Autonomic Function inNeurodegenerative Diseases

PhD Thesis

by Gertrud Laura SørensenDecember 2013

Danish Center for Sleep MedicineDepartment of Clinical NeurophysiologyCenter for Healthy AgingFaculty of Health SciencesUniversity of CopenhagenGlostrup HospitalMain supervisor:Poul J. Jennum

Preface

This report is the final result of the dissertation prepared by Gertrud Laura Sørensen inorder to achieve the degree of Doctor of Philosophy. The research in this dissertationhas been carried out at the Danish Center for Sleep Medicine, Department of ClinicalNeurophysiology, Glostrup University Hospital. The dissertation consists of this sum-mary report and four journal papers (all published at the time of oral defence).

The dissertation was submitted in May 2013 and defended in December 2013. ThePhD supervisor was Professor Poul Jennum, Danish Center for Sleep Medicine, Depart-ment of Clinical Neurophysiology, Faculty of Health Sciences, University of Copenhagen,Glostrup Hospital.

The PhD assessment committee consisted of:

• Professor Finn Sellebjerg, Danish Multiple Sclerosis Center, Department of Neuro-logy, Copenhagen University Hospital Rigshospitalet, Denmark

• Professor Giuseppe Plazzi, Department of Neurological Sciences, University of Bo-logna, Italy

• Professor Thomas Penzel, Sleep Medicine Center, Charité University Hospital, Ber-lin, Germany

i

Acknowledgement

The studies in this PhD dissertation would not have been possible without the generoushelp and assistance from a substantial number of people to whom I wish to express mygratitude.First of all, I wish to thank my supervisor Poul Jennum for providing expert guidance andfeedback throughout the whole project. I appreciate the meetings we have had, where wehave spend time discussing various aspects of the project and for his countless inspiringideas and never failing support.Next I would like to thank all the co-authors of my publications for their contributions tothe different studies. A special thank goes to Jacob Kempfner for the help with extractionof data and especially for the always great co-operation. Furthermore a thank to theneurophysiologists at Danish Center for Sleep Medicine, especially Helle L. Leonthin,who has spend hours thoroughly scoring the data for the project. Also a thank forthe support from Helge B. D. Sørensen and the Department of Electrical Engineering,Technical University of Denmark. Finally thanks to my family for their support andunderstanding the last three years where the workload at times has been above normal.

iii

Abstract

Neurodegenerative diseases are highly debilitating and often lead to severe morbidityand even death. Parkinson’s disease (PD) is the second most common neurodegenerativedisease after Alzheimer’s disease. According to the Braak staging study, the progressionof PD starts in the medulla oblongata, which includes the cardiac centre and controlsautonomic functions, and therefore autonomic dysfunction may be experienced early inthe disease course.Sleep disturbances are also common non-motor complications of PD, and therefore PDpatients undergo polysomnography at the Danish Center for Sleep Medicine to assessthe sleep disturbances. The aim of this PhD dissertation was to: 1) Develop a methodto investigate autonomic changes during sleep in neurodegenerative diseases, and applythis method on PD, iRBD and narcolepsy patients to evaluate the autonomic functionin these diseases. 2) Validate the method by applying standardized methods to measurethe autonomic function based on heart rate variability (HRV) measures. 3) Based onthe results, assess the validity of autonomic dysfunction as an early marker of a neuro-degenerative disease. 4) Evaluate the influence of hypocretin loss in narcolepsy patientson autonomic function and on the sleep transition rate.The results showed an attenuated heart rate response (HRR) in PD patients comparedto controls and early PD (iRBD patients). Also iRBD patients had an attenuated HRRcompared to control subjects, and the method to measure the HRR may be a possiblescreening tool to identify a neurodegenerative disease early in the disease course.Based on the HRV measures, when validating the method, it was found that sympatheticactivity was attenuated in iRBD patients which was further pronounced in PD patients.Parasympathetic activity, emanating from the nucleus ambiguus, may be relatively pre-served in patients with PD. The progressive reduction of sympathetic nervous activityis in line with the postganglionic sympathetic nervous dysfunction seen in early PD andmay represent an early manifestation of a neurodegenerative process involving brain stemareas, which is consistent with the Braak hypothesis.In the narcolepsy patients, it was shown that a reduced HRR to arousals was primarily

v

predicted by hypocretin deficiency in both rapid-eye-movement (REM) and non-REMsleep, independent of cataplexy and other factors. The results confirm that hypocretindeficiency affects the autonomic nervous system of patients with narcolepsy and that thehypocretin system is important for proper heart rate modulation at rest.Furthermore, it was shown that hypocretin deficiency and cataplexy are associated withsigns of destabilized sleep-wake and REM sleep control, indicating that the disorder mayserve as a human model for the sleep-wake and REM sleep flip-flop switches. The in-creased frequency of transitions may cause increased sympathetic activity during sleepand thereby increased heart rate, or the increased heart rate could be caused by decreasedparasympathetic activity due to the lack of hypocretin signalling.

vi

Danish summary

Neurodegenerative sygdomme er yderst invaliderende og fører ofte til alvorlig morbidi-tet og død. Parkinson’s sygdom er den næsthyppigste neurodegenerative sygdom efterAlzheimer’s sygdom. Ifølge Braaks stadie inddeling af sygdommen, starter Parkinson’ssygdom i den nederste del af hjernestammen (medulla oblongata), som bl.a. omfatterkredsløbscenteret og kontrollerer autonome funktioner. Derfor kan autonom dysfunktionopstå tidligt i sygdomsforløbet.Søvnforstyrrelser er også hyppige ikke-motoriske komplikationer, og derfor gennemgårParkinson patienter polysomnografi undersøgelser på Dansk Center for Søvn Medicin forat vurdere graden af disse søvnforstyrrelser. Formålet med denne afhandling var at: 1)Udvikle en metode til at undersøge autonome forandringer under søvn i neurodegener-ative sygdomme, og anvende denne metode på Parkinson, iRBD og narkolepsi patienterfor at evaluere deres autonome funktion. 2) Validere metoden ved at anvende stand-ardiserede metoder til måling af autonom funktion baseret på hjertefrekvens variabilitet.3) Baseret på resultaterne, vurdere gyldigheden af autonom dysfunktion som en tidligmarkør for neurodegenerative sygdomme. 4) Vurdere betydningen af hypokretin mangelhos narkolepsi patienter på autonom funktion og på frekvensen af søvn stadie skift.Resultaterne viste et dæmpet heart rate response (HRR) i Parkinson patienter sammen-lignet med kontroller og tidlig Parkinson (iRBD). Patienter med iRBD havde også etdæmpet HRR sammenlignet med kontroller, og metoden til at måle HRR kan være etmuligt screenings værktøj til at identificere neurodegenerative sygdomme tidligt i syg-domsforløbet.Ved validering af metoden, viste hjertefrekvens variabilitet at sympatisk aktivitet vardæmpet hos iRBD patienterne, hvilket var endnu mere udtalt hos Parkinson patien-terne. Parasympatisk aktivitet, stammende fra nucleus ambiguus, er sandsynligvis rela-tivt bevaret i patienter med Parkinson. Den progressive reduktion af sympatisk aktiviteter i overensstemmelse med den postganglionære sympatiske dysfunktion set i tidlig Par-kinson og kan repræsentere en tidlig manifestation af en neurodegenerativ proces, derinvolverer hjernestamme områder, der er i overensstemmelse med Braaks hypotese.

vii

I narkolepsi patienterne blev det vist, at hypokretin mangel er den primære prædiktorfor et dæmpet autonomt respons både i rapid-eye-movement (REM) og ikke-REM søvn,uafhængigt af katapleksi eller andre faktorer. Disse resultater bekræfter, at hypokretinmangel påvirker det autonome nervesystem hos patienter med narkolepsy, og at hypo-kretin systemet er vigtigt for korrekt hjertefrekvens modulation i hvile.Yderligere blev det vist, at hypokretin-mangel og katapleksi er forbundet med tegn pådestabiliseret søvn-vågen og REM-søvn kontrol, hvilket indikerer, at sygdommen kanudgøre en menneskelig model for en søvn-vågen og REM-søvns flip-flop kontakt. Denøgede hyppighed af søvn stadie transitioner kan muligvis medføre øget sympatisk aktivi-tet under søvn og dermed øget hjertefrekvens. Den øgede hjertefrekvens kan også væreforårsaget af nedsat parasympatisk aktivitet på grund af manglende hypokretin signale-ring.

viii

Contents

Contents ix

Nomenclature xi

List of Figures xiii

List of Tables xiv

1 Introduction 1

2 Clinical studies 52.1 Study 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Study 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152.3 Study 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232.4 Study 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 Discussion 353.1 Autonomic function in Parkinson’s disease . . . . . . . . . . . . . . . . . 353.2 Autonomic function in Narcolepsy . . . . . . . . . . . . . . . . . . . . . . 403.3 Methods, data quality and validity . . . . . . . . . . . . . . . . . . . . . 44

4 Conclusion and Perspectives 47

Bibliography 49

Appendix 59

A Paper 1 59

B Paper 2 67

ix

C Paper 3 77

D Paper 4 83

x

Nomenclature

AI Arousal Index

ARAS Ascending Reticular Activation System

AUC Area Under the Curve

BMI Body Mass Index

CSF cerebrospinal fluid

DCSM Danish Center for Sleep Medicine

DFA Detrended Fluctuation Analysis

EEG ElectroEncephaloGraphy

ECG ElectroCardioGraphy

EMG ElectroMyoGraphy

EOG ElectroOculoGraphy

GABAergic gamma-aminobutyric acidergic

hcrt-1 Hypocretin-1

hcrt-2 Hypocretin-2

HF High Frequency

HR Heart Rate

HRR Heart Rate Response

HRV Heart Rate Variability

HY Hoehn and Yahr

ICSD International Classification of Sleep Disorders

iRBD idiopathic REM sleep Behavior Disorder

LED Levodopa Equivalent Dose

LM Leg Movement

xi

LF Low Frequency

MIBG Metaiodobenzylguanidine

NC Narcolepsy with Cataplexy

non-REM Non-Rapid Eye Movement

NwC Narcolepsy without Cataplexy

PD Parkinsons’s Disease

PLM Periodic Leg Movement

PNS Parasympathetic Nervous System

PSA Power Spectrum Analysis

PSG Polysomnography

RBD REM sleep Behavior Disorder

REM Rapid Eye Movement

sd Standard Deviation

SE Sleep Efficiency

SL Sleep Latency

SNPC Substantia Nigra Pars Compacta

SNS Sympathetic Nervous System

VLF Very Low Frequency

TST Total Sleep Time

UPDRS Unified Parkinson Disease Rating Scale

xii

List of Figures

2.1 HRR to arousals calculated from non-REM 2 sleep. . . . . . . . . . . . . . . 122.2 HRR to arousals calculated from REM sleep. . . . . . . . . . . . . . . . . . . 122.3 HRR to LM calculated from non-REM 2 sleep. . . . . . . . . . . . . . . . . 132.4 HRR to LM calculated from REM sleep. . . . . . . . . . . . . . . . . . . . . 132.5 Scatterplot of the HRR to arousal for the iRBD patients. . . . . . . . . . . . 142.6 Scatterplot of the HRR to LM for the iRBD patients. . . . . . . . . . . . . . 142.7 Heart rate response calculated from non-REM 2 sleep. . . . . . . . . . . . . 192.8 Heart rate response calculated from REM sleep. . . . . . . . . . . . . . . . . 192.9 Heart rate response calculated from non-REM 2 sleep. . . . . . . . . . . . . 202.10 Heart rate response calculated from REM sleep. . . . . . . . . . . . . . . . . 202.11 Frequency of sleep-wake transitions . . . . . . . . . . . . . . . . . . . . . . . 252.12 Frequency of REM-NREM transitions . . . . . . . . . . . . . . . . . . . . . 26

