185. Kastrup A, Engelhorn T, Beaulieu C, de Crespigny A, Moseley ME. Dynamics of cerebral injury, perfusion, and blood-brain barrier changes after temporary and permanent middle cerebral artery occlusion in the rat. J Neurol Sci. 1999;166(2):91-9. 186. Jiang Q, Ewing JR, Ding GL, Zhang L, Zhang ZG, Li L, et al. Quantitative evaluation of BBB permeability after embolic stroke in rat using MRI. J Cereb Blood Flow Metab. 2005;25(5):583-92. 187. Preston E, Sutherland G, Finsten A. Three openings of the blood-brain barrier produced by forebrain ischemia in the rat. Neuroscience letters. 1993;149(1):75-8.

138 188. Belayev L, Busto R, Zhao W, Ginsberg MD. Quantitative evaluation of blood-brain barrier permeability following middle cerebral artery occlusion in rats. Brain Res. 1996;739(1- 2):88-96. 189. Beridze M, Sanikidze T, Shakarishvili R, Intskirveli N, Bornstein NM. Selected acute phase CSF factors in ischemic stroke: findings and prognostic value. BMC neurology. 2011;11:41. 190. Parakh N, Gupta HL, Jain A. Evaluation of enzymes in serum and cerebrospinal fluid in cases of stroke. Neurol India. 2002;50(4):518-9. 191. Brouns R, De Vil B, Cras P, De Surgeloose D, Marien P, De Deyn PP. Neurobiochemical markers of brain damage in cerebrospinal fluid of acute ischemic stroke patients. Clin Chem. 2010;56(3):451-8. 192. Yamasaki H, Sugino M, Ohsawa N. Possible regulation of intracranial pressure by human atrial natriuretic peptide in cerebrospinal fluid. Eur Neurol. 1997;38(2):88-93. 193. Kunz M, Nussberger J, Holtmannspotter M, Bitterling H, Plesnila N, Zausinger S. Bradykinin in blood and cerebrospinal fluid after acute cerebral lesions: correlations with cerebral edema and intracranial pressure. J Neurotrauma. 2013;30(19):1638-44. 194. Sorensen PS, Gjerris A, Hammer M. Cerebrospinal fluid vasopressin in neurological and psychiatric disorders. Journal of neurology, neurosurgery, and psychiatry. 1985;48(1):50- 7. 195. Sorensen PS, Gjerris F, Hammer M. Cerebrospinal fluid vasopressin and increased intracranial pressure. Ann Neurol. 1984;15(5):435-40. 196. Barbella YR, Keil LC, Wurpel JN, Severs WB. Cerebrospinal fluid pressure during cerebroventricular infusion of angiotensin and vasopressin. Experimental neurology. 1983;82(2):325-34. 197. Seckl JR, Lightman SL. Intracerebroventricular arginine vasopressin causes intracranial pressure to rise in conscious goats. Brain Res. 1987;423(1-2):279-85. 198. Sorensen PS, Gyring JA. Intracerebroventricular infusion but not bolus injection of vasopressin increases the cerebrospinal fluid pressure in awake rabbits. Neurol Res. 1990;12(2):83-8. 199. Kamiya T, Katayama Y, Kashiwagi F, Terashi A. The role of bradykinin in mediating ischemic brain edema in rats. Stroke. 1993;24(4):571-5; discussion 5-6. 200. Maktabi MA, Heistad DD, Faraci FM. Effects of central and intravascular angiotensin I and II on the choroid plexus. Am J Physiol. 1991;261(5 Pt 2):R1126-32. 201. Terris J, Ecelbarger CA, Nielsen S, Knepper MA. Long-term regulation of four renal aquaporins in rats. Am J Physiol. 1996;271(2 Pt 2):F414-22. 202. Nielsen S, Chou CL, Marples D, Christensen EI, Kishore BK, Knepper MA. Vasopressin increases water permeability of kidney collecting duct by inducing translocation of aquaporin-CD water channels to plasma membrane. Proceedings of the National Academy of Sciences of the United States of America. 1995;92(4):1013-7. 203. Raichle ME, Grubb RL, Jr. Regulation of brain water permeability by centrally-released vasopressin. Brain Res. 1978;143(1):191-4. 204. Liu X, Nakayama S, Amiry-Moghaddam M, Ottersen OP, Bhardwaj A. Arginine- vasopressin V1 but not V2 receptor antagonism modulates infarct volume, brain water content, and aquaporin-4 expression following experimental stroke. Neurocrit Care. 2010;12(1):124- 31. 205. Doczi T, Joo F, Szerdahelyi P, Bodosi M. Regulation of brain water and electrolyte contents: the opposite actions of central vasopressin and atrial natriuretic factor (ANF). Acta neurochirurgica Supplementum. 1988;43:186-8. 206. Doczi T, Laszlo FA, Szerdahelyi P, Joo F. Involvement of vasopressin in brain edema formation: further evidence obtained from the Brattleboro diabetes insipidus rat with experimental subarachnoid hemorrhage. Neurosurgery. 1984;14(4):436-41. 207. Doczi T, Szerdahelyi P, Gulya K, Kiss J. Brain water accumulation after the central administration of vasopressin. Neurosurgery. 1982;11(3):402-7. 208. Dickinson LD, Betz AL. Attenuated development of ischemic brain edema in vasopressin-deficient rats. J Cereb Blood Flow Metab. 1992;12(4):681-90.

139 209. Saladin LK, Bruni JE. The effects of intracerebroventricular versus intravenous administration of vasopressin on intracranial pressure in the rat. Neurol Res. 1993;15(3):198- 203. 210. Noto T, Nakajima T, Saji Y, Nagawa Y. Effect of vasopressin on intracranial pressure of rabbit. Endocrinol Jpn. 1978;25(6):591-6. 211. Dogterom J, van Wimersma Greidanus TB, De Wied D. Vasopressin in cerebrospinal fluid and plasma of man, dog, and rat. Am J Physiol. 1978;234(5):E463-7. 212. Kagerbauer SM, Martin J, Schuster T, Blobner M, Kochs EF, Landgraf R. Plasma oxytocin and vasopressin do not predict neuropeptide concentrations in human cerebrospinal fluid. J Neuroendocrinol. 2013;25(7):668-73. 213. Sorensen PS. Studies of vasopressin in the human cerebrospinal fluid. Acta neurologica Scandinavica. 1986;74(2):81-102. 214. Mens WB, Andringa-Bakker EA, Van Wimersma Greidanus TB. Changes in cerebrospinal fluid levels of vasopressin and oxytocin of the rat during various light-dark regimes. Neuroscience letters. 1982;34(1):51-6. 215. Schwartz WJ, Coleman RJ, Reppert SM. A daily vasopressin rhythm in rat cerebrospinal fluid. Brain Res. 1983;263(1):105-12. 216. Mather HM, Ang V, Jenkins JS. Vasopressin in plasma and CSF of patients with subarachnoid haemorrhage. Journal of neurology, neurosurgery, and psychiatry. 1981;44(3):216-9. 217. Vakili A, Kataoka H, Plesnila N. Role of arginine vasopressin V1 and V2 receptors for brain damage after transient focal cerebral ischemia. J Cereb Blood Flow Metab. 2005;25(8):1012-9. 218. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B, et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. 219. Tyanova S, Temu T, Sinitcyn P, Carlson A, Hein MY, Geiger T, et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat Methods. 2016;13(9):731-40. 220. Filippidis AS, Kalani MY, Nakaji P, Rekate HL. Negative-pressure and low-pressure hydrocephalus: the role of cerebrospinal fluid leaks resulting from surgical approaches to the cranial base. J Neurosurg. 2011;115(5):1031-7. 221. Hoffman WE, Edelman G, Kochs E, Werner C, Segil L, Albrecht RF. Cerebral autoregulation in awake versus isoflurane-anesthetized rats. Anesth Analg. 1991;73(6):753- 7. 222. Seckl JR, Lightman SL. Intracerebroventricular vasopressin reduces CSF absorption rate in the conscious goat. Experimental brain research. 1991;84(1):173-6. 223. Levine DN. Intracranial pressure and ventricular expansion in hydrocephalus: have we been asking the wrong question? J Neurol Sci. 2008;269(1-2):1-11. 224. Raabe A, Totzauer R, Meyer O, Stockel R, Hohrein D, Schoche J. Reliability of epidural pressure measurement in clinical practice: behavior of three modern sensors during simultaneous ipsilateral intraventricular or intraparenchymal pressure measurement. Neurosurgery. 1998;43(2):306-11. 225. Nornes H, Sundbarg G. Simultaneous recording of the ventricular fluid pressure and the epidural pressure. Eur Neurol. 1972;7(6):364-72. 226. Weinstabl C, Richling B, Plainer B, Czech T, Spiss CK. Comparative analysis between epidural (Gaeltec) and subdural (Camino) intracranial pressure probes. J Clin Monit. 1992;8(2):116-20. 227. Carson DS, Howerton CL, Garner JP, Hyde SA, Clark CL, Hardan AY, et al. Plasma vasopressin concentrations positively predict cerebrospinal fluid vasopressin concentrations in human neonates. Peptides. 2014;61:12-6. 228. Mens WB, Bouman HJ, Bakker EA, Van Wimersma Greidanus TB. Differential effects of various stimuli on AVP levels in blood and cerebrospinal fluid. European journal of pharmacology. 1980;68(1):89-92.

140 229. Berrouschot J, Sterker M, Bettin S, Koster J, Schneider D. Mortality of space- occupying ('malignant') middle cerebral artery infarction under conservative intensive care. Intensive Care Med. 1998;24(6):620-3. 230. Raimondi AJ. A unifying theory for the definition and classification of hydrocephalus. Childs Nerv Syst. 1994;10(1):2-12. 231. Tully HM, Dobyns WB. Infantile hydrocephalus: a review of epidemiology, classification and causes. Eur J Med Genet. 2014;57(8):359-68. 232. Edwards RJ, Dombrowski SM, Luciano MG, Pople IK. Chronic hydrocephalus in adults. Brain Pathol. 2004;14(3):325-36. 233. Banizs B, Pike MM, Millican CL, Ferguson WB, Komlosi P, Sheetz J, et al. Dysfunctional cilia lead to altered ependyma and choroid plexus function, and result in the formation of hydrocephalus. Development. 2005;132(23):5329-39. 234. Tarnaris A, Watkins LD, Kitchen ND. Biomarkers in chronic adult hydrocephalus. Cerebrospinal Fluid Res. 2006;3:11. 235. Alkan T, Korfali E, Kahveci N. Experimental subarachnoid haemorrhage models in rats. Acta Neurochir Suppl. 2002;83:61-9. 236. Hollig A, Weinandy A, Nolte K, Clusmann H, Rossaint R, Coburn M. Experimental subarachnoid hemorrhage in rats: comparison of two endovascular perforation techniques with respect to success rate, confounding pathologies and early hippocampal tissue lesion pattern. PLoS One. 2015;10(4):e0123398. 237. Schuller K, Buhler D, Plesnila N. A murine model of subarachnoid hemorrhage. J Vis Exp. 2013(81):e50845. 238. Conzen C, Becker K, Albanna W, Weiss M, Bach A, Lushina N, et al. The Acute Phase of Experimental Subarachnoid Hemorrhage: Intracranial Pressure Dynamics and Their Effect on Cerebral Blood Flow and Autoregulation. Transl Stroke Res. 2018. 239. Braun KP, van Eijsden P, Vandertop WP, de Graaf RA, Gooskens RH, Tulleken KA, et al. Cerebral metabolism in experimental hydrocephalus: an in vivo 1H and 31P magnetic resonance spectroscopy study. J Neurosurg. 1999;91(4):660-8. 240. Braun KP, Dijkhuizen RM, de Graaf RA, Nicolay K, Vandertop WP, Gooskens RH, et al. Cerebral ischemia and white matter edema in experimental hydrocephalus: a combined in vivo MRI and MRS study. Brain Res. 1997;757(2):295-8. 241. Braun KP, de Graaf RA, Vandertop WP, Gooskens RH, Tulleken KA, Nicolay K. In vivo 1H MR spectroscopic imaging and diffusion weighted MRI in experimental hydrocephalus. Magn Reson Med. 1998;40(6):832-9. 242. Klinge PM, Samii A, Muhlendyck A, Visnyei K, Meyer GJ, Walter GF, et al. Cerebral hypoperfusion and delayed hippocampal response after induction of adult kaolin hydrocephalus. Stroke. 2003;34(1):193-9. 243. Fencl V, Koski G, Pappenheimer JR. Factors in cerebrospinal fluid from goats that affect sleep and activity in rats. J Physiol. 1971;216(3):565-89. 244. Pappenheimer JR, Miller TB, Goodrich CA. Sleep-promoting effects of cerebrospinal fluid from sleep-deprived goats. Proceedings of the National Academy of Sciences of the United States of America. 1967;58(2):513-7. 245. Pappenheimer JR, Koski G, Fencl V, Karnovsky ML, Krueger J. Extraction of sleep- promoting factor S from cerebrospinal fluid and from brains of sleep-deprived animals. Journal of neurophysiology. 1975;38(6):1299-311. 246. Serafini T, Colamarino SA, Leonardo ED, Wang H, Beddington R, Skarnes WC, et al. Netrin-1 is required for commissural axon guidance in the developing vertebrate nervous system. Cell. 1996;87(6):1001-14. 247. Mehlen P, Rabizadeh S, Snipas SJ, Assa-Munt N, Salvesen GS, Bredesen DE. The DCC gene product induces apoptosis by a mechanism requiring receptor proteolysis. Nature. 1998;395(6704):801-4. 248. Chen YQ, Hsieh JT, Yao F, Fang B, Pong RC, Cipriano SC, et al. Induction of apoptosis and G2/M cell cycle arrest by DCC. Oncogene. 1999;18(17):2747-54.

141 249. Tsuchiya A, Hayashi T, Deguchi K, Sehara Y, Yamashita T, Zhang H, et al. Expression of netrin-1 and its receptors DCC and neogenin in rat brain after ischemia. Brain Res. 2007;1159:1-7. 250. Broughton BR, Reutens DC, Sobey CG. Apoptotic mechanisms after cerebral ischemia. Stroke. 2009;40(5):e331-9. 251. Zhang ZG, Chopp M. Exosomes in stroke pathogenesis and therapy. J Clin Invest. 2016;126(4):1190-7. 252. Camby I, Le Mercier M, Lefranc F, Kiss R. Galectin-1: a small protein with major functions. Glycobiology. 2006;16(11):137R-57R. 253. Qu WS, Wang YH, Ma JF, Tian DS, Zhang Q, Pan DJ, et al. Galectin-1 attenuates astrogliosis-associated injuries and improves recovery of rats following focal cerebral ischemia. J Neurochem. 2011;116(2):217-26. 254. Qu WS, Wang YH, Wang JP, Tang YX, Zhang Q, Tian DS, et al. Galectin-1 enhances astrocytic BDNF production and improves functional outcome in rats following ischemia. Neurochem Res. 2010;35(11):1716-24. 255. Akazawa C, Nakamura Y, Sango K, Horie H, Kohsaka S. Distribution of the galectin-1 mRNA in the rat nervous system: its transient upregulation in rat facial motor neurons after facial nerve axotomy. Neuroscience. 2004;125(1):171-8. 256. Coultrap SJ, Vest RS, Ashpole NM, Hudmon A, Bayer KU. CaMKII in cerebral ischemia. Acta Pharmacol Sin. 2011;32(7):861-72. 257. Vest RS, O'Leary H, Coultrap SJ, Kindy MS, Bayer KU. Effective post-insult neuroprotection by a novel Ca(2+)/ calmodulin-dependent protein kinase II (CaMKII) inhibitor. J Biol Chem. 2010;285(27):20675-82. 258. Waxham MN, Grotta JC, Silva AJ, Strong R, Aronowski J. Ischemia-induced neuronal damage: a role for calcium/calmodulin-dependent protein kinase II. J Cereb Blood Flow Metab. 1996;16(1):1-6. 259. Gunnarson E, Axehult G, Baturina G, Zelenin S, Zelenina M, Aperia A. Lead induces increased water permeability in astrocytes expressing aquaporin 4. Neuroscience. 2005;136(1):105-14. 260. Tourdias T, Mori N, Dragonu I, Cassagno N, Boiziau C, Aussudre J, et al. Differential aquaporin 4 expression during edema build-up and resolution phases of brain inflammation. J Neuroinflammation. 2011;8:143. 261. Narita K, Kozuka-Hata H, Nonami Y, Ao-Kondo H, Suzuki T, Nakamura H, et al. Proteomic analysis of multiple primary cilia reveals a novel mode of ciliary development in mammals. Biol Open. 2012;1(8):815-25. 262. Coutts SB, Dubuc V, Mandzia J, Kenney C, Demchuk AM, Smith EE, et al. Tenecteplase-tissue-type plasminogen activator evaluation for minor ischemic stroke with proven occlusion. Stroke. 2015;46(3):769-74. 263. Claassen J, Carhuapoma JR, Kreiter KT, Du EY, Connolly ES, Mayer SA. Global cerebral edema after subarachnoid hemorrhage: frequency, predictors, and impact on outcome. Stroke. 2002;33(5):1225-32. 264. Maas MB, Furie KL. Molecular biomarkers in stroke diagnosis and prognosis. Biomark Med. 2009;3(4):363-83. 265. Czosnyka M, Guazzo E, Whitehouse M, Smielewski P, Czosnyka Z, Kirkpatrick P, et al. Significance of intracranial pressure waveform analysis after head injury. Acta Neurochir (Wien). 1996;138(5):531-41; discussion 41-2. 266. Brinker T, Seifert V, Dietz H. Cerebral blood flow and intracranial pressure during experimental subarachnoid haemorrhage. Acta Neurochir (Wien). 1992;115(1-2):47-52. 267. Binder DK, Horton JC, Lawton MT, McDermott MW. Idiopathic intracranial hypertension. Neurosurgery. 2004;54(3):538-51; discussion 51-2. 268. Peng KP, Fuh JL, Wang SJ. High-pressure headaches: idiopathic intracranial hypertension and its mimics. Nat Rev Neurol. 2012;8(12):700-10. 269. Owler BK, Higgins JN, Pena A, Carpenter TA, Pickard JD. Diffusion tensor imaging of benign intracranial hypertension: absence of cerebral oedema. Br J Neurosurg. 2006;20(2):79-81.

142

143 Chapter 8 Appendix

Appendix A: Publications not included in this thesis

Publication 1

Ischemic penumbra as a trigger for intracranial pressure rise – A potential cause for collateral failure and infarct progression?

Beard, D.J*., Logan, C.L*., McLeod, D.D., Hood, R.J., Pepperall, D., Murtha, L.A., Spratt, N.J. (2016). Journal of Cerebral Blood Flow and Metabolism, 36(5):917-27

*Denotes equal contribution

144 Original Article Journal of Cerebral Blood Flow & Metabolism 0(00) 1–11 Ischemic penumbra as a trigger ! Author(s) 2016 Reprints and permissions: for intracranial pressure rise – A potential sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0271678X15625578 cause for collateral failure and infarct jcbfm.sagepub.com progression?

