Imaging Cerebrospinal Fluid Dynamics in Idiopathic Normal Pressure Hydrocephalus

Geir André Ringstad

Dissertation Submitted for the Degree of Doctor of Philosophy

Division of Radiology and Nuclear Medicine,

Oslo University Hospital - Rikshospitalet

Faculty of Medicine, University of Oslo,

Oslo, Norway

2018 © *HLU$QGUp5LQJVWDG, 2018

Series of dissertations submitted to the Faculty of Medicine, University of Oslo

ISBN 978-82-8377-

Cover: Hanne Baadsgaard Utigard. Print production: Reprosentralen, University of Oslo.

The great tragedy of science, the slaying of a beautiful theory by an ugly fact.

T. H. Huxley (1825-1895)

Contents

Acknowledgements...... 5 Abbreviations ...... 7

Summary in English ...... 9

Sammendrag på norsk ...... 11

List of publications ...... 13 1. Introduction ...... 14

1.1. Historical overview of cerebrospinal fluid ...... 14

1.2. From the bulk flow hypothesis to new concepts ...... 15

1.3 Presentation of NPH ...... 17 1.4 Pathogenesis ...... 18

1.5 Classification ...... 21

1.6 Epidemiology ...... 22

1.7 Symptoms and co-morbidities ...... 22

1.7.1 Gait disturbance ...... 23 1.7.2 Urinary incontinence ...... 23

1.7.3 Dementia ...... 23 1.8 Treatment ...... 24

1.9 Additional tests ...... 25

1.9.1 CSF tap test, infusion test and external lumbar drainage ...... 25

1.9.2 ICP monitoring ...... 26

1.9.3 Imaging ...... 26

1.10 Summary of key points from Introduction ...... 35

2. Aims of the thesis ...... 36

3. Methods ...... 37

3.1 Assessments of aqueductal CSF flow parameters (Paper 1 and 3) ...... 37

3.1.1. Study population and design ...... 37

3.1.2. Clinical management and ICP monitoring ...... 38

3.1.3 MRI ...... 39

3.1.4 Statistics ...... 39 3.2 Assessment of the PC-MRI derived pulse pressure gradient (MRI-dP) (Paper 2) ...... 40

3.2.1 Study population ...... 40 3.2.2 Clinical management and ICP monitoring ...... 40

3.2.3 PC-MRI ...... 40

3.2.4 Statistics ...... 41

3.3 Glymphatic MRI (Paper 4) ...... 41 3.3.1 Study population and design ...... 41

3.3.2 Clinical management ...... 41

3.3.3 MR imaging ...... 41 3.3.4 Statistics ...... 42

4. Results ...... 43

4.1 PC-MRI derived parameters of intracranial pulsatility ...... 43

4.1.1 Aqueductal Stroke Volume (Paper 1) ...... 43 4.1.2 Peak to peak pulse pressure gradient (MRI-dP) (Paper 2) ...... 43

4.2 Characteristics of CSF flow in iNPH ...... 45

4.2.1 Net Aqueductal Flow (Paper 3) ...... 45

4.2.2 Glymphatic MRI (Paper 4) ...... 46 5. Discussion ...... 48

5.1 PC-MRI derived parameters of intracranial pulsatility ...... 48

5.1.1 General methodological considerations (Paper 1 and 2) ...... 48

5.1.2 Aqueductal stroke volume (Paper 1) ...... 50 5.1.3 Peak to peak pulse pressure gradient (MRI-dP) (Paper 2) ...... 54

5.2 Characteristics of CSF flow in iNPH ...... 56 5.2.1 Net retrograde aqueductal flow and supra-aqueductal reflux of gadobutrol (Paper 3 and 4) ...... 56 5.2.2 Glymphatic MRI (Paper 4) ...... 62

5.3 Experiences with publications of negative results ...... 66

6. Conclusions ...... 68

7. Future developments ...... 69

References ...... 71

Acknowledgements

This thesis derives from studies carried out at Oslo University Hospital – Rikshospitalet in cooperation with the Department of Neurosurgery, Department of Radiology and The Intervention Centre.

First, I wish to express my sincere gratitude to all patients who participated in the studies and endured extensive MR imaging for the best of science. Many thanks also go to the MRI technicians at the Intervention Centre and nursing staff at the Department of Neurosurgery, who provided the best possible care for all patients.

I am deeply grateful to my main supervisor, Per Kristian Eide, for his invaluable support and for continuously sharing from his extensive experience and enthusiasm. Our countless, fruitful discussions and his open-mindedness to new perspectives and ideas have been of true inspiration to me. My warmest thanks also go to co-supervisor Kyrre Eeg Emblem, who has masterly tutored and encouraged me from the beginning of this project and always impressed me with his many skills and efficiency. Noam Alperin has also been my co-supervisor, and I thank him for his contributions.

The present work had not been possible without the time granted me from the Department of Radiology, and I would like to especially thank head of department, Paulina Due-Tønnessen, and head of neuroradiology, John K. Hald, for their support and believe in me through these years. For this I also thank all my fellow neuroradiologists at Rikshospitalet, and particularly Bård Nedregaard, Øivind Gjertsen and Ruth Sletteberg, who also have assisted with intrathecal contrast agent injections with impressive skills and flexibility.

I am further thankful to my co-authors Oliver Marcel Geier, Erika Kristina Lindstrøm, Svein-Are Sirirud Vatnehol and Kent-Andre Mardal for their

5 important contributions. I also appreciate all statistical advice given by Are Hugo Pripp.

From my heart I thank my parents, Turid and Gunnar, for their always thoughtful interest in my work, and for teaching values that I will always carry. I sincerely thank my wife Linn for her loving support, which is ever indispensable to me. Finally, I thank my children and sunshine of my life; Gyda, Håkon and Tarjei.

Abbreviations

Aβ Amyloid-β

AD Alzheimer`s disease

AQP4 Aquaporin 4

ASV Aqueductal stroke volume

CCJ Craniocervical junction

CNS Central nervous system

CSF Cerebrospinal fluid

DWI Diffusion weighted imaging

DTI Diffusion tensor imaging

ECS Extracellular space

EI Evan`s index

FLAIR Fluid attenuated inversion recovery

Gd Gadolinium

Gd-DTPA Gadiolinium-diethylenetriaminpentaacetate gMRI Glymphatic magnetic resonance imaging

ICP Intracranial pressure iNPH Idiopathic normal pressure hydrocephalus

ISF Interstitial space fluid sNPH Secondary normal pressure hydrocephalus

MRI Magnetic resonance imaging

MRI-dP MRI derived CSF peak to peak pulse pressure gradient

MWA Mean ICP wave amplitudes

OUH Oslo University Hospital

PACS Picture archiving and communication system

PC-MRI Phase-contrast MRI

REF Reference patients

ROI Region of interest

SAS Subarachnoid space

SU Signal unit

TR Repetition time

TE Echo time

VENC Velocity encoding gradient

Summary in English

This thesis explores short- and long-term characteristics of cerebrospinal fluid (CSF) flow in patients with idiopathic normal pressure hydrocephalus (iNPH), a condition with unknown cause, but typically characterized by gait disorder, urinary incontinence and dementia.

The first aim of the thesis was to investigate whether measurements of intracranial CSF flow based pulsatility by magnetic resonance imaging (MRI) compared with measurements of intracranial pressure (ICP) pulsatility in iNPH. While iNPH patients with pathologically elevated ICP pulsatility have shown to benefit from surgery with high probability (~ 90 %), ICP monitoring in itself demands surgery and carries a risk of serious complications like intracranial hemorrhage and infection. The threshold for undertaking such ICP measurements therefore remains high, and a search for non-surgical alternatives seems mandated.

CSF flow-sensitive MRI studies were performed at two levels. One was at the level of the aqueduct, which connects CSF within the brain ventricles and the exterior surface of the brain and spinal cord. The other measurement was done at level of the upper cervical spinal canal. From these, we calculated the aqueductal stroke volume and the peak to peak pulse pressure gradient, respectively. Neither of the MRI derived pulsatility measurements corresponded with over-night ICP monitoring. Their lack of association with ICP might be attributed to frequent pressure fluctuations observed at long-term ICP recordings as well as the influence on CSF flow by respiration. Long- and short term impacts from such physiological factors may not be well represented in a cardiac gated PC-MRI acquisition lasting over a few minutes.

The second aim of the thesis was to characterize CSF flow in iNPH on a more conceptual basis. Here, net aqueductal flow was shown to be in the upward (retrograde) direction, and thus into the ventricles above level of the aqueduct. The phenomenon was observed particularly frequently in patients with pathological ICP pulsatility. After shunting, from which 94 % of the patients responded, net flow increased in the downward direction. Net CSF flow measurements are sensitive to technical error, however, the evidence of retrograde net flow was further strengthened by observations made at long-term MRI, where it was demonstrated 9 ventricular reflux of a contrast agent that had been administered intrathecally. This direction of CSF flow may contradict previous evidence suggesting CSF production is limited to the ventricular compartment and might also carry implications for how we may better understand ventricular dilatation in iNPH.

Pulsations of intracranial arteries are typically restricted in iNPH. The pivotal role of arterial pulsations for CSF flow was further underlined by the particular distribution of contrast agent in CSF around large artery trunks on the brain surface. In iNPH, contrast agent propagation in CSF was also delayed. Contrast enhancement of CSF always preceded enhancement of adjacent brain parenchyma, particularly in the near vicinity of larger arteries, being most prominent at over-night scans. These observations support the existence of a brain-wide, perivascular pathway for clearance of macromolecular substances, which previously has been described in animals only and been denoted as the “glymphatic” system. Clearance of contrast agent was found delayed in iNPH compared to reference patients and may indicate that compromised glymphatic function is instrumental in iNPH pathogenesis.

Sammendrag på norsk

Avhandlingen utforsker bevegelser av cerebrospinalvæsken (væsken som omgir hjernen og ryggmargen) hos pasienter med voksenvannhode av ukjent årsak (idiopatisk normaltrykkshydrocephalus, iNPH). Dette syndromet kjennetegnes typisk av ustø gange, urinlekkasje og demens.

Noen iNPH-pasienter kan behandles ved å operere inn et rør (shunting) som drenerer cerebrospinalvæske vekk fra hjernens indre hulrom (ventriklene). Det er tidligere vist at man kan vente seg effekt av slik behandling hos cirka 90 % av pasientene med forhøyet pulstrykk intrakranielt (inne i hodeskallen). Ved selve trykkmålingen, som krever kirurgi, er det imidlertid risiko for komplikasjoner i form av hjerneblødning og infeksjon. Derfor ville en alternativ undersøkelsesmetodikk med lavere risiko være ønskelig. Et delmål for avhandlingen var derfor å undersøke om parametere utledet fra hastighetsmålinger av cerebrospinalvæsken med magnetisk resonans (MR)-teknikk (fasekontrast-MR) var sammenlignbare med kirurgiske målinger av pulstrykket i hjernen hos iNPH-pasienter.

Målingene med fasekontrast-MR ble utført på to ulike steder. Den ene målingen ble gjort der cerebrospinalvæsken strømmer gjennom akvedukten, en rørformet åpning som forbinder hjernens indre hulrom (ventriklene) med væskerommene utenfor hjernen og ryggmargen. Den andre målingen ble utført øverst i ryggmargskanalen. Ingen av MR-parameterne viste seg å være sammenlignbare med direkte trykkmålinger som ble utført i løpet av natten. Siden bildeopptaket med fasekontrast-MR varer i kun noen få minutter, kan det hende disse målingene er av for kort varighet til å fange opp velkjente trykkendringer som skjer over lenger tid. Det kan også hende at væskestrømmer som endres når pasienten puster, ikke oppdages med MR, siden bildeopptaket er synkronisert med hjerteslagene og ikke pusten.

Et annet delmål var å karakterisere strømninger av cerebrospinalvæsken i en mer fenomenologisk sammenheng. Med fasekontrast-MR fant vi ut at væsken strømmet gjennom akvedukten netto «baklengs» inn i ventriklene hos iNPH- pasienter. Dette funnet karakteriserte særlig pasienter med økt pulstrykk intrakranielt.

Etter kirurgisk behandling med shunting, som hadde effekt hos 94 % av pasientene, endret netto væskestrøm seg i retning ut av ventriklene. Slike målinger av netto væskestrøm er sårbare for tekniske feilkilder, men funnet ble vurdert som styrket da vi i en annen studie observerte at et MR-kontrastmiddel også fløt med cerebrospinalvæsken fra utsiden av hjernen og inn til ventriklene hos iNPH-pasienter. Dette flytmønsteret strider til dels mot tidligere oppfatninger om at cerebrospinalvæske bare produseres på innsiden av ventriklene, og kan kanskje også si noe om hvorfor ventriklene er utvidet ved iNPH.

Et vanlig fenomen ved iNPH antas å være at pulsårene på overflaten av hjernen ikke kan utvide seg normalt. Den viktige rollen pulserende blodårer har for å drive bevegelsen av cerebrospinalvæsken, ble understreket av at kontrastmiddelet i cerebrospinalvæsken typisk fordelte seg langs etter pulsårer, men langsommere hos iNPH-pasientene. Deretter fant vi at kontrastmiddelet gikk fra væskerommet på overflaten og inn i hjernevevet, særlig i områder nært inntil de største pulsårene. Dette var særlig tydelig på MR-bilder tatt dagen etter at kontrasten ble gitt. Basert på dyrestudier kan vi anta at kontrastmiddelet beveget seg inn i hjernevevet langs utsiden av blodårene. Funnet indikerer at det også hos mennesker kan finnes et «glymfatisk» system for utskillelse av avfallsstoffer fra hjernen, slik det tidligere bare er beskrevet hos dyr. Normal funksjon av det glymfatiske systemet antas å være viktig for hjernens evne til å kvitte seg med sykdomsfremkallende avfallstoffer. Hos pasienter med iNPH fant vi ut at utskillelsen av kontrastmiddelet fra hjernen var forsinket, og dette kan indikere at redusert utskillelse av avfallsstoffer spiller en rolle ved utvikling av iNPH, og kanskje særlig iNPH demens.

List of publications

I. Ringstad G, Emblem KE, Geier O, Alperin N, Eide PK. Aqueductal stroke volume: Comparisons with intracranial pressure scores in idiopathic normal pressure hydrocephalus. Am J Neuroradiol 2015; 36:1623-30 II. Ringstad G, Lindstrøm EK, Vatnehol SAS, Mardal K-A, Emblem KE, Eide PK. Non-invasive assessment of pulsatile intracranial pressure with phase-contrast magnetic resonance imaging. PLOS One 2017; 12(11):e0188896 III. Ringstad G, Emblem KE, Eide PK. Phase-contrast magnetic resonance imaging reveals net retrograde aqueductal flow in idiopathic normal pressure hydrocephalus. J Neurosurg 2016; 124:1850-1857 IV. Ringstad G, Vatnehol SAS, Eide PK. Glymphatic magnetic resonance imaging in idiopathic normal pressure hydrocephalus. Brain 2017; 140 (10): 2691-2705

1. Introduction

1.1. Historical overview of cerebrospinal fluid An awareness of a watery fluid within the skull can be traced to Hippocrates (460-375 BC), who commented on “water” surrounding the brain when he described congenital hydrocephalus [1]. Hippocrates considered, however, water merely to replace air as part of a pathological process [2], and adhered to the pneuma theory (air in circulation). Later, Galen (129-?216 AD) considered psychic pneuma to be stored in and distributed by the brain ventricles [3]. This understanding would prevail for more than a millennium, as no scientific autopsies were performed between ancient times and the Renaissance [4] due to restrictions from religious believes. In the early 16th century, Leonardo da Vinci revealed the anatomy of the ventricular system by producing a wax cast of bovine brain ventricles in his search for “senso commune” – which is probably better translated to “the soul” rather than the somehow more trivial term “common sense”. Leonardo da Vinci finally concluded that the seat of “senso commune” is the “middle” (third) ventricle, as opposed to other thinkers who had argued for its location in the heart [5]. While being still influenced by the pneuma theory, da Vinci did not address CSF in particular. It may have been the Swedish mining engineer, scientist, inventor and visionary Emanuel Swedenborg who provided the very first description of CSF between 1741 and 1744.

Figure 1. Emanuel Swedenborg (1688-1772), painted by Per Kraftt the older (1724-1793). In the anatomy book “Regnum Animale” (The Soul`s Domain) (1744-1745) Swedenborg identified the localization of cerebrospinal fluid [6]. (Picture from Store Norske Leksikon 9. februar 2018, https://snl.no/Emanuel_Swedenborg)

Swedenborg was a religious man, also searching for the seat of the soul and a connection between the spiritual and physical worlds. Due to his lack of medical credentials, he was unable to find a publisher, and his works were discovered in Stockholm one and a half centuries later and published in 1887 [1, 4]. In spite of this, many historians consider Domenico Cotugno to have first discovered CSF in 1764, and CSF was for some time referred to as “liquor cotugnii”. Francois Magendie was the first to use the term “cerebrospinal fluid” in 1842 [2], and also described CSF composition and flow between the ventricles and subarachnoid space through the foramen which still bears his name [4]. In 1891, Heinrich Quincke was the first to access the CSF compartment in living humans in order to treat increased CSF pressure [7]. With a fine cannula, he was also able to perform a chemical analysis of CSF [2]. Importantly, the method also allowed for measurement of CSF pressure and thereby an indirect assessment of intracranial pressure (ICP). Quincke further demonstrated that cinnabar injected into CSF of animals enmeshed within the Pacchionian granulations, already described in dissections by Pacchioni in 1705, and these observations were later confirmed in humans by Key and Retzius in 1875 [8].

1.2. From the bulk flow hypothesis to new concepts The neurosurgeon Walter E. Dandy, inventor of x-ray air encephalography in 1918, representing the first neuroradiological procedure, also conducted the first experimental hydrocephalus studies in 1914 together with Blackfan. They demonstrated that extirpation of the choroid plexuses in a dog modified the degree of internal hydrocephalus after occlusion of the Sylvian aqueduct. From this, they inferred that CSF is formed inside the ventricles, at least more rapidly than it is removed [9]. The division of hydrocephalus between obstructive and communicating type derives from their contribution. The principal understanding of CSF absorption, still accepted by many scientist today, was formed by the experiments of Dandy`s contemporary, Weed [10]. In these, CSF, together with its solutes, was found to escape the subarachnoid space via the Pacchionian granulations, as well as through the cribriform plate to lymphatic vessels of the nasal mucosa, and also along perineural spaces surrounding cranial nerves. Quite contrary to this, Dandy and Blackfan reported one month earlier that CSF is diffusely absorbed from the entire subarachnoid space and that resorption does not take place through Pacchionian granulations [9]. Still, Weed`s concept would prevail through the next decades, even

15 though Dandy refuted the role of Pacchionian granulations, or arachnoid villi, in later experiments [11].

Figure 2. To the left: Walter Edward Dandy (1886-1946) (Picture from: Congress of Neurological Surgeons. https://www.cns.org/about-us/history/walter-e-dandy).To the right: Harvey Williams Cushing (1869-1939) (Picture from: Wikipedia. https://en.wikipedia.org/wiki/Harvey_Cushing)

In 1926, Cushing introduced the bulk flow theory, where he hypothesized CSF to circulate with a direction, streaming from the ventricles towards drainage pathways at the brain surface [12]. This “third circulation”, as it was coined by Cushing, represented at the time a radical departure from the contemporary view that CSF moved by ebb and flow [13].

While the bulk flow paradigm may seem to still govern customary understanding of CSF flow in many aspects, several authors have called for a revision of concept due to an increasing body of opposing evidence [14-16]. First, the bulk flow hypothesis insufficiently incorporates the role of intracranial pulsations. Important evidence of their importance was given in 1962, when Bering found pulsatility from the choroid plexuses to be a force behind ventricular dilatation [17]. In 1978, Di Rocco produced hydrocephalus in lambs by increasing the intraventricular pulse pressure mechanically [18]. In vivo CSF pulsations were already observed using fluoroscopy in the 1960`s [19] and were later confirmed to be synchronous with the cardiac cycle using flow sensitive MRI [20]. The complexity of CSF flow dynamics has later been revealed and demonstrated to be dependent on various factors such as location [21], body posture [22] and respiration [23].

Second, and neither contained by the bulk flow hypothesis, is the vast capacity for fluid exchange throughout different compartments of the brain parenchyma [24]. The brain ISF is in communication with the ventricular and subarachnoid fluid via the perivascular fluid spaces that penetrate the brain parenchyma, and these spaces are carrier pathways for brain metabolites [25]. Movement of fluid and substances along these pathways seem to be driven by pulsations from the cerebral arteries [26], and an impaired ability of substances to move within the extravascular domain may contribute to a number of brain diseases [27]. However, there are still unresolved issues, one of them being opposing views about the direction of paravascular water and solute flow within the brain. While some authors have found signs of periarterial flow out of the brain parenchyma after intraparenchymal tracer injections [28-30], injections of tracer directly to the subarachnoid space have led to the conclusion that periarterial pathways are the route for transport of CSF tracers into the brain [26, 31].

1.3 Presentation of NPH In 1965, Hakim and Adams were the first to describe a symptom triade consisting of disabling dementia with psychomotor retardation, unsteadiness of gait and unwitting urinary incontinence, from which the authors reported a “dramatic” improvement of symptoms in three of their patients after surgical diversion of CSF (shunting) [32, 33]. As CSF pressure was within normal range upon lumbar puncture, the authors denoted the syndrome “Normal pressure hydrocephalus” (NPH). The hydrocephalus was referred to as “occult” as the patients` heads were of normal size, and as the ventricular system at the time could be imaged with x-ray air encephalography only (x-ray imaging of the head after lumbar injection of air). It was reported a particularly promising effect of surgery on cognitive disability, and the authors concluded that further cases should be sought within the large group of patients with late-life dementia. The great benefits of surgery were subsequently confirmed by some authors [34], but were not reproduced to the same extent in later, prospective studies [35-37], and rate of shunt insertion declined [38].

Despite extensive study of this syndrome (a recent PubMed search of “Hydrocephalus, Normal Pressure” [Mesh] yielded 1997 publications on the topic since 1965), NPH has to some degree remained controversial as a clinicopathological entity, and the term NPH has been proposed replaced by “chronic hydrocephalus” [15, 39]. To date, there is no common gold standard to establish 17 presence of NPH, nor are there uniform criteria for treatment indication. Even though shunt treatment undoubtedly may have a compelling effect on NPH symptoms, skepticism to the NPH diagnosis may probably be rooted in the presence of symptomatic overlap with other disease, and poor understanding of NPH etiology and mechanisms behind CSF circulation.

1.4 Pathogenesis Even though Hakim and Adams could not establish the etiology of NPH, they suggested the symptoms were result of decreased CSF resorption, and that ventricular enlargement occurred compensatory to increased pressure acting on the nervous tissue surrounding the ventricles, especially around the frontal horns adjacent to the frontal lobes. It was further postulated from “the hydraulic-press mechanism” that the intraventricular pressure exerted the greatest force against the widest part of the system. Further, expanded lateral ventricles were thought to promote tangential shearing forces on periventricular white matter fiber tracts associated with gait. With continued ventricular expansion, the cortex would be exposed to the same shearing forces, leading to dementia [40].

The most commonly identified causes of defective CSF resorption are meningitis and subarachnoid, or intraventricular, hemorrhage, which may cause inflammation and fibrosis of the arachnoid granulations, and thereby compromised CSF uptake. In the case of no identified cause, a previous head injury or subclinical viral infection may be considered, and biopsies in NPH patients without any known precipitating condition, have demonstrated leptomeningeal fibrosis [41]. There has also been established a significant association between NPH and vascular risk factors such as arterial hypertension and diabetes mellitus, and been suggested that these factors might be involved in pathophysiological mechanisms [42]. Stiff arteries and inability to dampen arterial pulsations in the subarachnoid compartment have been proposed to induce a “restricted arterial pulsation syndrome”, in which pulsations are rather propagated centripetally into the brain parenchyma. This “water hammer effect”, occurring as the pulsating brain repeatedly pounds against ventricular CSF, renders for compressive forces made upon periventricular white matter, and subsequent ventricular enlargement [15]. As restricted artery pulsations are a result of reduced intracranial compliance [43], the factors leading to reduced compliance have not been convincingly demonstrated. A pivotal role of ICP 18 pulsations in iNPH pathogenesis has been substantiated by high shunt response rates in patients who are selected to surgery based on ICP pulse pressure above established thresholds [44]. Also, intracranial pressure pulsations have been proposed instrumental to drive convective transport of fluid and solutes along brain perivascular and interstitial spaces, and restricted artery pulsations may reduce clearance of pathogenic macromolecules such as amyloid-β from the brain [31]. Whether such clearance pathways are compromised in NPH has never been addressed. Silverberg and colleagues hypothesized that NPH and AD share a common physiological basis with regards to clearance of toxic metabolites, which would lead to accumulation of Aβ and tau protein [45]. Contrary to previous, mechanistic approaches to understand NPH pathogenesis, a possible metabolic explanation for the dementia component in NPH was thereby introduced. Ventricular enlargement and stretching effects on associated white matter, particularly frontal fibers involved in micturition and executive motor function, was, however, maintained as the most probable cause of gait disturbance and urinary incontinence [46].

Furthermore, periventricular cerebral blood flow is found to be compromised in iNPH, which may promote watershed ischemia [47], and this is supported by observations of restored blood flow after shunting [48]. Deep white matter ischemia may compensate an initial CSF pressure increase and thereby explain normal CSF pressure in NPH [49].

Pathogenic factors at the venous side have also been hypothesized, and it has been suggested that elevation of cortical vein pressure affecting CSF absorption pathways to veins may be instrumental. In part, this is based on observations of restored cortical vein pulsatility in patients responding to shunting [50], and it has been claimed that NPH should rather be denoted “venous compressive ischemic encephalopathy [51]. However, the cortical vein compression theory has not received broad scientific acceptance, and it may seem plausible that restored venous pulsations after shunting is secondary to a general improvement in intracranial compliance provided by the shunt. It has even been proposed that lack of cortical vein compression is part of a viscous cycle in NPH [15].

With implications for a possible role of genetic, inborn factors behind NPH, some studies report larger head sizes in NPH patients than controls, and suggest the

19 possibility that internal [52, 53] or external [54] hydrocephalus of childhood may present with symptoms later in life. As for the latter (also called benign external hydrocephalus, or benign enlargement of subarachnoid spaces hydrocephalus), deep white matter ischemia of late adulthood has been hypothesized to represent a “second hit” by promoting increased resistance to CSF resorption into the ECS. Indeed, patients with NPH have higher prevalence and severity of periventricular signal intensity changes at MRI [55, 56]. Reports of siblings with NPH add to the hypothesis that there might be present an inborn, genetic factor [57, 58].

In 2012, the Danish researcher Maiken Nedergaard and co-workers described what they denoted the “glymphatic”, or glia-lymphatic, system [31]. In many ways, this may be regarded as a re-discovery of pathways described by Rennels et al. in 1985 [26], who visualized that the intrathecal CSF tracer horse-radish peroxidase spread from CSF along arteriolar paravascular spaces of the brain, with further penetration into the extracellular space, and finally accumulated around veins on prolonged scans. Nedergaard`s group extended Rennels` concept by demonstrating such flow to be dependent on AQP4, a protein which forms a water-specific channel, coded for by the AQP4 gene, and covers up to 40 % of astrocytic (glial) end feet surrounding brain capillary vessels [59]. They further provided evidence that clearance of brain macromolecules, such as Aβ, was dependent on AQP4 status. The Aβ peptide is a major constituent of extracellular aggregates known as neuritic plaques in AD [60].

Figure 3. The glymphatic system as illustrated by Iliff et al.[31] (reprinted with permission obtained through Copyright Clearance Center`s Rightslink®) (License No. 4285970774404).CSF and solutes enters and leave the brain along periarterial and perivenous spaces, respectively. CSF pulsations are a main force behind convective (net) flux through the brain interstitial space. This transport is dependent on AQP4 water channels polarized to astrocytic end feet.

Glymphatic transport has also been proposed instrumental for clearance of metabolic waste products particularly during sleep [61], for removal of excess fluid in brain edema [62], to have a role in normal ageing [63], and to be impaired after head trauma [64]. In iNPH, brain biopsies have recently demonstrated reduced levels of AQP4 and its anchoring protein dystrophin 71 in astrocytic end feet, which surround brain vessels, indicating a possible link between glymphatic function and iNPH pathology, in particular iNPH dementia [65].

1.5 Classification Currently, the causes of NPH and explanation for NPH symptoms are still debated [66] and should be considered incompletely described. NPH is typically

21 categorized into idiopathic NPH (iNPH) and secondary NPH (sNPH) [67], the latter term is used when precipitating conditions such as subarachnoid hemorrhage, meningitis, or neoplastic disease are identified. As from here, the acronym “NPH” without any prefix will only be used when NPH is uncategorized. Exactly how precipitating conditions produce chronic hydrocephalus without increased pressure is not completely understood. The reported proportion of iNPH to sNPH varies. There have been proposed stratifications of iNPH after levels of probability for having iNPH. The first Japanese guidelines introduced the categories possible, probable and definite. Possible iNPH included one or more of the classical symptom triad and ventricular dilation in a middle aged or elderly patient with effaced fluid spaces at the high convexity on MRI. Probable iNPH was defined as improvement of symptoms after CSF removal in a patient with possible iNPH. Definite iNPH demonstrated clinical improvement after shunt surgery [68]. The diagnosis was thus disproved when the patient had already been subjected to the risk of complications. Another group omitted shunt responsiveness as a diagnostic criterion and recommended that iNPH was classified into probable, possible and unlikely categories, depending on history, physical findings, and supporting studies [69].

1.6 Epidemiology A prevalence study carried out in Vestfold County, Norway, reported a prevalence of probable iNPH to 21.9/100 000 in the general population, however, prevalence increases with age to a peak of 181.7/100 000 in the age group 70-79 years [70]. A Japanese study indicated a prevalence of 2.9 % in community residents aged 65 years or older [71]. It has further been estimated that 9-14 % of nursing home residents suffer from iNPH [72]. INPH seem to be underdiagnosed and undertreated in Norway, as epidemiologic data suggest that less than one of five new cases each year receive surgical shunting [73], which is in line with surgery rates in Sweden [74].

1.7 Symptoms and co-morbidities INPH occurs with varying combinations or degrees of gait impairment, urinary incontinence, and dementia. Symptoms, as well as prognosis and treatment outcome, are prone to comorbidities, which include other causes of dementia, psychiatric and behavioral disorders, vascular disease, urinary problems and musculoskeletal conditions [75]. Contribution of comorbidity to overall morbidity, 22 mortality and long-term outcome may be considerable. In a study of NPH patients with mean age 72 years, only 36 % of patients were alive and able to meet for a 5- year follow-up evaluation [76]. A careful clinical examination is therefore necessary for the iNPH diagnosis. Grading scales incorporating at total score for different iNPH symptoms have been developed [44, 77] and are typically used to standardize the evaluation of treatment, but are insufficient as tools to decide which patients should be selected to surgical shunt treatment.

1.7.1 Gait disturbance Gait disturbance is the most common symptom in iNPH and the symptom with most favorable response to shunting [67], where improvement in up to 93 % of subjects is reported [78]. When untreated, it typically progresses from balance difficulty, to shortened stride length and arrestments with turning difficulties [67]. Gait problems may be associated with other motor symptoms such as hypokinesia, tremor and hypo- or hyperkinetic movement patterns in up to 86 % of elderly iNPH patients [79]. The differentiation between iNPH and Parkinson`s disease may therefore be challenging. Further, gait disorders are generally common in the elderly population and may derive from a number of causes, ranging from hip, knee and spine pathology to stroke and other neurological diseases.

1.7.2 Urinary incontinence Urinary incontinence has been reported to occur in between 60 and 79 % of iNPH patients [80, 81]. Detrusor over activity with urgency is postulated to be the basis for most urinary urgency or frequency [75]. Urinary symptoms are reported to improve with shunting in 36 to 76 % of cases (reviewed by [67]). These symptoms are not specific features of iNPH, as symptoms of bladder dysfunction are common in the adult population and have been reported in 57 % of patients consulting office- based primary care physicians [82].

1.7.3 Dementia While iNPH is found rather infrequently, it is estimated that an overall of 77 000 patients suffer from dementia in the Norwegian population, and may double within an ageing population by 2040 [83]. Imaging findings in iNPH may overlap significantly with other dementia types. INPH and AD can even occur in a mixed form

[84], and Aβ in cortical biopsies has been detected in 42 % of patients with possible NPH [85] and in up to 75 % of those with severe iNPH dementia [86]. Elevated pulsatile ICP, a hallmark of iNPH, has been shown associated with brain amyloid accumulation [87]. Clinically, hippocampal dysfunction and rapid forgetting of newly acquired information is the most common presenting symptom of AD. This may be present in iNPH as well, however, more common symptoms are frontal executive disturbances that can be improved by cues or reminders [75]. Other neurodegenerative diseases that may confound, or aggravate, iNPH dementia include Parkinson’s disease, dementia with Lewy bodies, and frontal lobe dementia. Moreover, a primary depressive disorder with mental slowness may also mimic symptoms of hydrocephalus. A wide range of neuropsychological deficits are found in iNPH patients, and these deficits may be aggravated by vascular comorbidity [88]. Indeed, vascular pathology is frequent among post-mortem NPH findings [89], and vascular risk factors are overrepresented in iNPH individuals [90]. In the overall population, subcortical vascular dementia is a more common cause to the classical symptom triad typical for NPH than NPH itself [91]. Dementia is only half as likely to improve following shunt surgery as compared to gait function [78].

1.8 Treatment Diagnostic efforts should not be unnecessarily delayed, as treatment delay is associated with reduced treatment effect. Probably, there exists a therapeutic window in early phase of the disease process, and iNPH has shown more responsive to treatment of milder symptoms of less than 2 years` duration [92]. Surgical CSF diversion is the only treatment option for NPH. Different types of CSF diversion include placement of a shunt from the cerebral ventricles to the peritoneum (VP-shunt), to the right atrium of the heart (VA-shunt), or between the lumbar subarachnoid space and peritoneum (LP-shunt). While endoscopic third ventriculostomy may also be considered, VP- or VA-shunting are considered the mainstay of surgical treatment [93]. However, a Cochrane Database Review reported that there had never been conducted a single, randomized controlled trial to assess the effect of surgical shunting compared to no shunt [94]. Ventricular shunting has indeed also been described as variable, short-lived and unpredictable [67]. Patient heterogeneity, and different assessment of treatment outcome, may well explain why shunt response rates for iNPH are reported to range from 15 to 96 % [95, 96]. Wide

24 variability in use and interpretation of diagnostic tests to decide treatment exist between different neurosurgical centers [97]. This has also rendered for scientific controversy, as some authors call for shunting in “many more patients” [38], while others claim that shunt dependable dementia accounts for only 0,4 % of all dementia patients [95], and that shunt insertion carries a risk disproportionate to the potential benefit [98]. A meta-analysis of 44 articles found that the pooled, mean rate of shunt complication (including death, infection, seizures, shunt malfunction, subdural hemorrhage or effusion) was 38% [67]. It has therefore been an urgent need for developing prognostic tests with low risk that identify iNPH patients who will respond to shunting, and also to spare non-responding patients from unnecessary surgical risks.

1.9 Additional tests A diagnosis of iNPH requires a synthesis of converging evidence from clinical history, physical examination, and brain imaging. Because of iNPH symptom heterogeneity and overlap with more common conditions, additional methods of assessment are mandated and with the primary aim to select iNPH patients who will benefit from surgical shunting. In practical terms, clinical work-up of iNPH is therefore more about evaluating the possibility of shunt response, rather than diagnosing its cause.

1.9.1 CSF tap test, infusion test and external lumbar drainage In the 2005 Guidelines from the International NPH Consulting Group, it was recommended that all patients with possible and probable iNPH (based on clinical exam and imaging) should be considered for, in a stepwise order, a CSF tap test, an infusion test and external lumbar drainage [99]. The positive result of a CSF tap test with removal of 40-50 ml CSF by spinal puncture may increase probability for a favorable shunt response compared to clinical examination only, but cannot be used as an exclusionary test, and have low sensitivity (26 – 61 %). By lumbar infusion test, an isotonic solution is typically infused at constant rate, and the resistance to outflow

Rout determined. There are several infusion methods with diverging results, however, infusion tests are generally considered to have higher sensitivity (57 – 100 %) compared to the tap test, while prolonged external lumbar drainage of more than 300 ml has the highest positive predictive value [100]. From the European iNPH multicenter study, it was concluded that infusion test (Rout) and CSF tap test did not 25 correlate with results of surgery at 12 months, and should not be used to exclude patients from treatment [101]. This conclusion was later maintained by the subcommittee of the American Academy of Neurology [96].

1.9.2 ICP monitoring ICP monitoring is carried out in local anesthesia, where an ICP sensor is placed in the frontal lobe parenchyma through a burr hole in the scull. Increased pulsatile ICP in iNPH is common due to reduced intracranial compliance. When intracranial compliance is low, only a small intracranial volume increase, such as the one induced by a heartbeat, will lead to a significant pressure increase [43, 102]. Compliance may thus be defined as the ability of the intracranial compartment to accommodate a volume change. Unlike invasive monitoring of mean ICP, which relates to atmospheric pressure, over-night monitoring of pulsatile ICP (mean wave ICP amplitudes, MWA) represents the absolute per-cardiac-beat pressure change induced by the temporary intracranial volume increase. Pulsatile ICP has demonstrated to predict shunt response in 9 of 10 iNPH patients when MWA in average >4 mmHg and/or percentage of MWA >5 mmHg in >10% of recording time are set as threshold levels for shunting [44, 103]. Software for automatic assessment of cardiac induced ICP single waves has not been readily available, and is still in use at only a few centers world-wide. Invasive pressure monitoring carries a risk of complications, this risk may, however, be considered low, as infections and bleedings occur in 1-2 % [103, 104].

1.9.3 Imaging

1.9.3.1 Structural imaging When NPH was first described in the mid 1960`s, methods for assessments of ventricular size and CSF compartments were restricted to use of x-ray air encephalography. Since then, a tremendous development in imaging has occurred, and modern imaging can now reveal far more features of the entire intracranial compartment. In radiology, which historically has been a discipline utilizing mainly subjective image interpretation, there is presently a tendency towards a search for quantifiable imaging biomarkers to better standardize the diagnostic process and the assessment of treatment effect.

The mainstay of radiological work-up in iNPH is computed tomography (CT) and MRI. Both CT and MRI provide volume acquisitions of the entire brain and its surroundings, where images are typically presented as consecutive image slices to cover the region of interest in any plane. CT depends on x-rays that pass through the object of interest, and where a detector at the opposite end of the radiation source register the proportion of radiation that has been absorbed, or attenuated, within the tissue. Degree of attenuation in every voxel element is presented on a greyscale image, in which the densest tissue is given the whitest shade of grey, and vice versa. CT provides reliable information about ventricular size in patients with suspected iNPH and is also faster, cheaper and less sensitive to patient motion than MRI. MRI, on the other hand, is not dependent of ionizing radiation, but magnetism. The physical principles behind MRI technology are complex and explained in detail elsewhere [105]. In short, an MRI scanner images protons, which are present throughout the body, mainly as constituents of fat and water. When the patient is placed in a strong external magnetic field, protons spin around their central axis (precession) at a certain frequency and add up to create a net magnetization in tissues along the direction of the external magnetic field. When a radiofrequency (RF) pulse is emitted against protons precessing with similar frequency (hence the term “resonance”), the direction of the net magnetization changes. The degree of angulation of the net magnetization from the direction of the external field is called the flip angle of the RF pulse. The component of the net magnetization that is perpendicular to the external field causes a signal that can be detected by antennas (coils) placed close to the body surface. MR sequences can be configured to reflect different properties in tissues. In a T1 weighted sequence the image signal reflects how quickly the net magnetization vectors recover their longitudinal magnetism after the RF excitation (T1 relaxation). The T1 relaxation time varies between different biological and pathological tissues, providing for image contrast. T2 weighted image contrast is dependent upon tissue specific loss of the transversal component of the net magnetization after the emission of a RF pulse (T2 relaxation). An MR image of the brain may further be weighted in a number of other ways to focus on other properties of tissue, such as water diffusion (diffusion weighted imaging) and suppression of unbound water (fluid attenuated inversion recovery, FLAIR). Soft tissue image contrast is superior at MRI compared to CT.

An essential part in evaluation of iNPH patients is to assess ventricular size with computed tomography CT or MRI, and to rule out a non-communicating hydrocephalus. Evans’ index (EI) is the ratio of the transverse diameter of the anterior horns of the lateral ventricles to the greatest internal diameter of the skull, and was developed for use in pneumoencephalograms [106]. A ratio > 0.3 is considered as definite ventricular enlargement. It has later been adapted for application on CT images [107]. A CT study may thereby exclude iNPH, while MRI provides more diagnostic information and better identifies causes of CSF obstruction. In the 2005 guidelines for diagnosis of idiopathic normal pressure hydrocephalus, an EI > 0.3 combined with gait dysfunction plus either urinary or cognitive dysfunction is required prior to consideration of treatment with ventriculo-periteoneal shunt [69]. EI can vary significantly between iNPH patients, and is not an ideal method for estimating ventricular volume [108], but neither has ventricular volume shown predictive value in differentiating between NPH patients who will respond to shunt surgery, or not [109]. The European multicenter study on iNPH demonstrated, however, a shunt response in up to 84 % of subjects after one year when being diagnosed solely on clinical and MRI criteria using EI [110]. Furthermore, new EI threshold values, incorporating the range of EI values in the elderly population, have proven good sensitivity for the iNPH diagnosis [111].