xiii

List of Tables

2.1 Demographic data of subjects in Study 1. . . . . . . . . . . . . . . . . . . . 72.2 Sleep parameters for PD and iRBD patients. . . . . . . . . . . . . . . . . . . 102.3 Demographic data of subjects included in Study 2. . . . . . . . . . . . . . . 162.4 Sleep parameters for narcolepsy patients. . . . . . . . . . . . . . . . . . . . . 172.5 Uni- and multivariate analysis of HRR, non-REM. . . . . . . . . . . . . . . . 212.6 Uni- and multivariate analysis of HRR, REM. . . . . . . . . . . . . . . . . . 222.7 Demographic data of subjects included in Study 3. . . . . . . . . . . . . . . 242.8 Demographic data of subjects in Study 4. . . . . . . . . . . . . . . . . . . . 272.9 Heart rate variability measures for control subjects. . . . . . . . . . . . . . . 302.10 Heart rate variability measures for iRBD patients. . . . . . . . . . . . . . . . 312.11 Heart rate variability measures for PD patients without RBD. . . . . . . . . 322.12 Heart rate variability measures for PD patients with RBD. . . . . . . . . . . 33

xiv

Chapter 1

Introduction

Neurodegenerative diseases are highly debilitating with serious detrimental impact onwork, social-, and family life [43]. The diseases often lead to severe morbidity and evendeath. Parkinson’s disease (PD) is the second most common neurodegenerative diseaseafter Alzheimer’s disease [60] and the prevalence in industrialised countries is thought tobe approximately 0.3 % [21]. This is increased to 1 % in people over the age of 60 and 3% in people over the age of 80 [21, 60], and an even larger number of people are affectedby the disease as relatives.The clinical diagnosis of PD rests on the identification of the characteristics related todopamine deficiency that is a consequence of degeneration of the substantia nigra parscompacta (SNPC) [17]. Main symptoms of PD are bradykinesia, rigidity, rest tremorand postural instability [53]. In addition to these motor symptoms, mental disorders,autonomic and gastrointestinal dysfunction often occur, which considerably impair thequality of life of PD patients [51]. Although the cerebral structures, undergoing neurode-generation in PD, are well characterized, the causing mechanisms of the disease are stillunknown [51], and no cure has been found to stop the progressive course of the disease.Non-dopaminergic and non-motor symptoms are sometimes present before diagnosis andalmost inevitably emerge with disease progression. Indeed, non-motor symptoms domi-nate the clinical picture of advanced PD and contribute to severe disability, impairedquality of life, and shortened life expectancy, but non-motor symptoms are often in-adequately treated [17]. The development of treatments that can slow or prevent theprogression of PD and its multicentric neurodegeneration provides the best hope of cur-ing non-motor symptoms.The most elusive goal in neurodegeneration is a neuroprotective agent; a therapy to slowthe underlying degenerative process. If neuroprotective treatment becomes available, it

1

will become essential to identify patients as early as possible. A partially effective agentwith minor utility late in disease could, if given before symptoms develop, slow or evenprevent the onset of the disease. This is the principal motivation for learning how topredict PD and other neurodegenerative diseases [70].In addition to the motor manifestations that define PD, patients have numerous abnor-malities of other systems, which may occur before motor signs. In the quest to detectneurodegeneration at its early stage, this has encouraged to investigate the non-motoraspects such as cognitive decline, behavioral changes and autonomic dysfunction. Al-though there is considerable promise for clinical markers of neurodegeneration, no singlemarker has been able to detect neither PD nor other neurodegenerative diseases at earlystages with good reliability and sensitivity [70].

In the Braak staging procedure for the pathological process of PD it is proposed that,the formation of intraneuronal Lewy bodies and Lewy neurites begins at two sites inthe brain and continues in a topographically predictable sequence in 6 stages, duringwhich components of the olfactory, autonomic, limbic, and somatomotor systems becomeprogressively involved [11]. In stages 1-2, the Lewy body pathology is confined to themedulla oblongata/pontine tegmentum and anterior olfactory structures. In stages 3-4,the substantia nigra, other nuclei of the basal mid- and forebrain, and the mesocortexbecome the focus of initially subtle and, then, severe changes. During this phase, theillness probably becomes clinically manifest. In the final stages 5-6, the lesions appear inthe neocortex.According to the Braak staging system, the progression of PD starts in the medulla ob-longata, which includes the cardiac centre and controls autonomic functions, and thereforeautonomic dysfunction may be experienced early in the disease course. Several studieshave already examined cardiovascular responses as a marker of autonomic dysfunctionin neurodegenerative diseased patients [4, 35, 45, 52, 55]. Many of these studies havefocused on orthostatic hypotension.Sleep disturbances are also common non-motor complications of PD and have been re-ported in up to 90 % of patients [98]. This may also be explained by the Braak stagingsystem, as the sleep and wakefulness cycle is regulated by brainstem structures like thecaudal raphe nuclei and ceruleus-subceruleus complex in the pons, where the accumula-tion of Lewy bodies starts in the early stages of PD [98].At the Danish Center for Sleep Medicine (DCSM), patients with PD undergo a polysom-nography (PSG) to assess the sleep disturbances. In this dissertation the aim is to develop

2

a simple method to measure the autonomic changes during sleep in PD patients from theinformation already available from the PSG. Arousal and leg movement (LM) will beused as stimuli to measure the heart rate response (HRR) during rapid-eye-movement(REM) and non-REM sleep. In order to evaluate autonomic function in early PD, pa-tients with idiopathic REM sleep behavior disorder (iRBD) will also be included, sincethis disease may precede PD by many years [41, 42, 70, 71, 81, 82, 83]. Based on theresults for autonomic function during sleep in PD and iRBD, it will be assessed whetherthe method has future potential for being part of the PSG assessment for clinical testingof patients suffering from neurodegenerative diseases and sleep disorders.Same method will further be applied on narcolepsy patients to evaluate the HRR inanother neurodegenerative disorder, were PSG data also are available at DCSM. Nar-colepsy is a complex disorder characterized by excessive daytime sleepiness, instabilityof sleep-wake regulation (short sleep latency, disrupted nocturnal sleep with awakenings)and instability of rapid eye movement (REM) sleep regulation and motor tonus regulation,causing hypnagogic hallucinations, sleep paralysis, cataplexy, and REM sleep behaviordisorder (RBD) [29, 47, 65]. The major pathophysiology is loss of the sleep-wake patternand motor tonus regulating hypocretinergic neurons in the hypothalamus [67, 91], res-ulting in low or undetectable levels of cerebrospinal fluid (CSF) hypocretin-1 (hcrt-1) inmost patients with cataplexy (muscle weakness triggered by emotions) [47]. The etiologyof human narcolepsy is still under debate. The disorder is closely associated with theleukocyte antigen subtype HLA-DQB1*0602, suggesting an autoimmune destruction ofthe hypocretin neurons [56].Overall hypocretin neurons are thought to sustain wakefulness and suppress REM sleep[78], which explains the severe sleepiness and frequent transitions into REM sleep experi-enced by patients with narcolepsy. It has been proposed that frequent transitions betweenstates, odd mixtures of states, and poor control of REM sleep are consistent with destabi-lization of the flip-flop switches that regulate REM-non-REM and sleep-wake transitionsbecause of the loss of hypocretin signalling [78]. Therefore it is also the aim to investigatethe transition rate in narcolepsy patients to reveal whether this could influence the HRR.Finally it is desired to validate the HRR method by measuring the heart rate variability(HRV) according to standardized methods.To summarize, the aim of this PhD thesis was to:

• Develop and implement a method to investigate autonomic changes during sleep inneurodegenerative diseases, and apply this method on PD, iRBD and Narcolepsypatients to evaluate the autonomic function in these diseases.

3

• Validate the method to investigate autonomic changes, by applying standardizedmethods to measure the autonomic function based on HRV measures.

• Based on above mentioned methods, assess the validity of autonomic dysfunctionas an early marker of a neurodegenerative disease.

• Evaluate the influence of hypocretin loss in narcolepsy patients on autonomic func-tion and on the sleep transition rate

4

Chapter 2

Clinical studies

This PhD thesis has included 4 studies that have resulted in 4 papers, included in Ap-pendix A to D. The following section will describe the methods and results of each of the4 studies. A joint discussion of the results of all studies can be found in Chapter 3.

2.1 Study 1

In the first description of PD in 1817, published by James Parkinson, the motor pro-blems of the patients were accurately described, but also the non-motor symptoms werenoticed [63], which dominate the clinical picture of advanced PD and contribute to severedisability, impaired quality of life, and shortened life expectancy. By contrast with thedopaminergic symptoms of the disease, for which treatment is available, non-motor symp-toms are often poorly recognized and inadequately treated [17]. Autonomic dysfunction isexperienced by 40-70 % of PD patients [17]. The degeneration and dysfunction include thenuclei mediating autonomic functions such as the dorsal vagal nucleus, nucleus ambiguus,and other medullary centres, which exert differential control on the sympathetic pregan-glionic neurons via descending pathways [17], but both parasympathetic and sympatheticfunctions may be impaired [1]. Several studies have examined cardiovascular responsesas markers of autonomic nervous system dysfunction in PD patients [4, 35, 45, 52, 55].This study focuses on the cardiac response to arousal and LM during sleep, which hasnot previously been assessed in PD patients.Wakefulness and arousals are regulated by two ascending neuron networks in the brain,referred to as the ascending reticular activation system (ARAS) [80]. The first activatingneuron network consists of the pedunculo-pontine and laterodorsal tegmental nuclei inthe brainstem, which are active during wakefulness and REM sleep [78]. The second

5

activating neuron network consists of the locus coeruleus, dorsal end median Raphe nuc-lei, ventral periaquaductal gray matter and tuberomammilary nucleus, which are activeduring wakefulness, slightly active during non-REM sleep and inactive during REM sleep[78]. Sleep is promoted by activation of the ventrolateral preoptic nucleus in the basalforebrain, which acts through inhibition of hypocretin neurons and the neurons in ARAS[78]. Transition from sleep to wakefulness and vice versa is regulated by the sleep-wakeflip-flop switch, which involves the mutual inhibition of the sleep promoting nuclei andthe wakefulness-promoting nuclei [79]. This flip-flop ensures rapid sleep-wake transitions.Arousal from sleep involves activation of the sympathetic nervous system (SNS), includ-ing spinal motor activations. Therefore arousal from sleep including motor activationsin sleep is associated with an increased heart rate (HR) and blood pressure in healthyadults [84], although the increase in HR gets lower with age [33]. An intact SNS is a keycomponent of the HRR associated with LM and arousal from sleep [30].PD is a progressive neurodegenerative disease believed to commence in the medulla ob-longata and progressing upwards [11]. iRBD is a strong predictor of the later developmentof PD, probably due to the involvement of the sublaterodorsal nucleus, which is involvedin atonia during REM sleep [41, 42, 70, 71, 81, 82, 83]. Due to the early involvement ofmedulla oblongata in iRBD and PD, it could be hypothesized that autonomic nervoussystem deficiency should be present early in the disease course and that, therefore, thesepatients will demonstrate an attenuated HRR associated with arousal and LM. SinceiRBD is reported to herald the onset of PD by years, attenuation of HRR in these pa-tients would support this hypothesis.