Daniel J Beard1,2,*, Caitlin L Logan1,2,*, Damian D McLeod1,2, Rebecca J Hood1,2, Debbie Pepperall1,2, Lucy A Murtha1,2 and Neil J Spratt1,2,3

Abstract We have recently shown that intracranial pressure (ICP) increases dramatically 24 h after minor intraluminal thread occlusion with reperfusion, independent of edema. Some of the largest ICP rises were observed in rats with the smallest final infarcts. A possible alternate mechanism for this ICP rise is an increase of cerebrospinal fluid (CSF) volume sec- ondary to choroid plexus damage (a known complication of the intraluminal stroke model used). Alternatively, sub- maximal injury may be needed to induce ICP elevation. Therefore, we aimed to determine (a) if choroid plexus damage contributes to the ICP elevation, (b) if varying the patency of an important internal collateral supply to the middle cerebral artery (MCA), the anterior choroidal artery (AChA), produces different volumes of ischemic penumbra and (c) if presence of ischemic penumbra (submaximal injury) is associated with ICP elevation. We found (a) no association between choroid plexus damage and ICP elevation, (b) animals with a good internal collateral supply through the AChA during MCAo had significantly larger penumbra volumes and (c) ICP elevation at &24 h post-stroke only occurred in rats with submaximal injury, shown in two different stroke models. We conclude that active cellular processes within the ischemic penumbra may be required for edema-independent ICP elevation.

Keywords Collaterals, intracranial pressure, middle cerebral artery occlusion, penumbra, photothrombosis

Received 17 September 2015; Revised 11 November 2015; Accepted 4 December 2015

Introduction elevation 24 h post-stroke. While developing this model, we discovered that ICP only appeared to In recent studies, we have shown that there is a dramatic increase following submaximal, not maximal injury. intracranial pressure (ICP) elevation occurring 18–24 This chance finding led us to our primary hypothesis hafter temporary middle cerebral artery occlusion that submaximal injury (penumbra) is necessary for (MCAo) in rats, which is edema independent.1,2 A pos- sible alternate mechanism for this ICP rise is an increase of cerebrospinal fluid (CSF) volume.3 Ischemia of the 1School of Biomedical Sciences and Pharmacy, University of Newcastle, anterior choroidal artery (AChA) and subsequently the New South Wales, Australia 2Hunter Medical Research Institute, New Lambton, New South Wales, choroid plexus is a known complication of the thread 4 Australia occlusion model. Choroid plexus produces CSF, and 3Department of Neurology, John Hunter Hospital, Hunter New England hence we hypothesised that choroid plexus damage may Local Health District, New South Wales, Australia increase CSF production, leading to edema-independent *These authors contributed equally to this work ICP elevation. To test this hypothesis, our first aim was Corresponding author: to measure ICP following cortical photothrombotic Neil J Spratt, School of Biomedical Sciences and Pharmacy, University of (PT) stroke, without choroid plexus damage to deter- Newcastle, University Drive, Callaghan, NSW 2308, Australia. mine if choroid plexus injury is necessary for ICP Email: [email protected]

145 2 Journal of Cerebral Blood Flow & Metabolism the delayed ICP elevation following stroke. To test this (Protocol # A-2011-112) and were in accordance with hypothesis, we aimed to induce PT stroke with varying the requirements of the Australian Code of Practice for degrees of injury in order to determine if submaximal the Care and Use of Animals for Scientific Purposes. injury is necessary for ICP elevation at 24 h post-stroke. The studies were conducted and the manuscript pre- ICP is not routinely measured in patients who have pared in accordance with the ARRIVE guidelines.10 small infarcts, due to the invasiveness of ICP monitor- ing equipment. Hence, we do not know whether a tran- Experimental protocols sient ICP increase occurs in patients with minor stroke as we see in rats.1,2 However, recent acute stroke Study I. The aim of study I was to investigate whether imaging studies indicate that the neurological deterior- choroid plexus ischemia or submaximal stroke is ation in patients with delayed infarct expansion most required for ICP elevation post-stroke. Baseline ICP commonly occurs at a very similar time interval.5 and mean arterial pressure (MAP) were monitored for Moreover, deterioration is rarely accompanied by wor- 1 h in all animals. PT stroke was induced by intraven- sening or recurrence of the arterial occlusion as was ous infusion of 10 mg/kg Rose Bengal (Sigma-Aldrich, previously supposed.5 Rather, it is associated with fail- Sydney, Australia) followed by illumination of the right ure of leptomeningeal collateral supply, in patients with parietal bone with a cold focal light source (n 16). stable proximal arterial occlusion, excellent collaterals Animals were randomised by sealed numbered¼ enve- and large penumbra at baseline imaging.6 In our recent lope to receive either low light exposure (LLE) or studies, we showed that ICP elevation reduces collateral standard light exposure (SLE). The skull was illumi- blood flow.7 This suggests that ICP elevation may be nated with a 12-inch optic fibre articulating arm light the cause of the collateral failure. Since a large penum- pipe with 6 mm diameter light tip either for 2 min at bra at baseline imaging was one of the defining features 0.13 W/cm2 (LLE, to produce submaximal injury) or of patients who developed stroke-in-progression, we 20 min at 0.3 W/cm2 (SLE, to produce maximal wished to try to replicate the situation of such patients injury). Sham animals (n 2) received the same light as far as possible in our experimental model and deter- exposure as the standard¼ exposure group, but Rose mine whether it resulted in a significant ICP rise. The Bengal was not administered. ICP and MAP were presence of penumbra in the PT model remains contro- monitored continuously from 18 h to 26 h post-stroke versial. Therefore, we wished to use a model with more (Figure 1a). clear-cut penumbra. We recently demonstrated the importance of an internal collateral vessel, the AChA, Study II. The aim of study II was to determine whether to infarct volume in the rat thread occlusion model. the patency of the AChA, would produce different vol- Preserved patency of the AChA resulted in significantly umes of ischemic penumbra 1 h following MCAo in smaller final infarct volumes.8 The patency of the Wistar rats, using computed tomography perfusion AChA is manipulable by altering the silicon tip length (CTP) imaging. Animals were subject to intraluminal of the thread occluder to either occlude or maintain pMCAo (n 28).11 CTP scans were taken at baseline patency of the AChA.9 Therefore, we also sought to and 1 h after¼ MCAo (Figure 1b). determine whether animals with patent AChA had larger volumes of acute ischemic penumbra. Our final Study III. In study III, we aimed to investigate the asso- aim was to determine whether such rats with permanent ciation between the presence of penumbra and ICP ele- MCAo (pMCAo) with excellent collaterals (mimicking vation. We altered the length of the silicone occluder tip the scenario of patients with delayed stroke progres- to produce different degrees of collateral perfusion sion) had 24-h ICP elevation. through the AChA and hence penumbra. Baseline ICP and MAP were monitored for 2 h in all animals. All rats received intraluminal pMCAo.11 Just prior Materials and methods to thread insertion, rats were randomized by sealed Ethics statement numbered envelope to a short-length (1.5 mm) silicon tip on the occluding thread to maintain patency of All animal experiments were performed on male the AChA during MCAo, or to a 4-mm tip-length outbred Wistar rats (aged 7–12 weeks, body weight: thread to simultaneously occlude the origin of the 300–500 g; ASU breeding facility, University of AChA (n 6/group). ICP and MAP were monitored Newcastle, Australia). All animals were housed in for 2 h at 24¼ h and 72 h post-stroke (Figure 1c), respect- cages in an animal-holding facility in the University ively. As per our usual practice, lack of laser Doppler of Newcastle with rat chow and water available ad lib- flowmetry (LDF) drop >50 % during MCAo or evi- itum. Experiments were approved by the Animal Care dence of subarachnoid hemorrhage were prespecified and Ethics Committee of the University of Newcastle exclusion criteria.7

146 Beard et al. 3

Figure 1. Experimental timelines. (a) Study I. ICP was monitored at baseline and between 18 and 26 h after PT stroke. (b) Study II. CTP imaging was carried out to measure whole brain perfusion changes during MCAo. For the purposes of this study, scans taken 1 h post- MCAo were used to create core and penumbra maps using previously established CTP-defined CBF thresholds for this model. (c) Study III. ICP was monitored at baseline, 24 and 72 h after permanent MCAo. ABG arterial blood gas; CTP: computed tomography perfusion; ICP: intracranial pressure; MAP: mean arterial pressure; MCAo: middle cerebral¼ artery occlusion; PT: photothrombotic stroke.

Anesthesia and monitoring continuous arterial pressure monitoring. Heart rate The anesthetic and monitoring protocols were as pre- and respiratory rate were calculated from the clearly viously reported.12 Rats were anesthetised with 5% iso- discernible cardiac and respiratory waveforms on the 13 flurane in O2/N2 (1:3) and maintained with 1–2% arterial pressure tracing. After stroke surgery, ani- isoflurane. Animals spontaneously breathed through a mals were injected subcutaneously with saline laboratory-manufactured low dead-space nose cone. (2 2.5 ml) to prevent dehydration and returned to  Core temperature was maintained at 37C by a thermo- their cages with free access to softened laboratory couple rectal probe and warming plate. Incision sites chow and water. were shaved, cleaned and injected subcutaneously with 2 mg/kg 0.05 % Bupivacaine (Pfizer, Sydney, ICP measurement Australia). Blood gases were monitored at baseline and 24 h in Study I and at baseline, 24 and 72 h in Epidural ICP was measured using a fibre optic pressure Study III. Blood samples (0.1 ml) were taken from a transducer (420LP, SAMBA Sensors, Sweden, femoral arterial line. This line was also used for accuracy 0.4 mmHg) according to our published ¼Æ

147 4 Journal of Cerebral Blood Flow & Metabolism method.1,2,7,13 Briefly, a hollow poly-ethyl ether ketone using MIStar imaging software (Apollo Medical screw (Solid Spot LLC, Santa Clara, CA, USA) of Imaging Technology, Melbourne, Australia) as previ- 2 mm diameter 5 mm length was inserted into the ously described12 with some minor modifications. left parietal bone. Our experience indicates that the Animals had already been classified as having side of probe placement does not influence ICP read- MCAo alone (MCAo, n 19) or MCAo with concomi- ings. The sensor was placed above the dura and sealed tant AChA occlusion (MCAo¼ AChAo, n 9) based in place with a caulking material (Silagum, DMG on cerebral blood volume (CBV)þ maps in¼ our previ- Dental, Hamburg, Germany). Recent studies indicate ous publication.8 Infarct core and penumbra vol- negligible differences between ICP measurements taken umes were determined in this cohort of animals by epidurally or intraparenchymally in rats, and the epi- applying our previously established cerebral blood dural method avoids brain trauma and the risk of creat- flow (CBF) thresholds for core and penumbra (<55 ing a CSF leak.14,15 Probe location was validated by % and <75 % of contralateral CBF, respectively) ensuring clear cardiac and respiratory waveforms and to all CBF maps.16 The total volume for each param- responsiveness of signal to abdominal compression.13 eter for each animal was calculated using MiStar Cerebral perfusion pressure (CPP) was calculated imaging software. using the formula CPP MAP Mean ICP. ¼ À Statistics Histological analysis Statistical tests were performed using Graphpad Prism Histological analysis was performed according to our 5.04 (La Jolla, USA) unless otherwise stated. Sample published method.12 After 24 h (Study I) and 72 h sizes calculated for between-group differences in ICP, (Study III), animals were sacrificed and underwent based on pilot studies, were <4. Slightly larger cohorts transcardiac perfusion with saline followed by 4% par- were used to allow for the possibility of outliers, given aformaldehyde in 0.2 M phosphate buffer. Brains were the novelty of the study. D’Agostino and Pearson then fixed in neutral buffered formalin before being omnibus normality tests were performed on all data; processed and paraffin embedded. Coronal sections of appropriate statistical tests were performed based on 5 mm were cut and stained with hematoxylin and eosin. the normality of the data. Images were scanned using a digital slide scanner Student’s t-tests were used to compare differences (Aperio Technologies, Vista, CA, USA) and analyzed between groups (unpaired t-test) or changes from base- by an investigator blind to experimental group alloca- line (paired t-test). In Study III, ICP was measured at tion. Infarct (corrected for edema) was calculated by baseline, 24 and 72 h, therefore an analysis of covari- subtracting the measured interhemispheric volume dif- ance (ANCOVA) was performed using SPSS 21.0 ference (edema volume, ipsilateral minus contralateral) (IBM, Armonk, USA) to test whether thread length from the measured infarct volume for each slice. had a significant effect on the primary outcome of Lateral ventricle choroid plexus damage was assessed change in ICP. Post-hoc Bonferroni’s t-tests were per- morphologically on three brain slices (1.3, 2.3 and formed to test the differences in ICP between groups at 3.3 mm caudal to bregma) using the following scale; each time-point. Mann–Whitney U tests were per- normal morphology (0) (Figure 2a), cellular shrinkage formed to compare choroid plexus damage scores or vacuolation (1) (Figure 2b) and epithelial desquam- between groups. Choroid plexus damage scores are pre- ation (2) (Figure 2c). The scores from each bregma level sented as median (25th–75th interquartile range; IQR). were summed to give a total score out of 6. Significant differences were accepted at the p < 0.05 level. Data are presented as mean standard deviation Æ CTP image acquisition, processing and infarct core (SD) unless otherwise stated. and penumbra volume analysis Detailed methods and CTP data have previously been Results 8,12,16 reported for this cohort of animals; however, the Study I CTP-derived infarct core and penumbra volumes 1 h after stroke were not previously analyzed. In brief, Three animals in total were excluded, due to epidural imaging was performed using a Siemens 64 slice hemorrhage from ICP catheter insertion (1 LLE and CT scanner (Siemens, Erlangen, Germany) with a 1 SLE) and equipment difficulties preventing collection 512 512 matrix, 50 mm field of view with twelve of ICP data on day 2 (1 SLE). Physiological variables 2.4 mm slices. Radio-opaque contrast for perfusion are presented in Table 1. In animals subjected to imaging was injected through a jugular venous cathe- LLE, mean ICP rose significantly above baseline ter.12 Post-processing of perfusion data was performed (10.2 3.6 mmHg baseline, 23.1 5.4 mmHg peak Æ Æ

148 Beard et al. 5

Figure 2. Choroid plexus damage scoring. Hematoxylin and eosin staining of lateral ventricle choroid plexus at 20 (left panels) and 100 (right panels). (a) Normal choroid plexus histology (given a score of 0) showing a continuous epithelial lining of the choroidal villus with healthy epithelial cells (arrow). (b) Moderate choroid plexus damage (given a score of 1) showing an intact epithelial lining with shrunken and vacuolated epithelial cells (arrow). (c) Severe choroid plexus damage (given a score of 2) showing areas of epithelial desquamation (arrow).

ICP, p < 0.001, n 9; Figure 3a) and above the highest 43 18 mm3, respectively, p < 0.01; Figure 3b). Sham mean ICP in¼ the SLE group (11.8 3.7 mmHg, animalsÆ had no signs of infarction. Representative p < 0.01; 9.9 2.4 mmHg baseline, n 7).Æ ICP did not infarct size from each group is shown in Figure 3c. rise in shamÆ light exposure animals¼ (Figure 3a). LLE Edema volumes in both groups were very small and animals had significantly smaller 24-h infarct volumes not significantly different (LLE: 0.2 0.4 mm3 vs. compared to those with SLE (9 8 mm3 vs. SLE: 2.8 3.8 mm3, p > 0.05). There wasÆ no evidence Æ Æ

149 6 Journal of Cerebral Blood Flow & Metabolism

Table 1. Physiological parameters for Study I.

Baseline During peak ICP 24 h post-stroke

Study I LLE SLE LLE SLE LLE SLE

MAP (mmHg) 102 13 94 14 96 9 94 14 96 9 91 11 Æ Æ Æ Æ Æ Æ CPP (mmHg) 91 13 85 16 73 13## 82 12 80 145 81 12 Æ Æ Æ Æ Æ Æ RR (BPM) 71 11 71 14 70 18 63 18 68 10 55 9 Æ Æ Æ Æ Æ Æ HR (BPM) 402 58 399 36 409 27 380 38 Æ Æ Æ Æ pCO (mmHg) 56 7 50 5 – – 55 87 58 12 2 Æ Æ Æ Æ pO 208 90 155 27 – – 188 37 170 54 2 Æ Æ Æ Æ pH 7.36 0.05 7.39 0.04 – – 7.39 0.05 7.39 0.06 Æ Æ Æ Æ Temp (C) 37.6 0.6 37.8 0.9 37.8 0.6 37.5 0.6 Æ Æ Æ Æ

Values for MAP, CPP, RR, HR, pCO2, pO2, pH and temperature. CPP: cerebral perfusion pressure; HR: heart rate; LLE: low light exposure; MAP: mean # ## arterial pressure; pCO2: partial pressure of carbon dioxide; pO2: partial pressure of oxygen; SLE: standard light exposure. p < 0.001 and p < 0.0001 versus baseline and *p < 0.001 versus LLE peak ICP. p values are for illustrative purposes and uncorrected for multiple comparisons.

Figure 3. Study I – ICP, CPP and infarct volume following photothrombotic stroke. (a) ICP 0–0.5 h and 18–26 h post-photo- thrombotic stroke in Wistar rats using either LLE (black line; n 9) or SLE (broken black line; n 7) and sham animals (grey lines; n 2), mean SEM. Each data point represents ICP averaged over¼ 10 min, at 30-min time intervals.¼ (b) Peak ICP versus 24-h infarct volume¼ for LLEÆ (filled squares, unbroken error bars) or SLE (open squares, broken error bars). (c) Infarct probability map at 4.3 mm Bregma, where the maximal extent of the photothrombotic lesions were located. Lighter regions represent areas more commonlyÀ infarcted. CPP: cerebral perfusion pressure; ICP: intracranial pressure; LLE: low light exposure; SEM: standard error mean; SLE: standard light exposure. of significant choroid plexus injury in any animals in Study II either group. Three animals in the LLE group and three animals in the SLE group had scores suggesting minor Physiological variables for this cohort of animals have injury, however median (IQRs) in both groups were been reported previously.7 Representative 1-h CBF 0 (0–1). maps for MCAo and MCAo AChAo groups are þ

150 Beard et al. 7

Figure 4. Study II – CTP core and penumbra volume analysis in animals with or without concomitant AChAo during intraluminal MCAo. Representative 1-h CBF maps for animals with MCAo (a) or MCAo AChA (b). Penumbra volume (c), core volume (d) at 1 h post-MCAo in animals with MCAo only (black bars) and animals with MCAo þAChAo (white bars). **p < 0.01. AChA: anterior choroidal artery; AChAo: anterior choroidal artery occlusion; CTP: computer tomograpþ hy perfusion; MCAo: middle cerebral artery occlusion. shown in Figure 4a and b, respectively. Penumbra vol- post-stroke (p < 0.01, Bonferroni’s t-test). One short ume was significantly larger in the MCAo group com- thread animal was observed to have severe stroke signs pared to the MCAo AChAo group (112 22 mm3 vs. shortly after recovery from anesthesia. This animal sub- 81 6mm3,respectively,þ p < 0.01; Figure 4c).Æ There was sequently had a large infarct volume at 72 h (221 mm3) noÆ significant difference in core volume between MCAo and had no 24-h ICP elevation. Infarction was seen within and MCAo AChAo groups (114 44 mm3 vs. the AChA territory. This animal was included in the short 134 20 mm3,þp > 0.05; Figure 4d). Æ thread (excellent collaterals group) in all analysis, accord- Æ ing to original group allocation. Study III There were no significant differences in infarct or edema volumes between short- and long thread groups. Twelve animals were excluded in total, due to technical (Infarct: 93 89 mm3 vs. 87 43 mm3,Figure5b;edema: difficulties during surgery (three long thread), mortality 1.0 2.8 mmÆ3 vs. 2.0 2.7 mmÆ 3,respectively;p > 0.05). during surgery (one prior to randomization and one ThereÆ was a noticeableÆ difference in distribution of infarct short thread), subarachnoid hemorrhage (two short volumes between groups with a dichotomous distribution and one long thread), or lack of LDF drop (two in the short thread group. Infarct probability maps short and two long thread). One short thread animal revealed that there was a greater probability of infarction died following 24-h ICP monitoring. It was included in within the region supplied by the AChA in the long ICP but not infarct volume analysis. Physiological vari- thread group compared to the short thread group ables are reported in Table 2. (Figure 5c). There was no significant difference in the There was a significant main effect of thread length on degree of lateral ventricle choroid plexus damage scores ICP, F (1,3) 6.33, p 0.023, n 6pergroup.Inthe between the short thread group, median score of 1 (0–1.5) short thread¼ group, baseline¼ ICP¼ was 6.7 2.2 mmHg and the long thread group, median score of 0 (0–1.25), and increased to 17.3 7.5 mmHg at 24 hÆ and subse- p 0.11. quently decreased toÆ 8.8 3.3 mmHg at 72 h. In the ¼ Æ long thread group, ICP was 8.3 4.4 mmHg at baseline, Discussion 6.4 3.8 mmHg at 24 h and 5.2Æ 2.3 mmHg at 72 h (FigureÆ 5a). ICP in the short threadÆ group was signifi- This study indicates that ICP elevation at 24 h post- cantly higher than the long thread group at 24 h stroke is not a product of the choroid plexus ischemia

151 8 Journal of Cerebral Blood Flow & Metabolism

Table 2. Physiological parameters for Study III.