EI does not discriminate between different causes of ventricular enlargement. The callosal angle (CA, the angle between the lateral ventricles on a coronal image through the posterior commissure) has been found to be steeper (< 120 degrees) at air encephalography in NPH than in ventricular enlargement due to brain atrophy [112] and is also proven steeper at MRI in shunt responders compared to non- responders [113]. Recently, it was reported that CA and EI combined provided good accuracy as a screening tool to differentiate patients with NPH from AD and healthy controls, and that patients within given threshold levels should be further evaluated with automated MRI brain tissue segmentation, i.e. labeling brain and brain subregions by use of computerized methods [114]. Such segmentation has demonstrated that shunt-responsive NPH is characterized by high preoperative ventricular and near normal grey matter volume compared to AD and healthy controls [115]. This is, however, in contradiction to a previous study, where no predictive value of shunt response in NPH could be demonstrated using volumetric

28 assessments of different imaging variables, including ventricular and brain volume [109].

A commonly seen feature in those with probable iNPH is also ventricular enlargement associated with effaced CSF spaces outside the high convexities and medial subarachnoid spaces of the brain, while the Sylvian CSF volume is prominent [116]. As distribution of CSF here is disproportionate between the inferior and superior subarachnoid spaces, the term DESH (disproportionate enlargement of subarachnoid spaces hydrocephalus) was coined, and was shown to have high positive predictive value in identifying shunt responders [80]. A broader assessment of several MRI features demonstrated that small CA, wide temporal horns and DESH, each independently, predicted a positive shunt response [117]. In the extensive, second edition of the Japanese guidelines for management of iNPH, iNPH is classified into DESH and non-DESH [118]. Negative predictive value of the DESH sign is, however, low, and absence of DESH should therefore not exclude patients from further diagnostic tests, or from receiving a shunt [119]. This is important, as non-DESH is the most frequent finding among patients with probable iNPH (70 %) [119]. With regards to differentiate the ventricular enlargement of iNPH from AD related brain atrophy, findings of a milder hippocampal atrophy and less widening of the parahippocampal sulci may be useful [118].

1.9.3.2 Imaging of CSF flow with phase-contrast MRI While structural imaging maintains to be a cornerstone in imaging of NPH, a frozen image can never reveal, or embrace, the complexity of a CSF in continuous motion. PC-MRI extracts quantitative velocity information from images. In short, a bipolar magnetic gradient (velocity encoding gradient, VENC) is applied to change magnetic precession phase of moving protons, opposite to stationary protons, which remain unchanged. The MRI scanner calculates the phase difference for each picture element (voxel) between phase images with, and without, use of VENC. This phase difference is proportional to flow velocity and flow in both directions through the image plane is given. Zero flow is displayed by a medium image grey tone, and flows of opposite directions are lighter and darker, respectively. When flow velocity exceeds velocities covered by the set VENC, aliasing will occur due to phase differences of more than 360 degrees in magnetic spin [105, 120].

After its first applications for use in blood flow measurements, a phase sensitive technique for the study of CSF flow was reported by Edelman et al. [20] and has later been further developed and sophisticated [121, 122]. By synchronizing image acquisition to the cardiac cycle, velocity information as function of the cardiac cycle is obtained, so-called cine phase contrast. This synchronization (cardiac gating) may be performed prospectively or retrospectively with respect to sampled R waves from the cardiac cycle. Retrospective gating has been shown to be more accurate as it enables continuous measurements throughout the cardiac cycle [122] and is typically the method of choice in clinical investigations. The flow velocity curve represents, however, an average measure over many cardiac cycles and is therefore not a real-time measurement, unlike, for example, ICP monitoring. Cardiac contraction may be registered with an electrocardiogram (ECG), or with peripheral triggering using a finger plethysmograph. With the latter, a broadened velocity distribution and a slightly lower maximum amplitude may be expected [122], but does not change the temporal relationship between cardiac systole (cardiac contraction) and diastole (cardiac relaxation) [121].

In healthy, PC-MRI technique and its use for investigating the relationship between cardiac dependent intracranial blood- and CSF flow has been extensively reviewed [123]. The net intracranial volume increase induced by arterial inflow is balanced by early displacement of CSF into the compliant spinal canal. This balances the ICP increase during the systolic phase, as venous blood, with its higher viscosity than CSF, is not drained instantaneously, but supplements the dampening of the ICP increase at a later phase of the cardiac cycle, and is also volumetrically larger than the CSF component. Compared to the craniocervical junction level, aqueductal CSF flow is almost ten times smaller, and occurs later in the cardiac cycle. As the aqueduct is narrow, and relatively long, the impact of aqueductal flow on ICP change is low in healthy subjects [124]. In NPH, however, the ventricular (aqueductal) CSF flow contributes more to CSF flow exiting the intracranial compartment through the foramen magnum than in healthy [125].

PC-MRI derived CSF flow velocity measurements have mainly been undertaken at level of the aqueduct in order to predict shunt response in iNPH. One important reason for this is the aqueductal hyper dynamic flow artifact (signal void) at T2 weighted images frequently observed in hydrocephalus [126]. Presence of this 30 artifact was shown to possibly predict shunt response [127], particularly when flow compensation algorithms (gradient moment nulling) were turned off in the MR scanner [40]. Secondly, the aqueduct is distinguished from many other regions of the CSF by its tubular-like shape and simple geometry, and by being the only macrostructural connection between the supra-aqueductal ventricles at the interior brain CSF compartment and the external surface of the brain and spinal cord. PC- MRI have demonstrated feasibility [121] and good reproducibility for pulsatile CSF flow measurements in healthy subjects, both at level of the aqueductal and upper cervical spine [128].

Bradley et al. were the first in 1996 to demonstrate possible utility of PC-MRI derived aqueductal CSF flow as marker to select NPH patients for shunting, where pre-surgical ASV > 42 μl was observed in 12/12 shunt responders [40]. Later, several studies confirmed that aqueductal flow parameters, and in particular ASV, could be useful for selection of patients to surgical shunting [129-134] and in follow-up [135]. Other studies have not been able to reproduce these beneficial results [136-139]. A common weakness of many studies assessing aqueductal CSF flow is the lack of direct comparison with invasive measures from the intracranial compartment. One study compared ICP with aqueductal flow measured with PC-MRI in the sagittal plane, but unfortunately without the possibility to quantify flow velocities, as flow velocities should be measured in an image plane as perpendicular to flow direction as possible [140]. Another study found association between ASV and a temporal sub-peak of the ICP wave in a small number of patients [141], but its significance has later been disputed [142].

While aqueductal measures of flow based pulsatility, such as the ASV, has been regarded as a possible indicator of intracranial ICP pulsatility [15, 143], other PC-MRI methods have been proposed with the aim to directly measure pressure- volume relations and compliance, the latter being typically reduced in shunt- responsive NPH [44], and characterized by increased ICP pulsatility [43]. Wåhlin et al. demonstrated that a combination of lumbar CSF infusion and PC-MRI proved feasible for assessment of the craniospinal cavity pressure-volume index in healthy elderly, and thereby enabled for a test of how well this compartment dampens the volume load represented by arterial pulsations [144]. While this test does not seem to have been taken widely into use clinically, a method for assessment of intracranial 31 elastance (dP/dV), the inverse of compliance, has shown more promise by means of clinical utility [145]. Positive experiences with the method have been retrieved from studies with baboons [145], healthy subjects [146], as well as patients with hydrocephalus [147, 148], Chiari 1 malformation [149] and NPH [150, 151]. However, this non-invasive method utilizes measurements of blood and CSF flow below level of the craniocervical junction, and has never been compared directly with invasive ICP monitoring in ill patients.

Moreover, PC-MRI derived velocity parameters, including flow phase, have been investigated for use in other anatomical locations than described above (prepontine cistern, 4th ventricle, etc) [21], and combined with blood flow measurements [125]. These procedures are time consuming, and may be considered to have more of an explorative character to assess features of hydrocephalus, and have not come to wide use outside the setting of scientific studies.

1.9.3.3 Diffusion weighted imaging and diffusion tensor imaging With diffusion weighted imaging (DWI) and diffusion tensor imaging (DTI), free (isotropic) and directed (anisotropic) diffusion of water molecules, respectively, can be studied [152]. The apparent diffusion coefficient (ADC) is a measure of isotropic diffusion, and values above normal in the brain may indicate increased Virchow- Robin spaces, increased extracellular brain water fraction, and changes in myelin- associated bound water. Increased ADC has been demonstrated in several brain regions of NPH [153] as well as other hydrocephalic conditions [154], most typically in the periventricular region [54, 155, 156] and with a decline after shunt response [154, 156]. It therefore seems likely that brain water content, and/or extracellular brain water fraction, is increased in NPH.

DTI allows for further assessment of microstructural changes in cerebral white matter utilizing several parameters of anisotropic diffusion along brain fibers, including fractional anisotropy (FA) and mean diffusivity (MD). In a recent review of 19 studies, where DTI was used for the identification and differentiation of iNPH from other neurodegenerative diseases, it was reported that FA had sensitivity of 94 % and specificity of 80 % for diagnosing iNPH. FA was typically increased in the corticospinal tract and negatively correlated with gait abnormality, whereas it was reduced after CSF drainage or shunting. Compared to healthy controls, MD was

32 higher in the corticospinal tract and corpus callosum [157]. The relative importance of individual DTI measures with regard to predict response to CSF drainage remains, however, a matter for debate [49].

1.9.3.4 Imaging of the glymphatic system To date, it has not been possible to undertake human CSF tracer experiments in which the brain paravascular and interstitial spaces can be assessed [158]. However, it has been shown in studies of rodents that an MRI contrast agent would be of suitable size to enter these pathways, constituting crucial elements of the glymphatic system, and that measures of macromolecular clearance from brain parenchyma may be obtained by MRI at multiple time points [159]. The technique has previously been used to demonstrate impaired glymphatic clearance after subarachnoid hemorrhage and ischemic stroke in mice [160] and later in nonhuman primates [161].It has furthermore been suggested that lumbar delivery of intrathecal contrast agent in conjunction with multiphase MRI may serve as a useful approach also in the study of human glymphatic function [162]. A human case report confirmed that an MRI contrast agent could be traced within brain parenchyma after subarachnoid administration [163], but no studies have yet applied this method in a patient cohort. For iNPH in particular, the method would on one hand have the potential to assess pathological distribution patterns of tracer substance in the interior and exterior of brain fluid compartments. Moreover, it could possibly reveal metabolic compromise by means of reduced parenchymal clearance of substances and thereby possibly shed new light on mechanisms behind neurodegeneration and dementia.

1.9.3.5 Nuclear medicine NPH has been characterized by using radioactive isotope cisternography [164, 165], where a typical feature has been ventricular reflux of radiotracer, and lack of ability to reach the high convexities. However, a study reported no more than a 55 % shunt response rate in iNPH patients with typical signs at cisternography [166], and the test was shown inferior to lumbar external drainage and CSF tap test [167]. Cisternography is not regarded necessary for the iNPH diagnosis [118]. Today, it is therefore rarely performed as part of the routine imaging work-up for iNPH, but its use is still sporadically reported [168].

Nuclear medicine studies such as PET and SPECT have been used to demonstrate changes in cerebral blood flow (CBF) and metabolism associated with

15 iNPH. Several PET studies have utilized O-H2O to assess CBF in iNPH. Klinge et al. demonstrated that global CBF was lower in iNPH than controls, but even lower in shunt responders than non-responders, and early improvement in the cerebrovascular reserve after shunting (assessed with acetazolamide) indicated a good prognosis [169]. Later, they revealed that cognitive impairment is associated with reduced CBF in mesial frontal and temporal areas, and that symptom improvement after shunting was paralleled by a CBF increase in some of the same areas [170]. In a group of patients with both iNPH and sNPH, Owler et al. found reduced CBF in the cerebrum and cerebellum, as well as basal ganglia and thalamus [171]. The authors could, however, not conclude whether the changes were primary or secondary to NPH disease. While no reduction in CBF of white matter could be demonstrated in this study, another work utilizing the same method showed that white matter CBF was indeed reduced in NPH, with a gradient stretching from the lateral ventricles (poorest) to the subcortical white matter (least poor) [47]. There are fewer reports on NPH from SPECT studies, but worth noticing is a paper which reported CBF changes in the anterior cingulate gyrus and hippocampus associated with impaired wakefulness, a symptom typically relieved after shunting [172].

More recently, PET has shown able to detect amyloid-β in vivo (amyloid-PET) and possibly facilitate diagnosis of AD in patients with suspected NPH, both utilizing the imaging agents [18F]flutemetamol [173-175] and [11C]Pittsburg Compound B [176]. In these quite small cohorts, amyloid-β pathology in biopsies was found in 23.5 to 50 % of patients with probable NPH.

1.10 Summary of key points from Introduction

x Mechanisms behind CSF circulation and disease are incompletely understood. x iNPH is characterized by enlarged brain ventricles and symptoms such as gait ataxia, urinary incontinence and dementia. Considerable influence from comorbidities that often affect patients of same age may occur. x More than 50 years after iNPH was first described as an entity, its etiology remains unknown, and iNPH may therefore be better defined by its response to treatment by surgical shunting than its cause. x While a single standard for the prognostic evaluation of iNPH is lacking, the diagnosis typically requires a synthesis of converging evidence from clinical history, physical examination, and brain imaging. x A positive CSF tap test, infusion test, or extended lumbar drainage test, will each increase probability for a positive shunt response, but should not be used to exclude patients from treatment. x Invasive ICP monitoring has proven to have high sensitivity and specificity, but its use is limited by complication risks. x A search for non-invasive methods to better characterize iNPH disease and identify shunt responders with high accuracy is warranted. x Structural imaging to demonstrate communicating hydrocephalus is considered mandatory for the iNPH diagnosis, and assessment of EI, CA and DESH may improve identification of shunt responders. x Imaging of intracranial fluid flow dynamics utilizing PC-MRI has been proposed to identify iNPH patients suitable for shunting (ASV) and to non-invasively assess intracranial elastance (dP/dV). x Multiphase MRI with contrast agent as CSF tracer is a new method that may characterize long-term CSF flow characteristics and detect reduced clearance of brain macromolecules in iNPH.

2. Aims of the thesis

The overall aim of this thesis was to study short- and long-term CSF flow characteristics and pathologic alterations in the pre-surgical work-up of iNPH patients.

More specifically, the aims were

a) To study whether PC-MRI derived imaging biomarkers of intracranial pulsatility obtained at the aqueduct (ASV) and CCJ (MRI-dP) are associated with invasive ICP monitoring b) To assess net aqueductal flow in iNPH and compare findings with invasive ICP c) To utilize an MRI contrast agent as CSF tracer and from this, image long-term, intracranial distribution of tracer in the CSF and brain compartment of iNPH patients and reference subjects

3. Methods

The conclusions of this thesis derive from prospective studies of two iNPH cohorts: In the first, we explored CSF flow based imaging biomarkers with PC-MRI at the level of the aqueduct and CCJ. In the second, we studied CSF flow patterns in a wider perspective by obtaining consecutive MRI acquisitions covering the intracranial compartment as a whole over an extended period of time using an MRI contrast agent as CSF tracer.

All patients were consecutively included in the studies and retrieved from referrals to the Department of Neurosurgery at Oslo University Hospital (OUH) for clinical and radiological work-up of suspected iNPH. Study related MR imaging sequences were performed at the final stage of a conventional imaging protocol for iNPH, and all pre- surgical imaging preceded ICP monitoring or other invasive tests, typically within a few weeks.

While patients were included and imaged prospectively, radiologic post-processing of images and comparisons with clinical data were performed in retrospect.

Written and oral informed consent to participate in the studies was retrieved from all patients, and the studies were approved by the Institutional Review Board and Regional Ethics Committee. The study incorporating intrathecal administration of MRI contrast agent also received approval from the National Medicines Agency of Norway.

3.1 Assessments of aqueductal CSF flow parameters (Paper 1 and 3)

3.1.1. Study population and design For all patients, at the time of PC-MRI, ventricular enlargement had already been confirmed by a prior CT or MRI, and symptoms had been considered suggestive of iNPH at the referring hospital. An inclusion criterion for the study was the combination of a technically successful PC-MRI at the aqueduct level and overnight ICP monitoring. In addition, patients were assessed clinically with determination of iNPH symptom severity. Patients, who were subsequently treated with surgical shunting, were invited to a second PC-MRI after one year with identical imaging protocol.

The study of net aqueductal flow (Paper 3) also included PC-MRI of two healthy individuals.

3.1.2. Clinical management and ICP monitoring

3.1.2.1 Clinical examination Severity of symptoms was graded by use of an NPH grading scale developed in the Department of Neurosurgery, which grades each component of the iNPH triade (gait disorder, urinary incontinence and dementia) from 1 to 5, rendering for a possible score from 3 (worst) to 15 (best) [44].

3.1.2.2 ICP monitoring At day-time, before the over-night ICP monitoring, the ICP sensor was inserted through a burr-hole in the skull to the inside of the brain surface in a parasagittal, frontal location using local anesthesia. The ICP recording was pre-defined to cover the time period from 23 p.m. to 7 a.m. and was performed in the patient-ward using a system for automatic detection of single ICP waves (Sensometrics AS, dPCom, Oslo). The difference between maximum and minimum pressure in cardiac systole and diastole, respectively, defines the ICP wave amplitude (pulsatile ICP). The average of all single ICP waves during consecutive 6-second intervals defines the mean ICP wave amplitude (MWA). In addition, mean ICP (relative to a zero pressure level) and percentage of MWA ≥ 5 mm Hg and mean ICP ≥ 15 mm Hg were recorded.

3.1.2.3 Ventriculoperitoneal shunting The decision for surgical shunting was based on the combined analysis of clinical examination, interpretation of radiological findings, and results of ICP monitoring. From the latter, primarily MWA is used to select patients for shunting with threshold levels of MWA ≥ 4 mm Hg and/or MWA ≥ 5 mm Hg in ≥ 10 % of recording time. A beneficial shunt response was defined by an increase of two points or more at the NPH grading scale, and this score was obtained at 3, 6 and 12 months after shunting. Complications to shunting were noted.

3.1.3 MRI

3.1.3.1 MRI protocol and image post-processing All MRI acquisitions were obtained in a 3 Tesla Philips Achieva scanner with a 16- channel head coil at the Intervention Centre, OUH. Main parameters of the PC-MRI obtained perpendicular to the mid-aqueductal level were TR/TE = 24/16 ms, voxel size 0.6/0.8/4.0 mm3, VENC = 10 cm/s and 30 – 40 phases per cardiac cycle with retrospective peripheral cardiac gating. For Paper 1, a T1-volume gradient echo MRI was obtained to render for volumetric analysis of supratentorial ventricles, using the medical imaging software ITK-SNAP 2.4 (www.itksnap.org) [177].

The aqueduct was manually defined with a ROI in dedicated software (Philips Q-flow) and potential aliasing was corrected by sinusoid curve fitting before ASV and net ASV were estimated. In Paper 1, ASV was defined as the mean of systolic and diastolic volumetric CSF flow during one cardiac cycle, minus net ASV. Net ASV was, as in Paper 3, defined as the absolute difference between systolic and diastolic volumetric CSF flow, and CSF flow volumetric rate (ml/min) was estimated by calculating net ASV with heart rate per minute. Flow in the caudo-cranial direction was defined as retrograde, or negative, flow.

3.1.4 Statistics

3.1.4.1 Paper 1 Data were assumed normally distributed and parametric tests were applied. ASV, as well as ventricular volume, before and after shunting was compared using paired samples t-test. Independent samples t-test was used to compare ASV in shunted and non-shunted patients. Correlations were assessed by the Pearson correlation coefficient. Significance level was set to 0.05.

3.1.4.2 Paper 3 Due to one extreme outlier, data were assumed not normally distributed and non- parametric tests were used for comparisons. Pairwise Wilcoxon signed-rank test was used to compare net ASV and CSF flow rate before and after shunting, respectively. Mann-Whitney test was applied to compare different groups. Significance level was set to 0.05.

3.2 Assessment of the PC-MRI derived pulse pressure gradient (MRI-dP) (Paper 2)

3.2.1 Study population Patients enrolled for this study of the PC-MRI pulse pressure gradient included all iNPH patients with PC-MRI examinations at the aqueduct level (Paper 1 and 3), and in addition iNPH patients where PC-MRI had been performed solely at level C2. Patients with the combination of technically successful PC-MRI at level C2 and ICP monitoring were finally included. Further, healthy control subjects were examined with PC-MRI for comparison with patients.

3.2.2 Clinical management and ICP monitoring Clinical examination, ICP monitoring and surgical shunting were performed according to the same definitions and descriptions reported above for patients examined with PC-MRI at level with the aqueduct.

3.2.3 PC-MRI

3.2.3.1 PC-MRI protocol and image post-processing The MR scanner and head coil were the same as for aqueductal measurements. In iNPH patients, PC-MRI was performed at level C2 with main parameters being TR/TE = “shortest” (typically 16/11 ms), pixel size 0.56/0.56 mm2 to 0.63/0.63 mm2, slice thickness 7 mm, VENC = 6 cm/sec, and 32-40 phases with retrospective peripheral cardiac gating. Healthy controls were scanned at four different time points during one day in a 3 Tesla Philips Ingenia scanner with a 32 channel head coil, using identical acquisition parameters and image positioning as for iNPH patients.

The CSF compartment at level C2 was defined with a ROI in dedicated software (nordicICE®). Conversion of pixel values to flow velocities by applying the VENC was performed using the software MATLAB®. Aliased velocities were corrected using a filter as described in Paper 2.

3.2.3.2 Computation of peak to peak pulse pressure gradients (MRI-dP) The pressure gradient was estimated from the Navier-Stokes equation as previously reported [145] in each pixel and for each time period of the cardiac cycle based on recorded CSF flow velocities across the axial plane at level C2, and then averaged, resulting in one mean pressure gradient for each time point. The peak to peak pulse 40 pressure gradient (MRI-dP) was defined as the difference between the minimum and maximum pressure gradient within one cardiac cycle. The computation is described in detail in Paper 2.

3.2.4 Statistics Mann-Whitney U test was used for comparison between groups and One-way ANOVA for comparison of means of more than two independent groups. Pearson correlation coefficient was applied to determine associations. Agreement between methods was evaluated with Bland-Altman plot. Reliability of multiple quantitative measurements within individuals was estimated using the intraclass correlation coefficient. The significance level was set to 0.05.

3.3 Glymphatic MRI (Paper 4)

3.3.1 Study population and design The study included patients referred to OUH with clinical suspicion of iNPH. In addition a reference group (REF) consisting of patients referred to work-up of possible CSF leakage, or intracranial cysts, was included.

3.3.2 Clinical management Clinical examination, ICP monitoring and decision for surgical shunting in iNPH were performed as previously described for iNPH patients included in the PC-MRI studies. REF patients were also assessed with NPH-score.

3.3.3 MR imaging

3.3.3.1 Imaging protocol A T1-weighted gradient echo volume scan was obtained before and at multiple time points through 24 hours after intrathecal administration of the MRI contrast agent gadobutrol (Gadovist®, Bayer Pharma AG, Berlin, Germany) via lumbar puncture and using the radiopaque contrast agent iodixanol (Visipaque®, GE Healthcare, USA) for verification of correct needle position at fluoroscopy. MR imaging parameters were set the same at all time points. Pre contrast, study subjects were also imaged with a 3D FLAIR volume acquisition. After contrast agent injection, all iNPH and REF patients were instructed to remain in bed between scans the first day. The final MRI was carried out the next morning (24 hours).

3.3.3.2 Image analysis Circular ROIs were placed in axially reconstructed T1 weighted images at predefined locations in CSF, brain parenchyma, and the superior sagittal sinus, to assess the mean SU from each ROI. Perivascular gadobutrol enhancement along the main artery trunks at the brain surface was at all time points dichotomized as present or non-present. Periventricular signal increase on FLAIR images was classified into four grades of severity, ranging from 1 (least) to 4 (most). Parameters of gadobutrol enhancement and clearance were calculated based on SU changes over time.

3.3.4 Statistics Data were considered normally distributed, and independent samples t-test was used to determine differences between continuous data, whereas Pearson Chi-square test was applied to determine differences between categorical data. The significance level was set to 0.05.

4. Results

4.1 PC-MRI derived parameters of intracranial pulsatility

4.1.1 Aqueductal Stroke Volume (Paper 1) Of 21 included patients, 17 patients were subsequently treated with shunt surgery, from which 16/17 (94 %) improved clinically. There were no serious complications to surgery; minor side-effects such as headache and dizziness were noted in five and two patients, respectively. In addition, the one non-responder had an epileptic seizure during the 12 months period of clinical follow-up, but without any obvious association with the shunt procedure. All shunted patients were invited to a second MRI after one year, where 12 met, and of these, 11 were shunt responders.

Median preoperative ASV for the entire cohort was 111 μl, and thus well above the previously suggested threshold level to select patients for shunting (42 μl). ASV was, however, not different in shunted and conservatively treated patients. While MWA was elevated in the surgery group, there was no correlation between ASV and parameters of pulsatile or mean ICP. Neither did ASV correlate with clinical NPH score or duration of NPH symptoms. There was found an association between ASV and ventricular volume before (R = 0.60, P = 0.004) and after (R = 0.73, P = 0.007) shunt surgery, and with aqueduct area (R = 0.58, P = 0.006) pre-surgically. After shunting, ASV declined from median 111 to 68 μl (P = 0.01) and ventricular volume declined from 137 to 105 ml (P = 0.001).

4.1.2 Peak to peak pulse pressure gradient (MRI-dP) (Paper 2) A total of 34 patients referred for clinical work-up of iNPH were enrolled, of these, 22 had the combination of technically successful PC-MRI from level C2 and over-night ICP monitoring, and was therefore finally included in the study. Of note, 13 patients from this cohort also participated in the study reported in Paper 1. Shunt surgery was performed in 17/22, whereas 16 (94 %) responded. No serious complications to shunt surgery occurred. Additionally, the study included four healthy control subjects.

MRI-dP was not different in iNPH patients and healthy controls and did not correlate with pulsatile or static (mean) ICP. Neither did MRI-dP differ between 43 patients with MWA above or below established thresholds for shunting. There were not found any systematic differences in HR at PC-MRI and ICP monitoring as demonstrated by the Bland-Altman plot. In this, the correlation between HR registered at PC-MRI and ICP monitoring was high (R = 0.71, P = 0.001). Nor was there any sign of influence on MWA from HR by means of association (R = 0.02). Importantly, over-night recordings of pulsatile and static ICP demonstrated large fluctuations over time, where the median coefficients of variation (CV) were 26 % and 128 %, respectively, and where particularly CV for static ICP ranged widely from 19 % to an extreme of 5600 %.

Figure 4. Trend plots demonstrating long-term fluctuations of a) pulsatile ICP (MWA) b) static ICP (mean ICP) and c) HR. An ICP single wave with its amplitude (dP) and rise time (RT) is presented in d) [178] (By license of Creative Commons Attribution 4.0 International).

4.2 Characteristics of CSF flow in iNPH

4.2.1 Net Aqueductal Flow (Paper 3) This study included the same 21 iNPH patients as described in further detail under 4.1.1 (Paper 1). Additionally, two healthy controls were included.

Net retrograde aqueductal flow (negative net ASV) was found in 16 of 21 patients (76 %) with median and range for the entire group being -5 μl (-175, 27), whereas flow was antegrade in both healthy controls, being 2 and 3 μl, respectively. When patients were categorized to high/low MWA groups, and where high MWA typically is considered as sign of reduced intracranial compliance, 14/17 (82 %) in the “high MWA” group had net retrograde flow, while this was observed in only 1/4 (25 %) in the “low MWA” group. Net ASV was also lower (more negative) by magnitude in the high MWA group (P = 0.025).

At PC-MRI performed after one year in 12 shunted patients (11 responders), net retrograde aqueductal flow was now observed in 5/12 (42 %), and net ASV had increased in 9/12 (75 %) patients (P = 0.002), who were all in the “high MWA” group.

Before shunting, the CSF formation rate in 21 patients was calculated to -0.56 ml/min (-12.78, 0.58) (median and range) and had increased after shunting in 12 patients to 0.06 ml/min (-4.51, 1.93) (P = 0.003).

4.2.2 Glymphatic MRI (Paper 4) The MRI contrast agent gadobutrol was utilized as CSF tracer in this study, which included 15 patients referred to OUH for clinical work-up of suspected iNPH, and eight reference patients with suspicion of CSF leakage (n = 7) and headache possibly related to a pineal gland cyst (n = 1). The reference cohort was significantly different from the iNPH patients with regards to gender (P = 0.01) and age (P < 0.001). Shunt surgery was performed in 13/15 iNPH patients with a beneficial shunt response in 13/13, whereas in the remaining two patients, shunting was desisted because of other co-morbidities. In the reference cohort, CSF leakage was found and confirmed by surgery in 2/8 (25 %).

There were no serious adverse events attributed to intrathecal administration of the contrast agents.

4.2.2.1 Subarachnoid and ventricular distribution and clearance of gadobutrol The CSF tracer gadobutrol distributed primarily along the main cerebral artery trunks (anterior- , middle- and posterior artery) in all study subjects. In iNPH, propagation of gadobutrol was found delayed along these arteries and other predefined SAS locations. Ventricular reflux and sign of transependymal migration of

46 gadobutrol was a common feature of iNPH, indicating redistribution of CSF to the ventricular spaces and periventricular brain tissue. Over the brain high convexities, and thus nearby the arachnoid granulations, gadobutrol enhancement was detected in only 3/15 (20 %) of iNPH patients and 3/8 (38 %) of reference patients at any time point. Clearance of gadobutrol from CSF was found delayed in iNPH, both with respect to of Max SU reduction and Clearance coefficient (SU/min) (P < 0.05).

4.2.2.2 Distribution and clearance of gadobutrol within brain parenchyma Pre intrathecal gadobutrol, parenchymal T1 SU was lower in iNPH than reference patients. Gadobutrol enhancement in pre-defined locations of brain parenchyma was observed in all study subjects and peaked overnight at time point 24 hours, but was more pronounced in iNPH patients in peri-Sylvian (subcortical white matter of inferior frontal gyrus) (P = 0.03) and periventricular (P = 0.008) locations (compared to the 4 p.m. scan at day 1). In these locations were also the highest associations between CSF and parenchymal enhancement noted (R = 0.85, P < 0.001 and R = 0.84, P < 0.001, respectively). As the peak parenchymal enhancement occurred at the final image acquisition (24 hours post contrast), calculation of parenchymal clearance parameters could not be achieved. No enhancement in the superior sagittal sinus was detected at any time points.

5. Discussion Mechanisms behind CSF flow and circulation are complex and matters of continuous debate. Investigations of the intracranial compartment have always represented a challenge to scientists, but new prospects for research have arisen with modern imaging techniques such as CT and, particularly, MRI. While the daily routine of MR imaging is mainly focused on structural imaging, and where CSF flow typically might be considered merely a source of artifacts, static imaging can never reveal the full complexity and dynamics of a system always in motion. PC-MRI may contribute to close that gap by its ability to quantify CSF movements. In this, PC-MRI plays a role in an ongoing development where subjective interpretation of radiological images is evolving toward increased use of quantitative measures to retrieve new biomarkers from images [179]. Quantitative imaging may render for increased reproducibility, better standardization and pooled analyses of larger patient cohorts than has been possible in the past. Initiatives in this direction should therefore be warmly welcomed; however, imaging biomarkers must be validated through testing in a clinical environment.

5.1 PC-MRI derived parameters of intracranial pulsatility At Oslo University Hospital, as one of few centers world-wide, over-night ICP monitoring is performed routinely as part of the pre-surgical work-up of iNPH patients. This practice is based on empirical observations as it has been demonstrated a beneficial shunt response in 9 of 10 patients with increased ICP pulsatility, while response is seen in only 1 of 10 subjects with normal ICP pulsatility [44, 103]. A drawback with invasive ICP is, however, an about 1-2 % risk of severe complications such as intracranial hemorrhage or infection [103, 104]. One aim of this thesis was therefore to compare PC-MRI derived surrogate markers of ICP pulsatility (ASV and MRI-dP) with measurements from invasive ICP monitoring.

5.1.1 General methodological considerations (Paper 1 and 2) Neither ASV nor MRI-dP compared with results of ICP monitoring. In subsequent sections, specific issues relevant to each of these PC-MRI methods will be addressed. Here, methodological considerations about PC-MRI study design and PC-MRI for CSF flow quantification in general, is discussed.

In the studies of both ASV and MRI-dP, comparisons of over-night ICP were made with day-time PC-MRI acquisitions, and the measured variables were thus not retrieved synchronously, and could be prone to diurnal variations. This potential source of error should be limited, as it has previously been demonstrated that contrary to static (mean) ICP, measures of ICP pulsatility, which are the main tool for selection of patients to surgery, do not differ on average between day- and night-time [44]. It cannot be excluded that synchronous measurements might had proven to compare better, but the relevance of this for clinical decision making would still be uncertain, as short-term ICP measurements are not considered representative for selection of iNPH patients to shunting. Both mean ICP and ICP pulsatility are subject to considerable temporal physiological fluctuations that are well demonstrated in long-term ICP recordings provided by others [180] and in Paper 2 (see Figure 4 in section 4.1.2). It is therefore not likely that PC-MRI derived markers of intracranial pulsatility, which are obtained within a few minutes, are representative for long-term pressure fluctuations.

Moreover, while PC-MRI was synchronized with heart beats (cardiac gating), CSF flow is also influenced by other factors, where respiration is probably most important. Following our studies, a later report demonstrated that respiration is the main determinant of CSF flow, and its effect is most pronounced at the aqueduct level, and particularly under the influence of deep inspiration [23]. During inspiration, there is retrograde (inferior to superior) CSF movement into the cranial cavity and lateral ventricles, while reversal occurs with expiration [181]. Furthermore, forced respiration promotes net upward CSF flow into the intracranial compartment and through the aqueduct and is counterbalanced by downward flow in spinal epidural veins [182]. Another study found the cardiac related component of CSF flow to be largest at aqueduct level, while respiration contributed, however, most to displacement (integral of velocity over time) [183]. Since cardiac gated PC-MRI acquires data over several minutes, respiratory phase contributions are unpredictable. A study utilizing PC-MRI with a temporal resolution of 50 ms (“real- time PC-MRI”) found that respiration effects may by averaged out by cardiac gated PC-MRI, but also demonstrated that cardiac gated PC-MRI may heavily underestimate flow velocity at the CCJ level [184]. Future MRI based measurements

49 of CSF flow velocity should therefore aim to incorporate the respiratory component of CSF flow.

The size of the cohorts was limited to 21 and 22 patients in the ASV and MRI- dP studies, respectively. Even though the study samples should be regarded representative for the source population (patients with clinically suspected iNPH under work-up for shunt surgery), and represent the largest clinical studies yet to compare invasive ICP measurements with PC-MRI, it is possible that studies on larger cohorts might have yielded different results. Also, lack of association between PC-MRI biomarkers and invasive ICP monitoring cannot be inferred to conclude that these measures are not interrelated in any manner. An assessment of these methods` sensitivity and specificity to predict favorable shunt response would in the end have been preferable; however, the high shunt response (16/17 in both studies) made such an assessment not meaningful.

5.1.2 Aqueductal stroke volume (Paper 1) Use of ASV to predict shunt response in iNPH is in part founded on empirical observations. In early history of MR imaging, a hyper dynamic flow artifact, indicating high CSF flow velocity and turbulence, was noted to be typical in hydrocephalic patients [126], and to possibly be an indicator of shunt response in NPH [127]. An important step towards quantitative imaging was taken when ASV above a certain threshold level was suggested to identify NPH shunt responders [40]. Later, clinical utility of aqueductal flow quantification has been reported in several studies [129-134, 185].

There is furthermore a theoretical foundation for a link between ASV and features of iNPH. One typical observation in shunt-responsive iNPH is increased ICP pulsatility [44], which is interpreted as sign of reduced intracranial compliance [43]. Compliance is the ratio of volume change (dV) to pressure change (dP), and when intracranial compliance is reduced, a small, temporary volume increase induced by one heart beat induces a disproportionate intracranial pressure increase. Compliance of a compartment therefore corresponds to the ability to accommodate a pressure increase induced by a volume fluctuation [123]. Pressure and flow pulsatility may change with disease throughout the body, but the intracranial compartment represents a unique environment for pulsations and a challenge to measurements of

50 pulsatility. Intracranial compliance is typically reduced when mean pressure increases [102], but also when compliance in the tissue itself is decreased, as for example with vascular disease and stiffening of arteries. The latter is most probably the case for the reduced intracranial compliance typically seen in iNPH, as mean pressure is typically within a normal range, or only slightly elevated. Another “endogenic” cause to reduced intracranial compliance and increased ICP pulsatility in iNPH may be hampered transfer of pulsations out of the cranium due to compromised venous or CSF outflow pathways [43].

In the theoretical concept given by Greitz [15], reduced intracranial compliance restricts arteries at the brain surface from expanding normally, and the pulsations are instead transferred downstream into brain capillaries, redistributing pulsations from being propagated to CSF at the brain surface, to the brain parenchyma. Brain motion has been found to be a contributing force to CSF pulsations [186]. As the brain is restricted from expanding outwards by the stiff cranium, brain pulsations act like a water hammer against the incompressible fluid within the ventricles, leading to increased aqueductal flow, and subsequently also ventricular enlargement due to collapse of compressible periventricular white matter [15]. The theoretical link between reduced intracranial compliance and increased ASV is thus established.

Association between ASV and ICP pulsatility was not established in Paper 1. This may, however, have been due to technical limitations of the PC-MRI acquisitions. Spatial resolution is one important factor, as smaller pixel size reduces partial volume averaging of adjacent stationary tissue outside the aqueduct. Compared to the pixel size used by other investigators of aqueductal flow measurements (reviewed by [187]), the one we used (0.6 x 0.8 mm2) may be considered moderate, and quite average. It has been proposed that previous successful use of ASV [40] may be attributed to a smaller pixel size, and also to the use of differentiated (including higher) VENCs to avoid flow aliasing [188]. However, the large aqueductal lumen area associated with hydrocephalus (median 14 mm2 in our study) should contain sufficient number of voxels to minimize effects of partial averaging in voxels at the periphery of the aqueductal lumen, and this large area should also be expected to reduce the degree of error that may be inherent with manually defined ROIs [189]. A measurement error of less than 10 % has been demonstrated when the image resolution is 4 pixels per diameter length of the 51 aqueduct [128]. Partial averaging was also minimized by assuring the imaging plane was manually positioned perpendicular to the aqueduct and also by a relatively small slice thickness of 4 mm. Furthermore, larger pixel size renders for better signal-to- noise ratio, as is also expected from our use of 3 T magnetic field strength, as opposed to 1.5 T used in [40], where retrospective assessment of PC-MR images from the publication demonstrates a signal-to-noise ratio that may be considered as critically low, in spite of spending 14 minutes scan time. Further, while aliasing is unwanted in PC-MRI acquisitions and minimized by using high VENC, this strategy will necessarily decrease signal-to-noise ratio and could limit detection of lower velocities at the periphery of the aqueduct, given a laminar flow pattern [190], and may in the end induce error in measurement of total flow [21]. Therefore, low-VENC imaging with post-scan aliasing correction may not be regarded as inferior to high VENC scans.

Moreover, while partial averaging effects influence velocity measurements along the periphery of the aqueduct, Ragunathan et al. [187] recently demonstrated that radiofrequency saturation also contributes to velocity measurement bias in any flow voxel containing spin velocities below threshold for substantial saturation effects (critical velocity). In this, slower spins are saturated to a greater extent than the faster, returning an overestimation of velocity. A change of saturation effect by an increase of flip angle from 10 to 30 degrees increased ASV significantly (by 12 %). Furthermore, critical velocity is dependent on slice thickness and repetition time. An extensive review of PC-MRI based aqueductal flow studies [187] showed that the acquisition parameters which determines critical velocity, as well as spatial resolution and number of cardiac phases, are heterogeneously applied throughout different reports, and in no studies were critical velocity the same. Our PC-MRI technique was estimated to benefit from a critical velocity of 0.3 cm/s (indicating low susceptibility to saturation effects), which also was among the lowest they assessed (range 0.1 - 5.0 cm/s) [187].

An even more fundamental explanation as to why aqueductal flow-based pulsatility (ASV) and pressure based intracranial pulsatility (MWA) did not correlate in our study may be attributed to their expression of different features of the intracranial compartment, in spite of theoretical considerations of their possible close relationship as elaborated above. Instead of being associated with ICP pulsatility, clinical NPH- 52 score and symptom duration, ASV correlated with ventricular size and aqueduct area, and declined after shunting in parallel to decline of ventricular volume. These are discouraging findings, as a CSF flow derived imaging biomarker should add clinical useful information of more than what is already retrievable from anatomical images. The results are also contradictive to the proposed ability of ASV to discriminate atrophy from no atrophy in patients with ventricular enlargement [188, 191] and to assess the natural course of untreated iNPH [192] and shunt malfunction in treated iNPH [135]. Our findings are, however, in line with a study demonstrating a strong association between ASV and ventricular volume as well as aqueductal area [193]. Furthermore, the data add to several other reports where clinical utility of predicting shunt response using aqueductal flow measurements was not found [136-139, 180, 194].