Study 1 - Methods

This study included 30 PD patients and 11 iRBD patients. All patients were identi-fied from the patient database at the DCSM, Department of Clinical Neurophysiology,Glostrup University Hospital, Denmark. All patients were evaluated by neurological ex-amination and PD patients were further examined by a DAT-scan to confirm the PDdiagnosis. Exclusion criteria were additional neurological, psychiatric or cardiovasculardisorders, alcohol abuse, head trauma, dementia and an apnea-hypopnea index >5. Nosubjects were taking any agents known to affect the autonomic system.It was not possible to withdraw patients from their Parkinson medication, but in orderto rule out any influence of these drugs, patients treated with intraspinal medication,duodopa or intercerebral stimulation were excluded. Total Levodopa equivalent dose(LED) was calculated according to [92] in order to check for any correlations between the

6

results and the effect of medication.RBD were diagnosed according to the criteria of the International Classification of SleepDisorders (ICSD) [2]. Of the 30 included PD patients, 14 patients had RBD whereasthe last 16 did not show symptoms of RBD. iRBD patients had no other disorders ex-plaining the RBD. Both iRBD and PD patients were assessed with the Hoehn and Yahr(HY) scale and the Unified Parkinson Disease Rating Scale (UPDRS). Seventeen healthycontrols, matched to the patient group by age, gender and body mass index (BMI), werealso recruited to the study. All included subjects gave their signed informed consent andthe study was ethical approved (HA20070120). Demographic data are given in Table 2.1.

Controls PD with RBD PD without RBD iRBD

Total number 17 14 16 11

Female/male 7/8 3/11 8/8 2/9

Age [years] 62.4 (9.7) 63.4 (6.7) 62.0 (7.3) 60.2 (7.8)

BMI [kg/m2] 23.4 (3.1) 26.3 (3.2) 25.7 (5.1) 24.9 (2.9)

UPDRS score - 23 (11) 23 (10) 1 (1)

HY score - 1.6 (0.8) 1.8 (0.8) 0.0 (0.0)

LED score - 623 (360) 680 (552) -

Table 2.1: Demographic data of subjects included in Study 1. Age, BMI, UPDRS andHY are given as mean (and sd).

All subjects underwent one night of PSG recorded either under ambulant conditions orat the DCSM. All recordings included a minimum of 6 electroencephalography (EEG)leads (F3-A2, F4-A1, C3-A2, C4-A1, O1-A2, O2-A1), and inpatient recordings includedEEG electrodes according to the 10-20 system all in accordance with the gold standard[40]. Furthermore, surface electromyography (EMG) of the left and right anterior tibialismuscle and the submental muscle, electrocardiography (ECG) and vertical and horizontalelectrooculography (EOG) were included in all recordings. Impedances were kept below10 kHz. The amplifier was different for the ambulant and inpatient recordings, althoughthis only influenced freqeuncies above 60 Hz, which is irrelevant in the context of thisstudy. Sleep stages and arousals from sleep were manually scored according to standardcriteria [7, 40]. LM were quantified from the surface EMG in accordance with standardcriteria [40].

7

The HRR to arousals or LM was estimated by calculating the change in RR intervalsin the ECG signal, and determining the area under the curve (AUC) for the HR changefrom 10 beats before to 15 beats after onset of the events. Data were analyzed by thefollowing steps:

I) Detection of QRS complexes and R waves:Several studies [31, 36, 48, 69] have evaluated and compared different QRS-detectionalgorithm. In this study it was chosen to implement the Pan-Tompkins QRS detector[62] since it had already been tested by several experts and had been used to detectQRS complexes during sleep [36, 39, 69]. The algorithm was implemented in Matlab R⃝

(R2009b, The MathWorks, Natick, MA, USA).For each sleep event (arousal or LM), QRS complexes were then automatically detectedin the ECG signal. In each detected QRS complex, the R wave was determined as thelargest peak.

II) Validation of QRS detector and calculation of RR intervals:Arousals and LMs associated with noise-contaminated ECG signals were excluded fromthe calculation of the HRR, since detection of QRS complexes in these ECG signal seg-ments was impossible to validate. RR intervals were calculated as the inter-beat intervalbetween successive R waves by subtracting the sample positions, Ri, of succeeding Rwaves and dividing by the sample frequency, fs,:

RRi = Ri+1 − Ri

fs

, i = {−10, −9, ..., 14, 15} \ {0}, (2.1)

where a negative i and a positive i denotes RR intervals before and after onset of theevent respectively.

III) Calculation of HRR curves:For every included sleep event, the HR was analyzed from 10 heartbeats before to 15heartbeats following the onset of the event. The identified RR intervals were convertedto HR in beats/min according to the following equation:

HRi = 1RRi

· 60, i = {−10, −9, ..., 14, 15} \ {0}. (2.2)

Previous studies of HRR to arousal or motor activity had shown that the HR beganto rise two heartbeats before the event [27, 84]. Therefore, a baseline value of HR wascalculated by averaging the first seven beats out of the 10 beats before the event. To

8

determine the HR change, HRC, this baseline value was subtracted from all 25 HR values(from 10 beats before to 15 beats after the event):

HRCi = HRi −−4∑

j=−10

HRj

7, i = {−10, −9, ..., 14, 15} \ {0}. (2.3)

For each subject, the values of (2.3) were calculated for all events, and the average HRchange was calculated as the mean of the HR change related to all events, calculatedseparately for arousal and LM. A curve was created for each patient, displaying the av-erage HR change for each of the 25 heartbeats. By averaging individual curves, the HRchange associated with arousals and LM was obtained for each group. Such curves werecalculated for two cases: events from non-REM 2 sleep and events from REM sleep.

VI) Calculation of HRR:The AUCs for the HR curves were calculated for each event type for each subject toobtain a measure of the HRR. Student’s t-tests were performed to compare HRR/AUCbetween groups and Pearson Product moment correlations were calculated to evaluatethe relationship of HRR with age, BMI, LED, HY and UPDRS score within groups.

The following PSG variables were also measured: sleep latency (SL), total sleep time(TST), sleep efficiency (SE), time from sleep onset to first epoch of REM sleep (REMSL), percentage of stages 1, 2, 3 non-REM and REM sleep and arousal index (AI) (arous-als per hour of sleep). To evaluate between group differences, a log transformation wasperformed for each variable and two-sided t-tests were calculated for each variable for eachof the three patient groups compared with the control group. A Bonferroni correctionwas performed to address the problem of multiple comparisons.

Study 1 - Results

Sleep parameters are presented in Table 2.2. The analysis of sleep parameters showedsignificant differences (p < 0.05) in TST and SL for the PD group without RBD comparedwith the control group. Other studies have shown lower TST and a lower percentageof stage 3 sleep in PD patients [2, 53, 64]. The percentage of stage 3 sleep was notsignificantly reduced in the PD patients in this study. The AI index was not increasedfor any of the patient groups compared with the control group, although PD and iRBDpatients are known to have disrupted sleep with frequent arousals. However, all groupshad a lower AI than had previously been noted in groups of similar age[9, 33].

9

Controls PD with RBD PD without RBD iRBD

TST [min] 418.0 (48.3) 372.9 (74.1) 348.5 (59.9)∗ 449 (79.9)

SL [min] 9.8 (8.2) 13.3 (22.9) 3.7(3.5)∗ 8.0 (7.1)

REM SL [min] 100.4 (50.6) 139.9 (87.4) 124.4 (85.8) 109.3 (53.4)

SE [%] 88.2 (7.6) 80.8 (14.1) 81.5 (6.5) 84.1 (14.3)

non-REM 1 [%] 10.5 (7.8) 10.4 (6.8) 9.6 (7.1) 9.3 (5.1)

non-REM 2 [%] 46.7 (11.1) 48.0 (16.5) 51.7 (17.4) 48.7 (7.4)

non-REM 3 [%] 20.9 (9.6) 23.9 (18.9) 23.5 (14.3) 21.1 (4.9)

REM sleep [%] 22.0 (4.0) 17.7 (10.5) 15.2 (9.4) 22.2 (6.2)

AI [/hour] 14.5 (6.0) 12.1 (5.4) 10.5 (4.1) 14.4 (6.4)

Table 2.2: Sleep parameters expressed as mean (and sd). ∗p-value < 0.05 for comparisonwith control group.

Figure 2.1 shows the HRR associated with arousals from non-REM 2 sleep. The curvesfor the PD, iRBD and control groups all show an acceleration of HR after the onset ofarousal. However, the HRRs for the PD groups were significantly lower than both thecontrol (p < 0.01) and iRBD (p < 0.02) groups. There was no significant difference inthe HRR of the iRBD group compared with the control group, nor between the two PDgroups.Figure 2.2 shows the HRR associated with arousals from REM sleep. Two of 17 controlsubjects did not have sufficient arousals during REM sleep to merit inclusion in this study.Again, all groups showed an acceleration of HR after the onset of arousal. The HRRs forthe PD groups were significantly lower than those of the control (p < 0.01) and iRBD(p < 0.05) groups. There was no significant difference in the HRR of the iRBD groupcompared with that of the control group, nor between the two PD groups. No significantcorrelations were observed between HRR to arousals and age, BMI, LED, HY or UPDRSwithin any of the groups. Baseline values were similar for all groups in Figure 2.1 and2.2.Figure 2.3 shows the HRR associated with LM during non-REM 2 sleep. As with arousalsthere was an increase in HR as a response to LM. The HRRs were significantly lower forboth the PD groups (p < 0.02) and the iRBD group (p = 0.03) relative to controls. Therewas no significant difference in the HRR between the iRBD group and the PD groups

10

nor between the two PD groups.Figure 2.4 shows the HRR associated with LM from REM sleep. Two of 17 control sub-jects and 2/16 PD patients without RBD did not have enough LM during REM sleepto warrant inclusion. The HRR was significantly lower in both PD groups (p < 0.05)and the iRBD group (p = 0.05) compared with the controls. There was no significantdifference in the HRR between the iRBD group and the PD groups nor between the twoPD groups. No significant correlations were observed between HRR to LM and age, BMI,LED, HY or UPDRS within any of the groups. Baseline values were similar for all groupsin Figure 2.3 and 2.4.