Baseline 24 h post-stroke 72 h post-stroke

Study III Short thread Long thread Short thread Long thread Short thread Long thread

MAP (mmHg) 104 11 103 13 103 8 106 12 97 13 93 7 Æ Æ Æ Æ Æ Æ CPP (mmHg) 97 11 92 17 87 15 103 7* 88 14 88 8 Æ Æ Æ Æ Æ Æ RR (BPM) 67 12 71 3 59 7 59 10 56 8 55 4# Æ Æ Æ Æ Æ Æ HR (BPM) 434 65 468 31 411 41 460 26 373 48 394 58# Æ Æ Æ Æ Æ Æ pCO (mmHg) 46 6 43 4 49 5 43 8 49 6 44 6 2 Æ Æ Æ Æ Æ Æ pO 213 10 203 27 216 15 207 37 223 16 209 28 2 Æ Æ Æ Æ Æ Æ pH 7.41 0.04 7.42 0.03 7.39 0.03 7.45 0.05 7.34 0.16 7.41 0.03 Æ Æ Æ Æ Æ Æ Temp (C) 37.7 0.9 38.3 0.9 37.4 0.6 38.2 1.1 37.3 0.5 37.3 0.5 Æ Æ Æ Æ Æ Æ . MAP, CPP, RR, HR, pCO2, pO2, pH and temperature. CPP: cerebral perfusion pressure; HR: heart rate; LLE: low light exposure; MAP: mean arterial # ## pressure; pCO2: partial pressure of carbon dioxide; pO2: partial pressure of oxygen; SLE: standard light exposure p < 0.05, p < 0.01 and ###p < 0.001 versus baseline and *p < 0.05, **p < 0.01 versus short thread. p values are for illustrative purposes, uncorrected for multiple comparisons.

Figure 5. Study III – ICP, infarct volume and infarct probability following pMCAo in Wistar rats. (a) ICP at baseline, 24 h and 72 h post-pMCAo. Strokes were induced with intraluminal threads with silicon tip length designed either to maintain AChA patency (1.5 mm tips, short thread group, closed circles; n 6) or to occlude the AChA (4 mm tips, long thread group, open circles; n 6). Main effect of thread length on ICP, F (1,3) 6.33,¼p 0.023; ANCOVA. **p < 0.01 between groups; post-hoc Bonferroni t-tests¼ (b) Relationship between 72-h infarct volume and¼ 24-h ICP¼ in the short thread group (closed circles, unbroken error bars) and the long thread group (open circles, broken error bars). (c) Infarct probability map at 2.3 mm Bregma. Lighter regions represent areas more commonly infarcted. The region supplied by the AChA is outlined in red.17 AChA:À anterior choroidal artery; ANCOVA: analysis of covariance: ICP: intracranial pressure; pMCAo: permanent middle cerebral artery occlusion.

152

Beard et al. 9 that has previously been reported in the thread occlu- other studies, where a penumbra like ‘region at risk’ sion model, since we saw ICP elevation with purely can be induced with the PT model using LLE and inten- cortical PT strokes. Our results also indicate that a sities.20 In contrast, the standard PT model is well region of submaximal injury is required to trigger the known to induce rapid infarct evolution that lacks pen- recently described edema-independent ICP elevation umbra and is exacerbated by ischemia-independent 24 h after onset of ischemic stroke. We suggest that mechanisms including generation of reactive oxygen active cellular processes within the ischemic penumbra species and blood–brain barrier breakdown.21,22 This are required for ICP elevation, since no ICP rise was suggests to us that an area of submaximal injury may seen in animals with larger, more complete infarction, be necessary to induce ICP elevation at 24 h post- in which cellular products would be expected to be stroke. Alternative explanations are also possible. For released from necrotic cells. example, there could be additional mechanisms block- We found no association between the presence of ing ICP elevation in animals with a large unreperfused choroid plexus damage and 24-h ICP elevation in infarct core. Further studies are required to elucidate both cortical PT and intraluminal MCAo stroke the exact mechanism. models. Somewhat surprisingly, we saw minimal Inadvertent occlusion of the AChA during intralum- damage to the choroid plexus following thread occlu- inal MCAo was associated with smaller volume of pen- sion, which is known to occlude the AChA and has umbra 1 h after stroke onset. There are large differences been shown to cause acute damage to the choroid in the volume of penumbral tissue and the rate it is plexus in Sprague-Dawley rats.4 This was despite the incorporated into the infarct core following occlusion long thread group having a higher frequency of infarc- of the MCA in different rat strains.23–26 This is gener- tion within the region of the sub-cortex supplied by the ally thought to be due to differences in the residual AChA compared to the short thread group, indicating blood flow to the ischemic brain via the leptomeningeal a higher frequency of AChA occlusion in the long collateral circulation.27 We have previously demon- thread group.17 A potential explanation for the lack strated that inadvertent occlusion of the AChA, an of choroid plexus damage is that choroid plexus hist- important internal collateral supply to the MCA results ology was assessed three days after MCAo, giving no in larger 24-h histological infarct volumes.8 The current indication of the degree of ischemia and choroid plexus analysis indicates that those animals with patent AChA damage at the time of MCAo or at 24 h when ICP was have a larger penumbra. elevated. Acute damage to the choroid plexus may not Our finding of ICP elevation in those with short be visible at three days due to regenerative capacity of thread tip lengths (1.5 mm, aiming to keep the AChA the choroid plexus (occurring within 12–24 h after fore- patent) but not in those with long-tipped threads brain ischemia).18 Alternatively, perfusion to the chor- (4 mm, aiming to occlude the AChA) strongly suggests oid plexus may have been maintained in the presence of that the presence and potentially salvage of penumbra AChA occlusion by the posterior choroidal artery, influences ICP elevation. The potential translational which arises from the posterior cerebral artery.19 significance of this is that the model of pMCAo with Regardless of the mechanism of preserved choroid good internal collateral supply mimics the recently plexus morphology following MCAo, the lack of chor- described scenario of patients with delayed stroke pro- oid plexus damage and the 24-h ICP rise after cortical gression. The very similar timing of the two events is PT stroke (at a distance from the subcortical choroidal suggestive. However, it still remains to be shown that circulation), and the greater ICP rise in the short thread ICP elevation occurs in patients with minor stroke, and group with less AChA territory infarction, all indicate definitive proof of ICP elevation directly causing infarct that choroid plexus damage is not necessary to cause expansion is also still required. 24-h ICP rise. ICP elevation was observed following PT stroke, but Conclusions only after stroke designed to produce submaximal lesions. In our PT model, the area of the cortex exposed In the current study, we have made two important to the light was the same between the two groups; only observations regarding the trigger for the recently light intensity and duration were varied. Only the SLE described ICP elevation occurring approximately 24 h produced stroke lesions that encompassed the majority after MCAo. First, we have shown that it occurs inde- of the area exposed to the light, whereas the short dur- pendent of choroid plexus damage. Second, the data ation group has much smaller lesions. We speculate indicate that the ICP rise is associated with presence that the difference in infarction between the two of ischemic penumbra or submaximal tissue injury. groups represents tissue that was mildly ischemic or Perhaps more remarkably, and in marked contrast to had spontaneously reperfused, thus resulting in tissue previous thoughts about ICP elevation after stroke, we salvage (penumbra). This has been demonstrated in found that ICP did not rise in animals with larger and

153 10 Journal of Cerebral Blood Flow & Metabolism more completely evolved infarction. We also found that 4. Ennis SR and Keep RF. The effects of cerebral ischemia ICP elevation occurred 24 h after permanent vessel on the rat choroid plexus. J Cereb Blood Flow Metab occlusion, but only in the setting of good collaterals. 2006; 26: 675–683. Coupled with our previous data showing that such ICP 5. Coutts SB, Modi J, Patel SK, et al. CT/CT angiography elevation reduces collateral flow, we suggest that ICP and MRI findings predict recurrent stroke after transient elevation is a possible cause of collateral failure and ischemic attack and minor stroke: results of the prospect- ive CATCH study. Stroke 2012; 43: 1013–1017. infarct expansion in patients with stroke-in-progres- 6. Campbell BC, Christensen S, Tress BM, et al. Failure of sion. Multiple unanswered questions remain, however, collateral blood flow is associated with infarct growth in these findings suggest that we may be on the cusp of a ischemic stroke. J Cereb Blood Flow Metab 2013; 33: major change in our understanding of stroke evolution. 1168–1172. 7. Beard DJ, McLeod DD, Logan CL, et al. Intracranial Funding pressure elevation reduces flow through collateral vessels and the penetrating arterioles they supply. A possible The author(s) disclosed receipt of the following financial sup- explanation for ‘collateral failure’ and infarct expansion port for the research, authorship, and/or publication of this after ischemic stroke. J Cereb Blood Flow Metab 2015; 35: article: This work was supported in part by a National Health 861–872. and Medical Research Council Project Grant, APP1033461, 8. McLeod DD, Beard DJ, Parsons MW, et al. Inadvertent the Hunter Medical Research Institute from funds donated by occlusion of the anterior choroidal artery explains infarct the Greater Building Society and by the National Stroke variability in the middle cerebral artery thread occlusion Foundation (Australia). D. Beard and L. Murtha were both stroke model. PLoS One 2013; 8: e75779. supported by an Australian Postgraduate Award. N. Spratt 9. Guan Y, Wang Y, Yuan F, et al. Effect of suture proper- was supported by a NHMRC career development fellowship, ties on stability of middle cerebral artery occlusion eval- #1035465. uated by synchrotron radiation angiography. Stroke 2012; 43: 888–891. Acknowledgements 10. Kilkenny C, Browne W, Cuthill IC, et al. Animal research: reporting in vivo experiments – the ARRIVE We would like to thank the Faculty of Health Stores guidelines. J Cereb Blood Flow Metab 2011; 31: 991–993. Workshop of the University of Newcastle for manufacture 11. Spratt NJ, Fernandez J, Chen M, et al. Modification of of bespoke surgical and anesthetic equipment. the method of thread manufacture improves stroke induction rate and reduces mortality after thread- Declaration of conflicting interests occlusion of the middle cerebral artery in young or The author(s) declared no conflicts of interest with respect to aged rats. J Neurosci Meth 2006; 155: 285–290. the research, authorship, and/or publication of this article. 12. McLeod DD, Parsons MW, Levi CR, et al. Establishing a rodent stroke perfusion computed tomography model. Int J Stroke 2011; 6: 284–289. Authors’ contributions 13. Murtha L, McLeod D and Spratt N. Epidural intracra- DB, CL and DM performed the surgical and experimental nial pressure measurement in rats using a fiber-optic components of the study, analyzed and interpreted the data, pressure transducer. J Vis Exp: JoVE 2012. DOI: including statistical analysis and drafted the manuscript. RH, 10.3791/3689. DP conducted histological analysis for the study. RH, DP, 14. Hiploylee C and Colbourne F. Intracranial pressure mea- LM and NS participated in the design of the study and helped sured in freely moving rats for days after intracerebral draft the manuscript. DB, CL, DM and NS, conceived the hemorrhage. Exp Neurol 2014; 255: 49–55. study and participated in its design and coordination. All 15. Uldall M, Juhler M, Skjolding AD, et al. A novel method authors read and approved the final manuscript. for long-term monitoring of intracranial pressure in rats. J Neurosci Meth 2014; 227: 1–9. 16. McLeod DD, Parsons MW, Hood R, et al. Perfusion References computed tomography thresholds defining ischemic pen- 1. Murtha LA, McLeod DD, McCann SK, et al. Short-dura- umbra and infarct core: studies in a rat stroke model. Int tion hypothermia after ischemic stroke prevents delayed J Stroke 2015; 10: 553–559. intracranial pressure rise. Int J Stroke 2014; 9: 553–559. 17. He Z, Yang SH, Naritomi H, et al. Definition of the 2. Murtha LA, McLeod DD, Pepperall D, et al. Intracranial anterior choroidal artery territory in rats using intralum- pressure elevation after ischemic stroke in rats: cerebral inal occluding technique. J Neurol Sci 2000; 182: 16–28. edema is not the only cause, and short-duration mild 18. Johanson CE, Palm DE, Primiano MJ, et al. Choroid hypothermia is a highly effective preventive therapy. plexus recovery after transient forebrain ischemia: role J Cereb Blood Flow Metab 2015; 35: 592–600. of growth factors and other repair mechanisms. Cell 3. Burrows G. On disorders of the cerebral circiulation and on Mol Neurobiol 2000; 20: 197–216. the connection between affections of the brain and diseases 19. Moffat DB. The development of the posterior cerebral of the heart. London: Longman, 1846. artery. J Anat 1961; 95: 485–494.

154 Beard et al. 11

20. Hu X, Brannstrom T, Gu W, et al. A photothrombotic vascular occlusion patterns after embolic stroke in rats. ring stroke model in rats with or without late spontan- J Cereb Blood Flow Metab 2014; 34: 332–338. eous reperfusion in the region at risk. Brain Research 25. Letourneur A, Roussel S, Toutain J, et al. Impact of gen- 1999; 849: 175–186. etic and renovascular chronic arterial hypertension on the 21. Kleinschnitz C, Braeuninger S, Pham M, et al. Blocking acute spatiotemporal evolution of the ischemic penum- of platelets or intrinsic coagulation pathway-driven bra: a sequential study with MRI in the rat. J Cereb thrombosis does not prevent cerebral infarctions induced Blood Flow Metab 2011; 31: 504–513. by photothrombosis. Stroke 2008; 39: 1262–1268. 26. Reid E, Graham D, Lopez-Gonzalez MR, et al. 22. Watson BD, Dietrich WD, Busto R, et al. Induction of Penumbra detection using PWI/DWI mismatch MRI in reproducible brain infarction by photochemically a rat stroke model with and without comorbidity: com- initiated thrombosis. Ann Neurol 1985; 17: 497–504. parison of methods. J Cereb Blood Flow Metab 2012; 32: 23. Bardutzky J, Shen Q, Henninger N, et al. Differences 1765–1777. in ischemic lesion evolution in different rat strains using 27. Riva M, Pappada GB, Papadakis M, et al. Hemodynamic diffusion and perfusion imaging. Stroke 2005; 36: monitoring of intracranial collateral flow predicts tissue 2000–2005. and functional outcome in experimental ischemic stroke. 24. Bouts MJ, Tiebosch IA, van der Toorn A, et al. Lesion Exp Neurol 2012; 233: 815–820. development and reperfusion benefit in relation to

155 Publication 2

Tissue Plasminogen Activator for preclinical stroke research: Neither “rat” nor “human” dose mimics clinical recanalization in a carotid occlusion model

Amelia J. Tomkins, Rebecca J. Hood, Christopher R. Levi & Neil J. Spratt (2015). Scientific Reports, 5, 16026

156 www.nature.com/scientificreports

OPEN Tissue Plasminogen Activator for preclinical stroke research: Neither “rat” nor “human” dose mimics Received: 27 March 2015 Accepted: 08 October 2015 clinical recanalization in a carotid Published: 02 November 2015 occlusion model

Amelia J. Tomkins1,2, Rebecca J. Hood1,2, Christopher R. Levi2,3,4 & Neil J. Spratt1,2,3

Tissue plasminogen activator (tPA) is the only approved thrombolytic therapy for acute ischemic stroke, yet many patients do not recanalize. Enhancing thrombolytic efcacy of tPA is a major focus of stroke research. Traditionally, a “rat dose” of 10 mg/kg has been used in rodent models. Recent studies suggested that the clinical “human” dose (0.9 mg/kg) may better mimic clinical recanalization. These studies only compared the rat and clinical doses, and so we aimed to test recanalization efcacy of multiple tPA doses ranging from 0.9 to 10 mg/kg in a model of endothelial injury and vessel stenosis. The common carotid artery of rats was crushed and stenosed to allow in- situ occlusive thrombus formation (Folt’s model of ‘physiological’ thrombus). Intravenous tPA was administered 60 minutes post-occlusion (n = 6-7/group). Sustained recanalization rates were 0%, 17%, 67% and 71%, for 0.9, 1.8, 4.5, and 10 mg/kg, respectively. Median time to sustained recanalization onset decreased with increasing dosage. We conclude that 10 mg/kg of tPA is too efective, whereas 0.9 mg/kg is inefective for lysis of occlusive thrombi formed in situ. Neither dose mimics clinical tPA responses. A dose of 2x the clinical dose is a more appropriate mimic of clinical tPA recanalization in this model.