In conclusion, the observations made in Paper 1, and the challenges with ASV measurements discussed above, suggest the factors to determine which iNPH patients will respond to surgical shunting, are unlikely to be contained by this single CSF flow parameter. It was beyond the primary study objective of this observational study to perform an exhaustive search for any factor that in the end may contribute to ASV size. As ventricular volume and aqueduct area were associated with ASV, we may hypothesize these act as confounders, however, association by itself does not imply causality. A future study of ASV should allow for a sufficient number of patients to assess sensitivity and specificity for prediction of shunt response, and may also include a multivariate statistical method such as regression analysis to determine the relationship between ASV and other variables.

Finally, it could be argued that the non-invasive diagnostic approach (ASV) should always be preferred at the cost of the invasive alternative (ICP) [188]. However, this statement is only valid as long as the non-invasive test has comparable clinical utility to that of the invasive. This would be of particular importance in iNPH, where shunted patients quite frequently experience complications with their shunts [195], and one main benefit of invasive ICP monitoring is the high ability to decide which patients would not respond and therefore should not be exposed to the risks inherent with shunting [44].

5.1.3 Peak to peak pulse pressure gradient (MRI-dP) (Paper 2) While ICP pulsatility can be a useful marker of intracranial compliance (dV/dP), a method to assess intracranial elastance (dP/dV), the inverse relationship of compliance, was proposed almost two decades ago [145]. The dV (from here of referred to as MRI-dV) describes the net volume change of intracranial constituents during one cardiac cycle, whereas dP (from here referred to as MRI-dP), is the CSF peak to peak pulse pressure gradient measured at the upper cervical spinal canal. Our first ambition was to calculate elastance (MRI-dP/MRI-dV) in iNPH. The clinical feasibility of the method, however, turned out to be poor. Out of 34 enrolled iNPH patients, measurements deemed to be of sufficient quality for MRI-dV and MRI-dP combined were successful in eight patients only, particularly as MRI-dV could be calculated in merely 9 of the 22 patients with technically adequate data for MRI-dP analysis. Main challenges to the investigation of MRI-dV included the inability to detect all vessels of interest, difficulties with delineation of vessel borders (segmentation), and vessels traversing the imaging plane in an oblique fashion. For the latter, it has been shown that error of vessel volumetric flow measurement increases with degree of misalignment to the perpendicular plane [196]. In an idealized situation, all vessels of interest run in parallel; however, this is often not the situation in the elderly population, where arteriosclerotic degeneration typically is accompanied by a tortuous course of vessels.

Furthermore, PC-MRI was performed at level C2 for the vessel and CSF flow analysis, which is typically the closest level to the intracranial compartment where the assumption of the spinal canal as a rigid, cylindrical tube can be made when estimating MRI-dP. At this level, vascular wall pulsations can be significant, and we were not in possession of software tools for pulsatility based segmentation of vessels [197]. Neither did we have the tools to obtain simultaneous measurements of MRI-dV and MRI-dP (“dual VENC”), which would have been highly preferable given the short- term fluctuations in CSF flow and intracranial pressure discussed previously. Net intracranial volume change (MRI-dV) can be expected to be in the order of 0.1 % [145]. In the end, any further efforts to estimate MRI-dV, and thus elastance, were abandoned.

The gain of narrowing down the study objective to investigate MRI-dP only is the opportunity for a direct comparison of this non-invasive marker of intracranial 54 pressure change with its invasive counterpart, pulsatile ICP (MWA). The linear relationship between pulsatile ICP and MRI-dP has previously been reported in a baboon experiment [145], and our hypothesis was thus a correlation between these two measures. To ensure optimal analysis of PC-MRI data, a skilled mathematician performed the calculation of MRI-dP, as well as the preceding CSF flow aliasing correction, and after ROIs had been defined in PC-MR images by a trained neuroradiologist. The methodologies are described in detail in Paper 2.

In extension of the previous methodological considerations made in section 5.1.1, the validity of MRI-dP may be further questioned. As MRI-dP is calculated at level C2, outside the intracranial compartment, where geometrical assumptions (of a cylindrical tube) can be made more easily, the MRI-dP constitutes a surrogate parameter for invasively measured pulsatile ICP. Lower pressure values are, however, expected at level C2 than in the posterior cranial fossa, and pressure gradients may become steep in the cervical canal as it narrows (tapering effects) [198]. Moreover, assumptions about uniform, or laminar, CSF flow within the spinal canal are questionable, as flow patterns may be influenced by bidirectional flow, non- uniform distributions due to fine anatomical structures, and in-plane flow [198]. This may introduce error to MRI-dP estimates derived from 2D PC-MRI, where flow velocities within a ROI are averaged and in-plane flow is neglected. In this respect, 4D PC-MRI, in which time adds the fourth dimension to a 3D volume, has shown promise for better evaluation of complex CSF dynamics in the cervical spine [199]. This methodology has not been explored by our group.

The iNPH patients and healthy controls, respectively, were imaged with different hardware due to an upgrade from the Philips Achieva to the Philips Ingenia platform, which also included change of head coils. This could represent a limitation to the comparison between the two groups. There are, however, no reports claiming that the success of MRI-dP is hardware dependent. It is therefore discouraging when MRI-dP in healthy, younger controls did not differ from patients with proven pathological ICP pulsatility. The reliability of multiple measurements in healthy controls was estimated to be high with intra-class correlation coefficient (ICC) (95 % CI) = 0.88 (0.59, 0.99), suggesting measurement error was limited. Moreover, high reliability renders for better statistical strength to detect differences between groups [200]. The inability to detect a difference may be related to sample size. However, the 55 validity of MRI-dP may also be questioned. A trend towards an inverse correlation between MRI-dP and ROI area was noted (R = -0.32, P = 0.15), and a similar observation has also been reported previously [145]. Among our healthy controls, the subject with the smallest CSF area also had the highest MRI-dP, and larger than several of iNPH patients. It seems therefore justified to speculate that ROI area may confound MRI-dP measurements.

5.2 Characteristics of CSF flow in iNPH

5.2.1 Net retrograde aqueductal flow and supra-aqueductal reflux of gadobutrol (Paper 3 and 4) Net ASV is a very delicate parameter and susceptible to error in measurement technique. In addition to technical limitations inherent with calculations of ASV, an important challenge with net ASV is the subtraction of larger quantities of flow in opposite directions, where the signal might get lost in noise. Net ASV assessment was neither validated by calibration using a phantom device, nor was any baseline shift at adjacent stationary tissue corrected for. In particular the latter may be critical; when average velocity of bidirectional flow is close to zero, any phase offset errors due to eddy currents might represent an inherent source of error in a scanner. It was beyond the scope of this clinical study to explore the extent of such error. Nevertheless, the need for correction of background velocity offset has been emphasized, not at least as errors are dependent on the specific settings of the equipment used at different sites [201]. In a previous study on net flow, contribution of phase offset error was considerable, but did not change the conclusion of retrograde net flow in infants with external hydrocephalus [202].

These limitations in our observations of net retrograde flow in iNPH are, however, counteracted by some factors. First, it may be difficult to find an entirely non-moving nearby region for baseline correction as brain parenchyma also pulsates in synchrony with the heart [203, 204]. By default, baseline correction using a phantom does not improve flow quantification [205]. In one study, the background offset correction itself, using ROIs from adjacent stationary tissue, differed depending on region, while ASV remained unchanged whether such correction was applied or not [189].

Second, the MR scanner was equipped with a local phase correction filter provided by the manufacturer to subtract the background offset caused by eddy currents [206]. Moreover, the study included two healthy controls that were imaged in the same scanner as patients, and in both, net antegrade flow was detected. Even though a higher number of controls would have been desirable, any significant baseline shift from phase offset error should have been expected to become visible in these healthy subjects. Twelve of the shunted patients also served as their own controls as they were scanned with PC-MRI 12 months after surgery. The tendency was clear; net ASV increased in the antegrade direction in 75 % of shunted patients. This might be interpreted as a sign of restored intracranial compliance after shunting. Net retrograde aqueductal flow was also found significantly more prevalent among subjects with pathological ICP pulsatility indicative of reduced intracranial compliance (82 %) than among subjects with ICP pulsatility below threshold (25 %). Several studies have reported retrograde aqueductal flow at PC-MRI in recent years [125, 130, 207-210]. Evidence of supra-aqueductal ventricular reflux of intrathecal gadobutrol and gadobutrol migration through the ventricular wall ependyma (Paper 4), further strengthens these observations. Ventricular reflux in iNPH is also in line with previous radioactive isotope cisternography studies [164, 165] and early observations made by Walter Dandy [211]. Finally, while magnitude of aqueductal flow (ASV) may be influenced by respiration [23], and in the short term even influence net flow direction [182], direction of net aqueductal flow may in the end be more of a parameter expressing a fundamental property of the “system”. If so, the influence of temporary net CSF flow changes from respiration might be reduced by averaging within the minutes` duration of a PC-MRI scan. Net flow of CSF into the ventricles has implications for the understanding of principles behind CSF circulation. First, there have to be additional escape routes for CSF other than the arachnoid granulations along the dural venous sinuses. Second, one could infer there should be a capacity for production of CSF outside the ventricular system, either from the surface of the brain and/or spinal cord. Indication of this also derives from a rodent study, where AQP4, a regulator of water homeostasis around brain capillaries, acted as a main regulator of water influx into CSF, and not Aqp1 in the choroid plexus [212].

From the ventricular system, CSF might escape via the choroid plexuses, through ventricular ependyma with further resorption by the blood circulation, and/or through the ependyma to mix with interstitial fluid to further efflux along routes outside the brain-blood barrier, such as the perivascular spaces.

CSF resorption by the choroid plexus was first suggested almost one hundred years ago by Foley in 1921 (reviewed by [213]) and later in hydrocephalic infants [214], but has been contradicted by others [215]. The choroid plexus can, however, remove drugs such as penicillin and other endogenous and exogenous solutes from the CSF [216]. In a recent review [217], it was concluded that while the balance of evidence is against such a route of CSF resorption, it has not yet been excluded, and further investigation is required. We are presently in progress with a study where any gadobutrol enhancement in the choroid plexus (and other regions) after intrathecal administration will be compared with levels of gadobutrol in blood samples obtained at multiple time points.

Paper 4 provided evidence of CSF resorption through the ventricular ependyma. Even though the egress of CSF tracer (gadobutrol) is necessarily not synonymous to escape of CSF, the assumption is reasonable, because the water molecule is of much smaller size than gadobutrol, and gadobutrol is highly hydrophilic. Moreover, periventricular hyperintensity on FLAIR-images, indicating increased parenchymal water content, was comparable with areas of gadobutrol enhancement in periventricular white matter. Image resolution at MRI did not allow for an assessment of clearance of CSF tracer from the parenchyma in further detail. Of note is that clearance routes for water and substances are necessarily not the same. A growing number of researchers agree that interstitial fluid and CSF are mainly transported and reabsorbed across the blood-brain barrier by simple diffusion and vesicular transport, or by co-transport along with substrates [14]. It has also been demonstrated that vascular basement membranes surrounding CNS capillaries play a key role in the bulk transport of fluids and solutes within the CNS [218]. Moreover, CSF entry into the brain is facilitated by pulsations along paravascular spaces surrounding penetrating arterioles [26, 31, 218], allowing for exchange between CSF and interstitial fluid. Brain interstitial fluid thus represents a mix of solutes and fluid derived from blood, tissue metabolism and CSF [219].

How fluid and solutes are further cleared from the brain, is a matter of ongoing debate. Controversy currently exists in regards to whether the paravascular clearance pathways out of the brain, acting as a substitute for regular lymphatic vessels, are along arteries [28, 30, 220] or veins [26, 31, 159]. This controversy extends further into the debate of how brain solutes like amyloid-β are transported to cervical neck lymph nodes, where clearance along veins into the recently discovered true lymphatic vessels of dura walls [221, 222] has been proposed [221]. This is quite contrary to those who argue for clearance along the retrograde direction of arteriolar smooth muscle cells and further along the internal carotid artery [30].

This body of evidence implies that large water fluxes may take place continuously within the brain. Although the neurovascular unit allows for bidirectional water transport between blood, interstitial fluid and CSF, this transport may also generate net flux [14], and the brain surface could be a net producer of CSF to the subarachnoid compartment. Thus, while net retrograde aqueductal flow is rather incompatible with historical models incorporating the choroid plexus to be the only, or main, source for CSF production, as the subarachnoid compartment in the end would be depleted; modern knowledge of brain water exchange renders net retrograde aqueductal flow quite possible. This may also be consistent with our calculated aqueductal flow rates. In some cases these exceeded by far previously reported CSF production rates of about 0.3 ml/min in studies using ventricular infusion of non- diffusible reference material to assess bulk flow (reviewed by [213]) and in PC-MRI studies of healthy subjects [223]. Notably, invasive indicator techniques for estimating CSF formation rate used substances designed not to be absorbed into capillaries. Weed, who formed the basis for the principle understanding of CSF absorption at the arachnoid villi, excluded the use of solutions with “diffuse tissue staining” [10], i.e. substances that would be resorbed in brain parenchyma. Our observations from utilizing gadobutrol as CSF tracer suggest exchange of water and substances (indicated by tracer distribution) within the brain is more important than at the arachnoid villi, as presence of tracer at the upper brain convexities was found in a minor proportion of iNPH or reference subjects, and then typically on late scans.

Obstructed CSF flow in iNPH is indicated by several observations made in the studies of this thesis, but obstruction may well occur at another level than the arachnoid villi. First, propagation of gadobutrol at the brain surface was delayed in 59 iNPH compared to reference subjects. Furthermore, clearance of gadobutrol was delayed. From the Monroe-Kellie doctrine it follows that an increased intracranial volume load from accumulated CSF would be accompanied by increase pressure and decreased intracranial compliance, expressed in its turn by increased intracranial pulse pressure [43]. Increased pulsatile ICP (MWA) is indeed a characteristic feature of iNPH, and particularly shunt-responsive disease [44, 103]. The large proportion of net retrograde aqueductal flow (82 %) among patients with high MWA (Paper 3) suggests that net CSF flow into the ventricles might be a feature of obstructed flow within the brain paravascular compartment with redirected flow along pathways of less resistance through the ventricular ependyma.

Reflux of CSF tracer into the supraaqueductal ventricular system may be regarded as compelling evidence of net retrograde aqueductal CSF flow. A main contribution to the phenomenon may, however, be dispersion. Dispersion is the combined effects of diffusion and advection, the latter here being represented by the to-and-fro pulsatile movement of CSF through the aqueduct, typically revealed by PC-MRI. Pure diffusion of a highly hydrophilic solute (such as gadobutrol) is slow. Advection moves the solute much faster, but would without any component of diffusion confine the solute to the same region. Dispersion, however, results in net flow of a solute, such as a CSF tracer, even though there is no time-averaged net flow of liquid [224] (see Figure 5). In iNPH, with typically enlarged diameter of the aqueduct and increased ASV, the potential for dispersion of CSF tracer into the ventricles is clearly present.

Figure 5. Illustration of differences between solute transport enabled by diffusion, advection and dispersion in the absence of net flow. The temporal evolution of the solute concentration (red: high, blue: low concentration, white: no solute) is illustrated at equally spaced time points, ti=0…5 [224] (reprinted with permission obtained through Copyright Clearance Center`s Rightslink®) (License No.4296910992147).

However, even though a CSF tracer thus might be transported retrogradely into the ventricles by dispersion, and without any net CSF flow, this does by no means rule out the possibility of retrograde net aqueductal CSF flow. On the contrary; zero net flow has never been scientifically confirmed. Given the vast amount of fluid being continuously exchanged within the brain across the neurovascular unit, the rationale for assuming zero net aqueductal CSF flow should rather be questioned. Traditionally, net retrograde net aqueductal flow has typically been considered as sign of technical error in calculations of ASV and therefore subtracted [40]. Furthermore, dispersion alone may hardly explain why CSF tracer migrated early into the supra-aqueductal ventricles, and persisted to stay within, far more frequently in iNPH patients than in control subjects. We have also experienced hydrocephalic cases without prominent supra-aqueductal reflux, and vice versa. In the end, it could therefore be hypothesized that signs of net retrograde aqueductal CSF flow, either at PC-MRI or when utilizing gadobutrol as CSF tracer, may serve as predictor for reduced intracranial compliance, and thus also for shunt-responsive iNPH.

Further corroborating evidence of net retrograde aqueductal flow would also have implications for how mechanisms behind ventricular enlargement in iNPH are 61 understood, since a prerequisite for such flow would be a net pressure gradient into the ventricular system. Even if small, such a pressure gradient could be expected to exert a force upon the ventricular wall during every heart beat, and may be sufficient to produce ventricular enlargement [225, 226]. Reduced ventricular size after shunting, as demonstrated in Paper 1, could further be explained by the reversal of a positive pressure gradient over the ventricular ependyma.

5.2.2 Glymphatic MRI (Paper 4) Brain parenchymal enhancement after utilization of gadobutrol as CSF tracer may provide evidence for the existence of a glymphatic system in man [162, 163]. The image resolution of MRI is limited by voxel size of typically 1 mm for a T1 weighted volume scan, while transport of water and solutes occurs in paravascular and interstitial spaces with size at the order of μm. The interpretation of our study findings therefore leans considerably towards observations made previously in animal studies. In a rat study, where MRI contrast agent (Gd-DTPA) was coinjected with a fluorescent tracer (Tr-d3) of comparable size [159], both substances distributed with a similar pattern in the brain. Epifluorescent microscopy of Tr-d3 allowed for imaging at a much higher resolution than MRI, and demonstrated that the fluorescent tracer spread from paravascular spaces of penetrating arteries and into the brain interstitial space. A direct connection between the subarachnoid space and paravascular compartment of penetrating parenchymal arteries has also been confirmed in rats by other investigators [227], which confirm exchange of substances between cerebrospinal and interstitial fluid.

It has from modeling studies been argued that pulsations cannot explain CSF net flow within interstitial spaces (typically referred to as convective flow) [228]. Moreover, models suggest that diffusion is more likely to explain clearance of solutes from the interstitial space than convective flow [229]. Pulsations could rather contribute to a dispersion effect (combination of diffusion and advection) in the paravascular space, but without net flow [224]. Paravascular compartment diameter has been suggested to be at least a factor of 100 times larger than in the interstitial space between cells, and with intrinsic hydraulic permeability of the paravascular space to be at least 10 000-folds higher than the cortical interstitial space [230]. It is therefore likely that the initial parenchymal enhancement from gadobutrol derives mostly from enrichment of paravascular spaces. Moreover, the same experimental 62 design which led to the conclusion of a glymphatic system was recently reproduced. The study provided evidence that could explain movement of solutes in the brain interstitial space by diffusion only, and coupled diffusion in the interstitial space to convective or dispersive flow in the paravascular spaces [231]. An assessment of which forces that best account for the rate of gadobutrol propagation in human brain parenchyma is currently being examined by members of our research group.

The observation that gadobutrol enhancement in CSF preceded, and correlated with, brain parenchymal enhancement, clearly indicates that contrast agent entered the brain directly from the surface. Penetration of gadobutrol into the human subcortical white matter extends knowledge from rodents, in which contrast enhancement in deep white matter could not be detected at MRI [159]. Furthermore, these in vivo observations directly contradict previous electron microscopy studies of fixed human cortex, indicating that cortical perivascular spaces are merely potential spaces that are not functional in human brain [218, 220]. Brain parenchymal enhancement after administration of a CSF tracer also opposes the perception that drugs injected into the subarachnoid space have very limited access to the brain [232], but is well in line with a recent report showing widespread distribution of antibodies along perivascular routes into the entire brain after intrathecal administration [233]. Endogenous antibodies (IgG) are abundant in CSF, and perivascular trafficking of IgG and other proteins may have relevance for cerebral amyloid angiopathy as well as autoimmune diseases [233]. This further questions the immune privilege of CNS, previously also challenged by the discovery of true lymphatic vessels in dural sinus walls for drainage of CSF and solutes [221, 222]. Drainage to neck lymph nodes would allow for antigen presentation from CNS to the immune system and a subsequently evoked systemic immune response.

The ability of macromolecular substances to propagate over distances in the range of millimeters, and even centimeters, exceeds by far the range over which diffusion within synapses and between cells occur, and might favor the existence of so-called “volume transmission”, or “paracrine transmission” in the brain. Such information exchange between brain cells, other than what is mediated by synapses and the neuronal circuitry, has previously been characterized to a very limited extent and remained an elusive concept [27]. More recently, this has been integrated

63 conceptually with the CSF compartment and suggested important for learning and memory [234].

5.2.2.1 Delayed clearance of CSF tracer (gadobutrol) in iNPH An adaptive response for up-regulation of AQP4 aimed at clearing excess brain fluid was shown in a recent study [235]. In periventricular edema of hydrocephalus, AQP4 expression is increased [236]. A main histopathological finding in iNPH is, however, astrogliosis and decreased presence of AQP4 and its anchoring protein dystrophin 71 at astrocytic end feet [65]. The ability to up-regulate AQP4 therefore seems compromised in iNPH. Given the fact that large water fluxes take place continuously throughout the brain parenchyma across astrocytic end feet [24], and that the paravascular space is continuous with CSF [227], CSF flow obstruction in iNPH may therefore be hypothesized to occur at the astrocytic end foot level. Water may enter the brain parenchyma independently of AQP4, but exits the brain through AQP4 [237]. Reduced escape of water through AQP4 is expected to increase extracellular water content, as well extracellular volume fraction, which has indeed been demonstrated in AQP4-deficient mice [238]. Findings of increased ADC [157] and decreased T1 signal (Paper 4) in parenchyma of iNPH patients are well in line with these observations at the cellular level.

Within the glymphatic concept, convective (net) flow of water through the interstitial space is the stream that carries cellular waste products towards the perivenous space. While water molecules may enter the interstitial space through AQP4 channels, larger molecules, such as gadobutrol, must pass between inter end- feet gaps of adjacent astrocytes [27, 224]. It has been described that AQP4 deletion or depolarization may induce changes in astrocyte end foot morphology [239] and thereby induce reduced width of inter end-feet gaps and deteriorated permeability for solutes. This was also predicted from a modeling study [224]. Arguably, reduced clearance of gadobutrol in iNPH should be considered not more than an indirect sign of AQP4 loss and reduced capacity of water transport through the brain. Clearance rate of gadobutrol may, however, be representative for clearance rate of macromolecules with similar properties [159]. One such molecule may be Aβ, which is accumulated in the brain of AD and iNPH patients [85]. Clearance of Aβ has been shown dependent of AQP4 status [31].

An important role of vascular pulsations is suggested by the delayed propagation of gadobutrol along leptomeningeal artery trunks. Furthermore, gadobutrol enhancement at 24 hours signs was most pronounced in the parenchyma being closest to a large surface artery (inferior frontal gyrus near branches of the middle cerebral artery). Brain compartment paravascular flow has previously been shown dependent on arterial pulsations [26, 31, 159, 240, 241], most likely because pulsations are important for fast paravascular transport by dispersion [224]. In iNPH patients, reduced intracranial compliance, accompanied by restricted expansion of arteries residing in the subarachnoid CSF compartment, may therefore be part of a viscous cycle: Obstruction to CSF flow at the astrocytic end foot level may be a cause of reduced intracranial compliance and thus restricted artery pulsations, which in its turn may reduce paravascular and glymphatic flow.

The observations made in this study may have profound implications for understanding dementia in iNPH-patients. Should compromised glymphatic function play a major role in clearance of toxic substances such as Aβ [31] and tau [242], and clearance of these be comparable with that of gadobutrol, gMRI could serve as a pre- clinical test to measure susceptibility for dementia even before the deposition of pathogenic substances in toxic amounts occurs [243]. Variations in the APQ4 gene has proven to modulate progression of cognitive decline in AD [244], and perivascular AQP4 localization is associated with AD status and Aβ burden [245]. One issue that though needs to be better established is to what extent paravascular routes contribute to Aβ clearance in humans compared to transendothelial transport from brain to blood. In mice studies, there seems to be conflicting evidence, as Aβ transport both via the glymphatic pathway [31], as well as via the brain-blood-barrier [246], have been reported to dominate. Blood-brain-barrier dysfunction has further been shown associated with AD [247], but the role of such dysfunction in AD pathogenesis still seems elusive.

It should be underlined that assessment of gadobutrol clearance by addressing T1 signal intensity change over time does not allow for assessment of clearance in terms of absolute quantities, as T1 signal increase is not linearly associated with increase of contrast agent concentration. Moreover, direct comparison of T1 signal intensities between time points can be flawed due to uneven image scaling between scans, and use of signal units normalized to other reference

65 tissue would have been preferable. However, the enhancement of the entire intra- dural compartment renders for few suitable locations to serve as reference. An extra- dural ROI in the superior sagittal sinus demonstrated no detectable levels of circulating gadobutrol in blood that might have confounded measurements, and revealed no significant change in baseline of image grey scale. Presently, we aim at developing an extra-cranial reference device, where several containers of different gadobutrol concentration might allow for a more direct quantification of gadobutrol amount in brain tissue. Such a device could further be validated against methods for directly quantifying T1 at MRI (T1-maps).

While gMRI revealed reduced clearance of tracer substance (gadobutrol) in the iNPH cohort, it could not be established to which proportion reduced clearance was affected by age and/or iNPH disease, as iNPH patients were significantly older, and age may by itself be a cause of reduced glymphatic clearance [63]. As the most profound parenchymal enhancement was observed at over-night scans, which is remarkable, there may be a possible role of sleep for glymphatic function, as previously suggested [61]. However, since we were unable to perform MRI scans between the afternoon and next morning, such a role of sleep should not be considered established yet.

In the clinic, intrathecal gadobutrol is used “off-label” in single cases after the informed consent of a patient. For this reason, we did not apply for permission to study age-matched, healthy controls. For safety purposes, all patients were closely monitored in the neurosurgical department 24 hours after intrathecal gadobutrol administration, and were also invited to clinical follow-up and MRI after 4 weeks. No serious side-effects attributed to intrathecal gadobutrol have been observed, and no signs of gadobutrol retention in the brain parenchyma have been detected.

5.3 Experiences with publications of negative results This thesis derives from studies reporting both negative (Paper 1 and 2) and positive results (Paper 3 and 4). It is somewhat striking (but by no means conclusive) that the manuscripts reporting positive results were accepted for publication in the first journal they were submitted to, while those reporting negative results faced far more resistance among reviewers and editors. Paper 1 about ASV was resubmitted only once to another journal than it was originally intended, but was published along

66 with a negative commentary [188], to which we replied [248]. Paper 2 was rejected by two journals; the third submission process lasted for 13 months before acceptance was granted. We first submitted Paper 2 to a radiology journal of high visibility, and where a previous paper of related content, but with report of positive findings, had been published [147]. The written response from the editor-in-chief was the following:

«…On a more personal note, please do not take this letter as a negative assessment of your work. Your work appears very well done, and I highly suspect that your conclusions are very much correct. However the rejection comes due to the large number of high quality papers (of which I would consider yours one) that the journal receives, the limited number of papers accepted, and a long standing policy of mine generally to not consider papers with a 'negative' result - i.e. showing that a technique does not work - as you have so effectively done. I would encourage you with your work, but feel that [this] is simply not the correct venue/journal.”

This policy to generally not consider papers with negative results for publication is scientifically problematic and will inherently lead to publication bias. It is known that studies reporting a statistically significant result (P<0.05) are more likely to be published, and published sooner, than studies that do not demonstrate statistical significance [249]. Negative results should not be mistaken for negative research, and according to Karl Popper (1902-1994), true progress in scientific knowledge goes through the method of falsification rather than verification, or “enlarging the graveyard of falsified hypotheses” (reviewed by [250]). The importance of publishing negative results is further underlined by the fact that most published research within emerging fields of knowledge cannot be reproduced [251] and thus not be generalized for wider use. One may assume that public sponsors of research, as well as the public in general, expect scientific results to be reported in a neutral fashion, whether positive or negative. Signs of public mistrust to scientific research are already emerging [252].

6. Conclusions This thesis provides new insights into the characteristics of CSF flow in iNPH patients. The conclusions derive from a primary aim to study short- and long-term CSF flow characteristics and its pathologic alterations in pre-surgical iNPH. Quantifiable imaging parameters from PC-MRI with potential clinical benefit (ASV and MRI-dP) were explored, and the studies add new information to previous reports by comparisons to invasive monitoring. Furthermore, we assessed long-term CSF flow characteristics in iNPH by utilizing an MRI contrast agent as CSF tracer (gMRI). In vivo CSF tracer studies have not been performed in humans before.

Patients who underwent shunt surgery, benefitted from a high response rate (94 %), which is in line with previous experience using pre-operative ICP pulsatility assessment for selection of patients to surgical treatment. We could not, however, demonstrate clinical value of the CSF flow based pulsatility parameters retrieved from PC-MRI. In iNPH, ASV was typically increased, and declined after shunting, but was not associated with ICP or symptom scores, and seemed rather to be associated with ventricular volume. MRI-dP was not able to separate iNPH patients from healthy control subjects, nor was MRI-dP associated with pulsatile ICP in iNPH.

A typical feature of iNPH was signs of net retrograde CSF aqueductal flow, which was more evident in iNPH patients with pathologic ICP pulsatility and a beneficial response to shunting. A pivotal role of intracranial pulsations was further suggested by observations of contrast agent being propagated along large intracranial artery trunks of the subarachnoid space, but with a delay in iNPH compared to reference patients. The thesis also provides evidence to support the existence of a human, CSF pulsatility driven, brain-wide pathway for paravascular transport and clearance of solutes. This pathway has been denoted the glymphatic system and has previously been described in rodents only. Signs of reduced glymphatic clearance were found in iNPH, introducing a possible link between restricted pulsations and compromised clearance of neurotoxic brain metabolites, and thus a possible mechanism behind iNPH disease pathogenesis and dementia. The observed CSF flow characteristics may also shed new light on mechanisms behind ventricular enlargement in iNPH, and may further question previously established concepts of CSF production and resorption.

7. Future developments The evolvement from visual assessment of the hyper dynamic aqueductal flow artifact in T2 weighted images to PC-MRI derived measurement of ASV illustrates well ongoing efforts to transform radiology from a primarily subjective discipline to a more quantitatively oriented one. Quantifiable imaging biomarkers may render for more objective and reproducible patient data; however, these markers must also be clinically valid. In this respect, the experience with ASV and MRI-dP in this thesis is discouraging. Future studies should search for better standardization and take into account the influence of respiration during image acquisition. Moreover, reports from larger study cohorts are limited, as well as studies reporting diagnostic accuracies for clinical outcome variables. For ASV in particular, technical limitations might be mitigated by reducing pixel size, especially at higher magnetic field strengths, and imaging protocols should be harmonized in order to avoid bias from different acquisition parameters. Minimizing background velocity offset errors are warranted for any PC-MRI protocol, and especially when small CSF quantities are to be measured. In the end, whether these steps may enable PC-MRI to embrace the complexity of CSF flow, and thereby allow for more robust measurements, remains to be shown.

The future prospects of gMRI are yet uncertain, but the potential of the method justifies further studies. Presently, there is no other competing imaging method to measure the ability of the craniospinal compartment to clear itself from cellular waste products. Accordingly, gMRI represents an entirely new imaging approach to diagnose clinical and pre-clinical neurodegenerative disease. As we have shown that an MRI contrast agent passes from the subarachnoid space to the parenchymal extra-vascular compartment, the method might also have potential to visualize extravascular CNS disease. Until now, intravenously administered contrast agents have been excluded from this space by the intact blood-brain-barrier. In further developments, MRI contrast agents may be conjugated with other molecules having the ability to target disease in the extravascular compartment directly. None of these indications for intrathecal use of MRI contrast agents are, however, established, and the concern for retention of gadolinium in tissue might be a limitation for broad use until all safety issues have been properly addressed. An alternative approach might be substances such as iodinated contrast agents, which are approved for intrathecal

69 use, and also expected to be confined to the extra-vascular compartment and cleared by glymphatics. Measuring clearance rate from the craniospinal compartment by retrieving samples from blood or urine at certain time points might therefore be feasible.

The information derived from gMRI is complex and constitute T1 signal changes that are detectable from the intracranial compartment, sometimes subtle, and at multiple time points on a large time scale. Datasets from gMRI would therefore advantage from standardized image acquisitions and normalization of signal units against an extra-cranial device. Isometric voxels from volume scans allow for optimal image alignment and co-registration in dedicated post-processing software. Furthermore, anonymized datasets may be uploaded and transferred to a web-based platform and into an anatomical coordinate system to create data from large cohorts of individuals. Gathering of “big data” provides for the opportunity to compare individuals against larger patient cohorts than can be retrieved at any single center. Also for research purposes, subtle changes and complex patterns of signal change would be easier appreciated, and interpreted, when data are gathered in large clusters. Diagnostic imaging is probably yet in the slightest beginning of a transformation where utility of computers shift from rule-based diagnostic algorithms (computer aided detection) towards deep machine learning (artificial intelligence). Big data from gMRI should represent no exception to imaging procedures that may benefit from this development.

References

1. Hajdu SI. A note from history: discovery of the cerebrospinal fluid. Annals of clinical and laboratory science. 2003;33(3):334-6. PubMed PMID: 12956452. 2. Deisenhammer F. The History of Cerebrospinal Fluid. In: Deisenhammer F, Sellebjerg F, Teunissen CE, Tumani H, editors. Cerebrospinal Fluid in Clinical Neurology. Cham: Springer International Publishing; 2015. p. 3-16. 3. Rocca J. Galen and the ventricular system. Journal of the history of the neurosciences. 1997;6(3):227-39. doi: 10.1080/09647049709525710. PubMed PMID: 11619860. 4. Herbowski L. The maze of the cerebrospinal fluid discovery. Anatomy research international. 2013;2013:596027. doi: 10.1155/2013/596027. PubMed PMID: 24396600; PubMed Central PMCID: PMC3874314. 5. Del Maestro RF. Leonardo da Vinci: the search for the soul. Journal of neurosurgery. 1998;89(5):874-87. doi: 10.3171/jns.1998.89.5.0874. PubMed PMID: 9817431. 6. Hofberg H, Heurlin F, Millqvist V, Rubenson O. Svenskt biografiskt handlexikon1906. 7. Woollam DH. The historical significance of the cerebrospinal fluid. Medical history. 1957;1(2):91-114. PubMed PMID: 13417893; PubMed Central PMCID: PMC1034258. 8. Pollay M. The function and structure of the cerebrospinal fluid outflow system. Cerebrospinal fluid research. 2010;7:9. Epub 2010/06/23. doi: 10.1186/1743-8454-7-9. PubMed PMID: 20565964; PubMed Central PMCID: PMCPMC2904716. 9. Dandy W, Blackfan K. Internal hydrocephalus. An experimental, clinical and pathological study. Am J Dis Child. 1914;8:406-82. 10. Weed LH. Studies on Cerebro-Spinal Fluid. No. III : The pathways of escape from the Subarachnoid Spaces with particular reference to the Arachnoid Villi. The Journal of medical research. 1914;31(1):51-91. Epub 1914/09/01. PubMed PMID: 19972194; PubMed Central PMCID: PMCPMC2094443. 11. Dandy W. Where is cerebrospinal fluid absorbed. Jama. 1929;92:2012-14. 12. Black PM. Harvey Cushing at the Peter Bent Brigham Hospital. Neurosurgery. 1999;45(5):990- 1001. Epub 1999/11/05. PubMed PMID: 10549919. 13. Milhorat TH. The third circulation revisited. Journal of neurosurgery. 1975;42(6):628-45. Epub 1975/06/01. doi: 10.3171/jns.1975.42.6.0628. PubMed PMID: 167134. 14. Brinker T, Stopa E, Morrison J, Klinge P. A new look at cerebrospinal fluid circulation. Fluids and barriers of the CNS. 2014;11:10. doi: 10.1186/2045-8118-11-10. PubMed PMID: 24817998; PubMed Central PMCID: PMC4016637. 15. Greitz D. Radiological assessment of hydrocephalus: new theories and implications for therapy. Neurosurgical review. 2004;27(3):145-65; discussion 66-7. doi: 10.1007/s10143-004-0326-9. PubMed PMID: 15164255. 16. Oreskovic D, Klarica M. The formation of cerebrospinal fluid: nearly a hundred years of interpretations and misinterpretations. Brain research reviews. 2010;64(2):241-62. doi: 10.1016/j.brainresrev.2010.04.006. PubMed PMID: 20435061. 17. Bering EA, Jr. Circulation of the cerebrospinal fluid. Demonstration of the choroid plexuses as the generator of the force for flow of fluid and ventricular enlargement. Journal of neurosurgery. 1962;19:405-13. Epub 1962/05/01. doi: 10.3171/jns.1962.19.5.0405. PubMed PMID: 13867840. 18. Di Rocco C, Pettorossi VE, Caldarelli M, Mancinelli R, Velardi F. Communicating hydrocephalus induced by mechanically increased amplitude of the intraventricular cerebrospinal fluid pressure: experimental studies. Experimental neurology. 1978;59(1):40-52. Epub 1978/03/01. PubMed PMID: 627267. 19. Du Boulay GH. Pulsatile movements in the CSF pathways. The British journal of radiology. 1966;39(460):255-62. doi: 10.1259/0007-1285-39-460-255. PubMed PMID: 5295581.

20. Edelman RR, Wedeen VJ, Davis KR, Widder D, Hahn P, Shoukimas G, et al. Multiphasic MR imaging: a new method for direct imaging of pulsatile CSF flow. Radiology. 1986;161(3):779-83. Epub 1986/12/01. doi: 10.1148/radiology.161.3.3786732. PubMed PMID: 3786732. 21. Wagshul ME, Chen JJ, Egnor MR, McCormack EJ, Roche PE. Amplitude and phase of cerebrospinal fluid pulsations: experimental studies and review of the literature. Journal of neurosurgery. 2006;104(5):810-9. Epub 2006/05/18. doi: 10.3171/jns.2006.104.5.810. PubMed PMID: 16703889. 22. Alperin N, Hushek SG, Lee SH, Sivaramakrishnan A, Lichtor T. MRI study of cerebral blood flow and CSF flow dynamics in an upright posture: the effect of posture on the intracranial compliance and pressure. Acta neurochirurgica Supplement. 2005;95:177-81. Epub 2006/02/09. PubMed PMID: 16463846. 23. Dreha-Kulaczewski S, Joseph AA, Merboldt KD, Ludwig HC, Gartner J, Frahm J. Inspiration is the major regulator of human CSF flow. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2015;35(6):2485-91. doi: 10.1523/JNEUROSCI.3246-14.2015. PubMed PMID: 25673843. 24. Bonomini F, Rezzani R. Aquaporin and blood brain barrier. Curr Neuropharmacol. 2010;8(2):92-6. doi: 10.2174/157015910791233132. PubMed PMID: 21119879; PubMed Central PMCID: PMCPMC2923372. 25. Cserr HF, Harling-Berg CJ, Knopf PM. Drainage of brain extracellular fluid into blood and deep cervical lymph and its immunological significance. Brain pathology (Zurich, Switzerland). 1992;2(4):269-76. Epub 1992/10/01. PubMed PMID: 1341962. 26. Rennels ML, Gregory TF, Blaumanis OR, Fujimoto K, Grady PA. Evidence for a 'paravascular' fluid circulation in the mammalian central nervous system, provided by the rapid distribution of tracer protein throughout the brain from the subarachnoid space. Brain research. 1985;326(1):47-63. Epub 1985/02/04. PubMed PMID: 3971148. 27. Sykova E, Nicholson C. Diffusion in brain extracellular space. Physiological reviews. 2008;88(4):1277-340. Epub 2008/10/17. doi: 10.1152/physrev.00027.2007. PubMed PMID: 18923183; PubMed Central PMCID: PMCPMC2785730. 28. Carare RO, Bernardes-Silva M, Newman TA, Page AM, Nicoll JA, Perry VH, et al. Solutes, but not cells, drain from the brain parenchyma along basement membranes of capillaries and arteries: significance for cerebral amyloid angiopathy and neuroimmunology. Neuropathology and applied neurobiology. 2008;34(2):131-44. doi: 10.1111/j.1365-2990.2007.00926.x. PubMed PMID: 18208483. 29. Weller RO, Galea I, Carare RO, Minagar A. Pathophysiology of the lymphatic drainage of the central nervous system: Implications for pathogenesis and therapy of multiple sclerosis. Pathophysiology : the official journal of the International Society for Pathophysiology. 2010;17(4):295-306. Epub 2009/12/04. doi: 10.1016/j.pathophys.2009.10.007. PubMed PMID: 19954936. 30. Weller RO, Djuanda E, Yow HY, Carare RO. Lymphatic drainage of the brain and the pathophysiology of neurological disease. Acta neuropathologica. 2009;117(1):1-14. doi: 10.1007/s00401-008-0457-0. PubMed PMID: 19002474. 31. 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. Science translational medicine. 2012;4(147):147ra11. Epub 2012/08/17. doi: 10.1126/scitranslmed.3003748. PubMed PMID: 22896675; PubMed Central PMCID: PMCPmc3551275. 32. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH. Symptomatic Occult Hydrocephalus with "Normal" Cerebrospinal-Fluid Pressure.A Treatable Syndrome. The New England journal of medicine. 1965;273:117-26. doi: 10.1056/NEJM196507152730301. PubMed PMID: 14303656. 33. Hakim S, Adams RD. The special clinical problem of symptomatic hydrocephalus with normal cerebrospinal fluid pressure. Observations on cerebrospinal fluid hydrodynamics. Journal of the neurological sciences. 1965;2(4):307-27. Epub 1965/07/01. PubMed PMID: 5889177.