In Figure 2.5 and 2.6 the HRRs for the iRBD patients have been displayed in scat-terplots. These scatterplots shows that the iRBD group can be divided in two groups.One group with an attenuated HRR to arousals and LM both in REM and non-REM 2sleep and another group having a HRR to arousal and LM both in REM and non-REM2 sleep resembling the control subjects.

11

−10 −5 0 5 10 15−4

−2

0

2

4

6

8

10

12

14

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupPD with RBDPD without RBDiRBDArousal onset

Figure 2.1: HRR associated with arousals from non-REM 2 sleep stages. The verticallines indicate the standard error of the mean.

−10 −5 0 5 10 15−4

−2

0

2

4

6

8

10

12

14

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupPD with RBDPD without RBDiRBDArousal onset

Figure 2.2: HRR associated with arousals from REM sleep stages. The vertical linesindicate the standard error of the mean.

12

−10 −5 0 5 10 15−4

−2

0

2

4

6

8

10

12

14

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupPD with RBDPD without RBDiRBDLM onset

Figure 2.3: HRR associated with LM from non-REM 2 sleep stages. The vertical linesindicate the standard error of the mean.

−10 −5 0 5 10 15−4

−2

0

2

4

6

8

10

12

14

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupPD with RBDPD without RBDiRBDLM onset

Figure 2.4: HRR associated with LM from REM sleep stages. The vertical lines indicatethe standard error of the mean.

13

0 20 40 60 80 100 1200

50

100

150

HRR to arousal in non−REM 2

HR

R to

aro

usal

in R

EM

Figure 2.5: Scatterplot of the HRR to arousal for the iRBD patients. The scatterplotshows that the iRBD group can be divided in two groups. One group with an attenuatedHRR to arousals both in REM and non-REM 2 sleep and another group having a HRRto arousal both in REM and non-REM 2 sleep resembling the control subjects.

0 20 40 60 80 100

0

20

40

60

80

100

HRR to LM in non−REM 2

HR

R to

LM

in R

EM

Figure 2.6: Scatterplot of the HRR to LM for the iRBD patients. The scatterplot showsthat the iRBD group can be divided in two groups. One group with an attenuated HRRto LM both in REM and non-REM 2 sleep and another group having a HRR to LM bothin REM and non-REM 2 sleep resembling the control subjects.

14

2.2 Study 2

In this study, the same method as described in Study 1, was applied on PSG data fromnarcolepsy patients. Narcolepsy is a neurologic disease affecting one in 2,000 individu-als and is characterized by excessive daytime sleepiness and early onset of REM sleep.The disease is closely associated with low levels of hcrt-1 in the CSF, especially in thetwo-thirds of patients with cataplexy [47]. The neuropeptides hcrt-1 and hypocretin-2(hcrt-2) are produced and co-released by neurons in the lateral posterior hypothalamusof the brain, which project widely within the central nervous system [66].Animal studies have suggested that hcrt-1 may affect neuronal circuits that control thecardiovascular system [19, 22, 76, 86] as exemplified by intrathecal hypocretin adminis-tration resulting in increased blood pressure and HR mediated by enhancement of thesympathetic output [22, 86].Two smaller human studies of patients with hcrt-1-deficient narcolepsy with cataplexy(NC) found contradictory results of sympathetic tone during daytime wakefulness [32, 34].However, during sleep, reduced HR and sympathetic tone were found in NC patients ofunknown hcrt-1 status [20, 28], supporting the results of the animal studies.The aim of this study was to evaluate whether hcrt-1 deficiency is the main predictor/causeof autonomic dysfunction during sleep in patients with narcolepsy.

Study 2 - Methods

This study included 67 narcolepsy patients. PSG recordings had been obtained over a10-year period (2001-2011), at the DCSM. The patients fulfilled the ICSD criteria fornarcolepsy [2] and were evaluated by neurologic examination, determination of routineblood characteristics, PSG, the multiple sleep latency test, and determination of CSFhcrt-1. Thirty-nine of 67 patients were included in a previous published study [47], wherethe hypocretin measurement protocol and the applied definition of hcrt-1 deficiency aredescribed. Exclusion criteria were: additional neurologic, psychiatric, and cardiovasculardisorders, and an apnea index > 5/h. All patients (except three with severe cataplexy)were free of antidepressants and stimulants 7-14 days before inclusion; 44 of 67 werecompletely drug-naïve (i.e., they had never taken medication).During the 10-year period, a total of 81 narcolepsy patients, who fulfilled the inclusioncriteria, had undergone a PSG recording at DCSM, although because HRR is known tobe age-related [33, 40], patients were excluded to enable accurate age-matching of thenarcolepsy groups. Of the excluded group, 12 of 14 had cataplexy and 10 of 13 had

15

hcrt-1 deficiency.Of the 67 included narcolepsy patients, 46 of 67 had cataplexy while hcrt-1 deficiencywas present in 38 of 61 patients. Six patients (five of six with cataplexy) refused lumbarpuncture. Normal hcrt-1 levels were present in nine of 46 patients with cataplexy, andhcrt-1 deficiency was present in five of 21 patients without cataplexy.Control subjects were recruited by advertising for normal volunteers, and a total of 22healthy controls matched for age, sex, and BMI were included. None of the control sub-jects had any history of neurologic or sleep disorders, and their PSGs were evaluated bya neurologist as being normal.All subjects gave their signed informed consent and the study was ethical approved(KA03119). Demographic data of included subjects are given in Table 2.3.

Controls NC NwC +hcrt-1 -hcrt-1

Number 22 46 21 23 38

Female/Male 16/6 20/26 12/9 13/10 16/22

Age [year] 32.2 (8.4) 31.9 (7.9) 32.2 (8.9) 31.4 (9.3) 32.5 (7.9)

BMI [kg/m2] 24.2 (4.3) 26.3 (6.0) 25.4 (4.1) 25.4 (5.0) 24.9 (4.0)

Table 2.3: Demographic data of subjects included in Study 2. Values are expressed asmean (and sd). +hcrt-1 = narcolepsy patients with normal hcrt-1. -hcrt-1 = narcolepsypatients with hcrt-1 deficiency; NC = narcolepsy with cataplexy; NwC = narcolepsywithout cataplexy.

PSG recordings consisted of EEG (C3-A2, C4-A1), vertical and horizontal EOG, surfaceEMG of the submentalis and anterior muscles, ECG, nasal air flow, thoracic respiratoryeffort, and oxygen saturation. Subjects were instructed not to consume any caffeinatedor alcoholic drinks for at least 6 hours before the recordings were made. Sleep stages,LM, and arousals from sleep were manually scored according to standard criteria [7, 40].The raw sleep data, hypnograms, and sleep events were extracted from Nervus R⃝ (V5.5,Cephalon DK, Nørresundby, Denmark) or Somnologica Studio R⃝ (V5.1, Embla, Broom-field, CO USA), using the built-in export data tool, and imported to Matlab R⃝ for furtherprocessing.The HRR to arousal and LM was estimated by the same method as used in Study 1 (de-scribed in Chapter 2.1). Univariate and backward stepwise multivariate linear regressionwere performed to determine the relationship between HRR and age, sex, BMI, disease-

16

duration, disease-onset, +/- previous anticataplexy medication, +/- cataplexy, and +/-hcrt-1 status. The +/- anticataplexy medication indicates whether patients were pausingwith medication or had never been taking medication before inclusion (drug-naïve).Also the following PSG variables were measured: SL, TST, SE, REM SL, percentage ofstages 1, 2, 3 non-REM and REM sleep. To evaluate between group differences, a logtransformation was performed for each variable and two-sided t-tests were calculated foreach variable for each of the three patient groups compared with the control group. ABonferroni correction was performed to address the problem of multiple comparisons.

Study 2 - Results

Sleep parameters are presented in Table 2.4. All narcolepsy groups, except NwC, had alower percentage stage 2 non-REM sleep compared with the control group. Narcolepsywith hcrt-1 deficiency had a higher percentage stage 1 sleep compared with narcolepsywith normal hcrt-1. Furthermore, the REM SL was significantly lower for all patientswith narcolepsy than for the controls, except for narcolepsy with normal hcrt-1. SL wasonly significantly lower for the narcolepsy patients with hcrt-1 deficiency.

Controls NC NwC +hcrt-1 -hcrt-1 P value

TST [min] 430 (52) 397 (78) 427 (49) 415 (70) 399 (77) NS

SL [min] 26 (33) 10 (19) 7 (11) 11 (14) 8 (19) <0.05d

REM SL [min] 124 (98) 82 (153) 58 (34) 103 (204) 53 (54) <0.05a,b,d

SE [%] 87 (16) 87 (12) 93 (5) 87 (15) 89 (7) NS

non-REM 1 [%] 5 (3) 9 (6) 5 (4) 4 (3) 10 (7) <0.05f

non-REM 2 [%] 50 (7) 39 (13) 43 (11) 39 (13) 40 (12) <0.05a,c,d

non-REM 3 [%] 23 (8) 30 (14) 29 (9) 34 (13) 27 (12) NS

REM sleep [%] 22 (4) 23 (6) 23 (6) 22 (5) 24 (5) NS

Table 2.4: Sleep parameters expressed as mean (and sd). P-values are calculated fromtwo-sided t-tests after log-transformation. aNC vs. controls. bNwC vs. controls. c+hcrt-1 vs. controls. d-hcrt-1 vs. controls. eNC vs. NwC. f+hcrt-1 vs. -hcrt-1.+hcrt-1 = narcolepsy patients with normal hcrt-1. -hcrt-1 = narcolepsy patients withhcrt-1 deficiency; NC = narcolepsy with cataplexy; NwC = narcolepsy without cataplexy.