Tissue plasminogen activator (tPA) is the only approved thrombolytic therapy for acute ischemic stroke. Early recanalization of occluded vessels is associated with improved clinical outcome, yet less than 50% of all stroke patients treated intravenously with tPA will successfully recanalize1. In the setting of carotid artery occlusion, tPA is even less efective, with recanalization rates of 10–30%2,3. Tere is a great need for improved therapies for stroke and one approach has been to enhance thrombolysis with adjuvant therapies such as sonothrombolysis4. For any new or adjuvant thrombolytic therapy, rigorous preclinical testing should occur and aim to mimic the clinical conditions of ischemic stroke and tPA efcacy. Tere has been controversy regarding what a “human equivalent” dose of tPA is for preclinical research—par- ticularly in rodents, in which most such studies are performed. Traditionally, a dose of 10 mg/kg tPA has been used for rodents. Tis “rat dose” was based on an in vitro study from the early 1980’s that demonstrated that the fbrinolytic system of rats is 10-fold less sensitive than humans5. Two recent comparisons of rat and clinical doses in rats6 and mice7 both

1School of Biomedical Sciences & Pharmacy, Medical Sciences Building, University of Newcastle, University Drive, Callaghan, NSW, 2308, Australia. 2Hunter Medical Research Institute, Lot 1 Kookaburra Circuit, New Lambton Heights, NSW, 2305, Australia. 3Hunter New England Local Health District, Department of Neurology, Lookout Road, New Lambton Heights, NSW, 2305, Australia. 4School of Medicine and Public Health, University of Newcastle, Newcastle, NSW, Australia. Correspondence and requests for materials should be addressed to N.J.S. (email: [email protected])

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 1

157 www.nature.com/scientificreports/

100 )

75 Rate (%

50

25 Recanalization 0 0306090120 150180 210 Minutes Post tPA bolus

Figure 1. Recanalization rates at varying doses of tPA. Rats with carotid artery occlusion were administered intravenous tPA at the clinical dose (0.9 mg/kg, ● ), 2x the clinical dose (1.8 mg/kg, ▲ ), 5x the clinical dose (4.5 mg/kg, ■), and the rat dose (10 mg/kg, ♦ ). Treatment began 60 minutes post-occlusion. n = 6 per group for 0.9, 1.8 and 4.5 mg/kg, n = 7 for 10 mg/kg. Data presented as the percentage of animals with sustained recanalization per group. Fisher’s exact test, p = 0.015.

indicated that the clinical dose was a better mimic of the clinical situation6,7. Additional doses were not compared in these studies. Te typical method for determining a “human equivalent” dose of any ther- apeutic utilizes a conversion based on body surface area of humans to the target species8. For rats, this conversion requires multiplication of the human dose (0.9 mg/kg) by 6.2, indicating a “human equiva- lent” dose of 5.58 mg/kg tPA for rats. However, this conversion does not take in to account additional factors that may afect the fbrinolytic process other than body size, and is not generally used when converting doses of tPA for stroke research. Multiple methods of forming experimental thrombi exist for preclinical stroke models. Formation of thrombi in situ or ex vivo and the presence or absence of added pro-thrombotic factors, such as thrombin and/or CaCl2, leads to variability of fnal thrombus composition. All of these factors play a key role in the overall thrombolytic susceptibility of the thrombus and therefore the variability of tPA efcacy between studies9,10. Ideally, a preclinical model for testing stroke thrombolytics and thrombolytic enhancers should use thrombi that closely mimic human stroke thrombi and have similar recanalization rates. For this study, we developed a method of physiological thrombus formation by endothelial injury and stenosis of the carotid artery in rats. To determine which tPA dose best refects clinical recanalization rates in this model, we aimed to investigate sustained recanalization rates of varying doses of tPA rang- ing from the clinical dose to the traditional rat dose. Time to sustained recanalization was a secondary outcome.

Results To test the thrombolytic efcacy of varied doses of tPA on a physiological thrombus we used a rat model of carotid occlusion with a mild underlying stenosis. Recanalization was monitored every 30 minutes post-tPA delivery to 4.5 hours post-occlusion. Sustained recanalization, defned as recanalization with- out reocclusion, was observed in 0% of 0.9 mg/kg treated rats (0/6), 17% of 1.8 mg/kg treated rats (1/6), 67% of 4.5 mg/kg treated rats (4/6), and 71% of 10 mg/kg treated rats (5/7) (Fisher’s exact test, p = 0.015, Fig. 1). Recanalization/reocclusion was observed in 2 animals, both in the 10 mg/kg dosage group. In pilot experiments we found that recanalization was easily confrmed (as in Fig. 3B). However accurate quantifcation of the degree of recanalization was not possible because coupling of the fow probe to the vessel with saline caused fuctuations in the baseline of the fow trace (data not shown). We did not see evidence of major changes in fow once vessels did recanalize. Terefore, in the interests of accuracy, we chose sustained recanalization as the marker of recanalization success based on dose, rather than attempting to quantify percentage recanalization. Median times to recanalization (interquartile range) from start of tPA treatment were 210 (210–210) minutes for 0.9 mg/kg treated rats, 210 (210–210) minutes for 1.8 mg/kg treated rats, 65.5 (26–210) for 4.5 mg/kg treated rats, and 34 (27–210) minutes for 10 mg/kg treated rats (Log Rank Test, p = 0.017, Fig. 2). Earliest recanalization onset was 25 minutes (10 mg/kg), with no recanalization occurring beyond 87 minutes (27 minutes afer tPA treatment end) in any group. No animals were excluded. One animal died just prior to the fnal observation (10 mg/kg group). Tis animal had fuctuating body temperatures throughout surgery and high temperatures (> 39 °C) leading up to its death, with no other explanation found. It was included in the primary analysis. Re-analysis

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 2

158 www.nature.com/scientificreports/

10 mg/kg

4.5 mg/kg

1.8 mg/kg

0.9 mg/kg

0306090120 150180 210 Time (min)

Figure 2. Time to sustained recanalization. Sustained recanalization was defned as fow return without reocclusion to experiment end (4.5 hours post-occlusion). Individual animal times are presented as dots. Any animal still occluded at the end of the experiment was designated the maximal time to recanalization (210 min). Data presented as median (solid band) and interquartile range (grey box). Tere was a signifcant overall diference of dose, log rank test, p = 0.017.

A 100 80 ▼ Occlusion ▼ 60

40

Blood Flow (cm/s) 20

0 0 10 20 30 128 1301132 34 B Minutes 50

30

10

Blood Flow (cm/s) 130.4 130.6 130.8 131.0 131.2 Minutes

Figure 3. Doppler fow of crush and occlusion with recanalization. Representative example of Doppler fow indicating crush injury and stenosis followed by occlusion (A). Grey shaded bars indicate 3 × 30 second crush cycles, including a recrush afer 10 minutes of continuous fow. Triangles (▼ ) indicate tightening of silk suture to create stenosis and achieve 75% fow reductions. Dashed horizontal lines indicate baseline fow of 57 cm/s (A) and 25% baseline fow of 14.25 cm/s (A,B). Recanalization was observed at 130.6 minutes (A,B). Te red box in (A) indicates the area zoomed in for (B) to show clear change from occluded “noise” to fow trace, representing recanalization. Recanalization was also confrmed by the return of audible Doppler fow correlating with fow trace.

excluding this animal gave a recanalization rate of 83% (instead of 71%) for this group, but made no material diference to the primary fndings. Discussion In this study, we found that the traditional rat dose of tPA (10 mg/kg) was highly efective for carotid artery recanalization (71% recanalization rate). Tis is far superior to what is achievable in clinical stroke, where recanalization rates are < 50% for middle cerebral artery (MCA) occlusion1 and only 10–30% for occlusions of the carotid arteries2,3. Te clinical dose of tPA (0.9 mg/kg) caused no recanalization in our model, which is also not refective of the clinical situation. We found a 2x clinical dose (1.8 mg/kg) to better refect clinical recanalization rates. We observed a possible ceiling efect at doses at and above 4.5 mg/kg (5x clinical dose). Our fndings are in keeping with previous work showing that the rat fbrino- lytic system is less sensitive than the human system. However, they suggest that the 10-fold diference in sensitivity found in vitro5 may overestimate the in vivo situation.

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 3

159 www.nature.com/scientificreports/

A “human equivalent” dose of tPA for preclinical research that results in recanalization rates refective of the clinical setting has not previously been determined. We chose a range of doses spanning across those used in previous studies, with multiples of the clinical dose (i.e. 2x and 5x the clinical dose). Although the thrombolytic efcacy of tPA has been well established clinically, its efcacy is subobtimal1 and the growing feld of research studying thrombolytic enhancers requires a preclinical dose of tPA that refects the clinical response rates. Recent studies comparing clinical to rat dose reported the clinical dose to be a better clinical mimic, but these studies did not investigate additional doses6,7. Te high recanali- zation rates we saw with the 10 mg/kg “rat” dose are similar to rates reported in other studies using this dose (67–100% in various stroke models7,11–13). Tese rates are not refective of clinical recanalization and are particularly unsuited to studies of thrombolytic enhancer therapies, since there is little scope for additional beneft. Additionally, we found a possible ceiling efect of tPA dose at 4.5 mg/kg, with no signifcant increase in recanalization rates at the higher dose. Such an efect has also previously been reported for doses above 10 mg/kg14. For a “human equivalent” tPA dose, our study indicates that a 2x clinical dose (1.8 mg/kg) in this model with 17% recanalization rates best refects clinical recanalization of carotid artery occlusion for which clinical rates are between 10–30%2,3. To achieve the higher end of this clinical range, a dose between 1.8 mg/kg and 4.5 mg/kg may be necessary. However, this highlights that the previously accepted doses are not ideal to mimic clinical rates. In other situations, such as MCA occlusion, clinical recanalization rates are higher than carotid occlusion1. Difering clot compositions and co-morbidities also afect recanalization efcacy with tPA15,16. It is likely that tPA doses required to mimic clinical recanalization rates will difer with regards to the model choice, incorporating clot type, co-morbidities, and species. We recommend that researchers aiming to fnd improved thrombolytic or recanalization therapies over tPA alone, need to determine the “human equivalent” dose for use in their chosen model that generates recanalization rates that correlate with the clinical conditions being tested. We chose a model of naturally forming thrombi in order to create as ‘physiological’ a model as pos- sible. Tis clot composition is the likely explanation for the complete lack of recanalization we saw with clinical dose tPA, in contrast to previous studies. Such studies have tended to use spontaneously formed thrombi, or other clot types with high sensitivities to tPA thrombolysis. Trombus composition is well known to afect the efcacy of tPA lysis9,10. Our model, based on the Folt’s method, produces physiologi- cal thrombi that are platelet rich17. Platelet rich clots are histologically better mimics of clinical thrombi18 and are more resistant to thrombolysis than many other experimental thrombi19–21. Recanalization was chosen as the primary outcome for this study because it is the major efect by which tPA causes clinical improvement1. In rat models of stroke, recanalization is not always reported due to difculties in directly monitoring recanalization intracranially in the rat. Te extracranial nature of our chosen model allowed direct real time monitoring of recanalization. We observed earlier onset time to recanalization with increasing doses of tPA. With early recanalization being predictive of good clinical outcome1, it is benefcial for enhancer therapies to not only increase the rates of recanalization, but to reduce the time to sustained recanalization onset. At higher doses, recanalization onset is already early (median time 34 minutes for 10 mg/kg), leaving little room for improvement if testing alternate therapies. Functional outcome was intentionally not included in this study because there is already convincing evidence that tPA produces good functional outcome in rodents1,22. We know from clinical studies that functional improvement is highly correlated with recanalization of the occluded artery. Te purpose of this study was to determine whether the previously accepted “rat dose” causes recanalization rates too high to provide any scope for improvement when testing thrombolytic enhancers. In conclusion, we have found that both rat and clinical doses of tPA are not refective of clinical recanalization rates in a model of naturally forming thrombus. Te rat dose was above that producing a “maximal” efect of recanalization. Te no response observed with clinical tPA dose confrms that the rat fbrinolytic system is less sensitive to humans, but not to the 10-fold degree previously accepted. Neither dose appeared ideal for testing thrombolytic enhancers. For this model of carotid occlusion in rats, we propose a 2x clinical dose (1.8 mg/kg) to be best refective of clinical recanalization rates. Methods Animals. Tis study was approved by the Animal Care and Ethics Committee of the University of Newcastle, Australia (Approval No. A-2010-128) and performed in compliance with requirements of the Australian Code of Practice for the Care and Use of Animals for Scientifc Purposes23. Male outbred Wistar rats (n = 25; 338–433 g; Animal Resources Centre, Perth, Australia) were anesthetised with 5% Isofurane, and maintained with 1.5–2% in 30/70% O2/N2 through a nose cone and rectal temperature was maintained at 37 °C with a feedback controlled heat mat (Faculty of Health Workshop, University of Newcastle, Australia). Heart rate, respiration and oxygen saturation were monitored throughout surgery.

Carotid Artery Occlusion. To create a physiological thrombus to occlude the carotid artery, a mod- ifcation of the Folt’s model17 was used with a mild underlying stenosis. Te right carotid arteries were exposed and the internal carotid artery ligated to avoid risk of thromboembolism to the brain. Resultant strokes from carotid thromboembolism are highly variable, as in stroke patients, and could potentially confound experiments. A 20-MHz, 0.8 mm Doppler fow probe (Iowa Doppler Products, USA) was posi- tioned over the external carotid artery to monitor blood fow. Baseline fow was recorded for 5 minutes prior to injury using LabChart 7 sofware (ADInstruments, Australia). Te common carotid artery was

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 4

160 www.nature.com/scientificreports/

crushed three times (30 seconds with 30 second rest intervals, Fig. 3A) using haemostats guarded with tape to disrupt the endothelium, exposing pro-thrombotic factors. A double looped 5-0 silk suture was placed over the site of injury, and tightened to create stenosis following fnal crush. Tis stenosis reduced blood fow by 75% of pre-crush baseline fow. Flow was monitored for cyclic fow patterns until com- plete occlusion was achieved (Fig. 3A). An additional 30 second crush was made over the stenosis if: 1) continuous fow was maintained for 10 minutes afer fnal crush, or 2) if cyclic fow patterns were observed for 30 minutes post-crush with no complete occlusion. No more than two additional crushes were performed in any animal. Te tie was loosened 45 minutes afer stable occlusion to establish a mild stenosis. Flow spontaneously returned pre-treatment in two animals. In one, the vessel was re-crushed, while the other reoccluded spontaneously within minutes. Sixty minute stable occlusion was again mon- itored before treatment.

Treatment Groups. At 60 minutes post-occlusion, animals were intravenously administered tPA (Alteplase, Boehringer Ingelheim, Australia) at the clinical dose (0.9 mg/kg, n = 6), 2x the clinical dose (1.8 mg/kg, n = 6), 5x the clinical dose (4.5 mg/kg, n = 6), or the rat dose (10 mg/kg, n = 7) as a 10% bolus over 1 minute, with the remainder infused over 1 hour. Te 0.9 mg/kg and 10 mg/kg groups were part of a separate study in which they had been randomised to either tPA or tPA+ ultrasound treatment (tPA + ultrasound groups not presented here). Te additional animals were randomized to either 1.8 or 4.5 mg/kg tPA for this study.

Recanalization. Te primary outcome for this study was recanalization rates for each group. Recanalization was monitored every 30 minutes afer tPA onset until 4.5 hours post-occlusion and presented as the total number of animals with sustained recanalization at experiment end per group. Te end point of 4.5 hours was chosen based on the clinical time window for tPA treatment inclusion. Recanalization was reported as sustained when recanalization occurred and fow remained until the fnal observation time. Recanalization/reocclusion was reported when fow was observed at one time point and no fow observed at the next. A secondary outcome of time to sustained recanalization was determined retrospectively from Labchart fles by a blinded observer and was reported as the time that the fow trace returned to normal waveform post tPA treatment start, with the fow trace continuing as normal to experiment end (Fig. 3B).

Statistics. Sustained recanalization rates at endpoint were analysed for statistical signifcance using Fisher’s exact test. Time to sustained recanalization onset was analysed by survival analysis and log-rank test. Statistical signifcance was considered to be a p-value < 0.05.

References 1. Rha, J. H. & Saver, J. L. Te impact of recanalization on ischemic stroke outcome: a meta-analysis. Stroke 38, 967–973 (2007). 2. Christou, I. et al. Intravenous tissue plasminogen activator and fow improvement in acute ischemic stroke patients with internal carotid artery occlusion. J Neuroimaging 12, 119–123 (2002). 3. Pechlaner, R. et al. Recanalization of extracranial internal carotid artery occlusion afer i.v. thrombolysis for acute ischemic stroke. PLoS One 8, e55318 (2013). 4. Tsivgoulis, G. et al. Safety and efcacy of ultrasound-enhanced thrombolysis: a comprehensive review and meta-analysis of randomized and nonrandomized studies. Stroke 41, 280–287 (2010). 5. Korninger, C. & Collen, D. Studies on the specifc fbrinolytic efect of human extrinsic (tissue-type) plasminogen activator in human blood and in various animal species in vitro. Tromb Haemost 46, 561–565 (1981). 6. Haelewyn, B., Risso, J. J. & Abraini, J. H. Human recombinant tissue-plasminogen activator (alteplase): why not use the ‘human’ dose for stroke studies in rats? J Cereb Blood Flow Metab 30, 900–903 (2010). 7. El Amki, M. et al. Experimental modeling of recombinant tissue plasminogen activator efects afer ischemic stroke. Exp Neurol 238, 138–144 (2012). 8. U.S. department of Health and Human Services, Food and drug administration. Centre for drug Evaluation and Research (CDER). Guidance for Industry. Estimating the maximum safe starting dose in initial clinical trials for therapeutics in adult healthy volunteers. Pharmacology and Toxicology (2005). 9. Niessen, F., Hilger, T., Hoehn, M. & Hossmann, K. A. Diferences in clot preparation determine outcome of recombinant tissue plasminogen activator treatment in experimental thromboembolic stroke. Stroke 34, 2019–2024 (2003). 10. Overgaard, K. et al. Composition of emboli infuences the efcacy of thrombolysis with rt-PA in a rat stroke model. Fibrinolysis 7, 141–148 (1993). 11. Busch, E., Kruger, K. & Hossmann, K. A. Improved model of thromboembolic stroke and rt-PA induced reperfusion in the rat. Brain Res 778, 16–24 (1997). 12. Guluma, K. Z. & Lapchak, P. A. Comparison of the post-embolization efects of tissue-plasminogen activator and simvastatin on neurological outcome in a clinically relevant rat model of acute ischemic stroke. Brain Res 1354, 206–216 (2010). 13. Okubo, S. et al. Terapeutic time window of rt-PA on embolic stroke in rats. Int Congress Series 1252, 203–207 (2003). 14. Calandre, L., Grau, M. & Cabello, A. Lysis rates with rt-PB vary between diferent human emboli in a rat model of cerebral embolism. Fibrinolysis & Proteolysis 12, 107–111 (1998). 15. Ribo, M. et al. Acute hyperglycemia state is associated with lower tPA-induced recanalization rates in stroke patients. Stroke 36, 1705–1709 (2005). 16. Tsivgoulis, G. et al. Association of pretreatment blood pressure with tissue plasminogen activator-induced arterial recanalization in acute ischemic stroke. Stroke 38, 961–966 (2007). 17. Sturgeon, S. A., Jones, C., Angus, J. A. & Wright, C. E. Adaptation of the Folts and electrolytic methods of arterial thrombosis for the study of anti-thrombotic molecules in small animals. J Pharmacol Toxicol Methods 53, 20–29 (2006).