34. Ojemann RG, Fisher CM, Adams RD, Sweet WH, New PF. Further experience with the syndrome of "normal" pressure hydrocephalus. Journal of neurosurgery. 1969;31(3):279-94. doi: 10.3171/jns.1969.31.3.0279. PubMed PMID: 5811831. 35. Salmon JH. Adult hydrocephalus. Evaluation of shunt therapy in 80 patients. Journal of neurosurgery. 1972;37(4):423-8. doi: 10.3171/jns.1972.37.4.0423. PubMed PMID: 5070869. 36. Shenkin HA, Greenberg J, Bouzarth WF, Gutterman P, Morales JO. Ventricular shunting for relief of senile symptoms. Jama. 1973;225(12):1486-9. PubMed PMID: 4740719. 37. Stein SC, Langfitt TW. Normal-pressure hydrocephalus. Predicting the results of cerebrospinal fluid shunting. Journal of neurosurgery. 1974;41(4):463-70. doi: 10.3171/jns.1974.41.4.0463. PubMed PMID: 4479249. 38. Stein SC, Burnett MG, Sonnad SS. Shunts in normal-pressure hydrocephalus: do we place too many or too few? Journal of neurosurgery. 2006;105(6):815-22. doi: 10.3171/jns.2006.105.6.815. PubMed PMID: 17405250. 39. Bret P, Guyotat J, Chazal J. Is normal pressure hydrocephalus a valid concept in 2002? A reappraisal in five questions and proposal for a new designation of the syndrome as "chronic hydrocephalus". Journal of neurology, neurosurgery, and psychiatry. 2002;73(1):9-12. Epub 2002/06/26. PubMed PMID: 12082037; PubMed Central PMCID: PMCPMC1757288. 40. Bradley WG, Jr., Scalzo D, Queralt J, Nitz WN, Atkinson DJ, Wong P. Normal-pressure hydrocephalus: evaluation with cerebrospinal fluid flow measurements at MR imaging. Radiology. 1996;198(2):523-9. doi: 10.1148/radiology.198.2.8596861. PubMed PMID: 8596861. 41. Bech RA, Juhler M, Waldemar G, Klinken L, Gjerris F. Frontal brain and leptomeningeal biopsy specimens correlated with cerebrospinal fluid outflow resistance and B-wave activity in patients suspected of normal-pressure hydrocephalus. Neurosurgery. 1997;40(3):497-502. Epub 1997/03/01. PubMed PMID: 9055288. 42. Krauss JK, Regel JP, Vach W, Droste DW, Borremans JJ, Mergner T. Vascular risk factors and arteriosclerotic disease in idiopathic normal-pressure hydrocephalus of the elderly. Stroke. 1996;27(1):24-9. Epub 1996/01/01. PubMed PMID: 8553398. 43. Wagshul ME, Eide PK, Madsen JR. The pulsating brain: A review of experimental and clinical studies of intracranial pulsatility. Fluids and barriers of the CNS. 2011;8(1):5. doi: 10.1186/2045- 8118-8-5. PubMed PMID: 21349153; PubMed Central PMCID: PMC3042979. 44. Eide PK, Sorteberg W. Diagnostic intracranial pressure monitoring and surgical management in idiopathic normal pressure hydrocephalus: a 6-year review of 214 patients. Neurosurgery. 2010;66(1):80-91. Epub 2009/12/22. doi: 10.1227/01.neu.0000363408.69856.b8. PubMed PMID: 20023540. 45. 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. The Lancet Neurology. 2003;2(8):506-11. Epub 2003/07/25. PubMed PMID: 12878439. 46. Silverberg GD. Normal pressure hydrocephalus (NPH): ischaemia, CSF stagnation or both. Brain : a journal of neurology. 2004;127(5):947-8. doi: 10.1093/brain/awh178. 47. Momjian S, Owler BK, Czosnyka Z, Czosnyka M, Pena A, Pickard JD. Pattern of white matter regional cerebral blood flow and autoregulation in normal pressure hydrocephalus. Brain : a journal of neurology. 2004;127(Pt 5):965-72. Epub 2004/03/23. doi: 10.1093/brain/awh131. PubMed PMID: 15033897. 48. Kimura M, Tanaka A, Yoshinaga S. Significance of periventricular hemodynamics in normal pressure hydrocephalus. Neurosurgery. 1992;30(5):701-4; discussion 4-5. Epub 1992/05/01. PubMed PMID: 1584381. 49. Keong NC, Pena A, Price SJ, Czosnyka M, Czosnyka Z, Pickard JD. Imaging normal pressure hydrocephalus: theories, techniques, and challenges. Neurosurgical focus. 2016;41(3):E11. doi: 10.3171/2016.7.FOCUS16194. PubMed PMID: 27581307. 50. Bateman GA. The reversibility of reduced cortical vein compliance in normal-pressure hydrocephalus following shunt insertion. Neuroradiology. 2003;45(2):65-70. Epub 2003/02/20. doi: 10.1007/s00234-002-0901-0. PubMed PMID: 12592485. 73

51. Bateman GA. Vascular compliance in normal pressure hydrocephalus. AJNR American journal of neuroradiology. 2000;21(9):1574-85. Epub 2000/10/20. PubMed PMID: 11039334. 52. Graff-Radford NR, Godersky JC. Symptomatic congenital hydrocephalus in the elderly simulating normal pressure hydrocephalus. Neurology. 1989;39(12):1596-600. Epub 1989/12/01. PubMed PMID: 2586776. 53. Krefft TA, Graff-Radford NR, Lucas JA, Mortimer JA. Normal pressure hydrocephalus and large head size. Alzheimer disease and associated disorders. 2004;18(1):35-7. Epub 2004/06/16. PubMed PMID: 15195461. 54. Bradley WG, Jr., Bahl G, Alksne JF. Idiopathic normal pressure hydrocephalus may be a "two hit" disease: benign external hydrocephalus in infancy followed by deep white matter ischemia in late adulthood. Journal of magnetic resonance imaging : JMRI. 2006;24(4):747-55. Epub 2006/09/08. doi: 10.1002/jmri.20684. PubMed PMID: 16958056. 55. Bradley WG, Jr., Whittemore AR, Watanabe AS, Davis SJ, Teresi LM, Homyak M. Association of deep white matter infarction with chronic communicating hydrocephalus: implications regarding the possible origin of normal-pressure hydrocephalus. AJNR American journal of neuroradiology. 1991;12(1):31-9. PubMed PMID: 1899515. 56. Krauss JK, Regel JP, Vach W, Orszagh M, Jungling FD, Bohus M, et al. White matter lesions in patients with idiopathic normal pressure hydrocephalus and in an age-matched control group: a comparative study. Neurosurgery. 1997;40(3):491-5; discussion 5-6. Epub 1997/03/01. PubMed PMID: 9055287. 57. Cusimano MD, Rewilak D, Stuss DT, Barrera-Martinez JC, Salehi F, Freedman M. Normal- pressure hydrocephalus: is there a genetic predisposition? The Canadian journal of neurological sciences Le journal canadien des sciences neurologiques. 2011;38(2):274-81. Epub 2011/02/16. PubMed PMID: 21320833. 58. Takahashi Y, Kawanami T, Nagasawa H, Iseki C, Hanyu H, Kato T. Familial normal pressure hydrocephalus (NPH) with an autosomal-dominant inheritance: a novel subgroup of NPH. Journal of the neurological sciences. 2011;308(1-2):149-51. Epub 2011/06/28. doi: 10.1016/j.jns.2011.06.018. PubMed PMID: 21704338. 59. Nagelhus EA, Ottersen OP. Physiological roles of aquaporin-4 in brain. Physiological reviews. 2013;93(4):1543-62. Epub 2013/10/19. doi: 10.1152/physrev.00011.2013. PubMed PMID: 24137016; PubMed Central PMCID: PMCPmc3858210. 60. Ross CA, Poirier MA. Protein aggregation and neurodegenerative disease. Nature medicine. 2004;10 Suppl:S10-7. Epub 2004/07/24. doi: 10.1038/nm1066. PubMed PMID: 15272267. 61. Xie L, Kang H, Xu Q, Chen MJ, Liao Y, Thiyagarajan M, et al. Sleep drives metabolite clearance from the adult brain. Science. 2013;342(6156):373-7. doi: 10.1126/science.1241224. PubMed PMID: 24136970; PubMed Central PMCID: PMC3880190. 62. Thrane AS, Rangroo Thrane V, Nedergaard M. Drowning stars: reassessing the role of astrocytes in brain edema. Trends Neurosci. 2014;37(11):620-8. doi: 10.1016/j.tins.2014.08.010. PubMed PMID: 25236348; PubMed Central PMCID: PMCPMC4253572. 63. Kress BT, Iliff JJ, Xia M, Wang M, Wei HS, Zeppenfeld D, et al. Impairment of paravascular clearance pathways in the aging brain. Annals of neurology. 2014;76(6):845-61. doi: 10.1002/ana.24271. PubMed PMID: 25204284; PubMed Central PMCID: PMC4245362. 64. Ren Z, Iliff JJ, Yang L, Yang J, Chen X, Chen MJ, et al. 'Hit & Run' model of closed-skull traumatic brain injury (TBI) reveals complex patterns of post-traumatic AQP4 dysregulation. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism. 2013;33(6):834-45. doi: 10.1038/jcbfm.2013.30. PubMed PMID: 23443171; PubMed Central PMCID: PMCPMC3677112. 65. Eide PK, Hansson HA. Astrogliosis and impaired aquaporin-4 and dystrophin systems in idiopathic normal pressure hydrocephalus. Neuropathology and applied neurobiology. 2017. Epub 2017/06/20. doi: 10.1111/nan.12420. PubMed PMID: 28627088. 66. Espay AJ, Da Prat GA, Dwivedi AK, Rodriguez-Porcel F, Vaughan JE, Rosso M, et al. Deconstructing normal pressure hydrocephalus: Ventriculomegaly as early sign of 74 neurodegeneration. Annals of neurology. 2017;82(4):503-13. Epub 2017/09/12. doi: 10.1002/ana.25046. PubMed PMID: 28892572. 67. Hebb AO, Cusimano MD. Idiopathic normal pressure hydrocephalus: a systematic review of diagnosis and outcome. Neurosurgery. 2001;49(5):1166-84; discussion 84-6. PubMed PMID: 11846911. 68. Ishikawa M, Guideline Committe for Idiopathic Normal Pressure Hydrocephalus JSoNPH. Clinical guidelines for idiopathic normal pressure hydrocephalus. Neurologia medico-chirurgica. 2004;44(4):222-3. PubMed PMID: 15185767. 69. Relkin N, Marmarou A, Klinge P, Bergsneider M, Black PM. Diagnosing idiopathic normal- pressure hydrocephalus. Neurosurgery. 2005;57(3 Suppl):S4-16; discussion ii-v. PubMed PMID: 16160425. 70. Brean A, Eide PK. Prevalence of probable idiopathic normal pressure hydrocephalus in a Norwegian population. Acta neurologica Scandinavica. 2008;118(1):48-53. doi: 10.1111/j.1600- 0404.2007.00982.x. PubMed PMID: 18205881. 71. Hiraoka K, Meguro K, Mori E. Prevalence of idiopathic normal-pressure hydrocephalus in the elderly population of a Japanese rural community. Neurologia medico-chirurgica. 2008;48(5):197-99; discussion 9-200. Epub 2008/05/24. PubMed PMID: 18497491. 72. Marmarou A, Young HF, Aygok GA. Estimated incidence of normal pressure hydrocephalus and shunt outcome in patients residing in assisted-living and extended-care facilities. Neurosurgical focus. 2007;22(4):E1. doi: 10.3171/foc.2007.22.5.2. PubMed PMID: 17613187. 73. Brean A, Fredo HL, Sollid S, Muller T, Sundstrom T, Eide PK. Five-year incidence of surgery for idiopathic normal pressure hydrocephalus in Norway. Acta neurologica Scandinavica. 2009;120(5):314-6. doi: 10.1111/j.1600-0404.2009.01250.x. PubMed PMID: 19832773. 74. Tisell M, Hoglund M, Wikkelso C. National and regional incidence of surgery for adult hydrocephalus in Sweden. Acta neurologica Scandinavica. 2005;112(2):72-5. doi: 10.1111/j.1600- 0404.2005.00451.x. PubMed PMID: 16008530. 75. Malm J, Graff-Radford NR, Ishikawa M, Kristensen B, Leinonen V, Mori E, et al. Influence of comorbidities in idiopathic normal pressure hydrocephalus - research and clinical care. A report of the ISHCSF task force on comorbidities in INPH. Fluids and barriers of the CNS. 2013;10(1):22. doi: 10.1186/2045-8118-10-22. PubMed PMID: 23758953; PubMed Central PMCID: PMC3689166. 76. Kahlon B, Sjunnesson J, Rehncrona S. Long-term outcome in patients with suspected normal pressure hydrocephalus. Neurosurgery. 2007;60(2):327-32; discussion 32. doi: 10.1227/01.NEU.0000249273.41569.6E. PubMed PMID: 17290183. 77. Hellstrom P, Klinge P, Tans J, Wikkelso C. A new scale for assessment of severity and outcome in iNPH. Acta neurologica Scandinavica. 2012;126(4):229-37. doi: 10.1111/j.1600-0404.2012.01677.x. PubMed PMID: 22587624. 78. Gallia GL, Rigamonti D, Williams MA. The diagnosis and treatment of idiopathic normal pressure hydrocephalus. Nature clinical practice Neurology. 2006;2(7):375-81. doi: 10.1038/ncpneuro0237. PubMed PMID: 16932588. 79. Krauss JK, Regel JP, Droste DW, Orszagh M, Borremans JJ, Vach W. Movement disorders in adult hydrocephalus. Movement disorders : official journal of the Movement Disorder Society. 1997;12(1):53-60. doi: 10.1002/mds.870120110. PubMed PMID: 8990054. 80. Hashimoto M, Ishikawa M, Mori E, Kuwana N. Diagnosis of idiopathic normal pressure hydrocephalus is supported by MRI-based scheme: a prospective cohort study. Cerebrospinal fluid research. 2010;7:18. Epub 2010/11/03. doi: 10.1186/1743-8454-7-18. PubMed PMID: 21040519; PubMed Central PMCID: PMCPmc2987762. 81. McGirt MJ, Woodworth G, Coon AL, Thomas G, Williams MA, Rigamonti D. Diagnosis, treatment, and analysis of long-term outcomes in idiopathic normal-pressure hydrocephalus. Neurosurgery. 2005;57(4):699-705; discussion 699-705. PubMed PMID: 16239882. 82. Goepel M, Hoffmann JA, Piro M, Rubben H, Michel MC. Prevalence and physician awareness of symptoms of urinary bladder dysfunction. European urology. 2002;41(3):234-9. PubMed PMID: 12180221. 75

83. Nasjonalforeningen ff. Hva er demens? [Internet page]. http://nasjonalforeningen.no/demens/hva-er-demens/2017 [updated 10.4.2017]. Available from: http://nasjonalforeningen.no/demens/hva-er-demens/. 84. Cabral D, Beach TG, Vedders L, Sue LI, Jacobson S, Myers K, et al. Frequency of Alzheimer's disease pathology at autopsy in patients with clinical normal pressure hydrocephalus. Alzheimer's & dementia : the journal of the Alzheimer's Association. 2011;7(5):509-13. doi: 10.1016/j.jalz.2010.12.008. PubMed PMID: 21723206; PubMed Central PMCID: PMC3166980. 85. Leinonen V, Koivisto AM, Alafuzoff I, Pyykko OT, Rummukainen J, von Und Zu Fraunberg M, et al. Cortical brain biopsy in long-term prognostication of 468 patients with possible normal pressure hydrocephalus. Neuro-degenerative diseases. 2012;10(1-4):166-9. doi: 10.1159/000335155. PubMed PMID: 22343771. 86. Golomb J, Wisoff J, Miller DC, Boksay I, Kluger A, Weiner H, et al. Alzheimer's disease comorbidity in normal pressure hydrocephalus: prevalence and shunt response. Journal of neurology, neurosurgery, and psychiatry. 2000;68(6):778-81. Epub 2000/05/16. PubMed PMID: 10811706; PubMed Central PMCID: PMCPMC1736969. 87. Kojoukhova M, Vanha KI, Timonen M, Koivisto AM, Nerg O, Rummukainen J, et al. Associations of intracranial pressure with brain biopsy, radiological findings, and shunt surgery outcome in patients with suspected idiopathic normal pressure hydrocephalus. Acta neurochirurgica. 2016. doi: 10.1007/s00701-016-3025-8. PubMed PMID: 27878614. 88. Hellstrom P, Edsbagge M, Archer T, Tisell M, Tullberg M, Wikkelso C. The neuropsychology of patients with clinically diagnosed idiopathic normal pressure hydrocephalus. Neurosurgery. 2007;61(6):1219-26; discussion 27-8. doi: 10.1227/01.neu.0000306100.83882.81. PubMed PMID: 18162901. 89. Leinonen V, Koivisto AM, Savolainen S, Rummukainen J, Sutela A, Vanninen R, et al. Post- mortem findings in 10 patients with presumed normal-pressure hydrocephalus and review of the literature. Neuropathology and applied neurobiology. 2012;38(1):72-86. doi: 10.1111/j.1365- 2990.2011.01195.x. PubMed PMID: 21696417. 90. Israelsson H, Carlberg B, Wikkelso C, Laurell K, Kahlon B, Leijon G, et al. Vascular risk factors in INPH: A prospective case-control study (the INPH-CRasH study). Neurology. 2017;88(6):577-85. doi: 10.1212/WNL.0000000000003583. PubMed PMID: 28062721; PubMed Central PMCID: PMCPMC5304464. 91. Vanneste JA. Three decades of normal pressure hydrocephalus: are we wiser now? Journal of neurology, neurosurgery, and psychiatry. 1994;57(9):1021-5. PubMed PMID: 8089664; PubMed Central PMCID: PMC1073120. 92. Petersen RC, Mokri B, Laws ER, Jr. Surgical treatment of idiopathic hydrocephalus in elderly patients. Neurology. 1985;35(3):307-11. PubMed PMID: 3974888. 93. Bergsneider M, Black PM, Klinge P, Marmarou A, Relkin N. Surgical management of idiopathic normal-pressure hydrocephalus. Neurosurgery. 2005;57(3 Suppl):S29-39; discussion ii-v. Epub 2005/09/15. PubMed PMID: 16160427. 94. Esmonde T, Cooke S. Shunting for normal pressure hydrocephalus (NPH). The Cochrane database of systematic reviews. 2002;(3):CD003157. doi: 10.1002/14651858.CD003157. PubMed PMID: 12137677. 95. Vanneste J, Augustijn P, Dirven C, Tan WF, Goedhart ZD. Shunting normal-pressure hydrocephalus: do the benefits outweigh the risks? A multicenter study and literature review. Neurology. 1992;42(1):54-9. PubMed PMID: 1734324. 96. Halperin JJ, Kurlan R, Schwalb JM, Cusimano MD, Gronseth G, Gloss D. Practice guideline: Idiopathic normal pressure hydrocephalus: Response to shunting and predictors of response: Report of the Guideline Development, Dissemination, and Implementation Subcommittee of the American Academy of Neurology. Neurology. 2015;85(23):2063-71. doi: 10.1212/WNL.0000000000002193. PubMed PMID: 26644048; PubMed Central PMCID: PMC4676757. 97. Krauss JK, Droste DW, Vach W, Regel JP, Orszagh M, Borremans JJ, et al. Cerebrospinal fluid shunting in idiopathic normal-pressure hydrocephalus of the elderly: effect of periventricular and 76 deep white matter lesions. Neurosurgery. 1996;39(2):292-9; discussion 9-300. PubMed PMID: 8832666. 98. Brecknell JE, Brown JI. Is idiopathic normal pressure hydrocephalus an independent entity? Acta neurochirurgica. 2004;146(9):1003-6; discussion 6-7. doi: 10.1007/s00701-004-0332-2. PubMed PMID: 15340812. 99. Marmarou A, Black P, Bergsneider M, Klinge P, Relkin N. Guidelines for management of idiopathic normal pressure hydrocephalus: progress to date. Acta neurochirurgica Supplement. 2005;95:237-40. Epub 2006/02/09. PubMed PMID: 16463856. 100. Marmarou A, Bergsneider M, Klinge P, Relkin N, Black PM. The value of supplemental prognostic tests for the preoperative assessment of idiopathic normal-pressure hydrocephalus. Neurosurgery. 2005;57(3 Suppl):S17-28; discussion ii-v. PubMed PMID: 16160426. 101. Wikkelso C, Hellstrom P, Klinge PM, Tans JT. The European iNPH Multicentre Study on the predictive values of resistance to CSF outflow and the CSF Tap Test in patients with idiopathic normal pressure hydrocephalus. Journal of neurology, neurosurgery, and psychiatry. 2013;84(5):562-8. Epub 2012/12/20. doi: 10.1136/jnnp-2012-303314. PubMed PMID: 23250963. 102. Marmarou A, Shulman K, LaMorgese J. Compartmental analysis of compliance and outflow resistance of the cerebrospinal fluid system. Journal of neurosurgery. 1975;43(5):523-34. Epub 1975/11/01. doi: 10.3171/jns.1975.43.5.0523. PubMed PMID: 1181384. 103. Eide PK, Sorteberg W. Outcome of Surgery for Idiopathic Normal Pressure Hydrocephalus: Role of Preoperative Static and Pulsatile Intracranial Pressure. World neurosurgery. 2016;86:186-93 e1. doi: 10.1016/j.wneu.2015.09.067. PubMed PMID: 26428326. 104. Koskinen LO, Grayson D, Olivecrona M. The complications and the position of the Codman MicroSensor ICP device: an analysis of 549 patients and 650 Sensors. Acta neurochirurgica. 2013;155(11):2141-8; discussion 8. doi: 10.1007/s00701-013-1856-0. PubMed PMID: 24018980. 105. Abildgaard A. MR for radiografer og radiologer. 1st ed: Universitetsforlaget; 2016. 106. Evans WJ. An encephalographic ratio for estimating ventricular enlargement. Arch Neurol Psychiatry. 1942;47(6):931-7. 107. Synek V, Reuben JR, Du Boulay GH. Comparing Evans' index and computerized axial tomography in assessing relationship of ventricular size to brain size. Neurology. 1976;26(3):231-3. PubMed PMID: 1082559. 108. Toma AK, Holl E, Kitchen ND, Watkins LD. Evans' index revisited: the need for an alternative in normal pressure hydrocephalus. Neurosurgery. 2011;68(4):939-44. Epub 2011/01/12. doi: 10.1227/NEU.0b013e318208f5e0. PubMed PMID: 21221031. 109. Palm WM, Walchenbach R, Bruinsma B, Admiraal-Behloul F, Middelkoop HA, Launer LJ, et al. Intracranial compartment volumes in normal pressure hydrocephalus: volumetric assessment versus outcome. AJNR American journal of neuroradiology. 2006;27(1):76-9. Epub 2006/01/19. PubMed PMID: 16418361. 110. Klinge P, Hellstrom P, Tans J, Wikkelso C, European i NPHMSG. One-year outcome in the European multicentre study on iNPH. Acta neurologica Scandinavica. 2012;126(3):145-53. doi: 10.1111/j.1600-0404.2012.01676.x. PubMed PMID: 22571428. 111. Brix MK, Westman E, Simmons A, Ringstad GA, Eide PK, Wagner-Larsen K, et al. The Evans' Index revisited: New cut-off levels for use in radiological assessment of ventricular enlargement in the elderly. European journal of radiology. 2017;95:28-32. Epub 2017/10/11. doi: 10.1016/j.ejrad.2017.07.013. PubMed PMID: 28987681. 112. Sjaastad O, Nordvik A. The corpus callosal angle in the diagnosis of cerebral ventricular enlargement. Acta neurologica Scandinavica. 1973;49(3):396-406. PubMed PMID: 4542888. 113. Virhammar J, Laurell K, Cesarini KG, Larsson EM. The callosal angle measured on MRI as a predictor of outcome in idiopathic normal-pressure hydrocephalus. Journal of neurosurgery. 2014;120(1):178-84. doi: 10.3171/2013.8.JNS13575. PubMed PMID: 24074491. 114. Miskin N, Patel H, Franceschi AM, Ades-Aron B, Le A, Damadian BE, et al. Diagnosis of Normal-Pressure Hydrocephalus: Use of Traditional Measures in the Era of Volumetric MR Imaging.

Radiology. 2017:161216. Epub 2017/05/13. doi: 10.1148/radiol.2017161216. PubMed PMID: 28498794. 115. Serulle Y, Rusinek H, Kirov, II, Milch H, Fieremans E, Baxter AB, et al. Differentiating shunt- responsive normal pressure hydrocephalus from Alzheimer disease and normal aging: pilot study using automated MRI brain tissue segmentation. Journal of neurology. 2014;261(10):1994-2002. Epub 2014/08/02. doi: 10.1007/s00415-014-7454-0. PubMed PMID: 25082631. 116. Kitagaki H, Mori E, Ishii K, Yamaji S, Hirono N, Imamura T. CSF spaces in idiopathic normal pressure hydrocephalus: morphology and volumetry. AJNR American journal of neuroradiology. 1998;19(7):1277-84. Epub 1998/09/03. PubMed PMID: 9726467. 117. Virhammar J, Laurell K, Cesarini KG, Larsson EM. Preoperative prognostic value of MRI findings in 108 patients with idiopathic normal pressure hydrocephalus. AJNR American journal of neuroradiology. 2014;35(12):2311-8. Epub 2014/07/12. doi: 10.3174/ajnr.A4046. PubMed PMID: 25012669. 118. Mori E, Ishikawa M, Kato T, Kazui H, Miyake H, Miyajima M, et al. Guidelines for management of idiopathic normal pressure hydrocephalus: second edition. Neurologia medico-chirurgica. 2012;52(11):775-809. Epub 2012/11/28. PubMed PMID: 23183074. 119. Craven CL, Toma AK, Mostafa T, Patel N, Watkins LD. The predictive value of DESH for shunt responsiveness in idiopathic normal pressure hydrocephalus. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2016;34:294-8. Epub 2016/10/04. doi: 10.1016/j.jocn.2016.09.004. PubMed PMID: 27692614. 120. Battal B, Kocaoglu M, Bulakbasi N, Husmen G, Tuba Sanal H, Tayfun C. Cerebrospinal fluid flow imaging by using phase-contrast MR technique. The British journal of radiology. 2011;84(1004):758-65. Epub 2011/05/19. doi: 10.1259/bjr/66206791. PubMed PMID: 21586507; PubMed Central PMCID: PMCPMC3473435. 121. Enzmann DR, Pelc NJ. Normal flow patterns of intracranial and spinal cerebrospinal fluid defined with phase-contrast cine MR imaging. Radiology. 1991;178(2):467-74. Epub 1991/02/01. doi: 10.1148/radiology.178.2.1987610. PubMed PMID: 1987610. 122. Nitz WR, Bradley WG, Jr., Watanabe AS, Lee RR, Burgoyne B, O'Sullivan RM, et al. Flow dynamics of cerebrospinal fluid: assessment with phase-contrast velocity MR imaging performed with retrospective cardiac gating. Radiology. 1992;183(2):395-405. doi: 10.1148/radiology.183.2.1561340. PubMed PMID: 1561340. 123. Baledent O. Imaging of the cerebrospinal fluid circulation. Cambridge University Press; 2014. p. 121-38. 124. Baledent O, Henry-Feugeas MC, Idy-Peretti I. Cerebrospinal fluid dynamics and relation with blood flow: a magnetic resonance study with semiautomated cerebrospinal fluid segmentation. Investigative radiology. 2001;36(7):368-77. Epub 2001/08/10. PubMed PMID: 11496092. 125. Baledent O, Gondry-Jouet C, Meyer ME, De Marco G, Le Gars D, Henry-Feugeas MC, et al. Relationship between cerebrospinal fluid and blood dynamics in healthy volunteers and patients with communicating hydrocephalus. Investigative radiology. 2004;39(1):45-55. Epub 2004/01/01. doi: 10.1097/01.rli.0000100892.87214.49. PubMed PMID: 14701988. 126. Bradley WG, Jr., Kortman KE, Burgoyne B. Flowing cerebrospinal fluid in normal and hydrocephalic states: appearance on MR images. Radiology. 1986;159(3):611-6. Epub 1986/06/01. doi: 10.1148/radiology.159.3.3704142. PubMed PMID: 3704142. 127. Bradley WG, Jr., Whittemore AR, Kortman KE, Watanabe AS, Homyak M, Teresi LM, et al. Marked cerebrospinal fluid void: indicator of successful shunt in patients with suspected normal- pressure hydrocephalus. Radiology. 1991;178(2):459-66. Epub 1991/02/11. doi: 10.1148/radiology.178.2.1987609. PubMed PMID: 1987609. 128. Wahlin A, Ambarki K, Hauksson J, Birgander R, Malm J, Eklund A. Phase contrast MRI quantification of pulsatile volumes of brain arteries, veins, and cerebrospinal fluids compartments: repeatability and physiological interactions. Journal of magnetic resonance imaging : JMRI. 2012;35(5):1055-62. doi: 10.1002/jmri.23527. PubMed PMID: 22170792.

129. Al-Zain FT, Rademacher G, Meier U, Mutze S, Lemcke J. The role of cerebrospinal fluid flow study using phase contrast MR imaging in diagnosing idiopathic normal pressure hydrocephalus. Acta neurochirurgica Supplement. 2008;102:119-23. PubMed PMID: 19388301. 130. Kim DS, Choi JU, Huh R, Yun PH, Kim DI. Quantitative assessment of cerebrospinal fluid hydrodynamics using a phase-contrast cine MR image in hydrocephalus. Child's nervous system : ChNS : official journal of the International Society for Pediatric Neurosurgery. 1999;15(9):461-7. doi: 10.1007/s003810050440. PubMed PMID: 10502007. 131. Sharma AK, Gaikwad S, Gupta V, Garg A, Mishra NK. Measurement of peak CSF flow velocity at cerebral aqueduct, before and after lumbar CSF drainage, by use of phase-contrast MRI: utility in the management of idiopathic normal pressure hydrocephalus. Clinical neurology and neurosurgery. 2008;110(4):363-8. doi: 10.1016/j.clineuro.2007.12.021. PubMed PMID: 18282655. 132. Luetmer PH, Huston J, Friedman JA, Dixon GR, Petersen RC, Jack CR, et al. Measurement of cerebrospinal fluid flow at the cerebral aqueduct by use of phase-contrast magnetic resonance imaging: technique validation and utility in diagnosing idiopathic normal pressure hydrocephalus. Neurosurgery. 2002;50(3):534-43; discussion 43-4. PubMed PMID: 11841721. 133. Abbey P, Singh P, Khandelwal N, Mukherjee KK. Shunt surgery effects on cerebrospinal fluid flow across the aqueduct of Sylvius in patients with communicating hydrocephalus. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2009;16(4):514-8. doi: 10.1016/j.jocn.2008.05.009. PubMed PMID: 19195891. 134. El Sankari S, Fichten A, Gondry-Jouet C, Czosnyka M, Legars D, Deramond H, et al. Correlation between tap test and CSF aqueductal stroke volume in idiopathic normal pressure hydrocephalus. Acta neurochirurgica Supplement. 2012;113:43-6. doi: 10.1007/978-3-7091-0923-6_9. PubMed PMID: 22116421. 135. Scollato A, Gallina P, Gautam B, Pellicano G, Cavallini C, Tenenbaum R, et al. Changes in aqueductal CSF stroke volume in shunted patients with idiopathic normal-pressure hydrocephalus. AJNR American journal of neuroradiology. 2009;30(8):1580-6. doi: 10.3174/ajnr.A1616. PubMed PMID: 19461060. 136. Delwel EJ, de Jong DA, Avezaat CJ. The prognostic value of clinical characteristics and parameters of cerebrospinal fluid hydrodynamics in shunting for idiopathic normal pressure hydrocephalus. Acta neurochirurgica. 2005;147(10):1037-42; discussion 42-3. doi: 10.1007/s00701- 005-0570-y. PubMed PMID: 16047108. 137. Bateman GA, Loiselle AM. Can MR measurement of intracranial hydrodynamics and compliance differentiate which patient with idiopathic normal pressure hydrocephalus will improve following shunt insertion? Acta neurochirurgica. 2007;149(5):455-62; discussion 62. doi: 10.1007/s00701-007-1142-0. PubMed PMID: 17406777. 138. Kahlon B, Annertz M, Stahlberg F, Rehncrona S. Is aqueductal stroke volume, measured with cine phase-contrast magnetic resonance imaging scans useful in predicting outcome of shunt surgery in suspected normal pressure hydrocephalus? Neurosurgery. 2007;60(1):124-9; discussion 9-30. doi: 10.1227/01.NEU.0000249208.04344.A3. PubMed PMID: 17228260. 139. Dixon GR, Friedman JA, Luetmer PH, Quast LM, McClelland RL, Petersen RC, et al. Use of cerebrospinal fluid flow rates measured by phase-contrast MR to predict outcome of ventriculoperitoneal shunting for idiopathic normal-pressure hydrocephalus. Mayo Clinic proceedings. 2002;77(6):509-14. doi: 10.4065/77.6.509. PubMed PMID: 12059119. 140. Poca MA, Sahuquillo J, Busto M, Rovira A, Capellades J, Mataro M, et al. Agreement between CSF flow dynamics in MRI and ICP monitoring in the diagnosis of normal pressure hydrocephalus. Sensitivity and specificity of CSF dynamics to predict outcome. Acta neurochirurgica Supplement. 2002;81:7-10. Epub 2002/08/10. PubMed PMID: 12168360. 141. Hamilton R, Baldwin K, Fuller J, Vespa P, Hu X, Bergsneider M. Intracranial pressure pulse waveform correlates with aqueductal cerebrospinal fluid stroke volume. Journal of applied physiology. 2012;113(10):1560-6. Epub 2012/09/22. doi: 10.1152/japplphysiol.00357.2012. PubMed PMID: 22995390; PubMed Central PMCID: PMCPMC3524656.

142. Tain RW, Alperin N. Intracranial pressure dynamics are not linked to aqueductal cerebrospinal fluid stroke volume. Journal of applied physiology. 2013;114(11):1645. Epub 2013/06/05. doi: 10.1152/japplphysiol.00357.2013. PubMed PMID: 23729634; PubMed Central PMCID: PMCPMC3680830. 143. Mase M, Yamada K, Banno T, Miyachi T, Ohara S, Matsumoto T. Quantitative analysis of CSF flow dynamics using MRI in normal pressure hydrocephalus. Acta neurochirurgica Supplement. 1998;71:350-3. PubMed PMID: 9779227. 144. Wahlin A, Ambarki K, Birgander R, Alperin N, Malm J, Eklund A. Assessment of craniospinal pressure-volume indices. AJNR American journal of neuroradiology. 2010;31(9):1645-50. Epub 2010/07/03. doi: 10.3174/ajnr.A2166. PubMed PMID: 20595369. 145. Alperin NJ, Lee SH, Loth F, Raksin PB, Lichtor T. MR-Intracranial pressure (ICP): a method to measure intracranial elastance and pressure noninvasively by means of MR imaging: baboon and human study. Radiology. 2000;217(3):877-85. Epub 2000/12/09. doi: 10.1148/radiology.217.3.r00dc42877. PubMed PMID: 11110957. 146. Alperin N, Lee SH, Bagci AM. MRI measurements of intracranial pressure in the upright posture: The effect of the hydrostatic pressure gradient. Journal of magnetic resonance imaging : JMRI. 2015;42(4):1158-63. doi: 10.1002/jmri.24882. PubMed PMID: 25753157. 147. Muehlmann M, Koerte IK, Laubender RP, Steffinger D, Lehner M, Peraud A, et al. Magnetic resonance-based estimation of intracranial pressure correlates with ventriculoperitoneal shunt valve opening pressure setting in children with hydrocephalus. Investigative radiology. 2013;48(7):543-7. doi: 10.1097/RLI.0b013e31828ad504. PubMed PMID: 23695081. 148. Glick RP, Niebruegge J, Lee SH, Egibor O, Lichtor T, Alperin N. Early experience from the application of a noninvasive magnetic resonance imaging-based measurement of intracranial pressure in hydrocephalus. Neurosurgery. 2006;59(5):1052-60; discussion 60-1. doi: 10.1227/01.NEU.0000245586.23710.EF. PubMed PMID: 17143240. 149. Alperin N, Sivaramakrishnan A, Lichtor T. Magnetic resonance imaging-based measurements of cerebrospinal fluid and blood flow as indicators of intracranial compliance in patients with Chiari malformation. Journal of neurosurgery. 2005;103(1):46-52. doi: 10.3171/jns.2005.103.1.0046. PubMed PMID: 16121972. 150. Mase M, Miyati T, Yamada K, Kasai H, Hara M, Shibamoto Y. Non-invasive measurement of intracranial compliance using cine MRI in normal pressure hydrocephalus. Acta neurochirurgica Supplement. 2005;95:303-6. PubMed PMID: 16463870. 151. Miyati T, Mase M, Kasai H, Hara M, Yamada K, Shibamoto Y, et al. Noninvasive MRI assessment of intracranial compliance in idiopathic normal pressure hydrocephalus. Journal of magnetic resonance imaging : JMRI. 2007;26(2):274-8. Epub 2007/07/05. doi: 10.1002/jmri.20999. PubMed PMID: 17610284. 152. Le Bihan D, Iima M. Diffusion Magnetic Resonance Imaging: What Water Tells Us about Biological Tissues. PLoS biology. 2015;13(7):e1002203. Epub 2015/07/24. doi: 10.1371/journal.pbio.1002203. PubMed PMID: 26204162; PubMed Central PMCID: PMCPMC4512706. 153. Gideon P, Thomsen C, Gjerris F, Sorensen PS, Henriksen O. Increased self-diffusion of brain water in hydrocephalus measured by MR imaging. Acta radiologica. 1994;35(6):514-9. Epub 1994/11/01. PubMed PMID: 7946670. 154. Leliefeld PH, Gooskens RH, Braun KP, Ramos LM, Uiterwaal CS, Regli LP, et al. Longitudinal diffusion-weighted imaging in infants with hydrocephalus: decrease in tissue water diffusion after cerebrospinal fluid diversion. Journal of neurosurgery Pediatrics. 2009;4(1):56-63. Epub 2009/07/03. doi: 10.3171/2009.3.peds08337. PubMed PMID: 19569912. 155. Aygok G, Marmarou A, Fatouros P, Young H. Brain tissue water content in patients with idiopathic normal pressure hydrocephalus. Acta neurochirurgica Supplement. 2006;96:348-51. Epub 2006/05/05. PubMed PMID: 16671483. 156. Ng SE, Low AM, Tang KK, Chan YH, Kwok RK. Value of quantitative MRI biomarkers (Evans' index, aqueductal flow rate, and apparent diffusion coefficient) in idiopathic normal pressure 80 hydrocephalus. Journal of magnetic resonance imaging : JMRI. 2009;30(4):708-15. Epub 2009/09/30. doi: 10.1002/jmri.21865. PubMed PMID: 19787714. 157. Siasios I, Kapsalaki EZ, Fountas KN, Fotiadou A, Dorsch A, Vakharia K, et al. The role of diffusion tensor imaging and fractional anisotropy in the evaluation of patients with idiopathic normal pressure hydrocephalus: a literature review. Neurosurgical focus. 2016;41(3):E12. Epub 2016/09/02. doi: 10.3171/2016.6.focus16192. PubMed PMID: 27581308. 158. Bakker EN, Bacskai BJ, Arbel-Ornath M, Aldea R, Bedussi B, Morris AW, et al. Lymphatic Clearance of the Brain: Perivascular, Paravascular and Significance for Neurodegenerative Diseases. Cellular and molecular neurobiology. 2016;36(2):181-94. doi: 10.1007/s10571-015-0273-8. PubMed PMID: 26993512; PubMed Central PMCID: PMC4844641. 159. Iliff JJ, Lee H, Yu M, Feng T, Logan J, Nedergaard M, et al. Brain-wide pathway for waste clearance captured by contrast-enhanced MRI. The Journal of clinical investigation. 2013;123(3):1299-309. Epub 2013/02/26. doi: 10.1172/jci67677. PubMed PMID: 23434588; PubMed Central PMCID: PMCPmc3582150. 160. Gaberel T, Gakuba C, Goulay R, Martinez De Lizarrondo S, Hanouz JL, Emery E, et al. Impaired glymphatic perfusion after strokes revealed by contrast-enhanced MRI: a new target for fibrinolysis? Stroke. 2014;45(10):3092-6. Epub 2014/09/06. doi: 10.1161/strokeaha.114.006617. PubMed PMID: 25190438. 161. Goulay R, Flament J, Gauberti M, Naveau M, Pasquet N, Gakuba C, et al. Subarachnoid Hemorrhage Severely Impairs Brain Parenchymal Cerebrospinal Fluid Circulation in Nonhuman Primate. Stroke. 2017;48(8):2301-5. Epub 2017/05/21. doi: 10.1161/strokeaha.117.017014. PubMed PMID: 28526764. 162. Yang L, Kress BT, Weber HJ, Thiyagarajan M, Wang B, Deane R, et al. Evaluating glymphatic pathway function utilizing clinically relevant intrathecal infusion of CSF tracer. Journal of translational medicine. 2013;11:107. doi: 10.1186/1479-5876-11-107. PubMed PMID: 23635358; PubMed Central PMCID: PMC3665671. 163. Eide PK, Ringstad G. MRI with intrathecal MRI gadolinium contrast medium administration: a possible method to assess glymphatic function in human brain. Acta radiologica open. 2015;4(11):2058460115609635. doi: 10.1177/2058460115609635. PubMed PMID: 26634147; PubMed Central PMCID: PMC4652208. 164. Heinz ER, Davis DO, Karp HR. Abnormal isotope cisternography in sympatomatic occult hydrocephalus. A correlative isotopic-neuroradiological study in 130 subjects. Radiology. 1970;95(1):109-20. Epub 1970/04/01. doi: 10.1148/95.1.109. PubMed PMID: 5417029. 165. Tator CH, Murray S. A clinical, pneumoencephalographic and radioisotopic study of normal- pressure communicating hydrocephalus. Canadian Medical Association journal. 1971;105(6):573-9. Epub 1971/09/18. PubMed PMID: 5118676; PubMed Central PMCID: PMCPMC1931198. 166. Black PM. Idiopathic normal-pressure hydrocephalus. Results of shunting in 62 patients. Journal of neurosurgery. 1980;52(3):371-7. Epub 1980/03/01. doi: 10.3171/jns.1980.52.3.0371. PubMed PMID: 7359191. 167. Kilic K, Czorny A, Auque J, Berkman Z. Predicting the outcome of shunt surgery in normal pressure hydrocephalus. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2007;14(8):729-36. Epub 2007/01/16. doi: 10.1016/j.jocn.2006.03.028. PubMed PMID: 17223561. 168. Walker L, Gholamrezanezhad A, Bucklan D, Faulhaber PF, O'Donnell JK. SPECT/CT Detection of a Communicating Arachnoid Cyst in a Patient With Normal Pressure Hydrocephalus. Clinical nuclear medicine. 2017. Epub 2017/05/10. doi: 10.1097/rlu.0000000000001685. PubMed PMID: 28481794. 169. Klinge PM, Berding G, Brinker T, Knapp WH, Samii M. A positron emission tomography study of cerebrovascular reserve before and after shunt surgery in patients with idiopathic chronic hydrocephalus. Journal of neurosurgery. 1999;91(4):605-9. Epub 1999/10/03. doi: 10.3171/jns.1999.91.4.0605. PubMed PMID: 10507382.