Figures 2.7 and 2.8 show the mean HRR to arousal from non-REM sleep stage 2 andREM sleep, respectively. The HR after arousal onset was accelerated in a similar way

17

for patients with narcolepsy and controls, although to a lesser degree for the narcolepsygroups. The HRR was significantly reduced for NC (P < 0.02) and for patients withnarcolepsy with hcrt-1 deficiency (P < 0.01) compared with the controls in both REMand non-REM 2 sleep. Furthermore, in REM and non-REM 2 sleep, patients with cata-plexy and patients with hcrt-1 deficiency had significantly lower HRR to arousal thandid those with NwC (P < 0.04) and narcolepsy with normal hcrt-1 (P < 0.04). Likewise,the HRR for the NwC patients and the patients with narcolepsy with normal hcrt-1 hada lower, although non-significant HRR compared with controls (P = 0.10 and P = 0.09,respectively), placing them between the control group and NC and narcolepsy with hcrt-1deficiency groups. The baseline HR was significantly (P < 0.05) elevated for all narco-lepsy groups except NwC patients compared with the controls. There was no differencein the baseline HR within the narcolepsy group.Table 2.5 and 2.6 show the results for the univariate and multivariate backward stepwiselinear regression analysis. In the univariate model, hcrt-1 deficiency was the only pre-dictor of HRR to arousal in non-REM 2 sleep stages (P = 0.01), whereas both cataplexy(P < 0.01) and hcrt-1 deficiency (P < 0.001) were predictors of the HRR to arousal inREM sleep stages. In the multivariate analysis, the HRR in REM sleep was predictedby disease duration (P = 0.02), hcrt-1 deficiency (P < 0.01), and cataplexy status (P =0.01), but only by hcrt-1 deficiency in non-REM 2 sleep (P = 0.02).

Figures 2.9 and 2.10 show the HRR to LM during non-REM 2 sleep and REM sleep,respectively. Four of 67 patients with narcolepsy (three of four with cataplexy and twoof three with hcrt-1 deficiency; CSF not available in one of four) were excluded fromfurther analysis due to noisy EMG signals. Furthermore, eight of 22 controls and 10 of63 patients with narcolepsy (seven with and three without cataplexy) were excluded fromthe REM analysis due to insufficient numbers of LM during REM sleep.Patients with narcolepsy and controls both had an increased HRR to LM, but to a sig-nificant lesser degree in the narcolepsy groups (P < 0.05). However, unlike the HRR toarousal, the HRR to LM did not differ within the narcolepsy groups, when the cataplexyand hcrt-1 groups were analyzed separately. This was confirmed by the univariate andmultivariate linear regression analysis, in which none of the parameters were significant(data not shown). The baseline HR was significantly (P < 0.05) elevated for all narcolepsygroups, except the NwC group, compared with the controls. There was no difference inbaseline HR within the narcolepsy groups.

18

−10 −5 0 5 10 15

0

5

10

15

20

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupNCNwC−hcrt−1+hcrt−1Arousal onset

Figure 2.7: HRR associated with arousals from non-REM 2 sleep stages. The verticallines indicate the standard error of the mean.

−10 −5 0 5 10 15

0

5

10

15

20

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupNCNwC−hcrt−1+hcrt−1Arousal onset

Figure 2.8: HRR associated with arousals from REM sleep stages. The vertical linesindicate the standard error of the mean.

19

−10 −5 0 5 10 15

0

5

10

15

20

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupNCNwC−hcrt−1+hcrt−1LM onset

Figure 2.9: HRR associated with LM from non-REM 2 sleep stages. The vertical linesindicate the standard error of the mean.

−10 −5 0 5 10 15

0

5

10

15

20

RR−intervals

Hea

rt R

ate

Cha

nges

[bea

ts/m

in]

Control GroupNCNwC−hcrt−1+hcrt−1LM onset

Figure 2.10: HRR associated with LM from REM sleep stages. The vertical lines indicatethe standard error of the mean.

20

Univariate analysis Multivariate analysis

HRR, non-REM 2 Coefficient P-value Coefficient P-value

(95 % CI) (95 % CI)

Age [year] -0.989 0.14

(-2.313, 0.335)

Sex, male -6.269 0.57

(-28.41, 15.87)

BMI [kg/m2] -0.325 0.74

(-2.297, 1.646)

Disease duration [year] -0.924 0.13

(-2.130, 0.281)

Disease onset [year] 0.205 0.75

(-1.115, 1.525)

Previous anaticataplexy -1.345 0.89

medication, yes (-21.81, 19.12)

Cataplexy, yes -13.846 0.28

(-37.58, 9.891)

Low hcrt-1, yes 0.071 0.01 0.072 0.02

(-0.013, 0.129) (0.011, 0.132)

Table 2.5: Univariate and multivariate linear regression analysis of HRR to arousal innon-REM 2 sleep. Bold P values indicate significance (P < 0.05).

21

Univariate analysis Multivariate analysis

HRR, REM Coefficient P-value Coefficient P-value

(95 % CI) (95 % CI)

Age [year] -0.351 0.65

(-1.937, 1.234)

Sex, male -2.425 0.85

(-28.54, 23.69)

BMI [kg/m2] -1.837 0.11

(-4.101, 0.427)

Disease duration [year] -1.109 0.12 -1.590 0.02

(-2.545, 0.328) (-2.928, -0.253)

Disease onset [year] 0.910 0.23

(0.599, 2.419)

Previous anaticataplexy -14.782 0.20

medication, yes (-37.78, 8.217)

Cataplexy, yes -45.543 <0.01 -35.65 0.01

(-71.23, -19.85) (-63.90, -7.405)

Low hcrt-1, yes 0.071 <0.001 0.091 <0.01

(0.0134, 0.130) (0.025, 0.157)

Table 2.6: Univariate and multivariate linear regression analysis of HRR to arousal inREM sleep. Bold P values indicate significance (P < 0.05).

22

2.3 Study 3

Wakefulness and arousals are believed to be regulated mainly by two ascending neuronnetworks in the ARAS [80]. The first activating neuron network consists of the ped-unculopontine and laterodorsal tegmental nuclei in the brainstem, which are active duringwakefulness and REM sleep [78]. The second activating neuron network consists of thelocus coeruleus, dorsal end median raphe nuclei, ventral periaqueductal gray matter, andtuberomammillary nucleus, which are active during wakefulness, slightly active duringnon-REM sleep, and inactive during REM sleep [78]. Sleep is promoted by activation ofthe ventrolateral preoptic nucleus in the basal forebrain, which acts by inhibiting hypo-cretin neurons and the neurons in ARAS [78]. The currently prevailing theory is thattransition from sleep to wakefulness and vice versa is regulated by a sleep-wake flip-flopswitch, involving the mutual inhibition of the sleep promoting nuclei and the wakefulnesspromoting nuclei [79].The REM-non-REM flip-flop switch works through another mutually inhibitory system,in which gamma-aminobutyric acidergic (GABAergic) neurons in the sublaterodorsal re-gion fire during REM sleep and inhibit GABAergic neurons in the ventrolateral periaque-ductal gray matter and the adjacent lateral pontine tegmentum that in turn fire duringnon-REM states to inhibit REM sleep [78]. The hypocretin neurons play a central role inthe coordination and stabilization of the sleep-wake and REM-non-REM flip-flop switches[79]. Consequently, hypocretin deficiency may be the cause of the observed instability ofthe sleep-wake pattern of the human narcoleptic phenotype. Instability in these systemsdue to hypocretin loss may lead to sleep-wake and REM-non-REM transition instabil-ities, which is investigated in this study. If this is the case, narcolepsy may serve as ahuman model for the flip-flop switches.

Study 3 - Methods

This study included 63 of the 81 narcolepsy patients who met the inclusion criteria inStudy 2. Same inclusion criteria were used in this study, except that an additional fivepatients with a periodic leg movements (PLM) during sleep index > 4/h were excluded inthis study and six patients were excluded due to poor PSG data quality. For the last 70narcolepsy patients, seven patients were excluded in order to age match the narcolepsygroup in the same way as was done in Study 2.All of the 63 included patients (except three with severe cataplexy) were free of an-tidepressants and stimulants 7 to 14 days before inclusion; 44 of 63 were completely

23

drug-naïve (i.e., they had never taken medication). Cataplexy was present in 43 of 63patients, and hcrt-1 deficiency was present in 37 patients. Six patients (five of six withcataplexy) refused lumbar puncture. Normal hcrt-1 levels were present in eight of 43patients with cataplexy, and hcrt-1 deficiency was present in seven of 20 patients withoutcataplexy. Demographic data of included subjects are given in Table 2.7.

NC NwC +hcrt-1 -hcrt-1

Number 46 20 20 37

Female/Male 20/23 12/8 14/6 16/21

Age [year] 31.9 (7.9) 32.2 (8.9) 31.4 (9.3) 32.5 (7.9)

BMI [kg/m2] 26.3 (6.0) 25.4 (4.1) 25.4 (5.0) 24.9 (4.0)

Table 2.7: Demographic data of subjects included in Study 3. Values are expressed asmean (and sd). +hcrt-1 = narcolepsy patients with normal hcrt-1. -hcrt-1 = narcolepsypatients with hcrt-1 deficiency; NC = narcolepsy with cataplexy; NwC = narcolepsywithout cataplexy.

All patients had underwent 1 night of PSG recording with the setup as described inStudy 2. Sleep stages and events were manually scored for epochs of 30 sec accordingto standard criteria [7, 40] by experienced PSG technicians, supervised by Poul Jennumand without knowledge of the final diagnosis. The hypnograms and sleep events wereextracted from Nervus R⃝ or Somnologica Studio R⃝, using the built-in export data tool,and imported to Matlab R⃝ for further processing.The frequency of transitions in the hypnograms for each patient was measured as thenumber of transitions per hour of sleep of the investigated sleep stage. Analyzed trans-itions included were transitions between wake and sleep, transitions to/from REM sleep(hereafter noted as REM-NREM transitions), transitions to/from non-REM stage 1, 2,and 3, and transitions between all sleep stages. Wilcoxon rank-sum tests were performedto compare the frequencies of transitions between different narcolepsy groups.A backward-stepwise multivariate linear regression was performed to determine the re-lationship between transitions and age, sex, body mass index, disease duration, diseaseonset, +/- previous anticataplexy medication, +/- cataplexy, and +/- hcrt-1 status.

Study 3 - Results

The frequencies of sleep-wake transitions are illustrated in the box plots in Figure 2.11.Patients with narcolepsy and low hcrt-1 level had a significantly higher frequency of

24

sleep-wake transitions than patients with narcolepsy and normal hcrt-1 level (P = 0.014).Patients with NC also had a significantly higher frequency of sleep-wake transitions thanthose with NwC (P = 0.002).The frequencies of REM-NREM transitions are illustrated in the box plots in Figure2.12. Patients with narcolepsy and low hcrt-1 level had a significantly higher frequencyof REM-NREM transitions than patients with narcolepsy and normal hcrt-1 level (P =0.044). Patients with NC also had a trend for more REM-NREM transitions than thosewith NwC, although this difference was not statistically significant. For the other ana-lyzed transitions (to/from non-REM stage 1, 2, and 3, and transitions between all sleepstages), there was no significance between any of the groups (results not shown).In the multivariate analysis, the number of REM-NREM transitions were predicted onlyby hcrt-1 deficiency (P = 0.011), whereas the number of sleep-wake transitions were pre-dicted only by cataplexy (P = 0.001).