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 5

161 www.nature.com/scientificreports/

18. Roessler, F. C. et al. Development of a new clot formation protocol for standardized in vitro investigations of sonothrombolysis. J Neurosci Methods 237, 26–32 (2014). 19. Collet, J. P. et al. Disaggregation of in vitro preformed platelet-rich clots by abciximab increases fbrin exposure and promotes fbrinolysis. Arterioscler Tromb Vasc Biol 21, 142–148 (2001). 20. Gold, H. K. et al. Animal models for arterial thrombolysis and prevention of reocclusion. Erythrocyte-rich versus platelet-rich thrombus. Circulation 83, IV26–40 (1991). 21. Tomkins, A. J. et al. Platelet rich clots are resistant to lysis by thrombolytic therapy in a rat model of embolic stroke. Exp Transl Stroke Med 7, 2 (2015). 22. Sena, E.S. et al. Factors afecting the apparent efcacy and safety of tissue plasminogen activator in thrombotic occlusion models of stroke: systematic review and meta-analysis. JCBFM 30, 1905–1913 (2010). 23. National Health and Medical Research Council (NHMRC). Australian code for the care and use of animals for scientifc purposes, 8th edition. Canberra: National Health and Medical Research Council (2013). Acknowledgements Te authors would like to acknowledge the technical assistance of Mrs Debbie Pepperall. Ms Amelia Tomkins was supported by a National Heart foundation/National Stroke Foundation Postgraduate Scholarship. Dr Neil Spratt was supported by an Australian National Health and Medical Research Council Career Development Fellowship (APP1035465). Author Contributions A.T., C.L. and N.S. were responsible for study design. Experimental procedures were performed by A.T. and R.H. Data analysis and manuscript preparation was by A.T. and N.S. All authors reviewed the fnal manuscript. Additional Information Competing fnancial interests: Te authors declare no competing fnancial interests. How to cite this article: Tomkins, A. J. et al. Tissue Plasminogen Activator for preclinical stroke research: Neither “rat” nor “human” dose mimics clinical recanalization in a carotid occlusion model. Sci. Rep. 5, 16026; doi: 10.1038/srep16026 (2015). Tis work is licensed under a Creative Commons Attribution 4.0 International License. Te images or other third party material in this article are included in the article’s Creative Com- mons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Scientific RepoRts | 5:16026 | DOi: 10.1038/srep16026 6

162 Publication 3

Thrombolytic recanalization of carotid arteries is highly dependent on degree of stenosis, despite sonothrombolysis

Amelia J Tomkins, Rebecca J Hood, Debbie Pepperall, Christopher L Null, Christopher R Levi, Neil J Spratt (2016). Journal of the American Heart Association, 5(2) e002716

163 ORIGINAL RESEARCH

Thrombolytic Recanalization of Carotid Arteries Is Highly Dependent on Degree of Stenosis, Despite Sonothrombolysis Amelia J. Tomkins, PhD; Rebecca J. Hood, BBiomedSci(Hons); Debbie Pepperall, DipPathTech; Christopher L. Null, BS; Christopher R. Levi, MBBS, BMedSci, FRACP; Neil J. Spratt, FRACP, PhD

Background-—Stroke associated with acute carotid occlusion is associated with poor effectiveness of tissue plasminogen activator (tPA) thrombolysis and poor prognosis. Rupture of atherosclerotic plaques resulting in vascular occlusions may occur on plaques, causing variable stenosis. We hypothesized that degree of stenosis may affect recanalization rates with tPA. Ultrasound+tPA (sonothrombolysis) has been shown to improve recanalization for intracranial occlusions but has not been tested for carotid occlusion. Our primary aim was to determine thrombolytic recanalization rates in a model of occlusion with variable stenosis, with a secondary aim to investigate sonothrombolysis in this model. Methods and Results-—Rat carotid arteries were crushed and focal stenosis created (25% baseline Doppler flow) with a silk-suture tie invoking thrombosis and occlusion. To model mild or severe stenosis, the tie was released pretreatment or left in place. Animals were treated with tPA (10 mg/kg) or tPA+ultrasound (2-MHz) in each stenosis model (n=7/group). Recanalization was assessed by Doppler flow. Thrombolytic recanalization rates were significantly higher in mild stenosis groups (71% versus 0% with severe stenosis; P<0.0001). Recanalization rates were not significantly higher with additional ultrasound in either model. Conclusions-—In this model, the degree of carotid stenosis had a large effect on thrombolytic recanalization. Sonothrombolysis using standard parameters for intracranial sonothrombolysis did not increase recanalization. Further testing is warranted. The degree of underlying stenosis may be an important predictor of thrombolytic recanalization, and clinical correlation of these findings may provide new approaches to treatment selection for patients with carotid occlusion. ( J Am Heart Assoc. 2016;5: e002716 doi: 10.1161/JAHA.115.002716) Key Words: carotid arteries • rats • sonothrombolysis • stenosis • thrombolysis

arotid artery occlusion can lead to devastating conse- population frequency of mild stenosis is greater. Hence, acute C quences such as stroke. Patients with carotid occlusion– occlusion of the carotid artery that leads to stroke may occur associated stroke have high rates of death and disability but with varying degrees of carotid stenosis. We hypothesized that treatments are limited. Successful recanalization with intra- the degree of underlying stenosis of carotid artery occlusion venous tissue plasminogen activator (tPA), the current stroke may be an important predictor of successful thrombolytic thrombolytic, is only achieved in 10% to 30% of patients with recanalization. Given the low efficacy of tPA in carotid athero- carotid occlusion.1–3 Although a severe carotid stenosis thrombotic occlusion, approaches to enhance its effect are predicts a greater risk of stroke for any individual,4 the required. Ultrasound+tPA (sonothrombolysis) appears promis- ing for improving intracranial recanalization,5 but has not been tested in extracranial carotid artery occlusion. Therefore, our From the School of Biomedical Sciences & Pharmacy, University of Newcastle, Callaghan, NSW, Australia (A.J.T., R.J.H., D.P., N.J.S.); School of Medicine and primary aim was to determine whether the degree of carotid Public Health, University of Newcastle, Callaghan, NSW, Australia (C.R.L.); stenosis affected rates of sustained recanalization with throm- Hunter Medical Research Institute, New Lambton, NSW, Australia (A.J.T., R.J.H., bolysis at 4.5 hours postocclusion. Our secondary aim was to D.P., C.R.L., N.J.S.); Hunter New England Local Health District Newcastle, determine whether ultrasound enhances sustained tPA recanal- Newcastle, NSW, Australia (C.R.L., N.J.S.); Seventh Wave Laboratories, Chesterfield, MO (C.L.N.). ization rates in carotid occlusion. Correspondence to: Neil J. Spratt, FRACP, PhD, School of Biomedical Sciences & Pharmacy, University of Newcastle, Callaghan, NSW 2308, Australia. E-mail: [email protected] Received September 24, 2015; accepted January 22, 2016. Methods ª 2016 The Authors. Published on behalf of the American Heart Association, Inc., by Wiley Blackwell. This is an open access article under the terms of the General Animal Details Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is Experiments were approved by the Animal Care and Ethics properly cited and is not used for commercial purposes. Committee of the University of Newcastle, Australia (Approval

DOI: 10.1161/JAHA.115.002716 Journal of the American Heart Association 1

164 Carotid Recanalization Is Dependent on Stenosis Tomkins et al RGNLRESEARCH ORIGINAL

No. A-2010-128) and performed in compliance with require- through the Doppler instrumentation and visually via ments of the Australian Code of Practice for the Care and Use LabChart software (ADInstruments, Australia). Since the of Animals for Scientific Purposes.6 Male Wistar rats (n=28; primary outcome was recanalization of the occluded common 311–417 g; Animal Resources Centre, Perth, Australia) were carotid artery, the internal carotid artery was ligated. This was anesthetized with 5% isoflurane and maintained with 1.5% to done to avoid the potential confounding effects of variable

2% in 30/70% O2/N2. Rectal temperature was maintained at strokes that may result from thromboembolism. A 5-0 silk 37°C with a feedback-controlled heat mat (Faculty of Health suture with a double throw tie was placed around the common Workshop, University of Newcastle, Australia). Heart rate, carotid artery at the intended stenosis site, but not tightened. respiration, and oxygen saturation were monitored throughout Using hemostats guarded with tape, the common carotid surgery. artery was crushed 3 times (30-s crush, 30-s rest) to disrupt the endothelium, exposing prothrombotic factors. Crushes were made over and directly adjacent to the suture. Following Model of Carotid Artery Occlusion With Stenosis the third crush, a stenosis was created by tightening the To create carotid artery occlusion with variable degrees of suture over the site of injury, reducing blood flow by 75% of stenosis, a modification of the Folt’s model was used.7 A precrush baseline flow. Flow was monitored for cyclic timeline and surgical schematic are presented in Figure 1. patterns until complete occlusion was achieved. Complete The right carotid arteries were exposed and a 20-MHz, 0.8- occlusion was confirmed by cessation of audible flow and mm Doppler flow probe (Iowa Doppler Products, USA) was irregular flow trace (Figure 2A and 2B). Up to 2 additional positioned over the external carotid artery, distal to the crushes were made if: (1) continuous flow was maintained for intended injury/stenosis site. Flow was monitored audibly 10 minutes after the crush cycle, or (2) if cyclic flow patterns including occlusion/recanalization events were observed for 30 minutes postcrush with no complete occlusion. To model mild or severe stenosis, the tie was loosened (mild) or left in place (severe) after 45 minutes of stable occlusion. If flow spontaneously returned pretreatment (n=1), the vessel was recrushed and reobserved for 60-minute occlusion. For each stenosis protocol, animals were random- ized to tPA, or tPA+ultrasound. tPA (10 mg/kg; Alteplase, Boehringer Ingelheim, Australia) was administered intra- venously via the femoral vein at 60 minutes postocclusion (10% bolus over 1 minute, remainder over 1 hour). For ultrasound groups, the skin was sutured, covered with TegadermTM (3M, North Ryde, Australia), and the ultrasound probe was positioned over the occlusion using a laboratory- manufactured spacer containing ultrasound gel. Insonation began with the start of tPA delivery, continuing for 2 hours (2- MHz, 720 mW/cm2, 25-mm depth, sample volume 10 mm; EZ-Dopâ, DWL Compumedics, Germany). Recanalization was Figure 1. Experimental timeline (A) and surgery schematic (B) of monitored at 30-minute intervals after tPA onset until carotid artery occlusion with stenosis. Branching arteries were 4.5 hours postocclusion. Recanalization was confirmed by cauterized (superior thyroid and occipital arteries), and the internal audible flow return and a regular LabChart flow trace carotid artery (ICA) was ligated to prevent embolism to the brain. A (Figure 2A and 2C). flow probe was placed over the external carotid artery (ECA) and baseline flow monitored for 5 minutes (depicted in blue in [B]). The stenosis suture was placed around the common carotid artery Statistical Analysis (CCA) but not tightened. Three 30-s crushes ([A] red) were made over and immediately adjacent to the suture with 30 second rest Baseline time for analysis was 60 minutes postocclusion, periods between crushes. The suture was tightened to reduce flow immediately prior to tPA administration. Recanalization was fl by 75% and ow was monitored for occlusion. Following 45 min- reported as sustained when a return of flow was observed and utes of stable occlusion, the stenosis suture was either loosened to remained until the final observation time (4.5 hours postoc- create mild stenosis or left in place for a severe stenosis. Treatment began 60 minutes postocclusion. Flow was monitored clusion). Rates of sustained recanalization (number of animals every 30 minutes post–treatment onset for any signs of recanal- with sustained recanalization from the total cohort) were ization ([A] green arrows). compared using log-rank test with differences considered

DOI: 10.1161/JAHA.115.002716 Journal of the American Heart Association 2

165 Carotid Recanalization Is Dependent on Stenosis Tomkins et al RGNLRESEARCH ORIGINAL

Figure 2. Example flow trace of crush, occlusion, and recanalization. A, Doppler Flow recorded via LabChart (ADInstruments, Australia) demonstrating crush injury (▲), stenosis (at 10 min), flow decrease À to occlusion (gray shading, B), and recanalization (gray shading, C). B, Flow decrease to occlusion, correlating with gray shading labeled B in (A). At low flow rates, audible signal was a better indicator of flow due to “noise” of the flow trace, as can be seen around the point of occlusion in (B). C, Recanalization correlating with gray shading labeled C in (A). significant if P<0.05 (Graphpad Prism 6). Investigation of flow return was clearly discernible irrespective of skin closure degree of stenosis was performed by pooling tPA and (Figure 2C). Therefore, in the interests of accuracy, we chose tPA+ultrasound groups per stenosis model (Figure 3B). Effect sustained recanalization as the marker of recanalization of treatment on sustained recanalization was analyzed for success, rather than attempting to quantify percentage each stenosis model. recanalization. The carotid artery of all animals had been occluded for 60 minutes at tPA onset. High rates of sustained recanaliza- Results tion were observed in the mild stenosis model compared to No animals were excluded. One animal died prior to final severe stenosis (P<0.0001; Figure 3A and 3B). In the mild observation (mild stenosis-tPA group). This animal had stenosis-tPA group, all animals exhibited recanalization. fluctuating body temperatures throughout surgery and high Sustained recanalization was observed in 5/7 rats, and the temperatures (>39°C) leading up to its death. It was included remaining 2 animals had recanalization/reocclusion. Reoc- in the primary analysis, and re-analysis excluding this animal clusion occurred within 30 minutes of recanalization onset in did not change the reported results. both cases. In all, initial recanalization occurred between 30 Exact flow values relative to baseline are not presented and 90 minutes post–treatment onset. In severe stenosis- because closing the surgical site for ultrasound insonation tPA, no recanalization was observed. affected the Doppler flow trace relative to baseline, preventing Sonothrombolysis had limited effects on recanalization in accurate judgment of the degree of recanalization. However, either mild or severe stenosis groups (Figure 3A). In mild

DOI: 10.1161/JAHA.115.002716 Journal of the American Heart Association 3

166 Carotid Recanalization Is Dependent on Stenosis Tomkins et al RGNLRESEARCH ORIGINAL

patients with acute carotid occlusion. The conventional “rat dose” of tPA, 10 mg/kg, was chosen based on its common use in the literature. There is evidence that this dose exerts a maximum effect8 and further dosage increases do not enhance recanalization.9 This study did not demonstrate a benefit of sonothrombolysis over tPA alone for extracranial carotid occlusion. The results provide evidence of the influence of stenosis on thrombolysis as well as providing a preclinical model for testing thrombolytics in extracranial carotid occlusions. Correlation with clinical data is, however, needed to confirm this effect in patients. There are differences between our acute stenosis model and patients with carotid stenosis and/ or occlusion. In patients, stenosis due to atherosclerosis is generally due to chronic plaque build-up, while our model is an acute stenosis. Our model is not an absolute replicate of the human condition of atherothrombotic carotid stenosis and occlusion, yet it is an important step forward for experimental models of carotid occlusion. It mimics the stenotic narrowing of the artery, reduced blood flow, endothelial damage, and exposure of subendothelial matrix proteins that promote Figure 3. Carotid artery recanalization in the setting of mild or thrombosis and cause carotid occlusion. Previous models of severe stenosis. Recanalization rates are expressed as the carotid occlusion have used ex vivo prepared clots injected percentage of animals recanalized per group (n=7/group). 10 Recanalization was measured every 30 minutes post–treatment into the vessel, or other nonphysiological forms of throm- 11 onset for 240 minutes (4.5 hours postocclusion). A, Recanaliza- bosis induction, such as FeCl3. Additionally, none of these tion/reocclusion events for all animals with mild or severe models have investigated stenosis as a factor of thrombolytic stenosis treated with tissue plasminogen activator (tPA) alone or efficacy. The results of our study indicate a clear difference in tPA+ultrasound (n=7/group). B, Sustained recanalization (re- recanalization response to thrombolytic therapy with differing canalization at end point) in mild (open triangle) and severe degrees of stenosis. It is this finding that should prompt stenosis models, pooled treatment (*P<0.0001, hazard ratio=16.7 [4.2–67.4, 95% CI]; n=14/group). Presence of ultra- further investigation of this effect clinically, and whether sound made no significant difference to the rates of sustained techniques can be developed to quantify the degree of recanalization in either mild or severe stenosis models. underlying carotid stenosis. Our experimental data suggest that this may prove useful in the treatment allocation of patients with carotid occlusion-associated stroke—particu- stenosis-tPA+ultrasound, sustained recanalization occurred larly in centers that may need to transfer patients for within 30 minutes of treatment onset in 5/7 rats, as was also endovascular therapies. seen without the addition of ultrasound (P=0.67). Recanaliza- In this study, we used the standard ultrasound parameters tion/reocclusion occurred in 2/7 of severe stenosis-tPA+ul- used for transcranial sonothrombolysis;5,12 however, even in trasound rats, but none had sustained recanalization the absence of skull attenuation, there did not appear to be a (Figure 3A). large additive benefit over tPA alone. Ultrasound power is attenuated through the skull by as much as 99%,13,14 yet evidence that higher power increases clot lysis15 indicates Discussion that power attenuation likely limits the efficacy of intracranial This study demonstrates a strong effect of the degree of sonothrombolysis. Additionally, there is evidence that low- carotid stenosis on thrombolytic recanalization after throm- frequency ultrasound facilitates clot lysis.15,16 However, for botic carotid occlusion. Sustained recanalization rates were intracranial occlusion, low-frequency ultrasound has the >70% with mild stenosis, compared to 0% with severe potential to cause the propagation of standing waves within stenosis. These findings have potentially important clinical the skull, causing intracerebral hemorrhage.14,17 In the setting implications. If the degree of carotid stenosis has similar of carotid occlusion, where there is not the risk of standing effects clinically and new imaging techniques enable degree waves and brain bleeding, lower frequency and/or higher of stenosis to be assessed, this could guide the choice of ultrasound power may be more effective for thrombolysis. thrombolysis versus endovascular treatment for stroke This proposition requires further testing.