170. Klinge PM, Brooks DJ, Samii A, Weckesser E, van den Hoff J, Fricke H, et al. Correlates of local cerebral blood flow (CBF) in normal pressure hydrocephalus patients before and after shunting--A retrospective analysis of [(15)O]H(2)O PET-CBF studies in 65 patients. Clinical neurology and neurosurgery. 2008;110(4):369-75. Epub 2008/02/12. doi: 10.1016/j.clineuro.2007.12.019. PubMed PMID: 18262344. 171. Owler BK, Momjian S, Czosnyka Z, Czosnyka M, Pena A, Harris NG, et al. Normal pressure hydrocephalus and cerebral blood flow: a PET study of baseline values. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism. 2004;24(1):17-23. Epub 2003/12/23. doi: 10.1097/01.wcb.0000093326.88757.49. PubMed PMID: 14688613. 172. Tullberg M, Hellstrom P, Piechnik SK, Starmark JE, Wikkelso C. Impaired wakefulness is associated with reduced anterior cingulate CBF in patients with normal pressure hydrocephalus. Acta neurologica Scandinavica. 2004;110(5):322-30. Epub 2004/10/13. doi: 10.1111/j.1600- 0404.2004.00325.x. PubMed PMID: 15476461. 173. Rinne JO, Frantzen J, Leinonen V, Lonnrot K, Laakso A, Virtanen KA, et al. Prospective flutemetamol positron emission tomography and histopathology in normal pressure hydrocephalus. Neuro-degenerative diseases. 2014;13(4):237-45. Epub 2013/12/04. doi: 10.1159/000355256. PubMed PMID: 24296542. 174. Leinonen V, Rinne JO, Virtanen KA, Eskola O, Rummukainen J, Huttunen J, et al. Positron emission tomography with [18F]flutemetamol and [11C]PiB for in vivo detection of cerebral cortical amyloid in normal pressure hydrocephalus patients. European journal of neurology. 2013;20(7):1043-52. Epub 2013/02/13. doi: 10.1111/ene.12102. PubMed PMID: 23398333. 175. Wong DF, Moghekar AR, Rigamonti D, Brasic JR, Rousset O, Willis W, et al. An in vivo evaluation of cerebral cortical amyloid with [18F]flutemetamol using positron emission tomography compared with parietal biopsy samples in living normal pressure hydrocephalus patients. Molecular imaging and biology : MIB : the official publication of the Academy of Molecular Imaging. 2013;15(2):230-7. Epub 2012/08/11. doi: 10.1007/s11307-012-0583-x. PubMed PMID: 22878921; PubMed Central PMCID: PMCPMC3936021. 176. Leinonen V, Alafuzoff I, Aalto S, Suotunen T, Savolainen S, Nagren K, et al. Assessment of beta-amyloid in a frontal cortical brain biopsy specimen and by positron emission tomography with carbon 11-labeled Pittsburgh Compound B. Archives of neurology. 2008;65(10):1304-9. Epub 2008/08/13. doi: 10.1001/archneur.65.10.noc80013. PubMed PMID: 18695050. 177. Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, et al. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. NeuroImage. 2006;31(3):1116-28. doi: 10.1016/j.neuroimage.2006.01.015. PubMed PMID: 16545965. 178. Ringstad G, Lindstrom EK, Vatnehol SAS, Mardal KA, Emblem KE, Eide PK. Non-invasive assessment of pulsatile intracranial pressure with phase-contrast magnetic resonance imaging. PloS one. 2017;12(11):e0188896. doi: 10.1371/journal.pone.0188896. PubMed PMID: 29190788; PubMed Central PMCID: PMCPMC5708728. 179. Boone JM. Radiological interpretation 2020: toward quantitative image assessment. Medical physics. 2007;34(11):4173-9. doi: 10.1118/1.2789501. PubMed PMID: 18072481. 180. Jaeger M, Khoo AK, Conforti DA, Cuganesan R. Relationship between intracranial pressure and phase contrast cine MRI derived measures of intracranial pulsations in idiopathic normal pressure hydrocephalus. Journal of clinical neuroscience : official journal of the Neurosurgical Society of Australasia. 2016;33:169-72. Epub 2016/08/16. doi: 10.1016/j.jocn.2016.03.029. PubMed PMID: 27519145. 181. Chen L, Beckett A, Verma A, Feinberg DA. Dynamics of respiratory and cardiac CSF motion revealed with real-time simultaneous multi-slice EPI velocity phase contrast imaging. NeuroImage. 2015;122:281-7. Epub 2015/08/05. doi: 10.1016/j.neuroimage.2015.07.073. PubMed PMID: 26241682; PubMed Central PMCID: PMCPMC4716739.

182. Dreha-Kulaczewski S, Joseph AA, Merboldt KD, Ludwig HC, Gartner J, Frahm J. Identification of the Upward Movement of Human CSF In Vivo and its Relation to the Brain Venous System. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2017;37(9):2395-402. Epub 2017/02/01. doi: 10.1523/jneurosci.2754-16.2017. PubMed PMID: 28137972. 183. Takizawa K, Matsumae M, Sunohara S, Yatsushiro S, Kuroda K. Characterization of cardiac- and respiratory-driven cerebrospinal fluid motion based on asynchronous phase-contrast magnetic resonance imaging in volunteers. Fluids and barriers of the CNS. 2017;14(1):25. doi: 10.1186/s12987- 017-0074-1. PubMed PMID: 28950883; PubMed Central PMCID: PMCPMC5615451. 184. Yildiz S, Thyagaraj S, Jin N, Zhong X, Heidari Pahlavian S, Martin BA, et al. Quantifying the influence of respiration and cardiac pulsations on cerebrospinal fluid dynamics using real-time phase- contrast MRI. Journal of magnetic resonance imaging : JMRI. 2017;46(2):431-9. doi: 10.1002/jmri.25591. PubMed PMID: 28152239. 185. Stecco A, Cassara A, Zuccala A, Anoaica MB, Genovese E, Car PG, et al. Quantitative analysis of cerebrospinal fluid dynamics at phase contrast cine-MRI: predictivity of neurosurgical "Shunt" responsiveness in patients with idiopathic normal pressure hydrocephalus. Journal of neurosurgical sciences. 2017. Epub 2017/09/05. doi: 10.23736/s0390-5616.17.04092-9. PubMed PMID: 28869371. 186. Enzmann DR, Pelc NJ. Brain motion: measurement with phase-contrast MR imaging. Radiology. 1992;185(3):653-60. doi: 10.1148/radiology.185.3.1438741. PubMed PMID: 1438741. 187. Ragunathan S, Pipe JG. Radiofrequency saturation induced bias in aqueductal cerebrospinal fluid flow quantification obtained using two-dimensional cine phase contrast magnetic resonance imaging. Magnetic resonance in medicine. 2017. doi: 10.1002/mrm.26883. PubMed PMID: 28833454. 188. Bradley WG, Jr. Intracranial Pressure versus Phase-Contrast MR Imaging for Normal Pressure Hydrocephalus. AJNR American journal of neuroradiology. 2015;36(9):1631-2. doi: 10.3174/ajnr.A4507. PubMed PMID: 26251438. 189. Florez YN, Moratal D, Forner J, Marti-Bonmati L, Arana E, Guajardo-Hernandez U, et al. Semiautomatic analysis of phase contrast magnetic resonance imaging of cerebrospinal fluid flow through the aqueduct of Sylvius. Magma (New York, NY). 2006;19(2):78-87. Epub 2006/06/17. doi: 10.1007/s10334-006-0030-6. PubMed PMID: 16779563. 190. Kurtcuoglu V, Soellinger M, Summers P, Boomsma K, Poulikakos D, Boesiger P, et al. Computational investigation of subject-specific cerebrospinal fluid flow in the third ventricle and aqueduct of Sylvius. Journal of biomechanics. 2007;40(6):1235-45. Epub 2006/08/15. doi: 10.1016/j.jbiomech.2006.05.031. PubMed PMID: 16904117. 191. Bradley WG, Haughton V, Mardal KA. Cerebrospinal fluid flow in adults. Handbook of clinical neurology. 2016;135:591-601. Epub 2016/07/20. doi: 10.1016/b978-0-444-53485-9.00028-3. PubMed PMID: 27432684. 192. Scollato A, Tenenbaum R, Bahl G, Celerini M, Salani B, Di Lorenzo N. Changes in aqueductal CSF stroke volume and progression of symptoms in patients with unshunted idiopathic normal pressure hydrocephalus. AJNR American journal of neuroradiology. 2008;29(1):192-7. doi: 10.3174/ajnr.A0785. PubMed PMID: 17925364. 193. Chiang WW, Takoudis CG, Lee SH, Weis-McNulty A, Glick R, Alperin N. Relationship between ventricular morphology and aqueductal cerebrospinal fluid flow in healthy and communicating hydrocephalus. Investigative radiology. 2009;44(4):192-9. doi: 10.1097/RLI.0b013e31819a640b. PubMed PMID: 19300098. 194. Algin O, Hakyemez B, Parlak M. The efficiency of PC-MRI in diagnosis of normal pressure hydrocephalus and prediction of shunt response. Academic radiology. 2010;17(2):181-7. Epub 2009/11/17. doi: 10.1016/j.acra.2009.08.011. PubMed PMID: 19910214. 195. Hung AL, Vivas-Buitrago T, Adam A, Lu J, Robison J, Elder BD, et al. Ventriculoatrial versus ventriculoperitoneal shunt complications in idiopathic normal pressure hydrocephalus. Clinical neurology and neurosurgery. 2017;157:1-6. Epub 2017/03/30. doi: 10.1016/j.clineuro.2017.03.014. PubMed PMID: 28347957.

196. Zhao M, Charbel FT, Alperin N, Loth F, Clark ME. Improved phase-contrast flow quantification by three-dimensional vessel localization. Magnetic resonance imaging. 2000;18(6):697-706. Epub 2000/08/10. PubMed PMID: 10930779. 197. Alperin N, Lee SH. PUBS: pulsatility-based segmentation of lumens conducting non-steady flow. Magnetic resonance in medicine. 2003;49(5):934-44. Epub 2003/04/22. doi: 10.1002/mrm.10455. PubMed PMID: 12704777. 198. Haughton V, Mardal KA. Spinal fluid biomechanics and imaging: an update for neuroradiologists. AJNR American journal of neuroradiology. 2014;35(10):1864-9. doi: 10.3174/ajnr.A4023. PubMed PMID: 25012674. 199. Yiallourou TI, Kroger JR, Stergiopulos N, Maintz D, Martin BA, Bunck AC. Comparison of 4D phase-contrast MRI flow measurements to computational fluid dynamics simulations of cerebrospinal fluid motion in the cervical spine. PloS one. 2012;7(12):e52284. Epub 2013/01/04. doi: 10.1371/journal.pone.0052284. PubMed PMID: 23284970; PubMed Central PMCID: PMCPMC3528759. 200. Bartlett JW, Frost C. Reliability, repeatability and reproducibility: analysis of measurement errors in continuous variables. Ultrasound in obstetrics & gynecology : the official journal of the International Society of Ultrasound in Obstetrics and Gynecology. 2008;31(4):466-75. Epub 2008/02/29. doi: 10.1002/uog.5256. PubMed PMID: 18306169. 201. Gatehouse PD, Rolf MP, Graves MJ, Hofman MB, Totman J, Werner B, et al. Flow measurement by cardiovascular magnetic resonance: a multi-centre multi-vendor study of background phase offset errors that can compromise the accuracy of derived regurgitant or shunt flow measurements. Journal of cardiovascular magnetic resonance : official journal of the Society for Cardiovascular Magnetic Resonance. 2010;12:5. Epub 2010/01/16. doi: 10.1186/1532-429x-12-5. PubMed PMID: 20074359; PubMed Central PMCID: PMCPMC2818657. 202. Bateman GA. Correction to "External hydrocephalus in infants: six cases with MR venogram and flow quantification correlation". Child's nervous system : ChNS : official journal of the International Society for Pediatric Neurosurgery. 2011;27(12):2033-4. Epub 2011/10/19. doi: 10.1007/s00381-011-1610-y. PubMed PMID: 22006365. 203. Feinberg DA, Mark AS. Human brain motion and cerebrospinal fluid circulation demonstrated with MR velocity imaging. Radiology. 1987;163(3):793-9. Epub 1987/06/01. doi: 10.1148/radiology.163.3.3575734. PubMed PMID: 3575734. 204. Greitz D, Wirestam R, Franck A, Nordell B, Thomsen C, Stahlberg F. Pulsatile brain movement and associated hydrodynamics studied by magnetic resonance phase imaging. The Monro-Kellie doctrine revisited. Neuroradiology. 1992;34(5):370-80. Epub 1992/01/01. PubMed PMID: 1407513. 205. Meierhofer C, Lyko C, Schneider EP, Stern H, Martinoff S, Hess J, et al. Baseline correction does not improve flow quantification in phase-contrast velocity measurement for routine clinical practice. Clinical imaging. 2015;39(3):427-31. Epub 2015/02/11. doi: 10.1016/j.clinimag.2014.12.010. PubMed PMID: 25661574. 206. Kværness J. E-mail from Jørn Kværness, Clinical Science Manager Nordic in Philips Healthcare. 2017. 207. Bateman GA, Brown KM. The measurement of CSF flow through the aqueduct in normal and hydrocephalic children: from where does it come, to where does it go? Child's nervous system : ChNS : official journal of the International Society for Pediatric Neurosurgery. 2012;28(1):55-63. doi: 10.1007/s00381-011-1617-4. PubMed PMID: 22038153. 208. Penn RD, Basati S, Sweetman B, Guo X, Linninger A. Ventricle wall movements and cerebrospinal fluid flow in hydrocephalus. Journal of neurosurgery. 2011;115(1):159-64. doi: 10.3171/2010.12.JNS10926. PubMed PMID: 21275563. 209. Bateman GA, Napier BD. External hydrocephalus in infants: six cases with MR venogram and flow quantification correlation. Child's nervous system : ChNS : official journal of the International Society for Pediatric Neurosurgery. 2011;27(12):2087-96. doi: 10.1007/s00381-011-1549-z. PubMed PMID: 21833725.

210. Yin LK, Zheng JJ, Zhao L, Hao XZ, Zhang XX, Tian JQ, et al. Reversed aqueductal cerebrospinal fluid net flow in idiopathic normal pressure hydrocephalus. Acta neurologica Scandinavica. 2017;136(5):434-9. doi: 10.1111/ane.12750. PubMed PMID: 28247411. 211. Dandy W. The cause of so-called idiopathic hydrocephalus. Bulletin of the Johns Hopkins Hospital. 1921;32(361):67-75. 212. Igarashi H, Tsujita M, Kwee IL, Nakada T. Water influx into cerebrospinal fluid is primarily controlled by aquaporin-4, not by aquaporin-1: 17O JJVCPE MRI study in knockout mice. Neuroreport. 2014;25(1):39-43. doi: 10.1097/WNR.0000000000000042. PubMed PMID: 24231830; PubMed Central PMCID: PMCPMC4235386. 213. McComb JG. Recent research into the nature of cerebrospinal fluid formation and absorption. Journal of neurosurgery. 1983;59(3):369-83. Epub 1983/09/01. doi: 10.3171/jns.1983.59.3.0369. PubMed PMID: 6886750. 214. Milhorat TH, Mosher MB, Hammock MK, Murphy CF. Evidence for choroid-plexus absorption in hydrocephalus. The New England journal of medicine. 1970;283(6):286-9. Epub 1970/08/06. doi: 10.1056/nejm197008062830604. PubMed PMID: 5427058. 215. Eisenberg HM, McLennan JE, Welch K. Ventricular perfusion in cats with kaolin-induced hydrocephalus. Journal of neurosurgery. 1974;41(1):20-8. Epub 1974/07/01. doi: 10.3171/jns.1974.41.1.0020. PubMed PMID: 4834193. 216. Spector R, Keep RF, Robert Snodgrass S, Smith QR, Johanson CE. A balanced view of choroid plexus structure and function: Focus on adult humans. Experimental neurology. 2015;267:78-86. Epub 2015/03/10. doi: 10.1016/j.expneurol.2015.02.032. PubMed PMID: 25747036. 217. Hladky SB, Barrand MA. Mechanisms of fluid movement into, through and out of the brain: evaluation of the evidence. Fluids and barriers of the CNS. 2014;11(1):26. doi: 10.1186/2045-8118- 11-26. PubMed PMID: 25678956; PubMed Central PMCID: PMC4326185. 218. Morris AW, Sharp MM, Albargothy NJ, Fernandes R, Hawkes CA, Verma A, et al. Vascular basement membranes as pathways for the passage of fluid into and out of the brain. Acta neuropathologica. 2016;131(5):725-36. doi: 10.1007/s00401-016-1555-z. PubMed PMID: 26975356; PubMed Central PMCID: PMCPMC4835509. 219. Abbott NJ. Evidence for bulk flow of brain interstitial fluid: significance for physiology and pathology. Neurochem Int. 2004;45(4):545-52. doi: 10.1016/j.neuint.2003.11.006. PubMed PMID: 15186921. 220. Engelhardt B, Vajkoczy P, Weller RO. The movers and shapers in immune privilege of the CNS. Nat Immunol. 2017;18(2):123-31. doi: 10.1038/ni.3666. PubMed PMID: 28092374. 221. 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. doi: 10.1038/nature14432. PubMed PMID: 26030524; PubMed Central PMCID: PMC4506234. 222. 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. The Journal of experimental medicine. 2015;212(7):991-9. doi: 10.1084/jem.20142290. PubMed PMID: 26077718; PubMed Central PMCID: PMC4493418. 223. Huang TY, Chung HW, Chen MY, Giiang LH, Chin SC, Lee CS, et al. Supratentorial cerebrospinal fluid production rate in healthy adults: quantification with two-dimensional cine phase-contrast MR imaging with high temporal and spatial resolution. Radiology. 2004;233(2):603-8. doi: 10.1148/radiol.2332030884. PubMed PMID: 15516623. 224. Asgari M, de Zelicourt D, Kurtcuoglu V. Glymphatic solute transport does not require bulk flow. Sci Rep. 2016;6:38635. doi: 10.1038/srep38635. PubMed PMID: 27929105; PubMed Central PMCID: PMCPMC5144134. 225. Levine DN. Intracranial pressure and ventricular expansion in hydrocephalus: have we been asking the wrong question? Journal of the neurological sciences. 2008;269(1-2):1-11. doi: 10.1016/j.jns.2007.12.022. PubMed PMID: 18234229.

226. Linninger AA, Tsakiris C, Zhu DC, Xenos M, Roycewicz P, Danziger Z, et al. Pulsatile cerebrospinal fluid dynamics in the human brain. IEEE transactions on bio-medical engineering. 2005;52(4):557-65. doi: 10.1109/TBME.2005.844021. PubMed PMID: 15825857. 227. Bedussi B, van der Wel NN, de Vos J, van Veen H, Siebes M, VanBavel E, et al. Paravascular channels, cisterns, and the subarachnoid space in the rat brain: A single compartment with preferential pathways. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism. 2016. doi: 10.1177/0271678X16655550. PubMed PMID: 27306753. 228. Jin BJ, Smith AJ, Verkman AS. Spatial model of convective solute transport in brain extracellular space does not support a "glymphatic" mechanism. The Journal of general physiology. 2016;148(6):489-501. doi: 10.1085/jgp.201611684. PubMed PMID: 27836940; PubMed Central PMCID: PMC5129742. 229. Holter KE, Kehlet B, Devor A, Sejnowski TJ, Dale AM, Omholt SW, et al. Interstitial solute transport in 3D reconstructed neuropil occurs by diffusion rather than bulk flow. Proc Natl Acad Sci U S A. 2017;114(37):9894-9. doi: 10.1073/pnas.1706942114. PubMed PMID: 28847942; PubMed Central PMCID: PMCPMC5604020. 230. Wolak DJ, Thorne RG. Diffusion of macromolecules in the brain: implications for drug delivery. Molecular pharmaceutics. 2013;10(5):1492-504. Epub 2013/01/10. doi: 10.1021/mp300495e. PubMed PMID: 23298378; PubMed Central PMCID: PMCPMC3646902. 231. Smith AJ, Yao X, Dix JA, Jin BJ, Verkman AS. Test of the 'glymphatic' hypothesis demonstrates diffusive and aquaporin-4-independent solute transport in rodent brain parenchyma. eLife. 2017;6. Epub 2017/08/23. doi: 10.7554/eLife.27679. PubMed PMID: 28826498; PubMed Central PMCID: PMCPMC5578736. 232. Pardridge WM. Drug transport in brain via the cerebrospinal fluid. Fluids and barriers of the CNS. 2011;8(1):7. doi: 10.1186/2045-8118-8-7. PubMed PMID: 21349155; PubMed Central PMCID: PMCPMC3042981. 233. Pizzo ME, Wolak DJ, Kumar NN, Brunette E, Brunnquell CL, Hannocks M-J, et al. Intrathecal antibody distribution in the rat brain: surface diffusion, perivascular transport, and osmotic enhancement delivery. The Journal of Physiology Accepted Author Manuscript. 2017. doi: http://dx.doi.org/10.1113/JP275105. 234. Fuxe K, Borroto-Escuela DO. Volume transmission and receptor-receptor interactions in heteroreceptor complexes: understanding the role of new concepts for brain communication. Neural Regen Res. 2016;11(8):1220-3. doi: 10.4103/1673-5374.189168. PubMed PMID: 27651759; PubMed Central PMCID: PMCPMC5020810. 235. Li X, Han Y, Xu H, Sun Z, Zhou Z, Long X, et al. Aquaporin 4 expression and ultrastructure of the blood-brain barrier following cerebral contusion injury. Neural Regen Res. 2013;8(4):338-45. doi: 10.3969/j.issn.1673-5374.2013.04.006. PubMed PMID: 25206674; PubMed Central PMCID: PMCPMC4107528. 236. Tourdias T, Dragonu I, Fushimi Y, Deloire MS, Boiziau C, Brochet B, et al. Aquaporin 4 correlates with apparent diffusion coefficient and hydrocephalus severity in the rat brain: a combined MRI-histological study. NeuroImage. 2009;47(2):659-66. Epub 2009/05/05. doi: 10.1016/j.neuroimage.2009.04.070. PubMed PMID: 19409501. 237. Manley GT, Binder DK, Papadopoulos MC, Verkman AS. New insights into water transport and edema in the central nervous system from phenotype analysis of aquaporin-4 null mice. Neuroscience. 2004;129(4):983-91. Epub 2004/11/25. doi: 10.1016/j.neuroscience.2004.06.088. PubMed PMID: 15561413. 238. Yao X, Hrabetova S, Nicholson C, Manley GT. Aquaporin-4-deficient mice have increased extracellular space without tortuosity change. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2008;28(21):5460-4. doi: 10.1523/JNEUROSCI.0257-08.2008. PubMed PMID: 18495879; PubMed Central PMCID: PMCPMC2659334. 239. Amiry-Moghaddam M, Otsuka T, Hurn PD, Traystman RJ, Haug FM, Froehner SC, et al. An alpha-syntrophin-dependent pool of AQP4 in astroglial end-feet confers bidirectional water flow 86 between blood and brain. Proc Natl Acad Sci U S A. 2003;100(4):2106-11. Epub 2003/02/13. doi: 10.1073/pnas.0437946100. PubMed PMID: 12578959; PubMed Central PMCID: PMCPMC149966. 240. Iliff JJ, Wang M, Zeppenfeld DM, Venkataraman A, Plog BA, Liao Y, et al. Cerebral arterial pulsation drives paravascular CSF-interstitial fluid exchange in the murine brain. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2013;33(46):18190-9. doi: 10.1523/JNEUROSCI.1592-13.2013. PubMed PMID: 24227727; PubMed Central PMCID: PMC3866416. 241. Rennels ML, Blaumanis OR, Grady PA. Rapid solute transport throughout the brain via paravascular fluid pathways. Adv Neurol. 1990;52:431-9. PubMed PMID: 2396537. 242. Iliff JJ, Chen MJ, Plog BA, Zeppenfeld DM, Soltero M, Yang L, et al. Impairment of glymphatic pathway function promotes tau pathology after traumatic brain injury. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2014;34(49):16180-93. Epub 2014/12/05. doi: 10.1523/jneurosci.3020-14.2014. PubMed PMID: 25471560; PubMed Central PMCID: PMCPMC4252540. 243. Laske C. Clinical and Biomarker Changes in Alzheimer's Disease New England Journal of Medicine. 2012;367(21):2050-2. doi: 10.1056/NEJMc1211767. 244. Burfeind KG, Murchison CF, Westaway SK, Simon MJ, Erten-Lyons D, Kaye JA, et al. The effects of noncoding aquaporin-4 single-nucleotide polymorphisms on cognition and functional progression of Alzheimer's disease. Alzheimer's & dementia (New York, N Y). 2017;3(3):348-59. Epub 2017/10/27. doi: 10.1016/j.trci.2017.05.001. PubMed PMID: 29067342; PubMed Central PMCID: PMCPMC5651426. 245. Zeppenfeld DM, Simon M, Haswell J, et al. Association of perivascular localization of aquaporin-4 with cognition and alzheimer disease in aging brains. JAMA Neurology. 2017;74(1):91-9. doi: 10.1001/jamaneurol.2016.4370. 246. Shibata M, Yamada S, Kumar SR, Calero M, Bading J, Frangione B, et al. Clearance of Alzheimer's amyloid-ss(1-40) peptide from brain by LDL receptor-related protein-1 at the blood-brain barrier. The Journal of clinical investigation. 2000;106(12):1489-99. Epub 2000/12/20. doi: 10.1172/jci10498. PubMed PMID: 11120756; PubMed Central PMCID: PMCPMC387254. 247. Montagne A, Zhao Z, Zlokovic BV. Alzheimer's disease: A matter of blood-brain barrier dysfunction? The Journal of experimental medicine. 2017. Epub 2017/10/25. doi: 10.1084/jem.20171406. PubMed PMID: 29061693. 248. Ringstad G, Emblem KE, Geier O, Alperin N, Eide PK. Reply: To PMID 25977480. AJNR American journal of neuroradiology. 2015;36(9):1633-4. doi: 10.3174/ajnr.A4488. PubMed PMID: 26251437. 249. Fagerland M. Evidence-Based Medicine and Systematic Reviews. In: Laake P, Benestad H, Olsen B, editors. Research in Medical and Biological Sciences: Academic Press Elsevier Ltd.; 2015. p. 453-6. 250. Hofmann B, Holm S. Philosophy of Science. In: Laake P, Benestad H, Olsen B, editors. Research in Medical and Biological Sciences. 1st ed: Elsevier, Academic Press; 2015. p. 1-41. 251. Ioannidis JP. Why most published research findings are false. PLoS medicine. 2005;2(8):e124. Epub 2005/08/03. doi: 10.1371/journal.pmed.0020124. PubMed PMID: 16060722; PubMed Central PMCID: PMCPMC1182327. 252. Forskningsrådet. Tillit til forskningen: -Flere må ta ansvar. Forskningsrådet; 2017.

ORIGINAL RESEARCH ADULT BRAIN

Aqueductal Stroke Volume: Comparisons with Intracranial Pressure Scores in Idiopathic Normal Pressure Hydrocephalus

X G. Ringstad, K.E. Emblem, O. Geier, N. Alperin, and P.K. Eide

ABSTRACT

BACKGROUND AND PURPOSE: Aqueductal stroke volume from phase-contrast MR imaging has been proposed for predicting shunt response in normal pressure hydrocephalus. However, this biomarker has remained controversial in use and has a lack of validation with invasive intracranial monitoring. We studied how aqueductal stroke volume compares with intracranial pressure scores in the presurgical work-up and clinical score, ventricular volume, and aqueduct area and assessed the patient’s response to shunting.

MATERIALS AND METHODS: Phase-contrast MR imaging was performed in 21 patients with probable idiopathic normal pressure hydrocephalus. Patients were selected for shunting on the basis of pathologically increased intracranial pressure pulsatility. Patients with shunts were offered a second MR imaging after 12 months. Ventricular volume and transverse aqueductal area were calculated, as well as clinical symptom score.

RESULTS: No correlations between aqueductal stroke volume and preoperative scores of mean intracranial pressure or mean wave amplitudes were observed. Preoperative aqueductal stroke volume was not different between patients with shunts and conservatively treated patients (P ϭ .69) but was correlated with ventricular volume (R ϭ 0.60, P ϭ .004) and aqueductal area (R ϭ 0.58, P ϭ .006) but not with the severity or duration of clinical symptoms. After shunting, aqueductal stroke volume (P ϭ .006) and ventricular volume (P ϭ .002) were reduced. A clinical improvement was seen in 16 of 17 patients who had shunts (94%).

CONCLUSIONS: Aqueductal stroke volume does not reﬂect intracranial pressure pulsatility or symptom score, but rather aqueduct area and ventricular volume. The results do not support the use of aqueductal stroke volume for selecting patients for shunting.

ABBREVIATIONS: ASV ϭ aqueductal stroke volume; ICP ϭ intracranial pressure; iNPH ϭ idiopathic normal pressure hydrocephalus; MWA ϭ mean ICP wave amplitude; PCMR ϭ phase-contrast MR imaging

diopathic normal pressure hydrocephalus (iNPH) is character- clinical response to this treatment has, accordingly, been reported Iized by dementia, incontinence, and gait disturbance1 and can to range from 29% to 90%.3-5 Previous investigators have, there- be treated by ventriculoperitoneal shunt surgery. However, the fore, sought to establish noninvasive parameters from MR imag- disease can be difficult to separate from other neurodegenerative ing studies. After Bradley et al6 first reported an increased CSF disorders such as Alzheimer and Parkinson diseases.2 Selection flow void in the aqueduct associated with a favorable shunt re- criteria for surgical shunting have been heterogeneous, and the sponse, studies using phase-contrast MR imaging (PCMR) have demonstrated aqueductal flow parameters, and in particular the aqueductal stroke volume (ASV), to be useful in the diagnosis and 7-13 Received December 15, 2014; accepted after revision February 11, 2015. selection of patients for shunt surgery and in the follow-up of From the Department of Radiology and Nuclear Medicine (G.R.) and Intervention patients with shunts.14 However, other studies do not reproduce Centre (K.E.E., O.G.), Oslo University Hospital-Rikshospitalet, Oslo, Norway; De- 15-18 partment of Radiology (N.A.), University of Miami Miller School of Medicine, Mi- the beneficial utility of measuring aqueductal flow ; therefore, ami, Florida; Department of Neurosurgery (P.K.E.), Oslo University Hospital, Oslo, the use of ASV in iNPH remains disputed. Accordingly, compar- Norway; and Faculty of Medicine (P.K.E.), University of Oslo, Oslo, Norway. isons with invasive intracranial measurements are warranted. Paper previously presented in part at: Annual Meeting of the Radiological Society of North America, December 1–6, 2013; Chicago, Illinois (No. SSM14-05, RSNA ID: Continuous monitoring of intracranial pressure (ICP) and 13021973). single cardiac-induced ICP waves in patients with iNPH has re- Please address correspondence to Geir Ringstad, MD, Department of Radiology vealed elevated mean ICP wave amplitudes (MWAs) in those re- and Nuclear Medicine, Oslo University Hospital-Rikshospitalet, Postboks 4950 Ny- 19,20 dalen, 0424 Oslo, Norway; e-mail address: [email protected] sponding to shunts, compared with those not responding. Indicates open access to non-subscribers at www.ajnr.org Hence, in this hospital, diagnostic preoperative monitoring of http://dx.doi.org/10.3174/ajnr.A4340 MWA is used routinely and predicts a beneficial shunt response in

AJNR Am J Neuroradiol 36:1623–30 Sep 2015 www.ajnr.org 1623 9 of 10 patients with iNPH.19 Elevated MWA is indicative of re- The MR imaging protocol also included 3D T1-weighted ul- duced intracranial compliance (ie, reduced pressure-volume re- trafast gradient echo, acquisition matrix ϭ 256 ϫ 256 ϫ 192 with serve capacity),21,22 which may be a pathophysiologic mechanism voxel size 1.0/1.0/1.0 mm3, flip angle ϭ 7, TR/TE ϭ 8.6/2.3 ms, behind iNPH.19,23-25 Increased ASV in iNPH has been attributed which was used for segmentation of the supratentorial ventricles to reduced intracranial compliance.26 If ASV should express re- for the volumetric analysis. duced intracranial compliance and thereby predict shunt response, a positive correlation between ASV and MWA seems MR Image Postprocessing reasonable. All examinations were postprocessed by using Q-flow software The purpose of this study was therefore to compare ASV from (Philips Healthcare). The aqueduct was manually defined in all phase-contrast MR imaging with preoperative ICP scores, clinical the phase images with an ROI in each section (Fig 1D) by a neu- normal pressure hydrocephalus scores, and MR imaging–derived roradiologist with 7 years of experience who was blinded to clin- ventricular volume and aqueductal area in patients with iNPH ical data. Care was taken not to include nonmoving tissue ele- before and after shunting. ments in the imaging plane to avoid background noise in the MR imaging signal. MATERIALS AND METHODS ASV was estimated after correction for potential aliasing by Patient Inclusion and Study Design sinusoid curve fitting and was defined as the mean of systolic and The study was approved by the institutional review board of Oslo diastolic volumetric CSF flow during 1 cardiac cycle minus net University Hospital. Inclusion was by written and oral informed flow (Fig 1B). consent. Calculation of supratentorial ventricular volume (referred to as Of 28 consecutive patients with iNPH undergoing assessment “ventricular volume”) was performed by a 3D medical image seg- 28 for probable iNPH within the department of neurosurgery, 7 pa- mentation software ITK-SNAP 2.4 (www.itksnap.org), which pro- tients were excluded because of motion artifacts at PCMR and/or vides semiautomatic segmentation by using active contour methods. termination of the examination before completion of the PCMR For the segmentation, the region competition method by Zhu and 29 sequence. Accordingly, 21 patients with iNPH (10 women, 11 Yuille was used. The segmentation result was controlled visually men; range, 56–84 years) with successful PCMR were included in and, if necessary, corrected manually (in ITK-SNAP). this prospective observational study. The assessment included clinical examination with determination of iNPH symptom se- ICP Monitoring verity, PCMR, and overnight ICP monitoring. The patients with All patients underwent continuous overnight ICP monitoring. As 19 iNPH who underwent shunt surgery were invited to a second previously described in detail, an ICP sensor was placed in the PCMR after 12 months. brain parenchyma through a small burr-hole in the skull with the patient under local anesthesia. Overnight monitoring was done in Clinical Management the patient ward by using a computerized system (Sensometrics The assessment of patients with iNPH for shunt surgery fol- Technology; dPCom AS, Oslo, Norway) for automatic identifica- lowed the clinical routine at the department of neurosurgery. tion of individual cardiac-induced single ICP waves. The ampli- Clinical grading of the severity of symptoms was performed by tude of the ICP wave was identified as the pressure difference using the normal pressure hydrocephalus grading scale of this between the systolic maximum and diastolic minimum pressures department,19 which assesses the combined severity of gait dis- (Fig 1A). The mean ICP wave amplitude is determined as the turbance, urinary incontinence, and dementia. Each compo- average of all single ICP waves during consecutive 6-second time nent is graded from 1 to 5, with a possible total score of 3 intervals, while the mean ICP is the average of absolute ICP rela- (worst) to 15 (best). According to the institutional routine, the tive to a zero pressure level. For the patients in this study, the decision for shunt surgery is based on a combination of clinical MWA and mean ICP values were determined for the 6-second assessment, radiologic assessment, and the results of continu- time windows from 11 PM to 7 AM (ie, 4 800 6-second time win- ous ICP monitoring.19 dows). During this recording period, both the average of MWA The shunt response was defined as an increase by at least 2 points and mean ICP were determined, as well as the percentage of MWA on the normal pressure hydrocephalus grading scale, and the clinical of Ն5 mm Hg and the percentage of mean ICP of Ն15 mm Hg, score was assessed at regular intervals, 3, 6, and 12 months (including during the recording period. imaging after 12 months) following shunting. According to the institutional routine, primarily the MWA is used for selection to shunting. Threshold levels of MWA repre- MR Image Acquisition senting an indication for shunting are MWA, on average, of Ն4 The techniques for CSF velocity imaging with MR imaging have mm Hg and/or the percentage of MWA of Ն5mmHginՆ10% of previously been described in detail.27 recording time. MR imaging was performed on a 3T Achieva system (Philips Healthcare, Best, the Netherlands). The MR imaging parameters for Statistical Analysis aqueductal flow were TR ϭ 24 ms and TE ϭ 16 ms, voxel size ϭ Under the assumption of normal distribution, correlations were 0.60 ϫ 0.80 ϫ 4.00 mm3, velocity encoding ϭ 10 cm/s, and 30–40 determined by the Pearson correlation coefficient, and pre- and phases with retrospective peripheral cardiac gating. The scan was postsurgical values of ASV and ventricular volume were com- obtained in a section perpendicular to the aqueduct (Fig 1C). pared by using a paired-samples t test. Comparison of ASV be- 1624 Ringstad Sep 2015 www.ajnr.org FIG 1. The study compared aqueductal flow–based pulsatility, expressed by ASV, with pressure-based intracranial pulsatility, expressed by MWA. A, Single cardiac-induced ICP wave, the MWA, is determined as the average of amplitudes from single ICP waves during consecutive 6-second time intervals. B, ASV is defined as the mean of systolic and diastolic volumetric CSF flow in the aqueduct during 1 cardiac cycle (area under/over the flow curve) minus net aqueductal flow. C, The aqueductal flow curve is based on section orientation (red line) perpendicular to the aqueduct. D, PCMR with manually drawn ROI (red circle) defines the aqueduct. Table 1: Patient data Idiopathic Normal Pressure Hydrocephalus Total Population Shunt Group Conservative Group Median (Range) Median (Range) Median (Range) No. 21 17 4 Age (yr) 74 (56–84) 74 (56–84) 74 (60–79) Sex (female/male) 10:11 8:9 2:2 Preoperative clinical state NPH score19 10 (4–13) 10 (4–13) 11 (9–13) Duration of symptoms (yr) 2 (0.5–10) 2 (1.0–10.0) 0.8 (0.5–10.0) Postoperative clinical state 12 mo after surgery NPH score 12 (8–15) 12 (8–15) 8 Note:—NPH indicates normal pressure hydrocephalus. tween patients with shunts (shunt group) and conservatively MR Imaging Data and ICP Scores treated patients (conservative group) was performed with an in- Table 2 presents the PCMR-derived ASV and the ventricular vol- dependent samples t test. The significance level was set to .05. ume and aqueduct area before/after shunting and the preopera- Statistical analysis was performed by using SPSS Statistics, Ver- tive ICP scores of the 21 patients with iNPH. sion 20 (IBM, Armonk, New York). As further illustrated in Fig 2, ASV before surgery did not differ between patients found eligible (shunt group) or noneligible RESULTS (conservative group) for shunting (P ϭ .69). Patients In the shunt group, ASV was reduced from a median of 111 Table 1 presents patient data. ␮L before to a median of 68 ␮L after surgery (P ϭ .01, Fig 2A), Among the 17/21 (81%) patients selected for shunting, 16/17 while the ventricular volume was a median of 137 mL before patients with shunts (94%) improved clinically (shunt respond- and a median of 105 mL after surgery (P ϭ .001, Fig 2B). There ers), while 1 (6%) had no clinical improvement (shunt nonre- was no significant change in the aqueductal area after shunting sponder). Among the 12 patients with PCMR after shunting, 11 (P ϭ .94). patients (92%) were responders. There was a positive correlation between ASV and ventricular AJNR Am J Neuroradiol 36:1623–30 Sep 2015 www.ajnr.org 1625 Table 2: MRI-derived measures and ICP scores Idiopathic Normal Pressure Hydrocephalus Total Population (n = 21) Shunt Group (n = 17) Conservative Group (n =4) Median (Range) Median (Range) Median (Range) MRI measures Aqueduct stroke volume (␮L) Before shunt 111 (26–417) 109 (26–417) 130 (36–163) After shunt 68 (17–201) – Ventricular volume (ml) Before shunt 138 (41–266) 137 (41–266) 143 (105–152) After shunt 105 (34–230) – Aqueductal area (mm2) Before shunt 14 (9–38) 14 (9–38) 13 (9–36) After shunt 12 (9–58) – ICP scores Mean ICP (mm Hg) Average 6.1 (Ϫ1.8–11.9) 6.1 (Ϫ1.8–11.9) 5.8 (3.0–8.8) Percentage Ն15 mm Hg 0 (0–2) 0 (0–2) 0 (0–0) Mean ICP wave amplitude (MWA, mm Hg) Average 4.5 (3.1–7.9) 4.7 (3.5–7.9) 3.4 (3.1–5.2) Percentage Ն5 mm Hg 26 (1–100) 27 (2–100) 3 (1–58) volume before surgery (R ϭ 0.60, P ϭ .004; Fig 3A) and after response in 16 of 17 (94%) patients in this study. A positive shunting (R ϭ 0.73, P ϭ .007; Fig 3B). Moreover, there was also a shunt response can be considered a marker of “true” iNPH, positive correlation between ASV and aqueduct area before sur- and the high response rate suggests that our study cohort was gery (R ϭ 0.58, P ϭ .006). We did not find any significant corre- representative of iNPH and contained few patients with dis- lation either between ASV and preoperative normal pressure hy- eases that clinically might present similarly, with similar ASV drocephalus scores (R ϭ 0.29, P ϭ .21) or between ASV and the values.32 We reproduced previous findings of ASV being ele- duration of iNPH symptoms (R ϭ 0.26, P ϭ .26). vated in patients with iNPH compared with healthy controls33 While the mean ICP was comparable between the shunt and and in contrast to elderly patients in general, in whom ASV has conservative groups, MWA was elevated in the surgery group (Ta- been reported to be reduced.34 However, ASV varied over a ble 2). There were no correlations between ASV and mean ICP wide range in patients found both eligible and noneligible for (Fig 4A), or ASV and MWA (Fig 4B). In addition, no correlations surgical shunting on the basis of MWA, and ASV values over- Ն between ASV and the percentage of mean ICP of 15 mm Hg or lapped between the groups. Accordingly, our results question Ն between ASV and the percentage of MWA of 5 mm Hg were the clinical utility of the ASV parameter, both with respect to observed. its ability to diagnose iNPH and its predictive value for a clinically favorable shunt response. DISCUSSION There may be several reasons why ASV does not compare with The main observation of this study was that CSF flow–based MWA. First, as for all PCMR measurements, ASV is obtained pulsatility expressed by ASV did not compare with intracranial from a time window of just a few minutes, while MWA represents pressure pulsatility expressed by MWA. On the other hand, a mean value from an 8-hour time window with registration of ASV correlated with ventricular volume and aqueduct area. several thousand single waves, where frequent physiologic pres- Even though PCMR-based aqueductal flow has previously sure fluctuations during the recording period are typically ob- been extensively investigated and advocated by some as a tool served.19,35 For ASV to be a valid marker of intracranial pulsatil- for selection of patients for shunt surgery in iNPH, comparisons with invasive intracranial measurements have been very ity, one would have to assume that the limited time window, from limited. To our knowledge, there is 1 previous study that com- which it is obtained, demonstrates values that are representative pared PCMR-derived ASV with ICP monitoring.30 This study of the underlying condition. To our knowledge, no previous ob- reported an association between ASV and a temporal subpeak servations support such an assumption being valid. of the ICP wave, but the result was based on a small cohort of 7 Several factors might influence aqueductal flow. It has previ- patients and the clinical significance of the findings has been ously been suggested that the systolic, inward expansion of the disputed.31 brain against enlarged ventricles is a fundamental cause of in- 26 In our study, a high proportion of the patients in the iNPH creased ASV, before irreversible atrophy occurs. Other contri- cohort had signs of reduced intracranial compliance by in- butions might be from the difference in CSF pressure between the creased MWA after overnight ICP monitoring (17/21). MWA third and fourth ventricles, the heart rate, and the aqueductal of Ն4 mm Hg or the percentage of MWA of Ն5mmHgin geometry and mainly at which level flow in the aqueduct is as- Ն10% of recording time or both were previously reported to sessed.36 In this study, a strong association between aqueductal predict shunt response in iNPH19 and have been considered as area and ASV was found; however, this does not necessarily imply indicative of impaired intracranial compliance.21 Using the causality. As pointed out by Chiang et al,33 it seems reasonable MWA for selection for shunt treatment gave an excellent shunt that the aqueduct can adapt to increased flow, similar to the ad- 1626 Ringstad Sep 2015 www.ajnr.org FIG 3. For patients with iNPH, the correlations between ASV and ventricular volume before (n ϭ 21) (A) and after (n ϭ 12) (B) shunt surgery are presented. The Pearson correlation coefficients and significance levels are presented in the plots.