0

5

10

15

NC NwC Low HCRT−1 Normal HCRT−1

Tra

nsiti

ons/

hour

p = 0.002 p = 0.014

Figure 2.11: Frequency of sleep-wake transitions illustrated in box plots for each patientgroup. Each box plot shows the fifth, 25th, 50th, 75th, and 95th percentile. Signific-ant P values are shown. HCRT-1, hypocretin-1; NC, narcolepsy with cataplexy; NwC,narcolepsy without cataplexy

25

0

2

4

6

8

10

12

14

16

18

20

NC NwC Low HCRT−1 Normal HCRT−1

Tra

nsiti

ons/

hour

p = 0.044

Figure 2.12: Frequency of REM-NREM transitions illustrated in box plots for each pa-tient group. Each box plot shows the fifth, 25th, 50th, 75th, and 95th percentile. Signi-ficant P values are shown. HCRT-1, hypocretin-1; NC, narcolepsy with cataplexy; NwC,narcolepsy without cataplexy.

26

2.4 Study 4

In 1996, a task force composed of members of the European Society of Cardiology and theNorth American Society of Pacing and Electrophysiology defined the guidelines necessaryfor comparing different assessment models of HRV [90]. These guidelines have proved tobe the most valuable non-invasive test for assessing autonomic nervous system function[96]. However, due to great intra-individual variance the tests do not work individuallyand are recommended only for describing groups of patients.In Study 1, a method was described to measure the autonomic activity based on changesin HR during sleep. It was shown that iRBD and PD patients had an attenuated HRRduring sleep. However, it was not clear whether this expresses suppression of sympatheticor parasympathetic activities or combinations of the two. The aim of this study was tovalidate the HRR method described in Study 1 by applying the methods described inthe Task Force of 1996 to measure HRV and use these HRV measures to identify possiblechanges in autonomic function during both wakefulness and sleep in patients with iRBDand PD.

Study 4 - Methods

This study included 10 PD patients with RBD, 13 PD patients without RBD, 11 iRBDpatients and 10 control subjects. All subjects were included in Study 1, and same inclusioncriteria as used in Study 1 were used in this study. However, of the patients includedin Study 1, 4 PD patients with RBD and 3 PD patients without RBD had too manynocturnal events to permit the identification of stationary signals, as described below,and therefore these patients were excluded from this study. Furthermore, 7 of the controlsubjects included in Study 1 were excluded in this study due to insufficient samplingfrequency. Demographic data of the included subjects are given in Table 2.8.

Controls PD with RBD PD without RBD iRBD

Total number 10 10 13 11

Female/male 5/5 3/7 6/7 2/9

Age [years] 59.0 (9.7) 62.5 (6.9) 60.8 (6.7) 60.2 (7.7)

Table 2.8: Demographic data of subjects included in Study 4. Values are given as mean(and sd).

27

All subjects underwent one night of PSG recording as described in Study 1. Sleep stagesand events were manually scored for 30-second epochs according to standard criteria[7, 40] by experienced polysomnographic technicians supervised by Poul Jennum, withoutknowledge of the final diagnosis. The hypnograms and sleep events were extracted fromNervus R⃝, using the built-in export data tool, and imported to Matlab R⃝ for further pro-cessing.Five-minute ECG segments were selected from wakefulness, stage 2 non-REM and REMsleep during the night. The five-minute wake was selected from the pre-sleep period andthe five-minute NREM 2 was selected from the beginning of the night. Where possible,the five-minute REM sleep was selected from the last REM period, as this tended to bethe longest period of REM in most subjects.Segments included stationary signals (without arousals, LM, complex movements duringREM sleep, or apnoeas) recorded with a lead II derivation using torso electrode place-ment [40] with a sampling frequency of 512 Hz. HRV was analyzed with respect to timedomain, frequency domain, and nonlinear methods using Kubios HRV version 2.1 soft-ware.Time-domain variables included the mean value of normal RR intervals (NN), the stand-ard deviation of normal RR intervals (SDNN), the root mean square of successive RRinterval differences (RMSSDs), the number of successive intervals differing by more than50 ms (NN50) and the corresponding relative amount (pNN50).Frequency-domain variables were calculated by autoregressive modeling [54] and includedabsolute powers of very low frequency (VLF, 0 - 0.04 Hz), low frequency (LF, 0.04 - 0.15Hz) and high frequency (HF, 0.15 - 0.4 Hz) bands, LF and HF band powers in normalizedunits (LFnu and HFnu) and the LF/HF power ratio.The nonlinear properties of HRV were analyzed by Poincaré plots, the Shannon entropyindex and detrended fluctuation analysis (DFA). An ellipse was fitted to the Poincaréplots, and the standard deviation of the points perpendicular to the line-of-identity (SD1)and the standard deviation along the line-of-identity (SD2) were calculated. In the DFA,two correlation measures (α1 and α2) were calculated as described in the Kubios HRVUser’s guide [89].All HRV variables were calculated individually and, subsequently, between-group diffe-rences were evaluated with Wilcoxon rank-sum tests. A Bonferroni correction was appliedto take into account the problem of multiple comparisons, resulting in a significance levelp < 0.00625 for all statistical tests.

28

Study 4 - Results

All HRV measures for the control group are shown in Table 2.9 for the three stages(wakefulness, non-REM 2, REM). In the time-domain analysis, the values were compar-able during wakefulness and REM, while they tended to be larger during non-REM. Inthe frequency-domain analysis using normalized units, LF and HF were similar duringwakefulness and REM, whereas LF tended to be lower and HF higher during non-REM.Short-term variation reflecting parasympathetic activity tended to be more pronouncedduring non-REM than in wakefulness and REM, with higher levels of NN, RMSSD, NN50,HFnu, and SD1. The pattern for long-term variation was less clear in comparisons of thewakefulness, non-REM and REM stages.Tables 2.10, 2.11 and 2.12 show the HRV measures for the iRBD, PD without RBD, andPD with RBD patients, respectively. P-values for comparisons with the control group,that remained significant after the Bonferroni correction, are indicated. For the iRBDpatients, only the VLF component in the wakefulness stage was significantly differentfrom the control group. None of the other variables differed significantly between thesegroups during REM and non-REM sleep.For the PD patients, SDNN, VLF, LF and SD2 were significantly different from the con-trol group in the wakefulness stage. None of the variables differed significantly in theREM and non-REM stages for the PD patients without RBD compared with the controlgroup. For the PD patients with RBD, only SD2 differed significantly from the controlgroup in the REM stage, while none of the HRV variables were significantly different inthe non-REM stage.Across the spectrum of clinical presentation there was a pattern of progressive reductionin VLF during wakefulness from controls (158 ms2), through iRBD (47 ms2) and PDwithout RBD (30 ms2), to PD with RBD (27 ms2). A similar pattern was noted for thelong-term variation expressed by LF and SD2.

29

Wake non-REM 2 REM

NN [ms] 872 (126) 979 (160) 887 (132)

SDNN [ms] 28 (6) 29.3 (8.5) 27.5 (7.2)

RMSSD [ms] 19.5 (6.9) 23.5 (8.4) 20.8 (4.8)

NN50 [count] 8.7 (8.2) 18.4 (17.2) 10.1 (8.2)

pNN50 2.9 (3.3) 6.6 (6.6) 3.0 (2.4)

VLF [ms2] 158 (90) 148 (144) 131 (83)

LF [ms2] 492 (235) 550 (432) 527 (334)

HF [ms2] 150 (105) 240 (163) 126 (51)

LFnu 76 (14) 67 (18) 76 (15)

HFnu 24 (14) 33 (18) 24 (15)

LF/HF 5.3 (4.9) 3.9 (4.8) 4.7 (3.3)

SD1 [ms] 13.8 (4.9) 16.6 (5.9) 14.7 (3.4)

SD2 [ms2] 37.3 (8.7) 37.5 (11.6) 35.8 (10.0)

Shannon Entropy 3.15 (0.44) 2.94 (0.27) 3.2 (0.3)

DFA, α1 1.34 (0.29) 1.19 (0.23) 1.30 (0.21)

DFA, α2 0.57 (0.17) 0.43 (0.10) 0.48 (0.12)

Table 2.9: HRV measures for control subjects given as mean (and sd).

30

Wake non-REM 2 REM

NN [ms] 936 (161) 1066 (164) 954 (155)

SDNN [ms] 20 (8) 28 (8) 21 (8)

RMSSD [ms] 15.7 (6.9) 25.1 (5.9) 15.5 (4.5)

NN50 [count] 2.36 (3.32) 15.6 (15.3) 2.5 (5.3)

pNN50 0.79 (1.06) 5.5 (5.1) 0.9 (1.8)

VLF [ms2] 47 (34)∗ 122 (193) 131 (123)

LF [ms2] 293 (251) 454 (420) 250 (190)

HF [ms2] 88 (63) 266 (179) 79 (38)

LFnu 76 (15) 54 (28) 70 (16)

HFnu 24 (15) 46 (28) 30 (16)

LF/HF 5.4 (5.0) 2.7 (3.3) 3.4 (2.6)

SD1 [ms] 11.1 (4.9) 17.8 (4.2) 11.0 (3.2)

SD2 [ms2] 26.4 (10.2) 34.7 (12.6) 27.2 (11.5)

Shannon Entropy 3.02 (0.21) 2.96 (0.23) 2.91 (0.16)

DFA, α1 1.36 (0.27) 1.03 (0.34) 1.27 (0.27)

DFA, α2 0.47 (0.26) 0.33 (0.15) 0.51 (0.14)

Table 2.10: HRV measures for iRBD patients given as mean (and sd). ∗p-value < 0.00625for comparison with control group.

31

Wake non-REM 2 REM

NN [ms] 881 (121) 981 (135) 934 (112)

SDNN [ms] 14.9 (4.5)∗ 22.0 (6.0) 21.7 (4.2)

RMSSD [ms] 15.7 (7.3) 21.5 (8.2) 20.5 (6.6)

NN50 [count] 2.8 (4.6) 10.9 (13.5) 8.0 (11.3)

pNN50 0.9 (1.6) 4.0 (5.1) 2.7 (3.7)

VLF [ms2] 30 (28)∗ 96 (92) 76 (52)

LF [ms2] 101 (86)∗ 259 (213) 217 (141)

HF [ms2] 88 (66) 174 (89) 158 (106)

LFnu 55 (23) 56 (21) 57 (17)

HFnu 45 (23) 44 (21) 43 (17)

LF/HF 2.0 (1.9) 1.96 (1.67) 1.76 (1.43)

SD1 [ms] 11.1 (5.1) 15.2 (5.8) 14.5 (4.6)

SD2 [ms2] 17.7 (4.8)∗ 26.7 (8.0) 26.7 (5.5)

Shannon Entropy 2.87 (0.19) 2.87 (0.27) 2.84 (0.27)

DFA, α1 0.98 (0.25) 1.01 (0.28) 1.09 (0.25)

DFA, α2 0.49 (0.11) 0.43 (0.16) 0.44 (0.13)

Table 2.11: HRV measures for PD patients without RBD given as mean (and sd). ∗p-value < 0.00625 for comparison with control group.