DOI: 10.1161/JAHA.115.002716 Journal of the American Heart Association 4

167 Carotid Recanalization Is Dependent on Stenosis Tomkins et al RGNLRESEARCH ORIGINAL

Our data suggest that severe carotid stenosis is a likely 4. Brott TG, Halperin JL, Abbara S, Bacharach JM, Barr JD, Bush RL, Cates CU, Creager MA, Fowler SB, Friday G, Hertzberg VS, McIff EB, Moore WS, Panagos explanation for at least some of the failed tPA recanalization PD, Riles TS, Rosenwasser RH, Taylor AJ. 2011 ASA/ACCF/AHA/AANN/ AANS/ACR/ASNR/CNS/SAIP/SCAI/SIR/SNIS/SVM/SVS guideline on the in patients with carotid occlusion. In our model, as in human management of patients with extracranial carotid and vertebral artery disease: atherothrombotic carotid occlusion, the bulk of the thrombus executive summary. Circulation. 2011;124:489–532. forms distal to the stenosis, thereby limiting drug delivery to 5. Tsivgoulis G, Eggers J, Ribo M, Perren F, Saqqur M, Rubiera M, Sergentanis TN, Vadikolias K, Larrue V, Molina CA, Alexandrov AV. Safety and efficacy the thrombus. This is likely to be different from intracranial of ultrasound-enhanced thrombolysis: a comprehensive review and meta- analysis of randomized and nonrandomized studies. Stroke. 2010;41:280– occlusions, which in Western populations are predominantly 287. thromboembolic in nonstenosed vessels. Advanced computed 6. Council NHaMR. Australian Code for the Care and Use of Animals for Scientific tomography, magnetic resonance, and carotid duplex imaging Purposes. 8th ed. Canberra: National health and medical research council; fl 2013. can evaluate both intraluminal ow and vessel wall charac- 7. Sturgeon SA, Jones C, Angus JA, Wright CE. Adaptation of the Folt’s and teristics and are also now being used to evaluate patient electrolytic methods of arterial thrombosis for the study of anti-thrombotic molecules in small animals. J Pharmacol Toxicol Methods. 2006;53:20–29. suitability for reperfusion therapies in acute stroke. The ability 8. Tomkins AJ, Hood RJ, Levi CR, Spratt NJ. Tissue plasminogen activator for to determine the degree of carotid stenosis in patients with preclinical stroke research: neither “rat” nor “human” dose mimics clinical carotid occlusion may help guide the choice of therapy for recanalization in a carotid occlusion model. Sci Rep. 2015;5:16026. 9. Calandre L, Grau M, Cabello A. Lysis rates with rt-PA vary between different patients with this devastating form of stroke. human emboli in a rat model of cerebral embolism. Fibrinolysis Proteolysis. 1998;12:107–111. 10. Holscher T, Fisher DJ, Ahadi G, Voie A. Introduction of a rabbit carotid artery model for sonothrombolysis research. Transl Stroke Res. 2012;3:397–407. Sources of Funding 11. Surin WR, Prakash P, Barthwal MK, Dikshit M. Optimization of ferric chloride induced thrombosis model in rats: effect of anti-platelet and anti-coagulant Tomkins was supported by a National Heart Foundation/ drugs. J Pharmacol Toxicol Methods. 2010;61:287–291. National Stroke Foundation postgraduate scholarship. Spratt 12. Alexandrov AV, Molina CA, Grotta JC, Garami Z, Ford SR, Alvarez-Sabin J, Montaner J, Saqqur M, Demchuk AM, Moye LA, Hill MD, Wojner AW. was supported by an NHMRC career development fellowship, Ultrasound-enhanced systemic thrombolysis for acute ischemic stroke. N Engl #1035465. J Med. 2004;351:2170–2178. 13. Pfaffenberger S, Devcic-Kuhar B, Kollmann C, Kastl SP, Kaun C, Speidl WS, Weiss TW, Demyanets S, Ullrich R, Sochor H, Wober C, Zeitlhofer J, Huber K, Groschl M, Benes E, Maurer G, Wojta J, Gottsauner-Wolf M. Can a commercial Disclosures diagnostic ultrasound device accelerate thrombolysis? An in vitro skull model Stroke. 2005;36:124–128. None. 14. Baron C, Aubry JF, Tanter M, Meairs S, Fink M. Simulation of intracranial acoustic fields in clinical trials of sonothrombolysis. Ultrasound Med Biol. 2009;35:1148–1158. 15. Nedelmann M, Eicke BM, Lierke EG, Heimann A, Kempski O, Hopf HC. Low- References frequency ultrasound induces nonenzymatic thrombolysis in vitro. J Ultrasound 1. Rha JH, Saver JL. The impact of recanalization on ischemic stroke outcome: a Med. 2002;21:649–656. meta-analysis. Stroke. 2007;38:967–973. 16. Akiyama M, Ishibashi T, Yamada T, Furuhata H. Low-frequency ultrasound 2. Christou I, Felberg RA, Demchuk AM, Burgin WS, Malkoff M, Grotta JC, penetrates the cranium and enhances thrombolysis in vitro. Neurosurgery. Alexandrov AV. Intravenous tissue plasminogen activator and flow improve- 1998;43:828–832; discussion 832-823. ment in acute ischemic stroke patients with internal carotid artery occlusion. J 17. Daffertshofer M, Gass A, Ringleb P, Sitzer M, Sliwka U, Els T, Sedlaczek O, Neuroimaging. 2002;12:119–123. Koroshetz WJ, Hennerici MG. Transcranial low-frequency ultrasound-mediated 3. Pechlaner R, Knoflach M, Matosevic B, Ruecker M, Schmidauer C, Kiechl S, thrombolysis in brain ischemia: increased risk of hemorrhage with combined Willeit J. Recanalization of extracranial internal carotid artery occlusion after ultrasound and tissue plasminogen activator: results of a phase II clinical trial. i.v. thrombolysis for acute ischemic stroke. PLoS One. 2013;8:e55318. Stroke. 2005;36:1441–1446.

DOI: 10.1161/JAHA.115.002716 Journal of the American Heart Association 5

168 Publication 4

Heptanoate is neuroprotective in vitro but triheptanoin post-treatment did not protect against middle cerebral artery occlusion in rats

Kah Ni Tan, Rebecca Hood, Kirby Warren, Debbie Pepperall, Catalina Carrasco-Pozo, Silvia Manzanero, Karin Borges, Neil J Spratt (2018). Neuroscience Letters, 683:207-214

169 1HXURVFLHQFH/HWWHUV  ²

Contents lists available at ScienceDirect

Neuroscience Letters

journal homepage: www.elsevier.com/locate/neulet

Research article Heptanoate is neuroprotective in vitro but triheptanoin post-treatment did 7 not protect against middle cerebral artery occlusion in rats

Kah Ni Tana,d, Rebecca Hoodb, Kirby Warrenb, Debbie Pepperallb, Catalina Carrasco-Pozoc,d, Silvia Manzaneroe, Karin Borgesa,⁎,1, Neil J. Sprattb,f,1 a School of Biomedical Sciences, Faculty of Medicine, The University of Queensland, St. Lucia, Australia b School of Biomedical Sciences and Pharmacy, University of Newcastle and Hunter Medical Research Institute, Callaghan, Australia c Department of Nutrition, Faculty of Medicine, The University of Chile, Santiago, Chile d Discovery Biology, Griffith Institute for Drug Discovery, Griffith University, Nathan, Australia e Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, St. Lucia, Australia f Department of Neurology, John Hunter Hospital, New Lambton, Australia

ARTICLE INFO ABSTRACT

Keywords: Triheptanoin, the medium-chain triglyceride of heptanoate, has been shown to be anticonvulsant and neuro- Heptanoate protective in several neurological disorders. In the gastrointestinal tract, triheptanoin is cleaved to heptanoate, Ischemic stroke which is then taken up by the blood and most tissues, including liver, heart and brain. Here we evaluated the MCAo neuroprotective effects of heptanoate and its effects on mitochondrial oxygen consumption in vitro. We also Mitochondrial function investigated the neuroprotective effects of triheptanoin compared to long-chain triglycerides when administered Neuroprotection after stroke onset in rats. Heptanoate pre-treatment protected cultured neurons against cell death induced by Triheptanoin oxygen glucose deprivation and N-methyl-D-aspartate. Incubation of cultured astrocytes with heptanoate for 2 h increased mitochondrial proton leak and also enhanced basal respiration and ATP turnover, suggesting that heptanoate protects against oxidative stress and is used as fuel. However, continuous 72 h infusion of tri- heptanoin initiated 1 h after middle cerebral artery occlusion in rats did not alter stroke volume at 3 days or neurological deficit at 1 and 3 days relative to long-chain triglyceride control treatment.

1. Introduction ischemic core [2–4] and penumbra [4]. Studies using 13C-labeled glu- cose and acetate found that impairments in neuronal and astrocytic Ischemic stroke accounts for approximately 87% of stroke incidents metabolism occurred in both the ischemic core [5] and penumbra of [1], and is one of the leading causes of death and disability worldwide. rats subjected to middle cerebral artery occlusion (MCAo) [6]. In ad- Recent advances in endovascular and pharmacological reperfusion dition, a loss of glutamate into systemic circulation and cerebrospinal therapies have improved outcomes for some patients dramatically. fluid was observed in both humans [7] and rats [8], which could drain However, these therapies require early administration and access is α-ketoglutarate from the tricarboxylic acid (TCA) cycle. This can fur- variable. Approaches to protect the brain while transporting patients to ther exacerbate the underlying metabolic perturbations, as four- and reperfusion centres offer great promise for even greater reductions in five-carbon TCA cycle intermediates are needed for the TCA cycle to disability associated with stroke. run efficiently and to produce amino acids [9,10]. Please note that the Growing pre-clinical and clinical evidence suggests that metabolic four-carbon oxaloacetate is required to allow the two-carbon entry of treatment approaches might be a potential treatment for ischemic acetyl-CoA into the TCA cycle. stroke due to the metabolic dysfunctions observed in brain. Upon artery The entry of carbons into the TCA cycle can be increased by ana- occlusion, glucose-dependent ATP generation decreases rapidly, plerosis (refilling of four- and five-carbon TCA cycle intermediates. causing both ATP and phosphocreatine levels to decrease in the Anaplerosis occurs largely in astrocytes, namely by production of

fl Abbreviations: BHB, β-hydryoxybutyrate; CBF, cerebral blood ow; LCT, long-chain triglyceride; MCAo, middle cerebral artery occlusion; NMDA, N-methyl-D- aspartate; OCR, oxygen consumption rate; OGD, oxygen glucose deprivation; TCA, tricarboxylic acid ⁎ Corresponding author at: School of Biomedical Sciences, Faculty of Medicine, The University of Queensland, St. Lucia, QLD, 4072, Australia. E-mail address: [email protected] (K. Borges). 1 Denotes equal contribution. https://doi.org/10.1016/j.neulet.2018.07.045 Received 11 April 2018; Received in revised form 17 July 2018; Accepted 31 July 2018 $YDLODEOHRQOLQH$XJXVW ‹(OVHYLHU%9$OOULJKWVUHVHUYHG

170 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ² oxaloacetate from pyruvate via pyruvate carboxylase. The carbons from 2.2. Primary cortical neuronal and astrocyte cultures this reaction can be transferred to neurons via several metabolic reac- tions, including enzymes of the TCA cycle to produce α-ketoglutarate, All culture media and reagents were obtained from Life which is then turned into glutamate and glutamine in astrocytes. The Technologies (CA, USA), unless stated otherwise. Briefly, bilateral glutamine-glutamate cycle can transfer carbons to neurons and can cortices were removed from E15 CD1 mouse embryos, triturated and 6 replenish their α-ketoglutarate levels. However, after MCAo, the ac- seeded at a density of 2 × 10 cells per well in a 6-well plate (OGD) or tivity of pyruvate carboxylase was found to be decreased [6]. In addi- 7.5 × 105 cells per well in a 24-well plate (NMDA) with Neurobasal® tion, the entry of glucose-derived carbons into the TCA cycle seems to media containing 25 mM glucose, 2 mM glutamine, 10 μg/ml genta- be impeded, due to reduction in pyruvate dehydrogenase activity micin and serum-free B-27® Supplement. For astrocyte-enriched cul- [11,12]. Therefore, a metabolic treatment that bypasses the glycolytic tures, cells were seeded at 3 × 105 cells/ml in a 175 cm2 culture flask pathway, replenishes the TCA cycle, and thereby improves the entry of and cultured in Dulbecco's modified Eagle's medium (DMEM) with glucose-derived carbons into the TCA cycle has been thought to be a Ham's F12 nutrient (1:1) containing 10% foetal bovine serum and an- potential treatment for ischemic stroke. tibiotics penicillin and streptomycin (25 U/ml final). Neuronal cells Triheptanoin is a triglyceride containing three seven-carbon hep- were cultured for 8 or 12 days and astrocyte-enriched cultures for two fl tanoate molecules, which are released after cleavage by lipases in the weeks until con uent at 37 °C with 5% CO2. gastro-intestinal tract and directly taken up into blood. Heptanoate can enter the brain directly or after hepatic metabolism into C5-ketone 2.3. Oxygen glucose deprivation and NDMA-induced cell death bodies [13]. Beta-oxidation of heptanoate and C5 ketone bodies pro- duces acetyl-CoA and propionyl-CoA, the latter is able to replenish the After eight days in vitro, neuronal cell cultures were treated with 50 TCA cycle via the propionyl-CoA carboxylation pathway, which pro- μM or 200 μM of heptanoic acid (pH 7.4; Sigma Aldrich) for 16 h prior duces the four-carbon succinyl-CoA [14,15]. In addition to being cur- to OGD. Culture media were then replaced with Locke’s buffer con- rently used as the treatment for rare metabolic disorders in children and taining 154 mM NaCl, 5.6 mM KCl, 2.3 mM CaCl2, 1 mM MgCl2, 3.6 mM adults [16,17], triheptanoin has been shown to be neuroprotective in NaHCO3, and 5 mM HEPES (pH 7.4) before being incubated in an several animal models of neurological disorders. Triheptanoin pre- anaerobic incubator with 95% N2, 5% CO2 and < 1% O2 at 37 °C for 3 treatment (35E% of total caloric intake) for two weeks prior to MCAo or 9 h. In NMDA-induced cell death experiments, neuronal cultures (12 fi signi cantly reduced infarct area, improved neurological function, days in vitro) were treated with 50 μM of heptanoic acid for 16 h prior to prevented increases in extracellular glutamate levels and attenuated treatment with 25 μM of NMDA (Sigma Aldrich) for 24 h. Cell death reductions in ATP levels as well as mitochondrial complex II and IV was measured using a Cytotoxicity Detection Kit (Roche Applied activities in mice [18]. Triheptanoin treatment also reduced oxidative Science, Mannheim, Germany) and expressed as percentage of lactate stress, prevented the loss of oligodendrocytes and improved motor dehydrogenase released into the media relative to the total amount of function in a mouse model of Canavan Disease [19], an autosomal re- lactate dehydrogenase activity per well. cessive neurodegenerative disorder caused by a deficiency of the en- zyme aminoacylase 2. In addition, oral triheptanoin treatment was 2.4. Mitochondrial function assessment in cultured astrocytes found to delay motor neuron loss and the onset of symptoms in a mouse Amyotrophic Lateral Sclerosis model [20]. After two weeks in vitro, microglial cells were shaken off and as- Here, we aimed to investigate the protective effects of heptanoate trocytes (as prepared in “Primary cortical neuronal and astrocyte cul- pre-treatment against cell death induced by oxygen glucose deprivation tures” section) were replated at a high density of 1.5 × 105 cells/well in (OGD) or N-methyl-D-aspartate (NMDA) in cultured neurons. We also polyethylenimine-coated XFe96 Cell Culture Microplates (Seahorse assessed changes in mitochondrial function after incubation with hep- Bioscience, MA, USA) and cultured for three days. Then, cultures were tanoate in cultured astrocyes. In the rat MCAo model, we then de- treated with 1 mM sodium pyruvate or 0.2 mM heptanoic acid in XF termined the effects of triheptanoin in comparison to long-chain tri- Assay Medium Modified DMEM (Seahorse Bioscience) containing 2 mM glycerides when administered 1 h after cerebral blood flow reduction. glucose, 0.8 mM lactic acid and 0.4 mM glutamine for 2 h. Stroke volume was evaluated at 3 days and neurological deficit at 1 and Mitochondrial functions were assessed using Seahorse XFe96 Analyzer 3 days. and XF Cell Mito Stress Kit (Seahorse Bioscience) based on oxygen consumption rates (OCR) at 37 °C as previously described [22]. Briefly, 2. Materials and methods OCR was measured prior to and throughout sequential addition of 1.5 μM oligomycin, 1 μM FCCP and 1 μM each of antimycin A/rotenone to All chemicals and reagents were obtained from Sigma Aldrich (St. stimulate different states of mitochondrial respiration. Various para- Louis, MO, USA) unless stated otherwise. The final concentrations of meters of mitochondrial function including basal respiration, maximal chemical and reagents are listed. respiration, proton leak, ATP turnover, coupling efficiency and re- spiratory control ratio uncoupled were quantified as previously de- 2.1. Animals scribed [23] and expressed as percentages relative to 1 mM sodium pyruvate. Male Wistar rats (13–18 weeks old; 270–370 g; ASU Breeding Facility, University of Newcastle, NSW, Australia) were co-housed (a 2.5. Anesthesia and monitoring maximum of four rats per cage) under a 12-hour light-dark cycle with free access to food and water and all experiments involving animals Rats were anesthetized using 5% isoflurane and maintained with fl were conducted during the light cycle. All experiments involving mice 1%–2% iso urane in N2/O2 (70/30; vol/vol %) through spontaneous (Animal Resources Centre, WA, Australia) were approved by the animal breathing. Core temperature was maintained at 37 °C using a thermos- ethics committee of the University of Queensland (SBMS/128/14) and coupled rectal probe and warming mat while blood oxygen saturation experiments involving rats were approved by the Animal Care and levels (SpO2) was monitored throughout surgery. Animals were injected Ethics Committee of the University of Newcastle (A-2014-431). All with analgesic 0.05% bupivacaine (2 mg/kg; subcutaneous; Pfizer, experiments were performed in accordance with Australian Code of Sydney, Australia) prior to incision and were given bupivacaine Practice for the Care and Use of Animal for Scientific Purposes. Every (0.2 mg/kg) and 3 ml of 0.9% saline (both subcutaneous) post-surgery. effort was made to minimize animal suffering and all work was con- All animals were housed individually with free access to mushed ducted according to the ARRIVE guidelines [21]. standard rodent chow and water.



171 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ²

2.6. Surgical procedures oven temperature was set at 80 °C with pressure of 14.82 psi and flow velocity of 45 cm/s. The temperature was then increased linearly to In the first study (study I), the rats only underwent the jugular vein 220 °C and held for 1 min. The derivatized samples were injected into cannulation procedure described below to investigate whether con- the GC–MS in split mode (1:20) and the selected-ion monitoring mode tinuous infusion of 20% LCT or triheptanoin emulsion (i.v.) for 72 h is was used. The m/z ratios for each metabolite are as follow: heptanoic safe and tolerable. Blood was collected from the left ventricle 72 h post acid (m/z = 187), BHB (m/z = 159) and d6 succinic acid (m/z = 293). infusion using heparinized syringe and centrifuged at 2000 g for 10 min The total levels of metabolites in each sample were calculated using the at 4 °C. Plasma was aliquoted and stored at -80 °C until analyzed. In the respective standard curves. second study (study II), the rats underwent all the surgical procedures described below to determine whether triheptanoin is neuroprotective 2.8. Histological assessment of infarct volume in the MCAo model. Rats were euthanized under deep anesthesia with isoflurane 72 h 2.6.1. Femoral artery cannulation after MCAo and were subjected to transcardial perfusion with 200 ml The right saphenous branch of the femoral artery was cannulated each of cold 0.9% saline and 4% paraformaldehyde solution before using a laboratory-manufactured catheter which was connected to a being decapitated. The brains were post-fixed in the skull in neutral- blood pressure transducer (CODAN, Forstinning, Germany). Mean ar- buffered formalin at 4 °C until being dehydrated through ascending terial pressure, heart rate and respiratory rate were calculated from the concentrations of ethanol, embedded in paraffin and sectioned. Coronal arterial pressure tracing and all parameters were monitored throughout sections (5 μm) were stained using hematoxylin and eosin and imaged surgery. Blood samples (0.1 ml) were taken from the arterial line for at 20x objective using Aperio slide scanner (Aperio Technologies, CA, blood gas analysis using i-STAT CG8+ cartridges (Abbott, IL, USA) USA). The infarct area between 3.7 mm to −4.3 mm from bregma prior to occlusion. (1 mm interval) was determined by an experimenter blinded to treat- ment group using ImageScope (Leica Biosystems, CA, USA) with a 10x 2.6.2. Laser Doppler Flowmetry objective and infarct volume was calculated and corrected for oedema The right parietal bone was thinned at 2 mm caudal and 5 mm lat- as previously described [25]. A second blinded experimenter was em- eral to bregma and a hollow PEEK® screw (2 mm diameter × 4 mm ployed to determine the infarct area of random samples to ensure that height; Solid Spot, CA, USA) was inserted. A laser Doppler flowmetry the assessment was reproducible. The infarct areas determined by the probe (Moore Instruments, Sussex, UK), which was inserted into the two experimenters were less than 10% different. screw for the measurement of cerebral blood flow (CBF) of the middle cerebral artery (MCA) region, was held in place using a caulking ma- 2.9. Neurological deficit testing terial (Silagum, DMG, Hamburg, Germany). Animals with less than 50% decrease in CBF in the penumbra region upon occlusion and those with Rats were tested for neurological deficits using forelimb flexion and subarachnoid hemorrhage were excluded from the study. torso twist tests as previously described [26] at 24 h and 72 h post- MCAo by a blinded experimenter. A score of 0 was assigned when no 2.6.3. Jugular vein cannulation flexion or twist was observed while scores of 1 and 2 indicated mild and The right jugular vein was cannulated using a laboratory-made ca- severe flexion or twist respectively. Neurological deficit was reported as theter which was connected to a syringe pump (Kent Scientific, CT, the sum of both forelimb flexion and torso twist tests and higher score USA) to allow continuous infusion of 20% LCT or 20% triheptanoin indicates more severe deficit. emulsion (both from B. Braun Melsungen, Melsungen, Germany) at a constant rate of 1.36 ml/kg/h to provide 30% of total caloric intake 2.10. Statistical analysis from triheptanoin [[13]]. Power analysis was performed based on our previous experience 2.6.4. Middle cerebral artery occlusion with our stroke model and indicated that 10 rats/group were required ff Occlusion of the MCA was performed using a very minor mod- to be able to detect > 25% di erence in stroke volume with α = 0.05, fi i cation of the previously described silicone-tipped intraluminal thread β = 0.2. All graphs show mean ± SEM except for plots depicting in- occlusion technique [24]. In this study, a silicone tip of 4 mm long with dividual and median neurological deficit scores. Differences between 0.38 mm diameter was used and the right MCA was occluded for 2.5 h. two groups were compared using unpaired, two-tailed Student’s t-test Sham rats underwent all surgical procedures except that the occlusion and for experiments with two variables, two-way ANOVA followed by thread was not advanced past the internal carotid artery to prevent Bonferroni’s multiple comparison post hoc test was used except when occlusion of the origin of MCA. Both sham and stroke rats were ran- compared to 0 h (two-way repeated measures ANOVA). A Mann- domly assigned to receive 20% LCT or 20% triheptanoin emulsion (i.v.) Whitney test was used to assess neurological deficit and statistical by a blinded experimenter and the continuous 72 h infusion was in- significance was accepted at p < 0.05. The number of stars in the itiated 1 h post-MCAo. figures indicates the significance in Student’s t-tests or Bonferroni’s multiple comparison tests. All statistical analyses were conducted using 2.7. Plasma metabolite level measurement using GC–MS GraphPad Prism 7.02 (GraphPad Software, La Jolla, CA, USA).