be primarily influenced by reduced ventricular size, rather than reflecting a clinical response, contrary to what has been suggested 14 FIG 2. ASV (A) and ventricular volume (B) are presented for patients previously. ASV also correlated positively with ventricular vol- with shunts and iNPH before (n ϭ 17) and after (n ϭ 12) shunting ume before shunting; this correlation confirms previous observa- (surgery group) and for conservatively managed patients with iNPH 33 ϭ tions but is contradictory to a more recent study by Chaarani (conservative group, n 4) before management. Significance levels 36 are presented in the plots. et al. In our study, the statistical significance of this positive correlation was dependent on 1 patient with extreme values, both aptation of the blood vessel lumen area to maintain a wall shear ASV and ventricular volume. stress within a normal range.37 The lack of correlation between ASV and symptom severity ASV and ventricular volume declined after shunting; this could theoretically be due to a decline in ASV as a sign of long- change is consistent with that in previous studies,14,38 while aq- standing progressive cerebral ischemic changes and atrophy, ueductal area did not change. While ASV did not reflect the clin- making the iNPH irreversible, which was postulated by Scollato ical severity of iNPH preshunting and did not compare with any et al,39 who followed patients with unshunted iNPH and found a pressure parameters, reduced ASV after shunting might therefore decline in ASV after typically 18–20 months of symptom dura- AJNR Am J Neuroradiol 36:1623–30 Sep 2015 www.ajnr.org 1627 Limitations Some limitations with this study should be noted. The patients were too few to determine the accuracy of ASV as a diagnostic test in iNPH, especially due to the small number of patients in sub- groups such as the conservatively managed group (n ϭ 4) and the nonresponsive group with shunts (n ϭ 1). A lack of correlations between tested variables may have been a reflection of few study subjects. Finally, the statistical correlations that were demonstrated in the study do not necessarily imply causality between the variables. The reported accuracy of PCMR volumetric flow measurements in pulsatile flow is within 2.8%,42 and calculation of aqueductal stroke volume is less sensitive to inaccuracies from manual selection of ROIs than is the calculation of flow velocity.27,33 However, measurements of aqueductal flow can be influenced by flow aliasing, which is characterized by its apparent high velocity in the opposite direction to the average velocity in the defined area of interest. This was corrected for with the same algorithm applied to all patients, as described in the “Materials and Methods” section. Another limitation might be the PCMR resolution with a pixel size of 0.60 ϫ 0.80 mm2 in the transverse plane, which is lower than that in the previous study of Bradley et al,9 supporting ASV as a shunt predictor. The inability to find the utility of ASV in our study could therefore have been influenced by inferior image resolution. However, other studies demonstrating a beneficial use of ASV in iNPH have used a pixel size comparable8,10,11,13 or infe- rior12 to that applied in our study. While reducing the pixel size would reduce the number of pixels from nonmoving tissue elements being included in the ROI defining the aqueduct, the use of larger pixels improves the signal-to-noise ratio and even more by use of a 3T magnetic field strength, as in our study, compared with 1.5T. A reference ROI can be placed in the adjacent cerebral pedun- cle to rule out partial volume effect and mass brain movement during aqueductal flow measurement with PCMR. Such a correction was not applied in this study. Contribution from mass brain movement to the ASV value has been reported to be small though7 and should not be expected to influence the results of the current measurements substantially. FIG 4. For patients with iNPH (n ϭ 21), the correlations between ASV and mean ICP (A) and ASV and MWA (B) before shunt surgery are presented. The Pearson correlation coefficients and significance lev- CONCLUSIONS els are presented in the plots. In this cohort of patients with iNPH, ASV was not associated with invasively measured ICP scores or symptom severity of tion. We found no such tendency toward ASV being in the lower iNPH but was correlated with ventricular volume and aque- range among patients with long-standing symptoms; however, duct area. The results do not support the use of ASV as a non- the median of symptom duration in our cohort was 2 years, and invasive tool to diagnose reduced intracranial compliance in hence the number of patients with a longer disease history was patients with iNPH who are candidates for shunting. The com- limited. position of the study cohort, with a small fraction of nonre- While MWA was increased in shunt responders in the present sponders to shunting and few conservatively treated patients, study, as previously reported,19 no such relationship between the did not allow a more direct assessment of ASV as a marker for occurrence of so-called B waves and shunt response has been shunt responsiveness in iNPH. shown.19,40 In this context, the MWA is computed from the single cardiac-induced ICP waves, while B waves are short-lasting (Ͻ 1 Disclosures: Kyrre E. Emblem—RELATED: Grant: South-Eastern Norway Regional minute) increases in static ICP (mean ICP). It has previously been Health Authority Grant 2013069*; UNRELATED: Patents (planned, pending or issued): NordicNeuroLabs AS.* Oliver Geier—UNRELATED: Patents (planned, reported that occurrences of B waves and single ICP waves (ex- pending or issued): Siemens Germany, Comments: Patent attorney was paid pressed by the MWA) do not correlate.41 by Siemens. Noam Alperin—UNRELATED: Stock/Stock Options: Aplerin Nonin- 1628 Ringstad Sep 2015 www.ajnr.org vasive Diagnostics. Per Kristian Eide—RELATED: Stock/Stock Options: dPCom fluid flow rates measured by phase-contrast MR to predict outcome AS, Oslo, Norway, Comments: financial interest (shares) in the software company of ventriculoperitoneal shunting for idiopathic normal-pressure (dPCom AS, Oslo) manufacturing the software (Sensometrics software) used for hydrocephalus. Mayo Clin Proceed 2002;77:509–14 analysis of the ICP recordings. *Money paid to the institution. 19. Eide PK, Sorteberg W. Diagnostic intracranial pressure monitoring and surgical management in idiopathic normal pressure REFERENCES hydrocephalus: a 6-year review of 214 patients. Neurosurgery 2010;66:80–91 1. Adams RD, Fisher CM, Hakim S, et al. Symptomatic occult hydro- 20. Eide PK. Intracranial pressure parameters in idiopathic normal cephalus with “normal” cerebrospinal-fluid pressure: a treatable pressure hydrocephalus patients treated with ventriculo-perito- syndrome. N Engl J Med 1965;273:117–26 neal shunts. Acta Neurochir (Wien) 2006;148:21–29; discussion 29 2. Relkin N, Marmarou A, Klinge P, et al. Diagnosing idiopathic nor- 21. Eide PK, Sorteberg W. Association among intracranial compliance, mal-pressure hydrocephalus. Neurosurgery 2005;57:S4–16; discus- intracranial pulse pressure amplitude and intracranial pressure in sion ii–v patients with intracranial bleeds. Neurol Res 2007;29:798–802 3. Krauss JK, Droste DW, Vach W, et al. Cerebrospinal fluid shunting 22. Gonza´lez-Darder JM, Barcia-Salorio JL. Pulse amplitude and vol- in idiopathic normal-pressure hydrocephalus of the elderly: effect ume-pressure relationships in experimental hydrocephalus. Acta of periventricular and deep white matter lesions. Neurosurgery Neurochir (Wien) 1989;97:166–70 1996;39:292–99; discussion 299–300 23. Black PM, Ojemann RG, Tzouras A. CSF shunts for dementia, in- 4. Vanneste JA. Three decades of normal pressure hydrocephalus: are continence, and gait disturbance. Clin Neurosurg 1985;32:632–51 we wiser now? J Neurol Neurosurg Psychiatry 1994;57:1021–25 24. Bateman GA. Vascular compliance in normal pressure hydroceph- 5. Hebb AO, Cusimano MD. Idiopathic normal pressure hydrocephalus: alus. AJNR Am J Neuroradiol 2000;21:1574–85 a systematic review of diagnosis and outcome. Neurosurgery 2001;49: 25. Graff-Radford NR, Godersky JC, Jones MP. Variables predicting 1166–84; discussion 1184–86 surgical outcome in symptomatic hydrocephalus in the elderly. 6. Bradley WG Jr, Whittemore AR, Kortman KE, et al. Marked cerebrospi- Neurology 1989;39:1601–04 nal fluid void: indicator of successful shunt in patients with suspected 26. Greitz D. Radiological assessment of hydrocephalus: new theories normal-pressure hydrocephalus. Radiology 1991;178:459–66 and implications for therapy. Neurosurg Rev 2004;27:145–65; dis- 7. Al-Zain FT, Rademacher G, Meier U, et al. The role of cerebrospinal cussion 166–67 fluid flow study using phase contrast MR imaging in diagnosing 27. Nitz WR, Bradley WG Jr, Watanabe AS, et al. Flow dynamics of ce- idiopathic normal pressure hydrocephalus. Acta Neurochir Suppl rebrospinal fluid: assessment with phase-contrast velocity MR im- 2008;102:119–23 aging performed with retrospective cardiac gating. Radiology 8. Kim DS, Choi JU, Huh R, et al. Quantitative assessment of cerebro- 1992;183:395–405 spinal fluid hydrodynamics using a phase-contrast cine MR image 28. Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active in hydrocephalus. Childs Nerv Syst 1999;15:461–67 contour segmentation of anatomical structures: significantly 9. Bradley WG Jr, Scalzo D, Queralt J, et al. Normal-pressure improved efficiency and reliability. Neuroimage hydrocephalus: evaluation with cerebrospinal fluid flow mea- 2006;31:1116–28 surements at MR imaging. Radiology 1996;198:523–29 29. Zhu SC, Yuille AY. Region competition: unifying snakes, region 10. Sharma AK, Gaikwad S, Gupta V, et al. Measurement of peak CSF growing, and Bayes/MDL for multiband image segmentation. IEEE flow velocity at cerebral aqueduct, before and after lumbar CSF drainage, by use of phase-contrast MRI: utility in the management Trans Pattern Anal Mach Intell 1996;18:884–900 of idiopathic normal pressure hydrocephalus. Clin Neurol Neuro- 30. Hamilton R, Baldwin K, Fuller J, et al. Intracranial pressure pulse surg 2008;110:363–68 waveform correlates with aqueductal cerebrospinal fluid stroke 11. Luetmer PH, Huston J, Friedman JA, et al. Measurement of cerebro- volume. J Appl Physiol 2012;113:1560–66 spinal fluid flow at the cerebral aqueduct by use of phase-contrast 31. Tain RW, Alperin N. Intracranial pressure dynamics are not linked magnetic resonance imaging: technique validation and utility in to aqueductal cerebrospinal fluid stroke volume. J Appl Physiol diagnosing idiopathic normal pressure hydrocephalus. Neurosur- 2013;114:1645 gery 2002;50:534–43; discussion 543–44 32. Bateman GA, Levi CR, Schofield P, et al. The pathophysiology of the 12. Abbey P, Singh P, Khandelwal N, et al. Shunt surgery effects on cerebro- aqueduct stroke volume in normal pressure hydrocephalus: can co- spinal fluid flow across the aqueduct of Sylvius in patients with com- morbidity with other forms of dementia be excluded? Neuroradiol- municating hydrocephalus. J Clin Neurosci 2009;16:514–18 ogy 2005;47:741–48 13. El Sankari S, Fichten A, Gondry-Jouet C, et al. Correlation between 33. Chiang WW, Takoudis CG, Lee SH, et al. Relationship between tap test and CSF aqueductal stroke volume in idiopathic normal ventricular morphology and aqueductal cerebrospinal fluid pressure hydrocephalus. Acta Neurochir Suppl 2012;113:43–46 flow in healthy and communicating hydrocephalus. Invest Radiol 14. Scollato A, Gallina P, Gautam B, et al. Changes in aqueductal CSF 2009;44:192–99 stroke volume in shunted patients with idiopathic normal-pressure 34. Stoquart-ElSankari S, Baledent O, Gondry-Jouet C, et al. Aging hydrocephalus. AJNR Am J Neuroradiol 2009;30:1580–86 effects on cerebral blood and cerebrospinal fluid flows. J Cereb 15. Delwel EJ, de Jong DA, Avezaat CJ. The prognostic value of clinical Blood Flow Metab 2007;27:1563–72 characteristics and parameters of cerebrospinal fluid hydrodynam- 35. Wagshul ME, Eide PK, Madsen JR. The pulsating brain: a review of ics in shunting for idiopathic normal pressure hydrocephalus. Acta experimental and clinical studies of intracranial pulsatility. Fluids Neurochir (Wien) 2005;147:1037–42; discussion 1042–43 Barriers CNS 2011;8:5 16. Bateman GA, Loiselle AM. Can MR measurement of intracranial 36. Chaarani B, Capel C, Zmudka J, et al. Estimation of the lateral ven- hydrodynamics and compliance differentiate which patient with tricles volumes from a 2D image and its relationship with cerebro- idiopathic normal pressure hydrocephalus will improve following spinal fluid flow. Biomed Res Int 2013;2013:215989 shunt insertion? Acta Neurochir (Wien) 2007;149:455–62; discussion 37. Reneman RS, Arts T, Hoeks AP. Wall shear stress: an important 462 determinant of endothelial cell function and structure—in the ar- 17. Kahlon B, Annertz M, Stahlberg F, et al. Is aqueductal stroke volume, terial system in vivo. Discrepancies with theory. J Vasc Res measured with cine phase-contrast magnetic resonance imaging 2006;43:251–69 scans useful in predicting outcome of shunt surgery in suspected 38. Singer OC, Melber J, Hattingen E, et al. MR volumetric changes after normal pressure hydrocephalus? Neurosurgery 2007;60:124–29; dis- diagnostic CSF removal in normal pressure hydrocephalus. J Neu- cussion 129–30 rol 2012;259:2440–46 18. Dixon GR, Friedman JA, Luetmer PH, et al. Use of cerebrospinal 39. Scollato A, Tenenbaum R, Bahl G, et al. Changes in aqueductal CSF AJNR Am J Neuroradiol 36:1623–30 Sep 2015 www.ajnr.org 1629 stroke volume and progression of symptoms in patients with uns- 41. Eide PK, Sorteberg W. Simultaneous measurements of intracranial hunted idiopathic normal pressure hydrocephalus. AJNR Am J pressure parameters in the epidural space and in brain parenchyma Neuroradiol 2008;29:192–97 in patients with hydrocephalus. J Neurosurg 2010;113:1317–25 40. Stephensen H, Andersson N, Eklund A, et al. Objective B wave anal- 42. Frayne R, Steinman DA, Ethier CR, et al. Accuracy of MR phase ysis in 55 patients with non-communicating and communicating contrast velocity measurements for unsteady flow. J Magn Reson hydrocephalus. J Neurol Neurosurg Psychiatry 2005;76:965–70 Imaging 1995;5:428–31

1630 Ringstad Sep 2015 www.ajnr.org II

RESEARCH ARTICLE Non-invasive assessment of pulsatile intracranial pressure with phase-contrast magnetic resonance imaging

Geir Ringstad1,2 , Erika Kristina Lindstr¡m3, Svein Are Sirirud Vatnehol2,4, Kent- Andre´ Mardal3, Kyrre Eeg Emblem4, Per Kristian Eide2,5

1 Department of Radiology and Nuclear Medicine, Oslo University Hospital—Rikshospitalet, Oslo, Norway, 2 Faculty of Medicine, University of Oslo, Oslo, Norway, 3 Department of Mathematics, Faculty of Mathematics and Natural Sciences, University of Oslo, Oslo, Norway, 4 The Intervention Centre, Oslo a1111111111 University Hospital, Oslo, Norway, 5 Department of Neurosurgery, Oslo University Hospital, Oslo, Norway a1111111111 a1111111111 * [email protected] a1111111111 a1111111111 Abstract

Invasive monitoring of pulsatile intracranial pressure can accurately predict shunt response in patients with idiopathic normal pressure hydrocephalus, but may potentially cause compli- 23(1 $&&(66 cations such as bleeding and infection. We tested how a proposed surrogate parameter for Citation: Ringstad G, Lindstr¡m EK, Vatnehol SAS, Mardal K-A, Emblem KE, Eide PK (2017) Non- pulsatile intracranial pressure, the phase-contrast magnetic resonance imaging derived invasive assessment of pulsatile intracranial pulse pressure gradient, compared with its invasive counterpart. In 22 patients with sus- pressure with phase-contrast magnetic resonance pected idiopathic normal pressure hydrocephalus, preceding invasive intracranial pressure imaging. PLoS ONE 12(11): e0188896. https://doi. monitoring, and any surgical shunt procedure, we calculated the pulse pressure gradient org/10.1371/journal.pone.0188896 from phase-contrast magnetic resonance imaging derived cerebrospinal fluid flow velocities Editor: Luigi Maria Cavallo, Universita degli Studi di obtained at the upper cervical spinal canal using a simplified Navier-Stokes equation. Re- Napoli Federico II, ITALY peated measurements of the pulse pressure gradient were also undertaken in four healthy Received: October 26, 2016 controls. Of 17 shunted patients, 16 responded, indicating high proportion of “true” normal Accepted: November 2, 2017 pressure hydrocephalus in the patient cohort. However, there was no correlation between Published: November 30, 2017 the magnetic resonance imaging derived pulse pressure gradient and pulsatile intracranial

Copyright: 2017 Ringstad et al. This is an open pressure (R = -.18, P = .43). Pulse pressure gradients were also similar in patients and access article distributed under the terms of the healthy controls (P = .26), and did not differ between individuals with pulsatile intracranial Creative Commons Attribution License, which pressure above or below established thresholds for shunt treatment (P = .97). Assessment permits unrestricted use, distribution, and of pulse pressure gradient at level C2 was therefore not found feasible to replace invasive reproduction in any medium, provided the original author and source are credited. monitoring of pulsatile intracranial pressure in selection of patients with idiopathic normal pressure hydrocephalus for surgical shunting. Unlike invasive, overnight monitoring, the Data Availability Statement: All relevant data are within the paper and its Supporting Information pulse pressure gradient from magnetic resonance imaging comprises short-term pressure files. fluctuations only. Moreover, complexity of cervical cerebrospinal fluid flow and -pulsatility at

Funding: Kyrre Eeg Emblem received Grant the upper cervical spinal canal may render the pulse pressure gradient a poor surrogate #2013069 from South Eastern Norway Regional marker for intracranial pressure pulsations. Health Authority (http://www.helse-sorost.no/fag/ forskning-og-innovasjon). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. PKE has a financial interest (shares) in the software company (dPCom AS, Oslo)

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 1 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI manufacturing the software (Sensometrics Introduction Software) used for analysis of the ICP recordings. For the remaining authors none were declared. Idiopathic normal pressure hydrocephalus (iNPH) is a chronic condition characterized by gait This does not alter our adherence to PLOS ONE ataxia, dementia and urinary incontinence[1] and can be treated successfully with shunt sur- policies on sharing data and materials. gery. However, it remains a challenge to select which patients who will respond to such treat- Competing interests: PKE has a financial interest ment. Unlike invasively obtained monitoring of mean intracranial pressure (ICP), over-night (shares) in the software company (dPCom AS, monitoring of pulsatile ICP has previously demonstrated to predict a beneficial shunt response Oslo) manufacturing the software (Sensometrics in 9 of 10 patients [2, 3]. Increased pulsatile ICP is expected in iNPH while reduced pressure- Software) used for analysis of the ICP recordings. volume capacity (reduced intracranial compliance) is a common feature [2, 4, 5]. KEE has received a grant (#2013069) from South- The disadvantage with invasive monitoring of pulsatile ICP is the risk of severe complica- Eastern Norway Regional Health Authority. For the remaining authors none were declared. This does tions such as cerebral bleeds and infections, which may occur in about 1%[6], and trends not alter our adherence to PLOS ONE policies on towards higher risk in iNPH patients who are older and with more comorbidity[7]. Given the sharing data and materials. useful role of invasive monitoring in patient care, there is an urgent need to develop alternative, non-invasive methods. There have been many attempts to apply phase-contrast MRI (PC-MRI) at intracranial locations to guide clinical decision making, particularly at the aqueduct level, but with varying degrees of success [8–13]. A non-invasively obtained marker for pulsatile ICP has been proposed, namely the PC- MRI-derived peak-to-peak pulse pressure gradient (MRI-dP), which is calculated from the cerebrospinal fluid (CSF) flow velocity change during one cardiac cycle at the upper cervical spinal canal[14]. Image acquisition is thus obtained in close proximity to the intracranial compartment; however, any pressure assessment at this level must be considered a surrogate for direct monitoring of pulsatile ICP. Use of MRI-dP has previously shown promise in studies on computational fluid dynamics [15], baboons and humans[14, 16], as well as in patients with hydrocephalus[17, 18], Chiari malformation[19] and NPH[20]. MRI-dP has thus the potential to replace invasive monitoring in the pre-surgical diagnostic work-up of iNPH patients that are shunt candidates. Compari- sons with invasive monitoring have though been few, and direct comparison with its invasive counterpart, pulsatile ICP, has so far been carried out in one baboon only[14]. It has previously been validated that over-night measurements of pulsatile ICP are comparable with measurements at daytime and are therefore neither sensitive to head position of the patient, unlike mean ICP [2]. The purpose of this study was therefore to assess the feasibility of MRI-dP by comparison with overnight invasive monitoring of pulsatile ICP in iNPH patients under clinical work-up for surgical shunting. We also compared MRI-dP derived from patients and healthy controls.

Materials and methods Study population and design The study and patient consent procedure was approved by the Institutional Review Board of Oslo University Hospital (07/5869) and Regional Ethics Committee (REC), South-East, Norway (S-07237). Each study participant was given written and oral information about the study and signed the consent form prior to inclusion in the study. All research was conducted according to the principles expressed in the Declaration of Helsinki. The institution that granted permission was Oslo University Hospital. In this prospective study, a cohort of 34 consecutive patients with clinically probable iNPH, hos- pitalized within the department of neurosurgery as part of their preoperative diagnostic work-up, were enrolled. PC-MRI was done at the end of a routine MR image acquisition protocol and preceded invasive ICP monitoring performed during the same hospital admission. Additionally, four healthy controls were examined with PC-MRI acquisitions at different time points during one day.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 2 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Clinical management All patients were scored clinically using a clinical NPH grading scale [2], which assesses the combined severity of (I) gait disturbance, (II) urinary incontinence, and (III) dementia. Each of the three variables is graded from 1 to 5, giving a possible worst score of 3 and best score of 15. The decision for shunt surgery was based on a combination of clinical assessment, radiological assessment and results of continuous ICP monitoring [2]. The shunt response was defined as an increase by at least 2 points on the NPH grading scale, and clinical score was assessed at regular intervals, 3, and 6–12 months following shunting.

ICP monitoring The procedure for continuous overnight ICP monitoring has previously been described[2]. In short, an ICP sensor is placed in the frontal lobe of the brain through a small burr hole in the skull, using local anesthesia. The monitoring is carried out in the patient ward with the patient in the supine position (equal to patient position at PC-MRI) using a computerized system (Sensometrics AS, dPCom, Oslo) for automatic identification of individual cardiac-induced single ICP waves. The amplitude of the ICP wave (pulsatile ICP) is defined as the pressure difference between the systolic maximum and diastolic minimum pressures (Fig 1). Pulsatile ICP expressed by the mean ICP wave amplitude (MWA) is determined during consecutive 6-second time intervals, and over the entire observation period. Mean (static) ICP is the average of absolute ICP relative to a zero pressure level. The MWA was used for selection to shunting according to the department routine [2]. MWA in average 4 mmHg and/or percentage of MWA 5 mmHg in 10% of recording time was set as threshold levels for shunting.

Fig 1. Single cardiac-induced ICP wave. One single ICP wave is illustrated with its individual parameters. We measured pressure in mm Hg; with a heart rate of 60, the duration of a single wave is about 1 second. The mean ICP wave amplitude (MWA) is determined as the average of amplitudes from single ICP waves during consecutive 6-second time intervals. https://doi.org/10.1371/journal.pone.0188896.g001

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 3 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

PC-MRI acquisition protocol For the iNPH patients, all PC-MRI images were obtained in a Philips 3 Tesla Achieva system (Philips Medical Systems, Best, The Netherlands) with a 16-channel head coil and an acquisition plane perpendicular to the upper cervical spinal canal at level C2, which is typically the closest level to the intracranial compartment where the assumption of the spinal canal as a rigid, cylindrical tube can be made when estimating MRI-dP. The PC-MRI scanning parameters were as follows: Repetition time (TR) and echo time (TE) was shortest possible, typically TR/TE = 16/11 ms, pixel size 0.56 x 0.56 mm2 to 0.63 x 0.63 mm2, slice thickness 7 mm, velocity encoding 6 cm/sec, and 32–40 phases with retrospective peripheral cardiac gating based on pulse oximeter plethysmography. Potential sources of inaccuracy in velocity measurements, such as eddy currents, Maxwell terms and gradient field non-linearities[21] had been corrected for as far as possible by the vendor of the scanner before the study, and also sought minimized during the study by optimal patient positioning. Duration of the PC-MRI acquisition was approximately 6 minutes. MR imaging always preceded any surgical shunting procedure. The healthy controls were scanned in a Philips 3 Tesla Ingenia system (Philips Medical Systems, Best, The Netherlands) with a 32 Channel head coil, and with identical acquisition plane,—level and—parameters as for the patients. Each healthy control was scanned with four PC-MRI acquisitions over at least two different time points during one day and with duration between time points of approximately two hours. Acquisitions within one time point were obtained consecutively without pause. All patients and controls were studied at daytime between 8 a.m. and 4 p.m. Fig 2 A demonstrates the slice orientation and level for all PC-MRI acquisitions.

Fig 2. (A) Level of PC-MRI acquisitions. PC-MRI was performed perpendicular to the cervical spinal canal at level C2, which is typically the closest level to the intracranial compartment where the assumption of the spinal canal as a rigid, cylindrical tube can be made when estimating MRI-dP. (B) ROI outlining. The subarachnoid space was manually defined from phase images with a region of interest (ROI). https://doi.org/10.1371/journal.pone.0188896.g002

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 4 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

PC-MR image outlining The subarachnoid fluid compartment surrounding the spinal cord was manually defined from 1 phase images with a region of interest (ROI) by use of dedicated software, nordicICE (Fig 2B) (NordicNeuroLab AS, Bergen, Norway) and was performed by an experienced neuroradiologist, blinded to NPH score and ICP data.

Pixel transformation and treatment of aliasing Recorded velocities were converted by linear transformation from pixel values to centimeters 1 per second by applying the velocity encoding using MATLAB (Mathworks, Natick, United States). Aliased velocities were treated with a filter that was activated if the temporal variation between two consecutive time steps of the examined pixel velocity exceeded a threshold value v v ± × V v of 1.1 times the VENC. The aliased pixel was replaced by = a 2 enc, where is the fil- v V tered velocity, a is the aliased pixel value and enc is the velocity encoding.

Computation of pressure gradients The method for computation of MRI-dP (mm Hg/cm) has been described in detail previously [14]. Mean velocities were computed by summarizing velocities at each pixel over the domain, and dividing the sum with the number of pixels. MRI-dP was computed according to Navier-Stokes equations for incompressible and New- tonian fluid. Velocities were recorded across an axial (x,y) plane, which reduced the Navier- Stokes equations to the z-component. Further, the flow was assumed parallel to rigid walls of a cylindrically shaped tube and the convective term was neglected. The resulting equation for the pressure gradient is given by @p @V @2V @2V ¼r þ m þ @z @t @x2 @y2

where the first term on the right-hand side describes the transient inertial forces and the second term describes the viscous forces in the flow. ρ = 1.0007 g/cm3 is the density of the CSF, and μ = 1.1 cP is the dynamic viscosity, and which are considered not to be affected by the iNPH condition. Pressure gradients for all pixels in the ROI were computed before they were averaged over the domain, resulting in one mean pressure gradient for each period. MRI-dP was calculated as the difference between the maximum and minimum pressure gradient within the cycle (Fig 3).

Statistical analysis ICP data, counting several thousands of observations from over-night monitoring, were assumed normally distributed, and presented in each patient with mean and standard deviation. For patients and healthy controls (n = 22 and 4, respectively), PC-MRI data were assumed not normally distributed and presented with median and range. Correlations were determined by Pearson correlation coefficient, and agreement between methods was evaluated with a Bland-Altman plot. Comparisons between patients and healthy controls were performed with Mann-Whitney U test. One-way ANOVA was used to compare means of more than two groups. Reliability (absolute agreement, two-way mixed) of multiple quantitative measurements within individuals was estimated with the intraclass correlation coefficient (ICC). The

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 5 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 3. The PC-MRI derived pulse pressure gradient (MRI-dP). MRI-dP is calculated as the difference between the maximum and minimum pressure gradient within the cycle. https://doi.org/10.1371/journal.pone.0188896.g003

significance level was set to 0.05. Statistical analysis was performed using SPSS Statistics version 20 (IBM Corporation, Armonk, NY, US).

Results Study population Of 34 enrolled patients, 12 were excluded from the study due to motion artifacts at PC-MRI (5/12), termination of the examination by the patient or MRI technician before completion (5/ 12), PC-MRI obtained at a suboptimal level (1/12) or image artifacts (1/12). Thus, 22 patients were included. Shunt surgery was performed in 17 of these, of which 16 (94%) were clinical responders. Other patient data are given in Table 1 and S1 File. Age and gender of four healthy controls are also presented in Table 1.

ICP scores A summary of the ICP data from the iNPH patients are given in Table 2, and the full dataset from the patient cohort is presented in S1 Table. The coefficient of variation (CV = standard deviation/mean) illustrates the span of fluctuations in measured pulsatile and static ICP during overnight monitoring. Overnight monitoring of pulsatile and static (mean) ICP demonstrated large fluctuations, where CV was 26% (12, 41) and 128% (19, 5600), respectively (median and range). Fig 4 exemplifies recordings from a study patient demonstrating pulsatile ICP expressed by MWA (a), static ICP expressed by mean ICP (b) and heart rate (c) as a function

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 6 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Table 1. Demographic information about patients and healthy controls. NPH patients Number 22 Age (yrs) 71 (45–84) Gender (F/M) 8/14 BMI (kg/m2) 25.4 (20.3–30.9) Cardiovascular co-morbidity (n) 12 Treatment Shunt (Responders/Non-responders) 17 (16/1) No shunt 5 Healthy individuals Number 4 Age (yrs) 31 (23–40) Gender (F/M) 1/3

Results presented as numbers for categorical data and as median (range) for continuous data.

https://doi.org/10.1371/journal.pone.0188896.t001

of time. For patient specific measurements, MWA was to a very limited degree affected by HR, as the association between pulsatile ICP (MWA) and heart rate (HR) was low with median R = .02 (S1 Table).

PC-MRI data from patients and healthy controls A summary of the PC-MRI data from patients and healthy controls is shown in Table 3, and an extended set of data from the PC-MRI studies of iNPH patients and healthy controls are presented in S2 and S3 Tables, respectively, and in S2 File. MRI-dP at level C2 was not different in iNPH patients and healthy subjects (P = .39). Only area of the subarachnoid space (ROI area and hence number of pixels) differed between groups (P = .016).

Comparison of patient ICP scores and MRI-dP There were no associations between invasively measured pulsatile ICP and the non-invasive assessment of MRI-dP (R = -.18, P = .43) (Fig 5A) or mean ICP and MRI-dP (R = .10, P = .68) (Fig 5B). Moreover, the MRI-dP did not differ between individuals with MWA above or below established thresholds for shunting (P = .97), or healthy controls (P = .44) (Fig 6).

Table 2. Summary of ICP data from iNPH patients.

Patients Observations MWA MWA CVMWA Mean ICP Mean ICP CVMeanICP (6-s windows) (mm Hg) (mean) (mm Hg) (%) (mm Hg) (mm Hg) (%) (SD) (mean) (SD) All 13.283 4.4 1.1 26 2.8 4.4 128 (n = 22) (median with range) (1501, 21397) (2.2, 7.3) (.5, 2.2) (12, 41) (-4.5, 14.5) (1.9, 11.8) (19, 5600)

ICP: intracranial pressure iNPH: idiopathic normal pressure hydrocephalus MWA: mean ICP wave amplitude (pulsatile ICP) CV: coefficient of variance (SD/mean) SD: standard deviation https://doi.org/10.1371/journal.pone.0188896.t002

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 7 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 4. Trend plots of (a) MWA (6.61.8 mmHg), (b) mean ICP (2.84.8 mmHg), and (c) HR (76.213.7 beats/ min) for patient 20, who underwent prolonged invasive monitoring, illustrating the variation in the ICP-derived parameters over time. In (d) is shown one individual cardiac-beat induced ICP wave with its amplitude (dP) and rise time (RT). https://doi.org/10.1371/journal.pone.0188896.g004

Table 3. PC-MRI data from patients and healthy controls (median with range). Patients (n = 22) Healthy (n = 4) P-value MRI-dP (mmHg/cm) .041 (.015, .083) .053 (.031, .093) .39 ROI area (cm2) 1.74 (1.20, 3.50) 1.26 (.77, 1.58) .016 Number of pixels 446 (306, 895) 323 (198, 405) .016 Heart rate (/min) 68 (52, 92) 61 (49, 69) .25

PC-MRI: phase-contrast magnetic resonance imaging MRI-dP: MRI-derived peak to peak pulse pressure gradient ROI: region of interest *Mann-Whitney U-test.

https://doi.org/10.1371/journal.pone.0188896.t003

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 8 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 5. MRI-dP and ICP parameters. The association between MRI-dP and standardized over-night, invasive monitoring (11 p.m. to 7 a.m.) of (a) pulsatile ICP (MWA) and (b) static ICP (mean ICP).The Pearson correlation coefficient (R) and significance level are presented for each plot. https://doi.org/10.1371/journal.pone.0188896.g005

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 9 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 6. MRI-dP and pulsatile ICP (MWA) thresholds for shunting in iNPH. The MRI-dP values are shown for iNPH patients with preoperative MWA values either above (n = 16) or below (n = 6) the thresholds used for selection of patients for shunting, and also for the healthy reference subjects. There were no significant differences between groups (P!0.44). https://doi.org/10.1371/journal.pone.0188896.g006 Heart rate during PC-MRI and ICP monitoring There was a high correlation between heart rates (HR) from PC-MRI and invasive ICP monitoring (R = .71, P = .001) (Fig 7A), and the Bland-Altman plot revealed no systematic differences in HR registered during invasive ICP monitoring and PC-MRI (Fig 7B).