32

Wake non-REM 2 REM

NN [ms] 1005 (150) 1114 (199) 1006 (128)

SDNN [ms] 17.8 (7.4)∗ 23.4 (9.3) 17.5 (5.6)

RMSSD [ms] 18.8 (8.9) 21.4 (5.9) 16.1 (6.9)

NN50 [count] 7.2 (9.5) 8.4 (6.4) 3.3 (8.7)

pNN50 3.1 (4.1) 3.4 (2.6) 1.4 (3.7)

VLF [ms2] 27 (26)∗ 85 (66) 67 (72)

LF [ms2] 146 (130)∗ 409 (416) 191 (144)

HF [ms2] 117 (98) 114 (55) 78 (74)

LFnu 56 (14) 68 (21) 71 (12)

HFnu 43 (14) 32 (21) 29 (11)

LF/HF 1.5 (0.7) 4.2 (4.3) 3.2 (2.2)

SD1 [ms] 13.4 (6.3) 15.1 (4.2) 11.4 (4.9)

SD2 [ms2] 21.0 (8.7)∗ 29.2 (13.2) 21.7 (7.0)∗

Shannon Entropy 2.7 (0.3) 2.80 (0.08) 2.87 (0.28)

DFA, α1 1.00 (0.21) 1.08 (0.24) 1.15 (0.20)

DFA, α2 0.39 (0.15) 0.38 (0.16) 0.44 (0.17)

Table 2.12: HRV measures for PD patients with RBD given as mean (and sd). ∗p-value< 0.00625 for comparison with control group.

33

Chapter 3

Discussion

The 4 studies in this PhD dissertation describe methods for assessment of autonomicfunction in neurodegenerative diseases. The analyzed patients are PD patients, iRBDpatients and narcolepsy patients. This chapter is divided in three parts, first discussingthe findings in PD and iRBD patients, then discussing the findings in the narcolepsypatients, and finally a section discussing the methods, data quality and validity.

3.1 Autonomic function in Parkinson’s disease

In Study 1 it was shown that PD patients have an attenuated HRR to arousals and LMcompared with healthy controls. This result was expected as PD patients are knownto have autonomic dysfunction. Also the HRR to LM for the iRBD patients was signi-ficantly lower than in the control group and the HRR curves for the iRBD group wereintermediate between those of the control and PD groups.For the baseline HR, calculated from the prearousal period, there was no significant differ-ence between any of the groups. However, some of the graphs indicated that the increasein HR started earlier for the control group, indicating that the cardiac response to arousalor LM is not only attenuated but also delayed in PD patients.Studies of the HRR to arousal in healthy subjects have found that tachycardia startsapproximately 2 beats before arousal onset and is followed by bradycardia before the HRreturns to baseline [84]. The bradycardia was not observed in Study 1. In the study bySforza et al. [84], the bradycardia occurred approximately 10 beats after arousal onsetand reached baseline at 15 beats after arousal onset. However, the mean age of the sub-jects in the study by Sforza et al. was 28.8 years, and compared with the mean age of 62.4years of the control group in Study 1, tachycardia and bradycardia would be expected

35

to be of significantly lower amplitude, as Gosselin et al. demonstrated in middle-agedsubjects [33].The attenuated HRR is consistent with the reduced HRR associated with PLM experi-enced by RBD patients [27]. Because RBD often precedes the clinical manifestation ofPD [71], the change in autonomic function may represent an early manifestation of aneurodegenerative process involving brain stem areas, which is consistent with the Braakhypothesis [11].According to the Braak staging study the progression of PD starts in the medulla ob-longata, which includes the cardiac centre and controls autonomic functions. Braak etal. found that both parasympathetic structures (dorsal motor nucleus of vagus) andsympathetic structures (postganglionic neurons) demonstrate deposition of synuclein atearly stages of PD pathology [13]. As arousal from sleep involves activation of the SNS,including spinal motor activations, it is likely that the attenuated HRR reflect an atten-uated SNS activity. However, the attenuated HRR may be due to either suppression ofsympathetic or parasympathetic activities or a combination of both.

In Study 4 the autonomic function in PD and iRBD patients was further evaluatedby measuring HRV in both wakefullness, non-REM and REM sleep. The HRV valuesfound in the control group were similar to the values reported by others for similar agegroups [95]. Looking at the various sleep stages in the control subjects, it was foundthat the HRV patterns were comparable during wakefulness and REM sleep, whereasthere was a pattern of increased parasympathetic relative to sympathetic activity duringnon-REM sleep. These findings are in accordance with those of other studies analysingsleep stage differences with respect to spectral and time domain variables describing theHRV [3, 8, 94].Vaughn et al. [94] found clear sleep stage differences in the HRV from spectral analysisthat were consistent with increased parasympathetic nervous system (PNS) activity dur-ing non-REM stage 2, decreased PNS and SNS activity during the non-REM stage 3 andincreased SNS during the REM sleep stage [94].In the limited group of iRBD patients included in Study 4, significant differences withrespect to the control subjects were seen only during wakefulness, while no significantdifferences were seen in non-REM sleep, and only the PD patients with RBD had a sig-nificant difference for a single variable in REM. However, the SNS is more active duringwakefulness than during the non-REM and REM sleep stages, making it more likely todemonstrate a reduction in sympathetic activity during the wakefulness stage.

36

The iRBD patients only differed significantly from the control group with respect to theVLF component. As this component reflects slow regulatory mechanisms [15] and is re-lated to sympathetic activity [26], this indicates that the SNS activity of iRBD patientswas attenuated. The finding of reduced activity in the slow components of HRV is in linewith the fact that in diseases associated with Lewy bodies, the vulnerable neurons arethose that are long, thin and unmyelinated and therefore slow-conducting [12]. Lanfran-chi et al. investigated the HRV in non-REM and REM sleep and found that cardiac andrespiratory responses were absent from subjects with iRBD [50]. This implies that iRBDpatients lack the possibility of activating the SNS, which may explain why a significantdifference was seen for the VLF component in the wakefulness stage in Study 4. Further-more this may explain why in Study 1, no significant differences were seen in the HRRto arousal for the iRBD patients when compared to the control group. As sympatheticactivity is less active during sleep compared to wake, an attenuated sympathetic activitymay not be as easy to show during sleep than during wake.A similar study by Valappil et al. [93] found differences in both the LF and HF compon-ent but not the VLF component in the wake stage between iRBD and healthy controls.The study only included 11 iRBD patients, as is also the case in Study 4. Studies withlarger patient groups is recommended to clarify the exact autonomic findings in the HRVmeasures.For the PD patients, the LF component was found to be significantly lower than for thecontrol subjects. The LF component is generally thought to reflect both sympathetic andparasympathetic activity [26] and the difference between the PD and control subjects isprobably due to a more pronounced dysfunction of the SNS than in the iRBD patients.The SDNN was significantly lower in the PD groups than in the control group. SDNN isa marker of the total variance of HRV and reflects all long-term components responsiblefor variability and correlates with VLF [96], and therefore this result corresponds withthe results for LF and VLF and is indicative of a dysfunction of the SNS.No significant differences were seen for the HF component neither for the PD patients norfor the iRBD patients. The HF component is known to reflect parasympathetic activity[96], and therefore the attenuated HRR identified in Study 1 is probably mainly due toinvolvement of the SNS, but does not exclude the possibility that the PNS is involved.In this context, it is important to differentiate between "slow" parasympathetic activitymediated through thin unmyelinated fibres originating from neurons in the dorsal vagalmotor nucleus and terminating primarily in organs below the diaphragm, and "fast" para-sympathetic activity mediated through large myelinated nerve fibres originating in the

37

nucleus ambiguus and terminating primarily in the heart. In the post-mortem PD casesstudied to date, it is the unmyelinated vagal preganglionic projecting neurons connect-ing the central nervous system to the enteric nervous system that contain α-synucleinimmunoreactive inclusions [13]. In terms of HRV, the fast components (RMSSD, HF,SD1) are commonly ascribed to parasympathetic activity without stressing the fact thatthis only holds true for the myelinated fibres terminating in the heart. The importanceof distinguishing between the parasympathetic activity mediated through thin unmyelin-ated fibres and thick myelinated fibres is reflected in the study by Oka et al.[61], whoshowed that early-stage PD patients feature changes in HRV that are compatible withsympathetic but not parasympathetic dysfunction in spite of the well known disturbancesin intestinal motility and bladder function associated with early PD.Of the nonlinear variables, only SD2 was significantly lower for the PD patients than thecontrol subjects, which is also consistent with the other results, since SD2 is believed todescribe long-term variability and is related to the time-domain measure SDNN [14].Sleep in the non-REM stage is characterized by relative autonomic stability, a domin-ance of parasympathetic influences to the heart, and a reduction in sympathetic efferentvasomotor tone [50], while REM sleep is a state of autonomic instability, dominated byremarkable fluctuations between parasympathetic and sympathetic influences [50] and,probably, activation of both nervous systems [15]. Therefore, if mainly the SNS is affectedin iRBD and PD and unmyelinated parasympathetic nerve fibres in PD, this would ex-plain why the HRV variables reflecting the sympathetic nervous system differ in PD andiRBD compared with the control subjects during wakefulness, rather than in non-REMand REM sleep.

According to the Braak staging study [11] the progression of PD starts in the medullaoblongata, which includes the cardiac centre and controls autonomic functions. Braaket al. found that both parasympathetic structures (dorsal motor nucleus of vagus) andsympathetic structures (postganglionic neurons) demonstrate deposition of synuclein atearly stages of PD pathology [13]. This is supported by the progressive reduction in VLFand the long-term variations expressed by LF and SD2 from controls to iRBD and to PDfound in Study 4. Furthermore, other studies have shown that the baroreceptor reflexin early-stage PD without orthostatic hypotension is not severely impaired [16, 38, 52],implying that the cardiac PNS emanating from the nucleus ambiguus may be relativelypreserved in patients with PD.From the medulla oblongata, the Braak staging system gradually ascends to the more

38

rostral structures implicating dysfunction in the sublaterodorsal nucleus and pre-locuscoeruleus structures, which could lead to RBD [6]. This temporal sequence could in-dicate that SNS dysfunction precedes iRBD, which is followed by PD or other diseasesassociated with Lewy body pathologies. In Study 1 it was also seen how the attenuatedHRR worsens from the onset of iRBD to the onset of PD. If the attenuated HRR is anearly manifestation of a neurodegenerative process, this could be an early marker of thedisease.Reviews on prediction of PD have highlighted that premotor symptoms like iRBD andanosmia may predict PD [70, 73], but it cannot be concluded that iRBD with dysauto-nomia presents an increased risk of further neurodegenerative development and it remainsunknown, which patients with iRBD will develop PD or another synucleinopathy. Thiswill have to be addressed in future studies.Other studies have also established the involvement of the autonomic system in iRBD andthe link to PD. In a study using 123I-MIBG cardiac scintigraphy, a reduced heart/mediasti-num ratio was found for iRBD, PD, and dementia of Lewy bodies patients relative tocontrols. However, normal values were found in multiple system atrophy and progressivesupranuclear palsy [57]. R-R variability [75] and orthostatic testing [72] during wakeful-ness were affected in PD with RBD but not in PD without RBD compared with controlsand in the latter case PD without RBD had even less impairment than those with iRBD.These data support the hypothesis, that iRBD with dysautonomia presents an increasedrisk of development of PD with RBD.The fact that differences were found between PD patients with and without RBD couldindicate that disease progression are different in the two PD types and that brain stemareas are not involved to the same extent. However, neither in Study 1 nor in Study 4there were significant differences in the HRR/HRV for PD patients with RBD comparedwith PD patients without RBD, although some HRV parameters tended to be lower inthe PD patients with RBD. The PSG variables in Study 1 showed that TST and SL weresignificantly different from the control group, but only for the PD patients without RBD,which could imply that sleep may be affected more in this patient group.