The plasma levels of heptanoate and β-hydryoxybutyrate (BHB) 3. Results from study I were measured using an Agilent Mass Spectrometer (model 5975B) equipped with Agilent Gas Chromatography system (model 3.1. Neuroprotective effects of heptanoate in vitro 6870) and a VF-1 msec capillary column. Samples were extracted, de- rivatized and measured as previously described [13,22] with the fol- Pre-treatment with 200 μM of heptanoate reduced cell death by lowing modifications. Briefly, 40 nmol of D6-succinic acid (98% D; 34 ± 10% after 3 h of OGD in three out of four independent cortical Cambridge Isotope Laboratory, MA, USA) was added to 40 μl of plasma neuronal culture preparations (Mean ± SD; n = 3 independent ex- as internal standards. The samples were derivatized using 105 μl of N- periments). A representative experiment with 28% reduction is shown methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide + 1% tert-bu- in Fig. 1A (p = 0.01; n = 3 wells each). A lower concentration of tyldimetheylchlorosilane and reacted overnight on a shaker at 37 °C. heptanoate pre-treatment (50 μM) was also found to reduce neuronal For the GC–MS, a helium flow of 1 ml/min was used and the starting cell death induced by 9 h of OGD by 18% and 22% (p = 0.01; 18%



172 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ²

Fig. 1. Neuroprotective effects of heptanoic acid in vitro and effects on mitochondrial functions in cultured astrocytes. Representative examples of the protection of heptanoic acid pre-treatment against neuronal cell death (expressed as % lactate dehydrogenase (LDH) release) induced by (A) 3 h and (B) 9 h of oxygen glucose deprivation (OGD; n = 3 wells each) as well as (C) NMDA using primary cortical neuronal cultures (n = 4–5 wells each) are shown. Cultured astrocytes grown in 2 mM glucose were incubated with 0.2 mM heptanoic acid or 1 mM sodium pyruvate for two hours. Using an extracellular flux analyzer, various mitochondrial function parameters, including (D) basal respiration, maximal respiration, proton leak, ATP turnover, (E) coupling efficiency and respiratory control ratio (RCR) uncoupled, were determined based on oxygen consumption rates. These parameters were then calculated and expressed as percentages relative to 1 mM sodium pyruvate. (A, B, D, E) All unpaired, two-tailed Student’s t-tests, stars indicate t-test significances. (C) Two-way ANOVA followed by Bonferroni’s multiple comparison test, stars indicate Bonferroni’s post hoc test significances, n = 5–6 independent culture preparations. reduction shown in Fig. 1B; p = 0.02; all unpaired, two-tailed Student’s or between 20% LCT (0 h: 321 ± 35 and 72 h: 313 ± 29 g) and tri- t-tests; n = 3 wells each) in two independent culture preparations. heptanoin treatment (0 h: 323 ± 48 and 72 h: 303 ± 55 g; two-way NMDA (25 μM) induced cell death in both vehicle and heptanoate repeated measures ANOVA; both n = 4 rats). We did not observe sig- groups (p < 0.0001; n = 4–5 wells each), which was reduced by nificant changes in blood glucose levels within 20% LCT and trihepta- heptanoate pre-treatment (50 μM) by 20 ± 7% in all four independent noin groups neither between the two groups at 0, 1, 3 and 72 h post culture preparations tested (28% reduction shown in Fig. 1C; infusion (two-way repeated measures ANOVA; n = 3 rats/group; p < 0.0001; two-way ANOVAs; n = 4–5 wells each). Table 1). The levels of blood electrolytes (Fig. 2A) and hematocrit (Fig. 2B) were also not altered by continuous infusion and did not differ 3.2. Effects of heptanoate on mitochondrial function in cultured astrocytes between the two emulsions (all two-way repeated measures ANOVAs; n = 4 rats/group). Body weight, blood glucose and electrolytes were Changes in mitochondrial function were assessed following two- not altered 72 h post infusion, suggesting that continuous infusion of hour incubation with 0.2 mM of heptanoate in cultured astrocytes using both 20% LCT and triheptanoin emulsions for 72 h via the jugular vein an extracellular flux analyzer to measure OCR. The average basal re- is safe and tolerable in rats. spiration with 0.2 mM heptanoate was 7.1 ± 3.4 pmol oxygen/min/μg protein (n = 6 independent cultures). Heptanoate increased basal re- spiration by 27% (p = 0.002; Fig. 1D), proton leak by 21% Table 1 (p = 0.0002; Fig. 1D) and ATP turnover by 55% (p = 0.0003; Fig. 1D) Blood glucose levels at 0, 1, 3 and 72 h after 20% long-chain triglycerides or relative to 1 mM of sodium pyruvate. Supplementation with heptanoate triheptanoin emulsion infusion in normal rats. did not alter maximal respiration (p = 0.13; Fig. 1D), coupling effi- Time post infusion (h) Blood glucose - mM (Mean ± SD) p-Values ciency (p = 0.08; Fig. 1E) or respiratory control ratio uncoupled 20% LCT 20% TRIH (p = 0.84; Fig. 1E; all unpaired, two-tailed Student’s t-tests; n = 6 in- dependent culture preparations). 0 8.4 ± 0.5 8.0 ± 0.9 > 0.99 1 7.4 ± 0.7 7.2 ± 1.0 > 0.99 3.3. Safety and tolerability of continuous triheptanoin emulsion infusion 3 6.6 ± 0.6 6.6 ± 0.4 > 0.99 72 8.6 ± 0.6 9.4 ± 1.0 > 0.99

Safety and tolerability of continuous triheptanoin emulsion infusion Two-way repeated measures ANOVA was used. p-Values of Bonferroni’s mul- was first investigated in a small group of naïve rats in study I. Body tiple comparison tests are shown; n = 3 rats/group. LCT, long-chain triglycer- weights were not significantly different before and after 72 h of infusion ides; TRIH, triheptanoin.



173 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ²

Fig. 2. Blood electrolytes and hematocrit in normal rats after 72 h of 20% long-chain tri- glycerides or triheptanoin emulsion infusion. (A) Blood electrolytes, including sodium (Na+), potassium (K+) and ionized calcium (iCa2+) as well as (B) hematocrit (Hct), were measured before and after 72 h of continuous infusion of 20% long-chain triglyceride (LCT) or triheptanoin emulsion (i.v.) in rats. (A, B) Both two-way repeated measures ANOVAs, n = 4 rats/group.

3.4. Plasma metabolite levels Table 2 Body weights, blood electrolytes, hematocrit and blood glucose levels at 0 h and In study I, plasma heptanoate levels were increased by 40-fold after 72 h post-MCAo in rats given 20% long-chain triglycerides or triheptanoin triheptanoin emulsion (LCT: 6.6 ± 0.9 and triheptanoin: emulsion infusion. 264.0 ± 90.1 μM; p = 0.001; n = 4 rats/group) infusion, while the Time post-MCAo (Mean ± SD) p-Values plasma levels of BHB were similar in both groups (LCT: 1.57 ± 0.43 and triheptanoin: 1.44 ± 0.19 mM; p = 0.64; both unpaired, two- 0h 72 h tailed Student’s t-tests; n = 3 rats/group). Body weight Sham LCT 312.0 ± 22.6 283.4 ± 26.1 < 0.0001 3.5. Physiological parameters Sham TRIH 323.0 ± 23.7 286.6 ± 27.8 < 0.0001 Stroke LCT 323.5 ± 22.8 289.8 ± 19.3 < 0.0001 Stroke TRIH 324.4 ± 15.9 285.0 ± 15.2 < 0.0001 In study II, three rats were excluded by a blinded experimenter for Na (mM) the following reasons: less than 50% reduction in CBF upon occlusion Sham LCT 141.4 ± 2.6 136.2 ± 1.2 0.04 (one LCT; one triheptanoin) and subarachnoid hemorrhage (one tri- Sham TRIH 143.0 ± 1.8 135.6 ± 1.4 0.002 heptanoin-treated rat). Physiological parameters including body tem- Stroke LCT 142.5 ± 3.4 135.2 ± 2.4 < 0.0001 Stroke TRIH 142.4 ± 2.1 136.1 ± 1.8 0.0002 perature, arterial partial pressure of oxygen and carbon dioxide as well K (mM) fi ff as arterial blood pH were not signi cantly di erent between groups Sham LCT 3.7 ± 0.5 5.0 ± 0.3 0.03 immediately before the occlusion at 0 h (Supplementary Table 1; all Sham TRIH 3.4 ± 0.3 5.6 ± 0.9 < 0.0001 two-way ANOVAs; sham: n = 5 rats/group; stroke: n = 10–11 rats/ Stroke LCT 3.6 ± 0.4 5.0 ± 0.7 0.0002 Stroke TRIH 3.7 ± 0.5 5.1 ± 0.7 0.0003 group). Other physiological parameters including SpO2%, heart rate, fi iCa (mM) respiratory rate and mean arterial pressure were also not signi cantly Sham LCT 1.39 ± 0.08 1.18 ± 0.10 0.01 different between or within groups at 0 h or during reperfusion at 2.5 h Sham TRIH 1.31 ± 0.09 1.15 ± 0.05 0.08 after MCAo (Supplementary Table 2; all two-way repeated measures Stroke LCT 1.32 ± 0.15 1.16 ± 0.15 0.005 ANOVAs; sham: n = 5 rats/group; stroke: n = 9–11 rats/group). Stroke TRIH 1.36 ± 0.10 1.15 ± 0.10 0.0006 Hematocrit Body weights, blood sodium, potassium and ionized calcium levels Sham LCT 0.40 ± 0.02 0.48 ± 0.11 0.18 ff were not di erent between groups at 0 h or 72 h post occlusion Sham TRIH 0.36 ± 0.04 0.40 ± 0.03 0.95 (Table 2; all two-way repeated measures ANOVAs; sham: n = 5 rats/ Stroke LCT 0.37 ± 0.06 0.43 ± 0.08 0.16 group; stroke: n = 10–11 rats/group). Apart from blood potassium le- Stroke TRIH 0.38 ± 0.03 0.41 ± 0.03 > 0.99 vels, which increased in all groups at 72 h post-MCAo, body weights, Glucose (mM) fi Sham LCT 11.0 ± 2.8 11.9 ± 0.5 > 0.99 blood sodium and ionized calcium levels were signi cantly decreased at Sham TRIH 9.6 ± 1.2 12.9 ± 3.1 0.13 72 h post occlusion in most groups (Table 2; all two-way repeated Stroke LCT 10.5 ± 3.3 11.3 ± 1.3 > 0.99 measures ANOVAs; sham: n = 5 rats/group; stroke: n = 10–11 rats/ Stroke TRIH 10.6 ± 1.6 11.0 ± 1.7 > 0.99 group). Blood glucose levels and hematocrit were unaltered between and within groups (Table 2; all two-way repeated measures ANOVAs; Two-way repeated measures ANOVAs were used and p-values of Bonferroni’s multiple comparison tests (versus 0 h) are shown. Sham: n = 5 rats/group; sham: n = 5 rats/group; stroke: n = 10–11 rats/group) at both 0 h and stroke: n = 10–11 rats/group. LCT, long-chain triglycerides; MCAo, middle 72 h. cerebral artery occlusion; TRIH, triheptanoin.

3.6. Triheptanoin did not alter infarct volume and neurological deficit occlusion (median scores; LCT: 0 and triheptanoin: 1; p = 0.41; both Mann Whitney tests; p = 10–11 rats/group; Fig. 3B). Interestingly, Infarct volumes were assessed using hematoxylin and eosin staining small intracerebral hemorrhages were observed in the majority of at 72 h post-MCAo. High variability in stroke volume was observed in stroke rats from both emulsion groups (LCT: 8 out of 11 and trihepta- fi ff both groups and there was no signi cant di erence in infarct volume noin: 8 out of 10 rats). All were located in the ipsilateral hemisphere between LCT and triheptanoin groups (mean ± SD; LCT: 46.6 ± 32.9 within or in close proximity to the infarct. and triheptanoin: 60.6 ± 32.5 mm3; p = 0.37; unpaired, two-tailed Student’s t-test; n = 10–11 rats/group; Fig. 3A). The sum of two neu- rological tests in stroke rats, forelimb flexion and torso twist, were not 4. Discussion statistically different between LCT and triheptanoin groups at both 24 h (median scores; LCT: 1 and triheptanoin: 0; p = 0.37) and 72 h post In previous studies triheptanoin was anticonvulsant and



174 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ²

Fig. 3. Triheptanoin treatment did not alter stroke volume and neurological deficits when administered 1 h after stroke onset in rats. (A) Brain sections were stained using hematoxylin and eosin and infarct volume at 72 h post-MCAo was assessed from bregma 3.7 to −4.3 mm in rats given 20% long-chain triglycerides (LCT) or triheptanoin (TRIH) emulsion infusion (i.v.). (B) Neurological deficits were assessed at 24 h and 72 h post occlusion using forelimb flexion and torso twist tests and the sum of both tests are shown (n = 10–11 rats/group). (A) Unpaired, two-tailed Student’s t-test, n = 10–11 rats/group. (B) Mann Whitney tests, n = 10–11 rats/ group. neuroprotective in several animal models of neurological disorders that ATP synthesis was not compromised despite mild proton leak in [18–20]. Pre-treatment with triheptanoin appeared promising in a cultured astroyctes. mouse model of MCAo [18]. Here we found that pre-treatment with Continuous intravenous infusion with triheptanoin emulsion ap- heptanoate protected against cell death induced by oxygen glucose peared to be safe and tolerable in normal rats in this study. The changes deprivation and NMDA in cultured neurons, indicating heptanoate is in body weights and blood electrolytes at 72 h post-MCAo were in- neuroprotective in vitro. We also showed that basal respiration and ATP dependent of emulsion type and stroke. These changes could be at- turnover were increased in cultured astrocytes treated with 0.2 mM tributed to post surgery trauma, which have been observed in patients heptanoate relative to 1 mM sodium pyruvate. These findings suggest following major surgery [32]. Intriguingly, continuous infusion with that heptanoate is oxidized directly by astrocytes, consistent with a LCT and triheptanoin emulsion led to small intracerebral hemorrhage in previous study which demonstrated that other medium-chain fatty the majority of the rats in the MCAo groups. This has not previously acids, specifically octanoate and decanoate were utilized by cultured been seen in other studies using this method [24,33,34]. We do not astrocytes [22]. However, we did not observe any neuroprotective ef- understand the exact cause of hemorrhage. Since both groups were fects of triheptanoin when administered (i.v.) commencing 1 h after equally affected, we speculate that glycerol, which has been shown to stroke in rats in this study. be greatly increased (1.7 mM) following triheptanoin infusion [13], The protective effects of heptanoate against oxygen glucose depri- could contribute to the hemorrhage. High concentration of glycerol vation- and NMDA-induced cell death in cultured neurons indicate that given systemically induce intraventricular and subarachnoid hemor- heptanoate can be neuroprotective. The observed changes in mi- rhage in neonatal rabbit models [35,36]. These findings suggest that tochondrial function in cultured astrocytes provide direct evidence that care should be taken if considering administering lipid-based parenteral it can be oxidized by astrocytes. This finding is consistent with another nutrition to stroke patients. As intra-duodenal infusion avoids high study which also showed that heptanoate is oxidized in mouse brain glycerol levels [14,15] and is similar to oral administration used suc- mostly by astrocytes following i.v. infusion by tracing the fate of 13C- cessfully by Schwarzkopf, [18], this requires further investigation. labeled heptanoate [15]. Although it is unclear to which extent hep- Triheptanoin emulsion increased plasma heptanoate levels and al- tanoate and C5 ketones reached penumbral tissue, heptanoate logP is though brain heptanoate and C5 ketone levels were not measured in this between 2.1 and 2.4 indicating that it will easily diffuse into the brain study, it is known that heptanoate enters the brain directly or as 13 and also the penumbra. When radioactively labelled octanoate, a C5ketone bodies based on a previous study which infused C-labeled medium chain fat with similar properties to heptanoate (logP = 3), was heptanoate (i.v.) in conscious mice [15]. In our study, triheptanoin injected in rats (i.v.), labelled octanoate and metabolites were found in treatment for 72 h did not alter stroke volume and neurological deficit the brain within 0.5–2 min [27]. Taken together, our study is consistent when administered 1 h after stroke onset relative to LCT emulsion. This with others indicating that heptanoate can be utilized as fuel in the was unexpected given the promising neuroprotective effects found in brain. It is still unclear to what extent neurons can metabolize hep- vitro in this study and following pre-treatment in vivo [18]. There are tanoate. several limitations to our study. The variability in stroke volume was In addition, we found that heptanoate increased mitochondrial higher than in our previous studies, which were used to determine proton leak in cultured astrocytes, which is similar to effects of de- sample size. On the other hand, there was no protective trend and canoate described earlier [22]. The mechanisms underlying increased therefore we concluded that there was no protective effect. The high proton leak are unknown, although studies have shown that fatty acids variability does not appear to be explained by our > 50% threshold for are able to activate uncoupling proteins responsible for mitochondrial cerebral blood flow reduction with a LDF probe positioned over the proton leakage [28,29]. Further studies are necessary to determine to penumbral region (not infarct core). Additional earlier and later time which extent heptanoate alters the levels of uncoupling proteins. Mild points may have shed more light on the effects of triheptanoin. In ad- uncoupling has been shown to be beneficial by reducing the production dition to the high variability, it is also possible that LCT is neuropro- of free radicals in the mitochondria [30,31]. Furthermore, the finding tective and may have masked the effects of triheptanoin. Thus, future in that ATP turnover was enhanced after heptanoate treatment indicates vitro experiments should include long-chain fatty acids. We considered