Discussion Invasive monitoring of pulsatile ICP has previously proven to be a precise tool to select which iNPH patients who will respond to surgical shunting. In this study, we tested the utility of PC-MRI to non-invasively assess a surrogate marker for pulsatile ICP, MRI-dP, obtained at level C2. MRI-dP did not correlate with invasively obtained pulsatile ICP from over-night monitoring, and did not discriminate iNPH patients from healthy subjects. Increased pulsatile ICP due to reduced intracranial compliance is common in iNPH [7], and patients with this disease should therefore be well suited for a study of non-invasive assessment of pulsatile ICP using PC-MRI. In the present study cohort, increased pulsatile ICP expressed by MWA above threshold level was diagnosed in 17 of 22 patients. Surgical shunting yielded a clinical improvement in 16 of 17 treated patients, indicating a high proportion of what may be considered “true iNPH”.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 10 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 7. HR from PC-MRI and ICP monitoring. (a) The association between MRI-derived HR and ICP-derived HR, Pearson correlation coefficient (R) and significance level. (b) Bland-Altman plot of all MRI- and ICP- derived HR observations. The line in the middle is the mean difference (2.3 beats/min) and the upper and lower lines represent mean+2SD (standard deviations; 15.0 beats/min) and mean– 2SD (-10.4 beats/min), respectively. https://doi.org/10.1371/journal.pone.0188896.g007

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 11 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

MRI-dP has previously been applied in a proposed method for non-invasive estimates of static (mean) ICP [14]. With this method, the elastance index is estimated by dividing MRI-dP by the per cardiac cycle total intracranial volumetric change (MRI-dV), i.e. MRI-dP/MRI-dV. The linear relationship between elastance and ICP is expected owing to the monoexponential relationship between intracranial volume and pressure[22]. Precise assessment of MRI-dP is thus an important precondition for such estimation of static ICP as well. Should a non-invasive measure such as MRI-dP be valid, it would have to reflect its gold standard of invasive measurement (pulsatile ICP). Both pulsatile and static (mean) ICP demonstrated in this study large fluctuations during the overnight monitoring, and this phenomenon is also a determining factor as to why clinical decisions based on monitoring of pulsatile ICP are made after analysis of observations from an extended time span (Fig 4). A PC-MRI based method applied to obtain measures of pulsatile ICP through a surrogate parameter like MRI-dP from a short time interval should therefore be interpreted with great care. One previous study could not demonstrate different MRI-dP between Chiari malformation type 1 patients and controls[23], even though elevated pulsatile ICP is frequently observed when measured over-night[24].The aqueductal stroke volume is another, MRI derived, CSF velocity based parameter obtained from a short time interval, which has been proposed to serve as a surrogate for ICP recordings[8], and were also unable to demonstrate any association with its invasive counterpart[13, 25]. There may be several reasons as to why MRI-dP derived from one simple PC-MRI acquisition does not reflect pulsatile ICP measured overnight. First, MRI-dP is based on acquisition of CSF velocity change at the upper cervical spinal canal, and is thus a surrogate marker for intracranial pressure change. A simplified model of a rigid, cylindrically shaped tube is desirable in order to make valid assumptions in fluid dynamic equations used in post-processing of PC-MRI data, where attempts made intracranially, with much more complex anatomy, are mostly restricted to the Sylvian aqueduct. However, even though the upper spinal canal is anatomically in close proximity with the intracranial compartment, the geometry is funda- mentally different, and from computational fluid flow dynamics, it has been demonstrated that lower pressure values can be expected at level C2 than in the posterior fossa due to pressure gradients that become steep in the cervical canal[26]. Secondly, the CSF flow at level C2 is the result of a dyssynchronous pressure pulsation caus- ing a pulsating cranio-cervical pressure gradient pumping CSF in and out of the cranial vault. The pressure gradient estimated from MRI is relatively small, and in this study, less than 0.1 mm Hg per cm in all patients. This magnitude corresponds well with previous studies. From numerical studies, where the CSF flow has been modeled in rigid and impermeable surroundings, the fluid flow has been predominated by velocities orthogonal to the axial plane, and the observed pressure gradients at level C2 in healthy subjects is also less than 0.1 mm Hg per cm [27, 28]. Computations that take into account the elasticity of the cervical spinal cord, tonsillar motion and the presence of nerve roots and denticulate ligaments, all predict pressure gradients of similar magnitudes[29–31]. Hence, the pressure gradient between the cervical and cranial compartments is but a fraction of the total pressure pulsation. Moreover, the pressure gradient computation in this study is based on a number of simpli- fying assumptions such as rigid and impermeable surroundings and laminar flow. CSF flow patterns are, however, complex due to the anatomy of the cervical subarachnoid space[26], and flow velocities in axial sections of the spine have been shown to have non-uniform distributions throughout the cardiac cycle as demonstrated from computational simulations[28]. In our study, the pixel-by-pixel analysis revealed a wide dispersion of flow velocities from within the same PC-MRI slab (Fig 8). Averaging of flow velocities from a ROI defining the CSF space therefore does not fully describe the diversity or complexity of velocity changes that occurs through a section of the cervical spine during one cardiac cycle.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 12 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

Fig 8. Heterogeneity of CSF flow velocities. Patient 4, demonstrating spatial variations in flow velocity within the ROI at one time step (t = 0.70 s). The color bar represents velocities in cm/s. https://doi.org/10.1371/journal.pone.0188896.g008

Interestingly, it may be noted that the healthy control with smallest CSF ROI area also had the largest MRI-dP among controls (No. 2 in S3 Table), and larger than in many iNPH patients with invasively proven increased pulsatile ICP (S2 Table). Previously, the inverse and almost linear relationship between MRI-dP and CSF flow area has been reported[14], and very steep pressure gradients have also been demonstrated in Chiari patients with tonsillar crowding at the foramen magnum[32]. For the patient group in our study, there was a trend towards an inverse correlation between MRI-dP and area of the subarachnoid fluid area at level C2 (ROI area), but this did not reach significant levels (R = -.32, P = .15). However, we notice that measuring reliability (ICC) was almost identical for MRI-dP and ROI area measurements in healthy controls. Thus, while MRI-dP should ideally be a marker of pulsatile ICP only, it may be hypothesized that a narrow CSF space surrounding the spinal cord may influence directly on MRI-dP. However, CSF area is already implicitly accounted for when estimating pressure based on the Navier-Stokes equation. Finally, but not least, recent studies have demonstrated respiration to have a major influence on CSF flow [33, 34]. With cardiac gated PC-MRI, which can be considered to represent current state-of-the-art methodology in PC-MRI based measurements of intracranial and intraspinal CSF flow quantification, only cardiac beat induced CSF flow is detected. Patient breathing is not controlled for. The invasive ICP measurements of this study were not performed synchronously with PC-MRI, but overnight, as standardized according to previous empirical observations supporting this routine. In principle, our data do therefore not in principle contradict that MRI-dP may reflect pulsatile ICP in real time. The comparison with invasive monitoring for a prolonged time period is, however, highly relevant, because clinical decision making is rarely based on ICP monitoring from a time interval as short as the duration of a PC-MRI acquisition. It has previously also been validated that over-night and daytime measurements of pulsatile ICP are comparable, unlike measurements of static ICP[2]. Another limitation was that technically adequate PC-MRI was obtained in less than 2/3 of enrolled patients, while all examinations in healthy subjects were successful. NPH patients are

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 13 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

typically elderly and suffer from discomfort during long MRI scan times, and the experimental PC-MRI acquisitions were performed towards the end of the imaging protocol. In iNPH, dementia is a common feature, which may be a challenge to patient cooperation throughout the scan. MRI-dP from PC-MRI may therefore seem more feasible in subjects that cooperate well rather than in patients where compliance to instructions from the MR technician can be a challenge. The inability to exam patients with the poorest level of cooperation skills may have introduced bias to the study, because more severely affected patients might have been more likely to be excluded. Moreover, the velocity encoding gradient (venc) was set at a low level for the PC-MRI acquisitions in patients and healthy controls. Because of flow heterogeneities, flow aliasing was quite frequent. We consider, however, this potential source of error to be small because of the aliasing correction procedure and that flow velocity values from each time point represent the average of velocities from the entire region of interest. Pixels with any inaccuracies of flow velocity measurements should therefore not have contributed substantially to the calculated MRI-dP. As for PC-MRI based measurements of CSF flow in general, a major concern with the technique would be the extraction of data from a limited time interval to make a diagnosis in patients where well-known physiological fluctuations are present. In the present study, all PC-MRI exams may be assumed to have been performed within normal physiological bound- aries, as they were obtained within a normal range for heart rate (Tables 2 and 3), and the statistical analysis did not reveal any systematic differences in heart rate (HR) between PC-MRI and ICP monitoring. This indicates that influence from patient stress and discomfort was not very different during PC-MRI and ICP monitoring, and this source of bias in comparing the methods should therefore have been modest.

Conclusions Invasively obtained pulsatile ICP with level above thresholds has previously been demonstrated to predict a beneficial shunt response in 9 of 10 patients with iNPH. In our study cohort of iNPH patients, patient selection with this method yielded a beneficial shunt response in 94%. However, pulsatile ICP was not associated with its non-invasive counterpart, the PC-MRI derived MRI-dP. Moreover, MRI-dP did not discriminate between patients with either preoperative values of pulsatile ICP above or below thresholds, and neither patients from healthy subjects. MRI-dP obtained at level C2 was therefore not found feasible to replace invasive monitoring of pulsatile ICP in selection of iNPH patients for surgical shunting. Unlike invasive, overnight monitoring, MRI-dP comprises short-term pressure fluctuations only. Moreover, complexity of cervical CSF flow and -pulsatility at the upper cervical spinal canal may render MRI-dP a poor surrogate marker for intracranial pressure change.

Supporting information S1 Table. ICP-data: Patients. Results from ICP monitoring in each of the iNPH study patients. (DOCX) S2 Table. PC-MRI data: Patients. Results from PC-MRI in each of the iNPH study patients. (DOCX) S3 Table. PC-MRI data: Healthy controls. Results from PC-MRI in each healthy subject at each time point. PC-MRI exams (MRI-1 and MRI-2) within one time point (e.g. “Time1”) were performed with the possible closest proximity in time. The distance in time between two

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 14 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

time points (e.g. “Time1” and “Time2”) was approximately two hours. (DOCX) S1 File. Patient data. Extended set of clinical patient data. (XLS) S2 File. Pressure gradients and mean velocities. Calculated pressure gradients and mean velocities from single time points are given for all patients and healthy controls. (ZIP)

Author Contributions Conceptualization: Geir Ringstad, Kyrre Eeg Emblem, Per Kristian Eide. Data curation: Geir Ringstad, Erika Kristina Lindstrøm, Svein Are Sirirud Vatnehol, Per Kris- tian Eide. Formal analysis: Geir Ringstad, Erika Kristina Lindstrøm, Kent-Andre´ Mardal, Per Kristian Eide. Funding acquisition: Kyrre Eeg Emblem. Investigation: Geir Ringstad, Erika Kristina Lindstrøm, Svein Are Sirirud Vatnehol, Kyrre Eeg Emblem, Per Kristian Eide. Methodology: Geir Ringstad, Erika Kristina Lindstrøm, Svein Are Sirirud Vatnehol, Kent- Andre´ Mardal, Kyrre Eeg Emblem, Per Kristian Eide. Project administration: Geir Ringstad, Per Kristian Eide. Resources: Geir Ringstad, Erika Kristina Lindstrøm, Svein Are Sirirud Vatnehol, Kent-Andre´ Mardal, Kyrre Eeg Emblem, Per Kristian Eide. Software: Erika Kristina Lindstrøm, Kent-Andre´ Mardal, Per Kristian Eide. Supervision: Kent-Andre´ Mardal, Kyrre Eeg Emblem, Per Kristian Eide. Validation: Geir Ringstad, Erika Kristina Lindstrøm, Kent-Andre´ Mardal, Per Kristian Eide. Visualization: Geir Ringstad, Erika Kristina Lindstrøm, Per Kristian Eide. Writing – original draft: Geir Ringstad, Erika Kristina Lindstrøm, Kent-Andre´ Mardal, Per Kristian Eide. Writing – review & editing: Geir Ringstad, Erika Kristina Lindstrøm, Svein Are Sirirud Vat- nehol, Kent-Andre´ Mardal, Kyrre Eeg Emblem, Per Kristian Eide.

References 1. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH. Symptomatic Occult Hydrocephalus with "Normal" Cerebrospinal-Fluid Pressure.A Treatable Syndrome. The New England journal of medicine. 1965; 273:117–26. https://doi.org/10.1056/NEJM196507152730301 PMID: 14303656. 2. Eide PK, Sorteberg W. Diagnostic intracranial pressure monitoring and surgical management in idiopathic normal pressure hydrocephalus: a 6-year review of 214 patients. Neurosurgery. 2010; 66(1):80– 91. Epub 2009/12/22. https://doi.org/10.1227/01.NEU.0000363408.69856.B8 PMID: 20023540. 3. Eide PK, Sorteberg W. Outcome of Surgery for Idiopathic Normal Pressure Hydrocephalus: Role of Pre- operative Static and Pulsatile Intracranial Pressure. World neurosurgery. 2016; 86:186–93 e1. https:// doi.org/10.1016/j.wneu.2015.09.067 PMID: 26428326. 4. Greitz D. Radiological assessment of hydrocephalus: new theories and implications for therapy. 2004; 27(3):145–65; discussion 66–7. https://doi.org/10.1007/s10143-004-0326-9 PMID: 15164255.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 15 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

5. Wagshul ME, Eide PK, Madsen JR. The pulsating brain: A review of experimental and clinical studies of intracranial pulsatility. Fluids and barriers of the CNS. 2011; 8(1):5. https://doi.org/10.1186/2045-8118- 8-5 PMID: 21349153; PubMed Central PMCID: PMC3042979. 6. Koskinen LO, Grayson D, Olivecrona M. The complications and the position of the Codman MicroSen- sor ICP device: an analysis of 549 patients and 650 Sensors. Acta neurochirurgica. 2013; 155 (11):2141–8; discussion 8. https://doi.org/10.1007/s00701-013-1856-0 PMID: 24018980. 7. Eide PK, Sorteberg W. Outcome of surgery for idiopathic normal pressure hydrocephalus: Role of preoperative static and pulsatile intracranial pressure. World neurosurgery. 2015. Epub 2015/10/03. https://doi.org/10.1016/j.wneu.2015.09.067 PMID: 26428326. 8. Bradley WG Jr., Scalzo D, Queralt J, Nitz WN, Atkinson DJ, Wong P. Normal-pressure hydrocephalus: evaluation with cerebrospinal fluid flow measurements at MR imaging. Radiology. 1996; 198(2):523–9. https://doi.org/10.1148/radiology.198.2.8596861 PMID: 8596861. 9. Luetmer PH, Huston J, Friedman JA, Dixon GR, Petersen RC, Jack CR, et al. Measurement of cerebrospinal fluid flow at the cerebral aqueduct by use of phase-contrast magnetic resonance imaging: technique validation and utility in diagnosing idiopathic normal pressure hydrocephalus. Neurosurgery. 2002; 50(3):534–43; discussion 43–4. PMID: 11841721. 10. Sharma AK, Gaikwad S, Gupta V, Garg A, Mishra NK. Measurement of peak CSF flow velocity at cerebral aqueduct, before and after lumbar CSF drainage, by use of phase-contrast MRI: utility in the management of idiopathic normal pressure hydrocephalus. Clinical neurology and neurosurgery. 2008; 110 (4):363–8. https://doi.org/10.1016/j.clineuro.2007.12.021 PMID: 18282655. 11. Kahlon B, Annertz M, Stahlberg F, Rehncrona S. Is aqueductal stroke volume, measured with cine phase-contrast magnetic resonance imaging scans useful in predicting outcome of shunt surgery in suspected normal pressure hydrocephalus? Neurosurgery. 2007; 60(1):124–9; discussion 9–30. https://doi.org/10.1227/01.NEU.0000249208.04344.A3 PMID: 17228260. 12. Dixon GR, Friedman JA, Luetmer PH, Quast LM, McClelland RL, Petersen RC, et al. Use of cerebrospinal fluid flow rates measured by phase-contrast MR to predict outcome of ventriculoperitoneal shunting for idiopathic normal-pressure hydrocephalus. Mayo Clinic proceedings. 2002; 77(6):509–14. https:// doi.org/10.4065/77.6.509 PMID: 12059119. 13. Ringstad G, Emblem KE, Geier O, Alperin N, Eide PK. Aqueductal Stroke Volume: Comparisons with Intracranial Pressure Scores in Idiopathic Normal Pressure Hydrocephalus. AJNR American journal of neuroradiology. 2015; 36(9):1623–30. https://doi.org/10.3174/ajnr.A4340 PMID: 25977480. 14. Alperin NJ, Lee SH, Loth F, Raksin PB, Lichtor T. MR-Intracranial pressure (ICP): a method to measure intracranial elastance and pressure noninvasively by means of MR imaging: baboon and human study. Radiology. 2000; 217(3):877–85. Epub 2000/12/09. https://doi.org/10.1148/radiology.217.3. r00dc42877 PMID: 11110957. 15. Loth F, Yardimci MA, Alperin N. Hydrodynamic modeling of cerebrospinal fluid motion within the spinal cavity. Journal of biomechanical engineering. 2001; 123(1):71–9. PMID: 11277305. 16. Alperin N, Lee SH, Bagci AM. MRI measurements of intracranial pressure in the upright posture: The effect of the hydrostatic pressure gradient. Journal of magnetic resonance imaging: JMRI. 2015; 42 (4):1158–63. https://doi.org/10.1002/jmri.24882 PMID: 25753157. 17. Muehlmann M, Koerte IK, Laubender RP, Steffinger D, Lehner M, Peraud A, et al. Magnetic resonance- based estimation of intracranial pressure correlates with ventriculoperitoneal shunt valve opening pressure setting in children with hydrocephalus. Investigative radiology. 2013; 48(7):543–7. https://doi.org/ 10.1097/RLI.0b013e31828ad504 PMID: 23695081. 18. Glick RP, Niebruegge J, Lee SH, Egibor O, Lichtor T, Alperin N. Early experience from the application of a noninvasive magnetic resonance imaging-based measurement of intracranial pressure in hydrocephalus. Neurosurgery. 2006; 59(5):1052–60; discussion 60–1. https://doi.org/10.1227/01.NEU. 0000245586.23710.EF PMID: 17143240. 19. Alperin N, Sivaramakrishnan A, Lichtor T. Magnetic resonance imaging-based measurements of cerebrospinal fluid and blood flow as indicators of intracranial compliance in patients with Chiari malformation. Journal of neurosurgery. 2005; 103(1):46–52. https://doi.org/10.3171/jns.2005.103.1.0046 PMID: 16121972. 20. Mase M, Miyati T, Yamada K, Kasai H, Hara M, Shibamoto Y. Non-invasive measurement of intracranial compliance using cine MRI in normal pressure hydrocephalus. Acta neurochirurgica Supplement. 2005; 95:303–6. PMID: 16463870. 21. Lorenz R, Bock J, Snyder J, Korvink JG, Jung BA, Markl M. Influence of eddy current, Maxwell and gradient field corrections on 3D flow visualization of 3D CINE PC-MRI data. Magnetic resonance in medicine. 2014; 72(1):33–40. https://doi.org/10.1002/mrm.24885 PMID: 24006013; PubMed Central PMCID: PMC3872262.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 16 / 17 Non-invasive assessment of pulsatile ICP with PC-MRI

22. Marmarou A, Shulman K, LaMorgese J. Compartmental analysis of compliance and outflow resistance of the cerebrospinal fluid system. Journal of neurosurgery. 1975; 43(5):523–34. Epub 1975/11/01. https://doi.org/10.3171/jns.1975.43.5.0523 PMID: 1181384. 23. Alperin N, Loftus JR, Oliu CJ, Bagci AM, Lee SH, Ertl-Wagner B, et al. Magnetic resonance imaging measures of posterior cranial fossa morphology and cerebrospinal fluid physiology in Chiari malformation type I. Neurosurgery. 2014; 75(5):515–22; discussion 22. https://doi.org/10.1227/NEU. 0000000000000507 PMID: 25328981; PubMed Central PMCID: PMC4854794. 24. Fric R, Eide PK. Comparison of pulsatile and static pressures within the intracranial and lumbar compartments in patients with Chiari malformation type 1: a prospective observational study. Acta neurochirurgica. 2015; 157(8):1411–23; discussion 23. https://doi.org/10.1007/s00701-015-2465-x PMID: 26105759. 25. Jaeger M, Khoo AK, Conforti DA, Cuganesan R. Relationship between intracranial pressure and phase contrast cine MRI derived measures of intracranial pulsations in idiopathic normal pressure hydrocephalus. Journal of clinical neuroscience: official journal of the Neurosurgical Society of Australasia. 2016; 33:169–72. https://doi.org/10.1016/j.jocn.2016.03.029 PMID: 27519145. 26. Haughton V, Mardal KA. Spinal fluid biomechanics and imaging: an update for neuroradiologists. AJNR American journal of neuroradiology. 2014; 35(10):1864–9. https://doi.org/10.3174/ajnr.A4023 PMID: 25012674. 27. Martin BA, Kalata W, Shaffer N, Fischer P, Luciano M, Loth F. Hydrodynamic and longitudinal imped- ance analysis of cerebrospinal fluid dynamics at the craniovertebral junction in type I Chiari malformation. PloS one. 2013; 8(10):e75335. https://doi.org/10.1371/journal.pone.0075335 PMID: 24130704; PubMed Central PMCID: PMC3794956. 28. Rutkowska G, Haughton V, Linge S, Mardal KA. Patient-specific 3D simulation of cyclic CSF flow at the craniocervical region. AJNR American journal of neuroradiology. 2012; 33(9):1756–62. https://doi.org/ 10.3174/ajnr.A3047 PMID: 22517282. 29. Heidari Pahlavian S, Yiallourou T, Tubbs RS, Bunck AC, Loth F, Goodin M, et al. The impact of spinal cord nerve roots and denticulate ligaments on cerebrospinal fluid dynamics in the cervical spine. PloS one. 2014; 9(4):e91888. https://doi.org/10.1371/journal.pone.0091888 PMID: 24710111; PubMed Cen- tral PMCID: PMC3977950. 30. Cheng S, Fletcher D, Hemley S, Stoodley M, Bilston L. Effects of fluid structure interaction in a three dimensional model of the spinal subarachnoid space. Journal of biomechanics. 2014; 47(11):2826–30. https://doi.org/10.1016/j.jbiomech.2014.04.027 PMID: 25005435. 31. Pahlavian SH, Loth F, Luciano M, Oshinski J, Martin BA. Neural Tissue Motion Impacts Cerebrospinal Fluid Dynamics at the Cervical Medullary Junction: A Patient-Specific Moving-Boundary Computational Model. Annals of biomedical engineering. 2015; 43(12):2911–23. https://doi.org/10.1007/s10439-015- 1355-y PMID: 26108203; PubMed Central PMCID: PMC4626272. 32. Roldan A, Wieben O, Haughton V, Osswald T, Chesler N. Characterization of CSF hydrodynamics in the presence and absence of tonsillar ectopia by means of computational flow analysis. AJNR American journal of neuroradiology. 2009; 30(5):941–6. https://doi.org/10.3174/ajnr.A1489 PMID: 19299486. 33. Dreha-Kulaczewski S, Joseph AA, Merboldt KD, Ludwig HC, Gartner J, Frahm J. Inspiration is the major regulator of human CSF flow. The Journal of neuroscience: the official journal of the Society for Neuroscience. 2015; 35(6):2485–91. Epub 2015/02/13. https://doi.org/10.1523/jneurosci.3246-14.2015 PMID: 25673843. 34. Daouk J, Bouzerar R, Baledent O. Heart rate and respiration influence on macroscopic blood and CSF flows. Acta radiologica. 2017; 58(8):977–82. https://doi.org/10.1177/0284185116676655 PMID: 28273732.

PLOS ONE | https://doi.org/10.1371/journal.pone.0188896 November 30, 2017 17 / 17

III

doi:10.1093/brain/awx191 BRAIN 2017: 140; 2691–2705 | 2691

Glymphatic MRI in idiopathic normal pressure hydrocephalus

Geir Ringstad,1,2 Svein Are Sirirud Vatnehol3 and Per Kristian Eide2,4

The glymphatic system has in previous studies been shown as fundamental to clearance of waste metabolites from the brain interstitial space, and is proposed to be instrumental in normal ageing and brain pathology such as Alzheimer’s disease and brain trauma. Assessment of glymphatic function using magnetic resonance imaging with intrathecal contrast agent as a cerebrospinal fluid tracer has so far been limited to rodents. We aimed to image cerebrospinal fluid flow characteristics and glymphatic function in humans, and applied the methodology in a prospective study of 15 idiopathic normal pressure hydrocephalus patients (mean age 71.3 Æ 8.1 years, three female and 12 male) and eight reference subjects (mean age 41.1 + 13.0 years, six female and two male)

with suspected cerebrospinal fluid leakage (seven) and intracranial cyst (one). The imaging protocol included T1-weighted magnetic resonance imaging with equal sequence parameters before and at multiple time points through 24 h after intrathecal injection of the contrast agent gadobutrol at the lumbar level. All study subjects were kept in the supine position between examinations during the first day. Gadobutrol enhancement was measured at all imaging time points from regions of interest placed at predefined locations in brain parenchyma, the subarachnoid and intraventricular space, and inside the sagittal sinus. Parameters demonstrating gadobutrol enhancement and clearance in different locations were compared between idiopathic normal pressure hydrocephalus and reference subjects. A characteristic flow pattern in idiopathic normal hydrocephalus was ventricular reflux of gadobutrol from the subarachnoid space followed by transependymal gadobutrol migration. At the brain surfaces, gadobutrol propagated antegradely along large leptomeningeal arteries in all study subjects, and preceded glymphatic enhancement in adjacent brain tissue, indicating a pivotal role of intracranial pulsations for glymphatic function. In idiopathic normal pressure hydrocephalus, we found delayed enhancement (P 5 0.05) and decreased clearance of gadobutrol (P 5 0.05) at the Sylvian fissure. Parenchymal (glymphatic) enhancement peaked overnight in both study groups, possibly indicating a crucial role of sleep, and was larger in normal pressure hydrocephalus patients (P 5 0.05 at inferior frontal gyrus). We interpret decreased gadobutrol clearance from the subarachnoid space, along with persisting enhancement in brain parenchyma, as signs of reduced glymphatic clearance in idiopathic normal hydrocephalus, and hypothesize that reduced glymphatic function is instrumental for dementia in this disease. The study shows promise for glymphatic magnetic resonance imaging as a method to assess human brain metabolic function and renders a potential for contrast enhanced brain extravascular space imaging.

1 Department of Radiology and Nuclear Medicine, Oslo University Hospital, Rikshospitalet, Oslo, Norway 2 Faculty of Medicine, University of Oslo, Oslo, Norway 3 The Intervention Centre, Oslo University Hospital-Rikshospitalet, Oslo, Norway 4 Department of Neurosurgery, Oslo University Hospital-Rikshospitalet, Oslo, Norway Correspondence to: Professor Per Kristian Eide, MD PhD Department of Neurosurgery Oslo University Hospital – Rikshospitalet Pb 4950 Nydalen N-0424 Oslo, Norway E-mail: [email protected] or [email protected] Keywords: glymphatic function; idiopathic normal pressure hydrocephalus; MRI; gadobutrol; cerebrospinal ﬂuid

Received May 25, 2017. Revised June 7, 2017. Accepted June 17, 2017. Advance Access publication August 18, 2017 ß The Author (2017). Published by Oxford University Press on behalf of the Guarantors of Brain. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact [email protected]

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2692 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

Abbreviations: FLAIR = ﬂuid attenuated inversion recovery; Gd-DTPA = gadolinium-diethylenetriaminpentaacetate; iNPH = idiopathic normal pressure hydrocephalus

patients with iNPH and reference subjects with MRI before Introduction and at multiple time points after intrathecal contrast agent The ‘glymphatic’ (glia-lymphatic) system denotes a brain- administration. wide pathway for clearance of brain metabolites, and depends on convective fluid transport through the interstitial space, mediated by AQP4 channels at astrocytic peri- Materials and methods vascular end feet (Iliff et al., 2012). Impaired glymphatic The study was approved by the Institutional Review Board transport with accumulation of cellular waste products b (2015/1868), Regional Ethics Committee (2015/96) and the such as amyloid- and tau aggregation has been proposed National Medicines Agency (15/04932-7). Inclusion was by as instrumental in conditions such as normal ageing (Kress written and oral informed consent. et al., 2014), brain trauma (Plog et al., 2015) and Alzheimer’s disease (Iliff et al., 2012). Glymphatic transport of solutes is highly dependent on sleep (Xie et al., 2013) Experimental design and body posture (Lee et al., 2015). In this prospective, observational study, consecutive patients While studies of the mouse brain have applied two-photon with iNPH and reference subjects were imaged with MRI in vivo imaging, assessment of glymphatic function has been before, and at multiple time points (10 min to 24 h) after performed with MRI in rat brain, using subarachnoid admin- intrathecal lumbar administration of the MRI contrast agent gadobutrol. Patients with iNPH and reference subjects were istration of the linear, gadolinium-based contrast agent Gd- prospectively enrolled from October 2015 to May 2016. DTPA (gadolinium-diethylenetriaminpentaacetate) (Iliff et al., After gadobutrol administration, all patients and reference sub- a et al. 2013 ;Yang , 2013). Gd-DTPA is of small molecular jects were admitted and observed in the hospital for at least weight (938 Da), which is an important precondition for 24 h. Exclusion criteria were: history of hypersensitivity reac- access of solutes to the interstitial space (Iliff et al., 2012). tions to contrast agents; history of severe allergy reactions in Gd-based contrast agents have already been used to some general, evidence of renal dysfunction, pregnant or breastfeed- extent to diagnose CSF leaks as cause of intracranial hypo- ing females, and age 518 or 480 years. tensioninhumans,andbeendocumentedtohavelowrisk According to the observational study design, randomization a priori (Jinkins et al., 2002; Reiche et al., 2002; Aydin et al., 2004; of patients was not relevant, neither was sample size Akbar et al., 2012). By administration of an intrathecal con- calculations. As ventricular enlargement is inherent with the iNPH diagnosis, image post-processing could therefore not trast agent at the lumbar level, ‘glymphatic MRI’ may be done blinded for the dichotomy iNPH/non-iNPH. thereby bridge the gap from animal studies to investigations of glymphatic clearance of solutes from the human brain interstitial space (Jessen et al., 2015). A case report demon- Subjects strated, for the first time, signs of glymphatic contrast Patients with idiopathic normal pressure enhancement in a patient under clinical work-up for CSF hydrocephalus leakage (Eide and Ringstad, 2015); however, no patient We included consecutive iNPH patients who were referred to cohort has yet been investigated with glymphatic MRI. the Department of Neurosurgery, Oslo University Hospital - Glymphatic convective solute transport through the inter- Rikshospitalet, Oslo, Norway based on clinical symptoms and et al. stitial space is dependent on arterial pulsations (Iliff , findings indicative of iNPH, and with imaging findings of ven- 2013b). Impaired glymphatic function may therefore be a triculomegaly. Within the Department of Neurosurgery, a clin- factor in poorly understood diseases characterized by ical assessment was done, and clinical severity graded based on reduced intracranial compliance and thus restricted intra- a previously described NPH grading scale (Eide and Sorteberg, cranial artery pulsations, such as in idiopathic normal pres- 2010). All patients underwent overnight intracranial pressure sure hydrocephalus (iNPH) (Eide and Sorteberg, 2010). monitoring for the purpose of selection for shunt surgery by INPH is a neurodegenerative disease with a typical symp- using threshold levels as previously defined (Eide and Sorteberg, 2010). tom triad of gait ataxia, urinary incontinence and dementia and can be treated surgically with CSF diversion proce- Reference subjects dures (Adams et al., 1965). Altered glymphatic function Reference subjects were, in parallel with iNPH patients, in iNPH could possibly be a mechanism behind the high recruited prospectively and consecutively from referrals to clin- comorbidity between iNPH and Alzheimer’s disease ical work-up of suspected CSF leakage syndrome or intracra- (Cabral et al., 2011). nial cysts. They underwent MRI with intrathecal gadobutrol The purpose of this study was therefore to assess CSF with the primary indication to define site of CSF leakage, or flow characteristics and glymphatic function in a cohort of cyst enhancement, respectively.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2693

MRI protocol move without any restrictions between the 4 pm examination at the end of Day 1 and the 24 h scan the next morning. Sagittal 3D T1-weighted volume scans were obtained exclusively While the MRI exams, for practical reasons, could not be in a 3 T Philips Ingenia MRI scanner (Philips Medical systems) obtained at identical time points for every study subject, all with equal imaging protocol settings at all time points. The main exams were categorized into the following time intervals: pre- imaging parameters were: repetition time = ‘shortest’ (typically contrast, 0–20 min, 20–40 min, 40–60 min, 1–2 h, 2–4 h, 4–6 h, 5.1 ms), echo time = ‘shortest’ (typically 2.3 ms), flip angle = 8, 6–9 h, and 24 h. field of view = 256 Â 256 cm and matrix = 256 Â 256 pixels (reconstructed 512 Â 512). We sampled 184 over-contiguous (overlapping) slices with 1-mm thickness, which was automati- Image analysis cally reconstructed to 368 slices. The total duration of each image For each time point, a neuroradiologist (G.R.) with 10 years’ acquisition was 6 min and 29 s. At each time point, an automated experience in neuroradiology, placed circular regions of inter- anatomy recognition protocol based on landmark detection in est at predefined locations directly on axially reconstructed T1- TM MRI data (SmartExam , Philips Medical Systems) was applied weighted images in the hospital PACS (picture archiving and Õ to secure consistency and reproducibility of the MRI studies. communication system (Sectra IDS7 , Sectra). Each region of Before gadobutrol administration, we also scanned patients and interest provides the mean signal unit from the image grey- reference subjects with a sagittal 3D fluid attenuated inversion scale, which can be compared between time points and study recovery (FLAIR) volume sequence, where the main imaging subjects when all MRI sequence parameters are kept equal. All parameters were: repetition time = 4800 ms, echo time = ‘shortest’ regions of interest were fitted to local anatomy to avoid partial (typically 318 ms), inversion recovery time = 1650 ms, field of volume effects from neighbouring tissue or CSF. As this Â Â view = 250 250 mm and matrix = 250 250 pixels (recon- human CSF tracer study is conceptually preceded by a similar Â structed 512 512). We sampled 184 over-contiguous slices study in rodents only (Iliff et al., 2013a), the position of with 1 mm thickness, which was automatically reconstructed to regions of interest were partly founded on observations made 365 slices. in animals, but also from previously recognized CSF flow patterns in iNPH derived from other sorts of imaging studies such Intrathecal administration of as radionuclide cisternography (Heinz et al., 1970) and phase- contrast MRI (Penn et al., 2011; Ringstad et al., 2016). gadobutrol For the ventricular and subarachnoid CSF spaces, a single region of interest was placed at the third and fourth ventricle, The study participants met at 8 am for a pre-contrast MRI and pontine cistern, and at the foramen magnum (cisterna magna). were thereafter transported on a mobile table to an adjacent Bilateral regions of interest were placed inside the frontal neurological surgery room, where an interventional neurora- horns of the lateral ventricles, Sylvian fissures, and in the cen- diologist performed X-ray guided lumbar puncture. Intrathecal tral sulci at the vertex of the brain. In addition, perivascular injection of 0.5 ml of 1.0 mmol/ml gadobutrol (GadovistTM, gadobutrol enhancement along the anterior, middle and pos- Bayer Pharma) was preceded by verifying the correct position terior cerebral arteries at the brain surface was dichotomized of the syringe tip in the subarachnoid space in terms of CSF as present and non-present, and the time point where such backflow from the puncture needle, and by injecting a perivascular enhancement could be observed, was registered. small amount (typically 3 ml) of 270 mg I/ml iodixanol For the brain parenchyma, a single region of interest was (VisipaqueTM, GE Healthcare) to confirm unrestricted distribu- placed in the central pons and bilaterally in thalami, inferior tion of radiopaque contrast agent in the lumbar subarachnoid frontal gyri anterior to the Sylvian fissures, periventricular white space. Following needle removal, the study subjects were matter outside the frontal horns, and precentral gyri, respectively. instructed to rotate themselves around the long axis of the A single region of interest was also placed in the posterior body once before transportation back to the MRI suite, part of the superior sagittal sinus. while remaining in the supine position. Supplementary Fig. 1 illustrates region of interest measurements from the Sylvian fissures and inferior frontal gyri in a Post-contrast MRI acquisitions patient. Finally, periventricular hyperintensity detected as any periven- Consecutive, identical MRI examinations with the previously tricular signal increase on FLAIR images was classified into four outlined MRI protocol parameters were initiated as soon as grades according to a modification (Iwamoto et al.,2001)ofthe possible after intrathecal gadobutrol administration (typically classification by van Swieten et al. (1990): grade 0, no periven- with 10-min delay), and performed approximately every tricular hyperintensity; grade 1, marginal periventricular hyper- 10 min during the first hour after contrast agent injection. intensity; grade 2, moderate between grade 1 and 3; grade 3, The study participants were thereafter instructed to remain severe periventricular hyperintensity extending to the subcortex. supine in bed, while one pillow under the head was allowed. Repeated, identical image acquisitions were then performed approximately every 2 h after intrathecal gadobutrol adminis- Parameters derived from gadobutrol tration until 4 pm. All transfer of study subjects between the enhancement neurosurgical department and the MRI suite, and between the bed and the MRI table, was performed by the hospital staff to For the enhancement phase, we calculated the time to peak allow for the patient to remain in the supine position. The final enhancement, maximum change in signal unit, and the MRI scanning was performed the next morning (24 h after enhancement coefficient (signal unit/min) for all regions of contrast agent injection). Patients and controls were allowed to interest within the CSF, venous space and brain parenchyma.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2694 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

For the same regions, we also assessed the signal unit decline Figure 2 provides graphic illustrations (trend plots) of from maximum to minimum (latest) enhancement (clearance gadobutrol enhancement at different locations in the sub- time), maximum decline in signal units and the clearance coef- arachnoid space and ventricular space. ficient (signal unit/min). In patients with iNPH, contrast enhancement was delayed in extra-parenchymal subarachnoid space locations Statistical analysis such as CSF in foramen magnum (Fig. 2A), pontine cistern (Fig. 2B), and Sylvian fissure (Fig. 2C). Table 1 provides Statistical analyses were performed using the SPSS software version 22 (IBM Corporation). Differences between categorical evidence that time to peak enhancement within the CSF data were determined using Pearson chi-square test, and dif- spaces was prolonged, and enhancement thus delayed, in ferences between continuous data were determined using inde- iNPH patients. This was significant for Sylvian fissure pendent samples t-tests. Statistical significance was accepted at (close to inferior frontal gyrus) (P 5 0.05) and at the the 0.05 level (two-tailed). vertex of the brain (precentral sulcus) (P 5 0.01). In iNPH, there was also a trend towards delayed enhancement at the foramen magnum, pontine cistern and Sylvian fis- Results sure, but this did not reach statistical significance. No differences between groups were seen for maximum Patients signal unit increase in subarachnoid space at the foramen magnum, pontine cistern, Sylvian fissure, or over vertex. Demographic data are presented in Supplementary Table 1. For the latter, enhancement of gadobutrol was low in The study includes 15 patients with iNPH and eight refer- both groups within the CSF in the precentral sulci. In ence subjects. The reference subjects underwent MRI due to 3/15 patients with iNPH and 3/8 reference subjects, gado- suspicion of intracranial hypotension associated with CSF butrol enhancement was not detected at the upper convex- leakage (n=7), and one had headache possibly associated ity of the brain (precentral sulci) at any time points. with a pineal gland cyst. The patient cohorts were significantly different in terms of age and distribution of symp- Delayed periarterial enhancement in idiopathic toms (P 5 0.001). normal pressure hydrocephalus In the iNPH cohort, indication for shunt surgery was At the outer surface of the brain, gadobutrol distributed found in 15/15 patients. In 2/15 patients, shunt surgery uniformly, and mainly along the outside of the anterior, was not recommended because of other co-morbidity, and middle and posterior cerebral arteries in all iNPH patients thus 13/15 patients were shunted. All shunted patients (13/ and reference subjects; however, this peri-arterial enhance- 13) responded with clinical improvement (Supplementary ment pattern was observed to be significantly delayed in Table 1). patients with iNPH (Fig. 3). In the reference subject cohort, one individual (1/8) was treated surgically for a pineal cyst with a good surgical Ventricular reflux in idiopathic normal pressure result and subsequent symptom improvement. CSF leakage hydrocephalus was found in 2/7 reference subjects (reference Subjects 4 Evidence of ventricular reflux in iNPH was indicated by and 5; Fig. 1B), and a dura leak was confirmed during larger contrast enhancement within the CSF of the fourth surgery in both. One patient with suspected leakage (refer- ventricle (Fig. 2E), third ventricle (Fig. 2F) and lateral ven- ence Subject 1) also underwent intracranial pressure mon- tricles (Fig. 2G). Moreover, the maximum increase in signal itoring, which demonstrated normal values. unit was significantly higher within the fourth (P 5 0.05), There were no serious adverse events to gadobutrol third (P 5 0.05) and lateral (P 5 0.01) ventricles of iNPH administration, either at the lumbar injection site, or sys- patients, indicating redistribution of CSF to the ventricular temically. One patient suffered a pulmonary embolism after spaces in iNPH (Table 1). 1 day; we considered the association with contrast agent administration as unlikely and rather due to a train journey of several hours the day preceding the MRI study. Gadobutrol enhancement within brain parenchyma

Distribution of gadobutrol within Glymphatic enhancement dependent of gadobutrol subarachnoid and ventricular CSF in nearby CSF

spaces Supplementary Table 2 presents T1 signal before gadobutrol as well as the maximum increase in signal unit after Delayed CSF enhancement in idiopathic normal gadobutrol for all individual patients. All individuals pressure hydrocephalus demonstrated the presence of brain parenchymal gadobu- Figure 1 exempliﬁes time-dependent distribution of trol enhancement. An increase of signal unit 5 10% was intrathecally administered gadobutrol in a patient with observed at all locations (pons: 10/15 iNPH patients, 2/8 iNPH (Fig. 1A) and a reference subject (Fig. 1B). reference subjects; thalamus: 12/15 iNPH patients, 3/8

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2695

Figure 1 CSF contrast enhancement at multiple time points. Reconstructed T1-weighted images in sagittal (top row), axial (middle row) and coronal (bottom row) planes from MRI at baseline (before contrast agent administration) and at four of the subsequent imaging time points demonstrating time-dependent contrast enhancement of subarachnoid and intraventricular spaces in iNPH patient (A) and reference subject (B). Reﬂux of gadobutrol to the lateral ventricles was a typical feature of iNPH. In B, retrodural contrast enhancement can be seen on sagittal images (top row) at time points 1 h, 3 h and 4.5 h as sign of a CSF leakage (reference Subject 5).

reference subjects; periventricular frontal horn: 14/15 iNPH Figure 4 demonstrates change in T1 signal (signal unit) patients, 2/8 reference subjects; inferior frontal gyrus: 14/15 after intrathecal gadobutrol within brain parenchyma at the iNPH patients, 6/8 reference subjects; precentral gyrus: group level in iNPH patients and reference subjects. 5/15 iNPH patients, 2/8 reference subjects). There were The brain parenchymal contrast enhancement underneath interindividual differences. The proportion of patients the brain surface correlated with the contrast available in with 510% increase of signal unit was higher in the the nearby CSF; Fig. 5 demonstrates maximum increase in iNPH cohort. The change in signal unit depended highly signal unit within parenchyma as related to nearby CSF. At on location, and with the most evident signal unit change the outer surface of the brain, this association was by far at the periventricular white matter of iNPH patients and at strongest between the Sylvian ﬁssure and adjacent inferior the inferior frontal gyrus of iNPH and reference subjects. frontal gyrus.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2696 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

Figure 2 Trend plots of CSF tracer (gadobutrol) enhancement depending on location. Gadobutrol enhancement was delayed in iNPH patients as compared to reference subjects in (A) foramen magnum, (B) nearby pons, (C) Sylvian ﬁssure, and (D) precentral sulcus. On the other hand, enhancement was signiﬁcantly stronger within CSF of (E) fourth ventricle, (F) third ventricle, and (G) lateral ventricles. Reference subjects: continuous lines; iNPH patients: dotted lines. Differences between groups at individual time points were determined by independent samples t-test (*P 5 0.05, **P 5 0.01, ***P 5 0.001).