175 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ² including another group of animals receiving other extra fuel or no online version, at doi:https://doi.org/10.1016/j.neulet.2018.07.045. extra fuel (i.v.) after stroke onset in the initial experimental design. However, animals do not feed well after experimental stroke and tend References to lose signinficant body weight. Therefore we felt that the latter was unethical. [1] A.S. Go, D. Mozaffarian, V.L. Roger, E.J. Benjamin, J.D. Berry, W.B. Borden, Although triheptanoin post-treatment did not alter infarct volume D.M. Bravata, S. Dai, E.S. Ford, C.S. Fox, S. Franco, H.J. Fullerton, C. Gillespie, S.M. Hailpern, J.A. Heit, V.J. Howard, M.D. Huffman, B.M. Kissela, S.J. Kittner, and neurological functions in this study, this does not imply that there D.T. Lackland, J.H. Lichtman, L.D. Lisabeth, D. Magid, G.M. Marcus, A. Marelli, is no benefit of triheptanoin or other metabolic approaches, several of D.B. Matchar, D.K. McGuire, E.R. Mohler, C.S. Moy, M.E. Mussolino, G. Nichol, which appear promising. The lack of effect of triheptanoin in our model N.P. Paynter, P.J. Schreiner, P.D. Sorlie, J. Stein, T.N. Turan, S.S. Virani, N.D. Wong, D. Woo, M.B. Turner, Heart disease and stroke statistics 2013 update: may be specific to the rat, as so far all neuroprotective effects were — a report from the American Heart Association, Circulation 127 (2013) e6–e245. described in mice. In addition, neuroprotective effects have been ob- [2] F.A. Welsh, V.R. Marcy, R.E. Sims, NADH fluorescence and regional energy meta- served in animal models of ischemic stroke following ketogenic state bolites during focal ischemia and reperfusion of rat brain, J. Cereb. Blood Flow induction [37,38] and supplementation of TCA cycle intermediates or Metab. 11 (1991) 459–465. [3] J. Folbergrova, H. Memezawa, M.L. Smith, B.K. Siesjo, Focal and perifocal changes their derivatives including pyruvate [39–42], α-ketoglutarate [43], in tissue energy state during middle cerebral artery occlusion in normo- and hy- fumarate [44,45] and oxaloacetate [46–49]. Therefore, the use of me- perglycemic rats, J. Cereb. Blood Flow Metab. 12 (1992) 25–33. tabolic treatments in ischemic stroke remains a potential avenue, al- [4] J. Folbergrova, Q. Zhao, K. Katsura, B.K. Siesjo, N-tert-butyl-alpha-phenylnitrone improves recovery of brain energy state in rats following transient focal ischemia, though further research is required. Proc. Natl. Acad. Sci. U. S. A. 92 (1995) 5057–5061. [5] A. Håberg, H. Qu, O. Saether, G. Unsgard, O. Haraldseth, U. Sonnewald, Differences 5. Conclusions in neurotransmitter synthesis and intermediary metabolism between glutamatergic and GABAergic neurons during 4 hours of middle cerebral artery occlusion in the rat: the role of astrocytes in neuronal survival, J. Cereb. Blood Flow Metab. 21 Here we found that heptanoate pre-treatment was protective against (2001) 1451–1463. oxygen glucose deprivation- and NMDA-induced cytotoxicity in cul- [6] A. Håberg, H. Qu, O. Haraldseth, G. Unsgard, U. Sonnewald, In vivo injection of [1- 13C]glucose and [1,2-13C]acetate combined with ex vivo 13C nuclear magnetic tured neurons. Heptanoate increased basal respiration and ATP turn- resonance spectroscopy: a novel approach to the study of middle cerebral artery over in cultured astrocytes, indicating that heptanoate can be utilized occlusion in the rat, J. Cereb. Blood Flow Metab. 18 (1998) 1223–1232. by astrocytes. Despite the promising neuroprotective effects found in [7] J. Castillo, A. Davalos, J. Naveiro, M. Noya, Neuroexcitatory amino acids and their relation to infarct size and neurological deficit in ischemic stroke, Stroke 27 (1996) vitro, triheptanoin treatment for 72 h did not alter stroke volume or 1060–1065. neurological deficit when administered 1 h after stroke onset relative to [8] O. Cataltepe, J. Towfighi, R.C. Vannucci, Cerebrospinal fluid concentrations of LCT emulsion. There were also noticeably frequent small haemorrhages glutamate and GABA during perinatal cerebral hypoxia-ischemia and seizures, seen in both triheptanoin and control emulsion treated animals. This Brain Res. 709 (1996) 326–330. [9] R.J. Haslam, H.A. Krebs, The metabolism of glutamate in homogenates and slices of result is in contrast to a previous report in a mouse model of experi- brain cortex, Biochem. J. 88 (1963) 566–578. mental stroke using triheptanoin pretreatment. However, due to the [10] H.S. Waagepetersen, U. Sonnewald, O.M. Larsson, A. Schousboe, Compartmentation of TCA cycle metabolism in cultured neocortical neurons re- high variability in this study, further investigation is required before 13 ff vealed by C MR spectroscopy, Neurochem. Int. 36 (2000) 349–358. any conclusion can be drawn about the neuroprotective e ects of tri- [11] Y.E. Bogaert, R.E. Rosenthal, G. Fiskum, Postischemic inhibition of cerebral cortex heptanoin post-MCAo. Also, other metabolic approaches remain pro- pyruvate dehydrogenase, Free Radic. Biol. Med. 16 (1994) 811–820. ff mising as potential treatments for ischemic stroke. Our results suggest [12] T. Fukuchi, Y. Katayama, T. Kamiya, A. McKee, F. Kashiwagi, A. Terashi, The e ect of duration of cerebral ischemia on brain pyruvate dehydrogenase activity, energy that any future studies using lipid emulsions should seek to also de- metabolites, and blood flow during reperfusion in gerbil brain, Brain Res. 792 termine potential effects of such emulsions on haemorrhage rates, as (1998) 59–65. well as effects on infarct volumes and long term behavioural outcomes. [13] L. Gu, G.F. Zhang, R.S. Kombu, F. Allen, G. Kutz, W.U. Brewer, C.R. Roe, H. Brunengraber, Parenteral and enteral metabolism of anaplerotic triheptanoin in normal rats. II. Effects on lipolysis, glucose production, and liver acyl-CoA profile, Funding Am. J. Physiol. Endocrinol. Metab. 298 (2010) E362–371. [14] R.P. Kinman, T. Kasumov, K.A. Jobbins, K.R. Thomas, J.E. Adams, L.N. Brunengraber, G. Kutz, W.U. Brewer, C.R. Roe, H. Brunengraber, Parenteral We are grateful for funding from NHMRC grants (project grant and enteral metabolism of anaplerotic triheptanoin in normal rats, Am. J. Physiol. 1044007 to KB), Fondecyt Initiation into Research (Grant 11130232 to Endocrinol. Metab. 291 (2006) E860–866. CC), UQ scholarships (KT) and UoN scholarships (RH and KW). NJS was [15] I. Marin-Valencia, L.B. Good, Q. Ma, C.R. Malloy, J.M. Pascual, Heptanoate as a the recipient of a co-funded NHMRC/NHF Career Development/Future neural fuel: energetic and neurotransmitter precursors in normal and glucose transporter I-deficient (G1D) brain, J. Cereb. Blood Flow Metab. 33 (2013) Leader Fellowship APP1110629/100827. 175–182. [16] C.R. Roe, H. Brunengraber, Anaplerotic treatment of long-chain fat oxidation dis- Declaration of conflicting interests orders with triheptanoin: review of 15 years experience, Mol. Genet. Metab. 116 (2015) 260–268. [17] J. Vockley, D. Marsden, E. McCracken, S. DeWard, A. Barone, K. Hsu, E. Kakkis, KB has filed for a US patent on triheptanoin as a treatment for Long-term major clinical outcomes in patients with long chain fatty acid oxidation seizures and has obtained a patent for the treatment of Amyotrophic disorders before and after transition to triheptanoin treatment—a retrospective chart review, Mol. Genet. Metab. 116 (2015) 53–60. Lateral Sclerosis (both licensed to Ultragenyx Pharmaceutical Inc.). All [18] T.M. Schwarzkopf, K. Koch, J. Klein, Reduced severity of ischemic stroke and im- other authors declare no conflict of interest. provement of mitochondrial function after dietary treatment with the anaplerotic substance triheptanoin, Neuroscience 300 (2015) 201–209. [19] J.S. Francis, V. Markov, P. Leone, Dietary triheptanoin rescues oligodendrocyte loss, Acknowledgements dysmyelination and motor function in the nur7 mouse model of Canavan disease, J. Inherit. Metab. Dis. 37 (2014) 369–381. We thank Tanya McDonald for optimizing the GC–MS protocol and [20] T.W. Tefera, Y. Wong, M.E. Barkl-Luke, S.T. Ngo, N.K. Thomas, T.S. McDonald, K. Borges, Triheptanoin protects motor neurons and delays the onset of motor Dr Chris Dayas and members of his laboratory for their assistance on the symptoms in a mouse model of amyotrophic lateral sclerosis, PLoS One 11 (2016) setup of continuous jugular vein infusion. We are grateful to Ultragenyx e0161816. Pharmaceutical Inc. and B. Braun Melsungen for providing triheptanoin [21] C. Kilkenny, W.J. Browne, I.C. Cuthill, M. Emerson, D.G. Altman, Improving oil and LCT and triheptanoin emulsion, respectively for research pur- bioscience research reporting: the ARRIVE guidelines for reporting animal research, PLoS Biol. 8 (2010) e1000412. poses. [22] K.N. Tan, C. Carrasco-Pozo, T.S. McDonald, M. Puchowicz, K. Borges, Tridecanoin is anticonvulsant, antioxidant, and improves mitochondrial function, J. Cereb. Blood Appendix A. Supplementary data Flow Metab. 37 (2017) 2035–2048. [23] C. Carrasco-Pozo, K.N. Tan, K. Borges, Sulforaphane is anticonvulsant and improves mitochondrial function, J. Neurochem. 135 (2015) 932–942. Supplementary material related to this article can be found, in the [24] N.J. Spratt, J. Fernandez, M. Chen, S. Rewell, S. Cox, L. van Raay, L. Hogan,



176 K.N. Tan et al. 1HXURVFLHQFH/HWWHUV  ²

D.W. Howells, Modification of the method of thread manufacture improves stroke [37] M.A. Puchowicz, J.L. Zechel, J. Valerio, D.S. Emancipator, K. Xu, S. Pundik, induction rate and reduces mortality after thread-occlusion of the middle cerebral J.C. LaManna, W.D. Lust, Neuroprotection in diet-induced ketotic rat brain after artery in young or aged rats, J. Neurosci. Methods 155 (2006) 285–290. focal ischemia, J. Cereb. Blood Flow Metab. 28 (2008) 1907–1916. [25] D.D. McLeod, M.W. Parsons, C.R. Levi, S. Beautement, D. Buxton, B. Roworth, [38] C.L. Gibson, A.N. Murphy, S.P. Murphy, Stroke outcome in the ketogenic state—a N.J. Spratt, Establishing a rodent stroke perfusion computed tomography model, systematic review of the animal data, J. Neurochem. 123 (Suppl 2) (2012) 52–57. Int. J. Stroke 6 (2011) 284–289. [39] J.Y. Lee, Y.H. Kim, J.Y. Koh, Protection by pyruvate against transient forebrain [26] D. Petullo, K. Masonic, C. Lincoln, L. Wibberley, M. Teliska, D.L. Yao, Model de- ischemia in rats, J. Neurosci. 21 (2001) Rc171. velopment and behavioral assessment of focal cerebral ischemia in rats, Life Sci. 64 [40] J.B. Kim, Y.M. Yu, S.W. Kim, J.K. Lee, Anti-inflammatory mechanism is involved in (1999) 1099–1108. ethyl pyruvate-mediated efficacious neuroprotection in the postischemic brain, [27] Y. Kuge, K. Yajima, H. Kawashima, H. Yamazaki, N. Hashimoto, Y. Miyake, Brain Brain Res. 1060 (2005) 188–192. uptake and metabolism of [1-11C] octanoate in rats: pharmacokinetic basis for its [41] J.S. Yi, T.Y. Kim, D. Kyu Kim, J.Y. Koh, Systemic pyruvate administration markedly application as a radiopharmaceutical for studying brain fatty acid metabolism, Ann. reduces infarcts and motor deficits in rat models of transient and permanent focal Nucl. Med. 9 (1995) 137–142. cerebral ischemia, Neurobiol. Dis. 26 (2007) 94–104. [28] E. Winkler, M. Klingenberg, Effect of fatty acids on H+ transport activity of the [42] M. Boyko, A. Zlotnik, B.F. Gruenbaum, S.E. Gruenbaum, S. Ohayon, R. Kuts, reconstituted uncoupling protein, J. Biol. Chem. 269 (1994) 2508–2515. I. Melamed, A. Regev, Y. Shapira, V.I. Teichberg, Pyruvate’s blood glutamate [29] K.D. Garlid, D.E. Orosz, M. Modriansky, S. Vassanelli, P. Jezek, On the mechanism scavenging activity contributes to the spectrum of its neuroprotective mechanisms of fatty acid-induced proton transport by mitochondrial uncoupling protein, J. Biol. in a rat model of stroke, Eur. J. Neurosci. 34 (2011) 1432–1441. Chem. 271 (1996) 2615–2620. [43] T.N. Kovalenko, G.A. Ushakova, I. Osadchenko, G.G. Skibo, S.G. Pierzynowski, The [30] D. Arsenijevic, H. Onuma, C. Pecqueur, S. Raimbault, B.S. Manning, B. Miroux, neuroprotective effect of 2-oxoglutarate in the experimental ischemia of hippo- E. Couplan, M.C. Alves-Guerra, M. Goubern, R. Surwit, F. Bouillaud, D. Richard, campus, J. Physiol. Pharmacol. 62 (2011) 239–246. S. Collins, D. Ricquier, Disruption of the uncoupling protein-2 gene in mice reveals a [44] R. Lin, J. Cai, E.W. Kostuk, R. Rosenwasser, L. Iacovitti, Fumarate modulates the role in immunity and reactive oxygen species production, Nat. Genet. 26 (2000) immune/inflammatory response and rescues nerve cells and neurological function 435–439. after stroke in rats, J. Neuroinflammation 13 (2016) 269. [31] P.G. Sullivan, N.A. Rippy, K. Dorenbos, R.C. Concepcion, A.K. Agarwal, J.M. Rho, [45] B.H. Clausen, L. Lundberg, M. Yli-Karjanmaa, N.A. Martin, M. Svensson, The ketogenic diet increases mitochondrial uncoupling protein levels and activity, M.Z. Alfsen, S.B. Flaeng, K. Lyngso, A. Boza-Serrano, H.H. Nielsen, P.B. Hansen, Ann. Neurol. 55 (2004) 576–580. B. Finsen, T. Deierborg, Z. Illes, K.L. Lambertsen, Fumarate decreases edema vo- [32] R.R. Singh, S. Shekhar, M.J. Akhtar, V. Shankar, Serum electrolyte changes in major lume and improves functional outcome after experimental stroke, Exp. Neurol. 295 surgical trauma, Int. J. Res. Med. Sci. 4 (2016) 2893–2896. (2017) 144–154. [33] L.A. Murtha, D.D. McLeod, D. Pepperall, S.K. McCann, D.J. Beard, A.J. Tomkins, [46] M. Marosi, J. Fuzik, D. Nagy, G. Rakos, Z. Kis, L. Vecsei, J. Toldi, A. Ruban- W.M. Holmes, C. McCabe, I.M. Macrae, N.J. Spratt, Intracranial pressure elevation Matuzani, V.I. Teichberg, T. Farkas, Oxaloacetate restores the long-term potentia- after ischemic stroke in rats: cerebral edema is not the only cause, and short- tion impaired in rat hippocampus CA1 region by 2-vessel occlusion, Eur. J. duration mild hypothermia is a highly effective preventive therapy, J. Cereb. Blood Pharmacol. 604 (2009) 51–57. Flow Metab. 35 (2015) 592–600. [47] D. Nagy, M. Marosi, Z. Kis, T. Farkas, G. Rakos, L. Vecsei, V.I. Teichberg, J. Toldi, [34] D.J. Beard, D.D. McLeod, C.L. Logan, L.A. Murtha, M.S. Imtiaz, D.F. van Helden, Oxaloacetate decreases the infarct size and attenuates the reduction in evoked re- N.J. Spratt, Intracranial pressure elevation reduces flow through collateral vessels sponses after photothrombotic focal ischemia in the rat cortex, Cell. Mol. Neurobiol. and the penetrating arterioles they supply. A possible explanation for 'collateral 29 (2009) 827–835. failure' and infarct expansion after ischemic stroke, J. Cereb. Blood Flow Metab. 35 [48] F. Campos, T. Sobrino, P. Ramos-Cabrer, B. Argibay, J. Agulla, M. Perez-Mato, (2015) 861–872. R. Rodriguez-Gonzalez, D. Brea, J. Castillo, Neuroprotection by glutamate ox- [35] C.O. Chua, H. Chahboune, A. Braun, K. Dummula, C.E. Chua, J. Yu, Z. Ungvari, aloacetate transaminase in ischemic stroke: an experimental study, J. Cereb. Blood A.A. Sherbany, F. Hyder, P. Ballabh, Consequences of intraventricular hemorrhage Flow Metab. 31 (2011) 1378–1386. in a rabbit pup model, Stroke 40 (2009) 3369–3377. [49] L. Knapp, L. Gellert, K. Kocsis, Z. Kis, T. Farkas, L. Vecsei, J. Toldi, Neuroprotective [36] C.M. Traudt, R.J. McPherson, C. Studholme, K.J. Millen, S.E. Juul, Systemic glycerol effect of oxaloacetate in a focal brain ischemic model in the rat, Cell. Mol. decreases neonatal rabbit brain and cerebellar growth independent of in- Neurobiol. 35 (2015) 17–22. traventricular hemorrhage, Pediatr. Res. 75 (2014) 389–394.



177 Appendix B: Copyright Permissions

I warrant that I have obtained, where necessary, permission from the copyright owners to use any third party copyright material reproduced in this thesis, or to use any of my own published work in which he copyright is held by another party.

178

179 20/04/2018 Rightslink® by Copyright Clearance Center

Title: Tissue Plasminogen Activator for preclinical stroke research: Neither “rat” nor “human” dose mimics clinical recanalization in a carotid occlusion model Author: Amelia J. Tomkins, Rebecca J. Hood, Christopher R. Levi, Neil J. Spratt Publication: Scientific Reports Publisher: Springer Nature Date: Nov 2, 2015 Copyright © 2015, Springer Nature

Creative Commons

This is an open access article distributed under the terms of the Creative Commons CC BY license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

You are not required to obtain permission to reuse this article.

Are you the author of this Springer Nature article?

To order reprints of this content, please contact Springer Nature by e­mail at [email protected], and you will be contacted very shortly with a quote.

https://s100.copyright.com/AppDispatchServlet?author=Amelia%20J.%20Tomkins%2C%20Rebecca%20J.%20Hood%2C

180

181

182