Before gadobutrol administration, the parenchymal T1 (Fig. 4 and Table 2). Overnight, at 24 h, the T1 signal had signal was signiﬁcantly lower within iNPH cases in pons, increased in all study subjects, but signiﬁcantly more in thalamus, periventricular frontal horn, and inferior frontal iNPH patients than reference patients as compared to the gyrus, and remained lower at all time points during Day 1 4 pm scan in the inferior frontal gyrus (P 5 0.05) and

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2697

Table 1 Information about the enhancement phase for different regions of interest within CSF and venous space and brain parenchyma

Enhancement phase Baseline (SU) Max increase (SU) Time to peak (min) Enhancement coefficient (SU/min) iNPH REF iNPH REF iNPH REF iNPH REF CSF space Foramen magnum 10.3 + 3.4 8.9 + 2.6 398.3 Æ 108.0 426.0 Æ 123.0 174.5 Æ 94.2 129.4 Æ 156.7 5.9 Æ 12.2 17.7 Æ 24.5 Sylvian fissure 7.0 + 1.6 8.6 + 3.6 226.2 Æ 100.2 204.2 Æ 104.4 689.0 Æ 536.3a 258.6 Æ 114.8 0.57 Æ 0.41 0.9 Æ 0.7 Pontine cistern 10.2 + 3.7 9.5 + 1.3 378.2 Æ 133.0 388.5 Æ 121.7 230.9 Æ 98.4 159.8 Æ 148.9 2.4 Æ 2.3 10.8 Æ 18.2 Fourth ventricle 11.3 + 3.1 12.6 + 3.0 386.5 Æ 138.2a 236.0 Æ 142.9 286.7 Æ 322.9 123.4 Æ 98.1 5.6 Æ 14.5 6.3 Æ 14.0 Third ventricle 14.4 + 2.6 14.5 + 2.8 243.6 Æ 136.9a 115.1 Æ 103.8 615.3 Æ 587.5 253.8 Æ 133.2 1.4 Æ 1.7 0.4 Æ 0.4 Lateral ventricle 6.0 + 1.2c 9.4 + 1.5 168.5 Æ 99.8b 42.5 Æ 51.9 843.4 Æ 560.2 402.6 Æ 446.1 0.41 Æ 0.39 0.13 Æ 0.16 Precentral sulcus 7.6 + 1.6 8.1 + 1.2 131.1 Æ 94.1 87.8 Æ 99.3 1254.7 Æ 483.8b 475.0 Æ 601.9 0.10 Æ 0.06 0.29 Æ 0.39 Central venous space Sagittal sinus/confluence 34.9 + 7.2 32.3 + 7.6 8.1 Æ 8.0 7.4 Æ 9.6 350.2 Æ 457.3 524.4 Æ 591.3 0.16 Æ 0.25 0.41 Æ 0.85 Brain parenchyma Inferior frontal gyrus 82.5 + 4.3b 89.5 + 3.2 39.1 Æ 25.3 23.8 Æ 20.9 1358.0 Æ 313.4 1421.3 Æ 53.1 0.03 Æ 0.02 0.02 Æ 0.02 Pons 96.8 + 4.2a 101.1 + 4.8 11.6 Æ 7.0 7.4 Æ 5.8 1090.0 Æ 602.3 929.4 Æ 676.9 0.03 Æ 0.08 0.08 Æ 0.17 Thalamus 86.7 + 4.8b 93.9 + 6.0 13.0 Æ 6.8 9.0 Æ 7.1 1261.7 Æ 462.6 1288.4 Æ 386.3 0.01 Æ 0.02 0.01 Æ 0.01 Frontal horn 63.9 + 18.4b 94.1 + 15.7 48.5 Æ 32.8b 7.9 Æ 7.9 1349.3 Æ 350.9 1109.8 Æ 585.6 0.04 Æ 0.02 0.05 Æ 0.12 Precentral gyrus 82.7 + 6.0 87.4 + 3.7 9.0 Æ 8.3 7.8 Æ 11.2 900.0 Æ 686.4 742.8 Æ 727.9 0.04 Æ 0.06 0.03 Æ 0.03

Signiﬁcant differences between iNPH and REF groups were determined by independent samples t-test: aP 5 0.05, bP 5 0.01, cP 5 0.001. REF = reference subject; SU = signal unit.

outside the frontal horn (P = 0.008) (Supplementary Table unit change from circulating contrast agent in cerebral ves- 3). Associations between enhancement in CSF and nearby sels (Table 2). parenchyma were also largest in these locations (R = 0.85, P 5 P 5 0.001 and R = 0.84, 0.001, respectively). Clearance of gadobutrol Transependymal CSF flow in idiopathic normal At 24 h, contrast enhancement was more profound in all pressure hydrocephalus CSF compartments in iNPH patients, which indicates Figure 4C demonstrates contrast enhancement within peri- delayed clearance of gadobutrol compared with reference ventricular white matter parenchyma outside the frontal subjects (Fig. 2). This is further detailed in Table 2. horns of the lateral ventricles. This enhancement was Accordingly, the maximum reduction of signal unit was strongly associated with ventricular reflux of contrast significantly lower within the CSF of foramen magnum media in the lateral ventricles (Fig. 5D) and significantly (P 5 0.01), Sylvian fissure (P 5 0.05) and pontine cistern higher in iNPH cases (Table 2 and Fig. 2 E–G). (P 5 0.01) in iNPH. The clearance coefficients were signifi- As illustrated in Fig. 6, the present data provide evidence cantly lower for iNPH cases within CSF of foramen that periventricular hyperintensity on FLAIR images is magnum (P 5 0.01), Sylvian fissure (P 5 0.05) and pontine dependent on transpendymal water flux from ventricles to cistern (P 5 0.01) (Table 2). the periventricular white matter. The degree of periventri- For the brain parenchyma, signal unit peaked at 24 h, cular hyperintensity, graded 0–3, was significantly related and the current observation period was therefore too to the contrast enhancement in the periventricular frontal short to determine clearance time of gadobutrol from the brain parenchyma. The maximum signal unit increase in brain. periventricular frontal brain parenchyma was significantly Moreover, and as already noted, there were no signs of higher in grade 2 (P = 0.02) and grade 3 (P = 0.03), as gadobutrol enhancement in the central venous space, and compared to grade 0 (Fig. 6). therefore estimation of gadobutrol clearance to venous blood was not relevant. Gadobutrol enhancement within central venous space Discussion There was no significant signal unit increase that could be observed in the superior sagittal sinus at any time point In this study, humans (patients with iNPH and reference after intrathecal administration of gadobutrol, and thus subjects) were for the first time investigated by use of an no evidence of any contribution to parenchymal signal intrathecally administered, small molecular weight contrast

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2698 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

concerns. Our observations demonstrate that gadobutrol propagates in subarachnoid space antegradely along the outside of large leptomeningeal arteries, and that presence of gadobutrol in subarachnoid space is a prerequisite for glymphatic contrast agent enhancement in adjacent brain parenchyma. Glymphatic enhancement was different between patients and brain regions, occurred much later in humans than reported in rats, and peaked at the final 24 h MRI exam. Clearance of gadobutrol was delayed in iNPH. Moreover, signs of net CSF ventricular reflux and transependymal flow from the ventricles towards periventricular white matter was a characteristic feature of iNPH. Enhancement of gadobutrol outside the upper brain convexities was sparse in both iNPH as well as reference subjects, even after 24 h, and in some subjects was not detectable.

Intrathecal administration of gadobutrol as CSF tracer The macrocyclic MRI contrast agent gadobutrol has low molecular weight (605 Da) and is a highly hydrophilic, non-ionic compound that easily distributes in water (Cheng, 2004) and is thus well suited as a CSF tracer. While a linear contrast agent (Gd-DTPA) was applied in a previous study of glymphatic function in rat brain (Iliff et al., a Figure 3 Perivascular enhancement. Gadobutrol enhance- 2013 ), it has recently been recommended that consideration ment in sulci traversed by the major cerebral arteries (anterior, should be made in terms of using macrocyclic Gd-based middle and posterior) was identified and categorized as present/ contrast agents rather than linear agents in humans absent for each time point. (A) MRI in axial plane 1 h after contrast (Malayeri et al., 2016), due to their superior stability in agent administration shows gadobutrol distributed to the interhe- biological tissue (Lohrke et al., 2016). Regarding glymphatic mispheric fissure, Sylvian fissure and ambient cistern (arrows), and transport, gadobutrol also has the advantages of even smal- reflux to the third ventricle (dotted arrow). (B) At later time points, ler molecular size (molecular weight 604 Da) than Gd- even though contrast subsequently distributed more freely in the DTPA, and higher T relaxation (Lohrke et al., 2016). subarachnoid spaces, there was a clear tendency for enhancement in 1 In studies of mice (Iliff et al., 2012; Bedussi et al., 2017), cerebral fissures traversed by the anterior, middle and posterior arteries in both iNPH and reference subjects. (C) Evidence of tracer has been injected into the cisterna magna at rates perivascular gadolinium enhancement was categorized as present/ and volumes that potentially could overwhelm the intracra- absent for each time point. The percentage of individuals in whom a nial compartment (Hladky and Barrand, 2014), and perivascular contrast enhancement was visualized, is plotted for thereby interfere with physiological CSF flow. Another each time point, showing significant differences between reference study has, however, demonstrated that a minor and tran- subjects and iNPH cohorts (20–40 min: P = 0.049; 40–60 min: sient increase in intracranial pressure is unlikely to affect P = 0.01; Pearson chi-square test). The plot in C suggests delayed glymphatic tracer transport (Yang et al., 2013). In this perivascular flow at the brain surface in iNPH patients. study of humans, we injected less than a total of 5 ml fluid into the subarachnoid space, level with the lumbar spine, and interference with intracranial pressure and CSF flow at later time points should therefore be negligible. agent (gadobutrol) as CSF tracer to assess brain glymphatic In accordance with previous studies using other MRI function with MRI (glymphatic MRI) and also CSF flow contrast agents for intrathecal use, gadobutrol also characteristics in the subarachnoid space and ventricles seemed safe for this purpose, as there were no observed over multiple time points. The iNPH cohort consisted of serious adverse events related to the administration of con- a high proportion of ‘true’, or definitive, iNPH defined by trast agent in the current study groups. the shunt response rate. Moreover, the reference patients can be considered suitable to serve as controls, as CSF leakage was found in only 2/8. Since intrathecal adminis- CSF flow characteristics tration of MRI contrast agent is merely used off-label on First signs of gadobutrol at the level of the foramen clinical indication, inclusion of non-symptomatic controls magnum occurred typically within 20 min for most study at this stage of research was not done due to ethical subjects, who all were kept in supine position during and

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2699

Figure 4 Glymphatic enhancement. Trend plots of gadobutrol enhancement in parenchyma depending on location, including (A) pons,

(B) thalamus, (C) periventricular frontal horn, (D) inferior frontal gyrus (IFG), and (E) precentral gyrus. The T1 signal (signal unit) was signiﬁcantly lower in iNPH cases at various locations in pons, thalamus, periventricular frontal horn, and IFG. In iNPH cases, the change was signiﬁcantly higher in periventricular frontal horn and IFG between last daytime exam and at 24 h (Supplementary Table 3), and remained at a higher level after 24 h, indicative of delayed clearance. Reference subjects (Ref): continuous lines; iNPH patients: dotted lines. Differences between groups at individual time points were determined by independent samples t-test (*P 5 0.05, **P 5 0.01, ***P 5 0.001).

after injection of contrast agent through the final MRI around sulcal vessels at a much wider zone (Fig. 3) than examination at Day 1. The tracer did primarily not distri- can be appreciated by the area covered with an electron bute freely around the brain convexities, but rather within microscope. Additional causes to this perivascular distribu- sulci and cisternal spaces traversed by the three major cer- tion pattern should therefore be explored in further studies. ebral leptomeningeal arteries (anterior, middle and poster- Arterial pulsations are the force behind CSF pulsations ior), and gadobutrol propagated further along their (Greitz, 2004). The finding of parenchymal enhancement antegrade directions. This distribution pattern in the sub- was strongly correlated with enhancement in adjacent subarachnoid space could possibly correspond to leptomenin- arachnoid space (Fig. 5), and therefore also strongly sup- geal perivascular sheaths described in electron microscopy ports antegrade CSF flow within paravascular spaces of studies (Weller, 2005). In the current in vivo observations, brain parenchymal penetrating arteries. This is consistent MRI demonstrated, however, presence of gadobutrol tracer with animal studies where tracer was administered in

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2700 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

Figure 5 Glymphatic enhancement as function of CSF enhancement. The association between maximum enhancement within the CSF spaces and nearby brain parenchyma was determined for different regions of interest. For all locations (A–E), there was a highly signiﬁcant correlation between contrast agent availability within the CSF space and enhancement of gadobutrol within nearby parenchyma.

Table 2 Information about the clearance phase for different regions of interest within CSF and venous spaces

Clearance phase Clearance time (min) Max reduction (SU) Clearance coefficient (SU/min) iNPH REF iNPH REF iNPH REF CSF space Foramen magnum 1264.3 Æ 116.1 1294.4 Æ 154.5 À206.7 Æ 90.8b À358.1 Æ 117.7 À0.16 Æ 0.07 b À0.28 Æ 0.08 Sylvian fissure 1123.7 Æ 41.1 1162.6 Æ 78.1 À68.0 Æ 51.2a À133.0 Æ 68.4 À0.06 Æ 0.05a À0.12 Æ 0.06 Pontine cistern 1207.9 Æ 114.8 1261.5 Æ 137.0 À177.9 Æ 95.5b À313.3 Æ 105.3 À0.15 Æ 0.08b À0.25 Æ 0.08 Fourth ventricle 1234.4 Æ 109.1 1287.6 Æ 105.9 À231.6 Æ 128.1 À232.6 Æ 103.7 À0.19 Æ 0.1 À0.18 Æ 0.08 Third ventricle 1234.3 Æ 116.2 1135.3 Æ 74.2 À114.4 Æ 76.8 À103.7 Æ 83.9 À0.10 Æ 0.07 À0.09 Æ 0.08 Lateral Ventricle 1115.1 Æ 46.3 1098.0 Æ 12.5 À57.3 Æ 48.9 À72.3 Æ 11.9 À0.05 Æ 0.04 À0.07 Æ 0.01 Precentral sulcus 1375.5 + 67.2 1261.7 + 168.6 À6.5 + 3.5 À57.0 + 63.2 À0.005 + 0.003 À0.05 + 0.06 Central venous space Sagittal sinus/confluence 1287.4 Æ 150.9 1247.3 Æ 175.9 À12.1 Æ 6.7 À9.5 Æ 3.9 À0.01 Æ 0.006 À0.008 Æ 0.003

Signiﬁcant differences between iNPH and REF groups were determined by independent samples t-test: aP 5 0.05, bP 5 0.01, cP 5 0.001. REF = reference subject; SU = signal unit. Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2701

convexities and thus a very limited role of arachnoid villi along the sagittal sinus for CSF resorption. While CSF seems to enter the brain driven by high-pressure arteries traversing the lower surface of the brain, it could be hypothesized that CSF exits at paravenous routes mainly along areas of the brain where presence of (and translated pressure from) larger arteries is more sparse, which is the case around the upper convexities. In iNPH, reduced intracranial compliance and increased intracranial pulse pressure is an important feature of the disease (Eide and Sorteberg, 2010), and it has also been described as a ‘restricted arterial pressure pulsation syndrome’ (Greitz, 2004), referring to decreased ability of arteries in subarachnoid space to expand secondary to reduced intracranial compliance. In mice, it was demonstrated that decreased arterial pulsatility reduced paravascular flow (Iliff et al., 2013b). In our cohort of iNPH patients with expected decreased arterial pulsatility due to reduced intracranial compliance, and consistent with the mentioned observations in mice, we found that the gadobutrol tracer distributed more slowly along external brain surface arteries than in reference subjects, and further, clearance of gadobutrol was decreased. It may therefore be hypothesized that a vicious cycle may be established in iNPH, where restricted arterial pulsations reduces glymphatic flow, which leads to reduced transport of CSF and solutes through glymphatic pathways and thereby further Figure 6 Transependymal flow of MRI contrast agent. reduces intracranial compliance. In Alzheimer’s disease, the (A) Periventricular hyperintensity (PVH) was identified and graded glymphatic system may contribute to a larger portion of into four grades from FLAIR images (right column). The maximum extracellular amyloid-b clearance than previously thought increase in signal unit within the periventricular brain parenchyma (Tarasoff-Conway et al., 2016), and a recent study demon- between pre-contrast MRI (left column) and 24 h post-contrast MRI strated Alzheimer’s disease-related biopsy findings in iNPH (middle column) was related to the periventricular hyperintensity patients with signs of reduced compliance (Kojoukhova categories. Top row: Axial sections; bottom row: coronal sections. et al., 2017). Reduced intracranial compliance in iNPH (B) ANOVA with Bonferroni corrected post hoc tests revealed that may, however, be a phenomenon secondary to obstructed periventricular enhancement was significantly larger in periventricular lucency grades 2 and 3 than grade 0. CSF flow. In light of the evolving knowledge of brain glymphatic function, we find it reasonable to attribute delayed propagation and clearance of the CSF tracer gadobutrol, as well as presumed restricted CSF pulsations, to an obstruc- subarachnoid space (Iliff et al., 2012, 2013a; Bedussi et al., tion of CSF flow between the paravascular and interstitial 2017), but contradictory to studies where intraparenchymal spaces. Increased resistance to glymphatic flow may also injection of tracers has been used (Carare et al., 2008; cause CSF to follow pathways of least resistance, among Bakker et al., 2016), where it was concluded with retro- them, the retrograde transventricular route. An obstruction grade flow. Presence of soluble amyloid-b deposits in the at this level would also explain disproportionate and loca- walls of leptomeningeal arteries has previously led to the lized expansion of subarachnoid space around large lepto- conclusion that CSF and solutes exit the brain along meningeal arteries in iNPH, which is referred to by some as arteries (Weller et al., 2009). It has, however, been pro- DESH (disproportionally enlarged subarachnoid-spaces posed that different directions of CSF flow along paravas- hydrocephalus) (Hashimoto et al., 2010), but not why cular pathways may co-exist (Bakker et al., 2016). While Virchow-Robin spaces are macroscopically not enlarged we found evidence for antegrade periarterial movement of (Ishikawa et al., 2015). the CSF tracer, there was in many iNPH and reference Astrocytic end feet enclose all parenchymal brain vessels, subjects, a strikingly sparse and sometimes absent enhance- and the water channel AQP4 occupies up to 40% of the ment of tracer at the upper brain convexities, even after end foot area, rendering a vast surface for water transport 24 h with patients remaining mainly in the supine position. throughout the entire brain (Nagelhus and Ottersen, 2013). One explanation to these observations might be a gradient While it has been demonstrated that glymphatic function of net CSF flow from the upper towards the lower brain and clearance of metabolites from the extracellular space is convexities, suggesting CSF secretion at the upper brain reduced in AQP4 knock-out mice (Iliff et al., 2012), no

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2702 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

studies have yet been published regarding AQP4 function compared to reference subjects. We attribute this to an in iNPH. In idiopathic intracranial hypertension perivascu- increase in brain water content in iNPH, since water is

lar AQP4 was, however, increased, and it was interpreted by far the most significant contributor to prolonged T1 as a possible compensatory mechanism to enhance brain relaxation in the brain. Signs of increased brain water con- fluid drainage (Eide et al., 2016). We expect upcoming tent in NPH has also previously been demonstrated in sev- studies of AQP4 content, function and polarization in eral brain regions by using MRI diffusion-weighted imaging iNPH patients. (Gideon et al., 1994). In light of the previously discussed potential role of AQP4, it is notable that increased brain Ventricular reflux of gadobutrol in idiopathic normal water content is also seen in mice with complete loss of pressure hydrocephalus AQP4, however, not in mice with selective removal of peri- In iNPH, gadobutrol entered the ventricular system early, vascular AQP4 or deletion of ependymal AQP4 (Vindedal remained in the ventricles over time, and correlated et al., 2016). strongly with contrast enhancement in adjacent periventri- Glymphatic enhancement was observed at all locations cular white matter, indicating transependymal migration of when preceded by enhancement of the adjacent subarach- the contrast agent, which in its turn was associated with noid space, but was not observed in every patient at all periventricular hyperintensity at FLAIR images. These locations, and there were substantial interindividual varia- observations support previous findings of net retrograde tions (Supplementary Table 2). The current study does not aqueductal flow in iNPH (Ringstad et al., 2016; Yin fully reveal why such differences exist, but we notice that et al., 2017), and this phenomenon has also been associated glymphatic enhancement was particularly apparent and fre- with restricted intracranial pulsations and change of net quent at the inferior frontal gyrus in both iNPH and refer- flow in the antegrade direction after surgical shunting ence subjects, which anatomically is adjacent to the middle (Ringstad et al., 2016). Aqueductal net flow is typically cerebral artery, consistent with a fundamental role of reported in microlitres, which might be within the range pulsations for glymphatic transport, as mentioned. of measurement inaccuracies, and has by some been con- Interestingly, this current in vivo observation is contradic- sidered a sign of measurement inaccuracy when it is tory to recent modelling studies, which conclude that arter- observed, and therefore been subtracted when estimating ial pulsations are an unlikely origin of the driving force that the more robust parameter aqueductal stroke volume to could account for convective solute transport through the predict shunt response in NPH (Bradley et al., 1996). interstitial space (Asgari et al., 2016; Jin et al., 2016). Our observations of ventricular reflux might contribute to In contradiction to previous data from anaesthetized mice explain the mechanism behind hydrocephalus in NPH, and rats, where glymphatic enhancement typically peaked since a precondition for such flow would be a minute posi- within 2 h, glymphatic enhancement in humans peaked at tive pressure gradient from the subarachnoid space to the 24 h, and was, except from in periventricular white matter ventricles, and also over the ventricular ependyma. in iNPH, not detectable at MRI exams obtained at first-day Ventricular dilatation may therefore constitute a manifesta- time intervals following administration of the intrathecal tion of how the brain adapts to such a minor pressure tracer (Fig. 4). Xie et al. (2013) demonstrated that natural gradient, while also increased area of the ependymal sur- sleep or anaesthesia is associated with a 60% increase in face and stretching of ependymal lining may facilitate trans- the interstitial space, which increased glymphatic clearance ependymal water transport. It seems reasonable to assume of amyloid-b 2-fold. Our observations indicate that glym- that the periventricular gadobutrol enhancement observed phatic function is also highly dependent on sleep in in periventricular hyperintensity grade 1–3 expresses trans- humans. ependymal migration by the same mechanisms as reported While enhancement peaked at the final MRI exam of the in previous studies (Bering and Sato, 1963; Sahar et al., study protocol, clearance of gadobutrol from brain par- 1969) rather than within the traditional defined glymphatic enchyma could not be calculated, and further studies pathway via periarterial spaces. A transependymal pressure should account for this observation. However, as clearance gradient, even small, may be sufficient to enlarge the ven- of gadobutrol was delayed in CSF, and glymphatic tricles (Linninger et al., 2005; Levine, 2008). In one study enhancement was found to be a function of CSF enhance- with invasive monitoring, the pulse pressure amplitude was ment (Fig. 5), we find it reasonable to hypothesize that median 0.4 mm Hg higher inside the ventricles than in the maximum parenchymal enhancement at 24 h is a sign of parenchyma (Eide, 2008). Another study was, however, not reduced clearance of gadobutrol via the glymphatic system able to measure a transependymal pressure gradient (Eide in iNPH. As failure of glymphatic function has previously and Saehle, 2010). been associated with reduced clearance of interstitial solutes like amyloid-b, it seems reasonable to hypothesize that the Brain enhancement and clearance of previously poorly understood iNPH dementia may share a pathogenic mechanism with Alzheimer’s disease. gadobutrol There were no signs of contrast agent enhancement at the Before glymphatic contrast enhancement, the parenchymal venous side (superior sagittal sinus) at any time point, indi-

T1 signal unit was lower in all regions of interest in iNPH cating no effect on other observations due to circulating

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2703

contrast agent in veins. The image resolution of a 3 T points, as well as segmentation techniques to differentiate human study is insufﬁcient to allow for detection of menin- brain subregions, and CSF from parenchyma. We are in the geal lymphatic vessels that have previously been proposed process of elaborating such postprocessing algorithms with to represent potential drainage routes from the glymphatic external partners. circulation (Louveau et al., 2015).

Limitations Conclusion CSF leakage was diagnosed in 2/8 reference subjects. These reference subjects may therefore not have a normal CSF Intrathecal administration of an MRI contrast agent ﬂow pattern, and intracranial hypotension may be assumed (gadobutrol) serving as CSF tracer, followed by multiple to affect paravascular transport as well, given the impor- MRI exams over 24 h (glymphatic MRI), demonstrated tant role of CSF pulsations. However, there was no obvious signs of delayed glymphatic clearance in iNPH patients

difference in T1 signal unit change (glymphatic enhance- compared with a reference group. In all study subjects, ment) in the two patients with leakage compared to the the CSF tracer propagated in the subarachnoid space ante- reference subject group. A second limitation is that refer- gradely along large leptomeningeal arteries, and presence ence subjects were significantly younger than the iNPH of tracer in subarachnoid space always preceded glympha- patients and of different gender. It can therefore not be tic enhancement in adjacent brain tissue, indicating a pivo- established to what degree iNPH disease or age/gender con- tal role of intracranial pulsations for glymphatic function. tributed to differences between groups, while reduced glym- Glymphatic enhancement peaked overnight, and we attri- phatic function should be expected with normal ageing bute this to increased glymphatic function during sleep, (Kress et al., 2014). However, in addition to the observed which has been suggested in previous animal experiments. signs of different glymphatic clearance between the two The sparse enhancement of CSF tracer at the upper brain groups, there are also substantial differences in speed of convexities in both study cohorts questions the role of gadobutrol propagation in CSF as well as ventricular CSF resorption at the arachnoid villi. Other features of reflux pattern, which may less likely represent age effects. iNPH were ventricular reflux and transependymal migra- We therefore hypothesize that signs of glymphatic impair- tion of the tracer from the lateral ventricles towards the ment in the iNPH group are more likely to be disease-spe- brain parenchyma. Given the ability of MRI contrast cific rather than purely due to the patient’sage. agents to access and be cleared from the brain extravas- Moreover, even though we standardized MRI routines cular space after intrathecal administration, glymphatic extensively, imaging time points after contrast agent injec- MRI may have the potential to assess brain metabolic tion were not identical due to practical limitations, and function. MRI exams were therefore rather categorized at defined time intervals. We did not think it reasonable to enforce restrictions on patient movement between the final MRI of Day 1 and the MRI of following morning, and could there- Acknowledgements fore not control data for patient activity and body position during that time period, nor for sleep. In these cohorts of We thank Dr Øivind Gjertsen, Dr. Ba˚rd Nedregaard and symptomatic patients, already burdened with an extensive Dr Ruth Sletteberg from the Department of Radiology, number of imaging procedures, it was considered unreason- Oslo University Hospital – Rikshospitalet, who performed able, and even unethical, to demand further exams during the intrathecal gadobutrol injections in all study subjects. evening or night-time. However, this also represents a We also sincerely thank the Intervention Centre and weakness of the study, as it cannot be established at Department of Neurosurgery at Oslo University Hospital which exact time point glymphatic enhancement first Rikshospitalet for providing valuable support with MR occurred. The hypothesis that sleep might have an impact scanning and care-taking of all study subjects throughout on glymphatic function may still be sustained by observa- the study. tions of no, or very sparse, parenchymal enhancement during Day 1, in spite of an adequate amount of contrast agent at the adjacent brain surface, but enhancement at MRI performed on the morning of Day 2 (24 h). Funding Finally, MRI analysis of signal unit at different time No funding was received towards this work. points was carried out manually from predefined areas of brain and CSF spaces, where regions of interest had to be fitted to adjust for local anatomy to avoid partial volume averaging effect. In future studies, a whole-brain and -CSF Supplementary material approach should be preferred. This would require perfect alignment and co-registration of images from different time Supplementary material is available at Brain online.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 2704 | BRAIN 2017: 140; 2691–2705 G. Ringstad et al.

Heinz ER, Davis DO, Karp HR. Abnormal isotope cisternography in References sympatomatic occult hydrocephalus. A correlative isotopic-neuroradiological study in 130 subjects. Radiology 1970; 95: 109–20. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH. Hladky SB, Barrand MA. Mechanisms of fluid movement into, Symptomatic occult hydrocephalus with "Normal" cerebrospinal- through and out of the brain: evaluation of the evidence. Fluids fluid pressure. A treatable syndrome. N Eng J Med 1965; 273: Barriers CNS 2014; 11: 26. 117–26. Iliff JJ, Lee H, Yu M, Feng T, Logan J, Nedergaard M, et al. Brain- Akbar JJ, Luetmer PH, Schwartz KM, Hunt CH, Diehn FE, Eckel LJ. wide pathway for waste clearance captured by contrast-enhanced The role of MR myelography with intrathecal gadolinium in local- MRI. J Clin Invest 2013a; 123: 1299–309. ization of spinal CSF leaks in patients with spontaneous intracranial Iliff JJ, Wang M, Liao Y, Plogg BA, Peng W, Gundersen GA, et al. A hypotension. AJNR Am J Neuroradiol 2012; 33: 535–40. paravascular pathway facilitates CSF flow through the brain paren- Asgari M, de Zelicourt D, Kurtcuoglu V. Glymphatic solute transport chyma and the clearance of interstitial solutes, including amyloid does not require bulk flow. Sci Rep 2016; 6: 38635. beta. Sci Transl Med 2012; 4: 147ra111. Aydin K, Guven K, Sencer S, Jinkins JR, Minareci O. MRI cisterno- Iliff JJ, Wang M, Zeppenfeld DM, Venkataraman A, Plog BA, Liao Y, graphy with gadolinium-containing contrast medium: its role, ad- et al. Cerebral arterial pulsation drives paravascular CSF-interstitial vantages and limitations in the investigation of rhinorrhoea. fluid exchange in the murine brain. J Neurosci 2013b; 33: 18190–9. Neuroradiology 2004; 46: 75–80. Ishikawa M, Yamada S, Yamamoto K. Three-dimensional observation Bakker EN, Bacskai BJ, Arbel-Ornath M, Aldea R, Bedussi B, Morris of Virchow-Robin spaces in the basal ganglia and white matter and AW, et al. Lymphatic clearance of the brain: perivascular, paravas- their relevance to idiopathic normal pressure hydrocephalus. Fluids cular and significance for neurodegenerative diseases. Cell Mol Barriers CNS 2015; 12: 15. Neurobiol 2016; 36: 181–94. Iwamoto T, Miyaji H, Kanaya K, Takasaki M. Periventricular lucency Bedussi B, van der Wel NN, de Vos J, van Veen H, Siebes M, evaluated by second-derivative finger plethysmography. Geriatr VanBavel E, et al. Paravascular channels, cisterns, and the subarach- Gerontol Int 2001; 1: 25–32. noid space in the rat brain: a single compartment with preferential Jessen NA, Munk AS, Lundgaard I, Nedergaard M. The pathways. J Cereb Blood Flow Metab 2017; 37: 1374–85. glymphatic system: a beginner’s guide. Neurochem Res 2015; 40: Bering EA Jr, Sato O. Hydrocephalus: changes in formation and ab- 2583–99. sorption of cerebrospinal fluid within the cerebral ventricles. Jin BJ, Smith AJ, Verkman AS. Spatial model of convective solute J Neurosurg 1963; 20: 1050–63. transport in brain extracellular space does not support a "glym- Bradley WG Jr, Scalzo D, Queralt J, Nitz WN, Atkinson DJ, Wong P. phatic" mechanism. J Gen Physiol 2016; 148: 489–501. Normal-pressure hydrocephalus: evaluation with cerebrospinal fluid Jinkins JR, Rudwan M, Krumina G, Tali ET. Intrathecal gadolinium- flow measurements at MR imaging. Radiology 1996; 198: 523–9. enhanced MR cisternography in the evaluation of clinically sus- Cabral D, Beach TG, Vedders L, Sue LI, Jacobson S, Myers K, et al. pected cerebrospinal fluid rhinorrhea in humans: early experience. Frequency of Alzheimer’s disease pathology at autopsy in patients Radiology 2002; 222: 555–9. with clinical normal pressure hydrocephalus. Alzheimers Dement Kojoukhova M, Vanha KI, Timonen M, Koivisto AM, Nerg O, 2011; 7: 509–13. Rummukainen J, et al. Associations of intracranial pressure with Carare RO, Bernardes-Silva M, Newman TA, Page AM, Nicoll JA, brain biopsy, radiological findings, and shunt surgery outcome in Perry VH, et al. Solutes, but not cells, drain from the brain paren- patients with suspected idiopathic normal pressure hydrocephalus. chyma along basement membranes of capillaries and arteries: sig- Acta Neurochir 2017; 159: 51–61. nificance for cerebral amyloid angiopathy and neuroimmunology. Kress BT, Iliff JJ, Xia M, Wang M, Wei HS, Zeppenfeld D, et al. Neuropathol Appl Neurobiol 2008; 34: 131–44. Impairment of paravascular clearance pathways in the aging brain. Cheng KT. Gadobutrol. Molecular Imaging and Contrast Agent Ann Neurol 2014; 76: 845–61. Database (MICAD). Bethesda, MD: National Center for Lee H, Xie L, Yu M, Kang H, Feng T, Deane R, et al. The effect of Biotechnology Information (US); 2004. body posture on brain glymphatic transport. J Neurosci 2015; 35: Eide PK. Demonstration of uneven distribution of intracranial pulsa- 11034–44. tility in hydrocephalus patients. J Neurosurg 2008; 109: 912–17. Levine DN. Intracranial pressure and ventricular expansion in hydro- Eide PK, Eidsvaag VA, Nagelhus EA, Hansson HA. Cortical astrogliosis and increased perivascular aquaporin-4 in idiopathic intracranial cephalus: have we been asking the wrong question? J Neurol Sci hypertension. Brain Res 2016; 1644: 161–75. 2008; 269: 1–11. Eide PK, Ringstad G. MRI with intrathecal MRI gadolinium contrast Linninger AA, Tsakiris C, Zhu DC, Xenos M, Roycewicz P, Danziger medium administration: a possible method to assess glymphatic Z, et al. Pulsatile cerebrospinal fluid dynamics in the human brain. function in human brain. Acta Radiol Open 2015; 4: IEEE Trans Biomed Eng 2005; 52: 557–65. 2058460115609635. Lohrke J, Frenzel T, Endrikat J, Alves FC, Grist TM, Law M, et al. 25 Eide PK, Saehle T. Is ventriculomegaly in idiopathic normal pressure years of contrast-enhanced MRI: developments, current challenges hydrocephalus associated with a transmantle gradient in pulsatile and future perspectives. Adv Ther 2016; 33: 1–28. intracranial pressure? Acta Neurochir 2010; 152: 989–95. Louveau A, Smirnov I, Keyes TJ, Eccles JD, Rouhani SJ, Peske JD, Eide PK, Sorteberg W. Diagnostic intracranial pressure monitoring and et al. Structural and functional features of central nervous system surgical management in idiopathic normal pressure hydrocephalus: a lymphatic vessels. Nature 2015; 523: 337–41. 6-year review of 214 patients. Neurosurgery 2010; 66: 80–91. Malayeri AA, Brooks KM, Bryant LH, Evers R, Kumar P, Reich DS, Gideon P, Thomsen C, Gjerris F, Sorensen PS, Henriksen O. Increased et al. National Institutes of Health perspective on reports of gado- self-diffusion of brain water in hydrocephalus measured by MR linium deposition in the brain. J Am Coll Radiol 2016; 13: 237–41. imaging. Acta Radiol 1994; 35: 514–19. Nagelhus EA, Ottersen OP. Physiological roles of aquaporin-4 in Greitz D. Radiological assessment of hydrocephalus: new theories and brain. Physiol Rev 2013; 93: 1543–62. implications for therapy. Neurosurg Rev 2004; 27: 145–65; discus- Penn RD, Basati S, Sweetman B, Guo X, Linninger A. Ventricle wall sion 166–7. movements and cerebrospinal fluid flow in hydrocephalus. Hashimoto M, Ishikawa M, Mori E, Kuwana N. Diagnosis of idio- J Neurosurg 2011; 115: 159–64. pathic normal pressure hydrocephalus is supported by MRI-based Plog BA, Dashnaw ML, Hitomi E, Peng W, Liao Y, Lou N, et al. scheme: a prospective cohort study. Cerebrospinal Fluid Res 2010; Biomarkers of traumatic injury are transported from brain to blood 7: 18. via the glymphatic system. J Neurosci 2015; 35: 518–26.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018 Glymphatic MRI in iNPH BRAIN 2017: 140; 2691–2705 | 2705

Reiche W, Komenda Y, Schick B, Grunwald I, Steudel WI, Reith W. membranes causes brain water accumulation. Mol Cell Neurosci MR cisternography after intrathecal Gd-DTPA application. Eur 2016; 77: 47–52. Radiol 2002; 12: 2943–9. Weller RO. Microscopic morphology and histology of the human Ringstad G, Emblem KE, Eide PK. Phase-contrast magnetic resonance meninges. Morphologie 2005; 89: 22–34. imaging reveals net retrograde aqueductal flow in idiopathic normal Weller RO, Boche D, Nicoll JA. Microvasculature changes and cere- pressure hydrocephalus. J Neurosurg 2016; 124: 1850–7. bral amyloid angiopathy in Alzheimer’s disease and their potential Sahar A, Hochwald GM, Ransohoff J. Alternate pathway for cerebro- impact on therapy. Acta Neuropathol 2009; 118: 87–102. spinal fluid absorption in animals with experimental obstructive Xie L, Kang H, Xu Q, Chen MJ, Liao Y, Thiyagarajan M, et al. Sleep hydrocephalus. Exp Neurol 1969; 25: 200–6. drives metabolite clearance from the adult brain. Science 2013; 342: Tarasoff-Conway JM, Carare RO, Osorio RS, Glodzik L, Butler T, 373–7. Fieremans E, et al. Clearance systems in the brain–implications for Yang L, Kress BT, Weber HJ, Thiyagarajan M, Wang B, Deane R, et al. Alzheimer diseaser. Nat Rev Neurol 2016; 12: 248. Evaluating glymphatic pathway function utilizing clinically relevant van Swieten JC, Hijdra A, Koudstaal PJ, van Gijn J. Grading white intrathecal infusion of CSF tracer. J Transl Med 2013; 11: 107. matter lesions on CT and MRI: a simple scale. J Neurol Neurosurg Yin LK, Zheng JJ, Zhao L, Hao XZ, Zhang XX, Tian JQ, et al. Psychiatry 1990; 53: 1080–3. Reversed aqueductal cerebrospinal fluid net flow in idiopathic Vindedal GF, Thoren AE, Jensen V, Klungland A, Zhang Y, Holtzman normal pressure hydrocephalus. Acta Neurol Scand 2017. [Epub MJ, et al. Removal of aquaporin-4 from glial and ependymal ahead of print]. doi: 10.1111/ane.12750.

Downloaded from https://academic.oup.com/brain/article-abstract/140/10/2691/4085293 by University of Oslo Library user on 14 February 2018