Postuma et al. [74] found no evidence that baseline cardiac autonomic dysfunction canpredict the risk of neurodegenerative diseases and that autonomic dysfunction is linkedwith RBD independent of associated PD or Lewy body dementia. From the two scatter-plots (Figure 2.5 and 2.6) of HRR in iRBD patients in Study 1, it was seen how the iRBDpatients could be divided in two groups: one group having a normal HRR to both arousal

39

and LM, and the other group having an attenuated HRR to arousal and LM. Followingthese iRBD patients in a prospective study to see if these 2 groups predict who will de-velop a neurodegenerative disease could be of interest, although the Postuma study [74]do not support this hypothesis. Larger prospective studies are needed to clarify the exactconnection between dysautonomia in iRBD and the risk and type of neurodegenerativedevelopment.

In summary the salient finding was that of a pattern of progressive reduction in HRparameters (HRR/HRV) associated with sympathetic nervous activity in line with thenotion that postganglionic sympathetic nervous dysfunction is an early component in thedevelopment of PD. The attenuated HRR in patients with iRBD and PD with/withoutRBD may be a manifestation of the autonomic deficits experienced in these patientsand could indicate a rather significant impact on multiple systems in the brain includingthe ARAS and the autonomic nervous system. The iRBD patients not only exhibitedimpaired motor function with REM sleep without atonia but also incipient autonomicdysfunction, revealed in the attenuated HRR to LM in and in the VLF component duringwake.It is essential to state that measurements of HRV based on the spontaneous variability inHR are known to vary greatly within individuals, which could blur the outcome of smallcohort studies. Future studies could include forced variations in HR by deep breathingat 0.1 Hz, the Valsalva manoeuvre and HR change during active standing, since theseparameters reflect the capacity of the autonomic system rather than the activity and aresubject to smaller intra-individual variation.

3.2 Autonomic function in Narcolepsy

In Study 2, cataplexy and hcrt-1 status both significantly predicted HRR to arousal inREM sleep, although only hcrt-1 status remained significant in non-REM 2 sleep. None ofthe possible biasing factors (age, sex, BMI, disease duration, disease onset, anti-cataplexymedication) predicted HRR to arousal, and the study hereby shows, for the first time,that a reduced autonomic response to arousals in patients with narcolepsy is primarilypredicted by hcrt-1 deficiency in both REM and non-REM sleep, independent of cata-plexy and other factors. The results confirm that hcrt-1 deficiency affects the autonomicnervous system of patients with narcolepsy and that the hypocretin system is important

40

for proper HR modulation at rest.The results are in accordance with the findings of reduced HRR associated with PLMwith and without arousals during non-REM 2 sleep in 14 HLA-DQB1*0602-positive (andpresumed hcrt-1-deficient) NC patients [20]. The Dauvilliers study analyzed the HRRbeat by beat only for PLM with and without arousals during non-REM 2 sleep, andfound a significantly lower amplitude of tachycardia and bradycardia after onset of PLM,although the hcrt-1 levels of the patients were not known, but they were presumed hcrt-1deficient.The HRR to LM did not differ within the narcolepsy groups, but was significantly reducedcompared with the control group both in REM and non-REM 2 sleep. A previous studyhas found hypocretin deficiency to be independently associated with the prevalence ofRBD during REM sleep in narcolepsy [47]. Therefore, the LM in REM sleep may reflectunderlying RBD in the narcolepsy patients.

Study 3 is the first study to show that the frequency of sleep-wake transitions and REM-NREM transitions are associated with cataplexy and hypocretin deficiency in patientswith narcolepsy. The study documents that sleep-wake and REM-NREM transitions aremore frequent in patients with NC and hcrt-1-deficient narcolepsy. These results areconsistent with the flip-flop model of transitions proposed by Saper et al. [79], and areconsistent with a human model in which hypocretin enforces general-state stability.Quantitative EEG studies suggest changes in frequencies in non-REM 2 sleep [37] as wellas sleep onset [46] in patients with NC. Another study has shown that the occurrence ofmultiple spontaneous daytime sleep onset REM periods clearly identified patients withnarcolepsy [68].Animal studies support that sleep-wake transition is under control by the hypocret-inergic system, which facilitates wakefulness and REM-NREM cyclicity. Studies withhypocretin-deficient knockout mice and animals with narcolepsy have shown sleep stagedysregulations [18] and considerably more transitions between all stages [25, 44, 58, 59].As hypocretin is believed to stabilize wake and sleep, the hypocretin deficiency in patientswith narcolepsy may cause a low threshold to transition between stages, causing the frag-mented sleep pattern seen in patients with narcolepsy. This supports Study 3, showingthat hypocretin-deficient patients with narcolepsy may present sleep stage instability ascompared with non-hypocretin-deficient patients with narcolepsy.Hypocretin neurons are known to suppress REM sleep by reinforcing the activity of themonoaminergic neurons in the locus coeruleus and dorsal raphe nucleus [10, 49], which in

41

turn activate REM sleep-suppressing neurons and inhibit REM on neurons. It can be hy-pothesized that loss of hypocretin neurons therefore also enables more frequent transitionsinto REM sleep and makes the REM-NREM switch unstable. In Study 3 it was detectedthat patients with narcolepsy and hcrt-1 deficiency had more REM-NREM transitionsthan those without hcrt-1 deficiency, and this higher frequency of REM-NREM transi-tions was primarily predicted by hcrt-1 deficiency, independent of cataplexy and otherfactors, supporting the theory that hypocretin destabilizes the sleep stage switch.

The HRV measures described in Study 4, were not assessed in narcolepsy patients dueto insufficient sampling frequency of the ECG in the majority of these patients. Howeverthree other studies have investigated the HRV in patients with narcolepsy using powerspectrum analysis (PSA) and obtained very different results [28, 32, 34].Fronczek et al. [32] found a higher power in all frequency bands in 15 hypocretin defi-cient NC patients. In contrast, normal cardiovascular changes during head-up tilt tests,the Valsalva manoeuvre, deep breathing, isometric handgrip, and cold face tests, but anincreased sympathetic drive on HR at supine rest was found in 10 NC patients (nine of 10with hcrt-1 deficiency) [34], and an increase in HRV and sympathetic activity using PSAwas found in NC patients of unknown hcrt-1 status [28], indicating normal and increasedsympathetic tones, respectively. The latter study also reported an increased LF/HF ratioduring wakefulness before sleep, suggesting an altered circadian autonomic function inpatients with narcolepsy.The great variability of these results may be due to the different study designs (day/nightand measurement methods), small populations, and inadequate knowledge of the hcrt-1level. Furthermore, as hypothesized by Grimaldi et al. [34], patients with narcolepsymay be highly vigilant during the daytime, when they are continuously fighting againstsleepiness, which might imply stronger daytime sympathetic activation and thereby ex-plain the different results obtained in previous studies. Not only during wakefullness,narcolepsy patients may experience increased sympathetic activity. In Study 3, it wasseen how hcrt-1 deficient narcolepsy patients had increased frequency of sleep-wake trans-itions which could increase sympathetic activity in sleep.

Hypocretin neurons in the lateral hypothalamus reinforce the activity in the wake-promotingprojections arising from neurons in the upper brainstem, whereas sleep is promoted byactivation of the ventrolateral preoptic nucleus in the basal forebrain, which acts by in-hibiting hypocretin neurons and the neurons in ARAS [78]. The hypocretin neurons are

42

therefore believed to play an important role in sustaining wakefulness and in stabilizingthe flip-flop switch. However, other important arousal-producing neurons are present inthe posterior lateral hypothalamus, which helps maintain wakefulness when hypocretinneurons are deficient.The hypocretin neurons may help to sustain activity in aminergic and cholinergic arousalregions, and in the absence of hypocretin, these arousal regions may have reduced activ-ity, resulting in inappropriately low thresholds to transition into non-REM sleep [58]. Inpatients with narcolepsy and hypocretin deficiency, one of the most important compon-ents in the flip-flop switch is absent, making the switch unstable and entailing frequentsleep/wake switches, as seen in Study 3.

Chronic lack of hypocretin signalling may have consequences for cardiovascular func-tion, which may be revealed in the HR and BP [5]. Hypocretin neurons and receptorshave been proposed as providing a link between central mechanisms that regulate arousaland sleep-wakefulness states and central control of autonomic functions [97], includingparasympathetic cardiac activity [23]. Hypocretin fibers have also been found in nucleithat are well known to be involved in cardiovascular regulation, and it has been shownthat hcrt-1 diminishes parasympathetic cardiac activity and thereby contributes to HRacceleration [23]. Also, hypocretin acting on neurons in the hypothalamic paraventricularnucleus is known to increase the cardiovascular response [87].In Study 2 the hcrt-1-deficient patients did not only have an elevated baseline HR butalso failed to increase their HR in response to arousals or LM. This may be explained bytwo separate mechanisms: patients with hypocretin deficiency had increased arousability,sleep stage shifts and daytime somnolence, which may affect autonomic activity; and thehypocretin system may directly influence autonomic activity on its own. The attenuatedHRR may be explained by the already elevated HR in these patients, their impaired hy-pocretin signalling, or by both factors.Hcrt-1 fibers have been found in the lateral paragigantocellular nucleus, which elicits aninhibitory pathway to preganglionic cardiac vagal neurons [23]. The hcrt-1 deficiencymay cause this pathway to be impaired and another explanation is that parasympatheticactivity could explain the attenuated HRR.

To summarize, Study 3 shows that patients with narcolepsy and hypocretin deficiencyor cataplexy have more frequent sleep-wake and REM-NREM transitions, which is con-sistent with the destabilization of the flip-flop switches that regulate sleep-wake and

43

REM-NREM transitions due to the loss of hypocretin neurons. The elevated frequencyof REM-NREM transitions is primarily predicted by hcrt-1 deficiency, independent ofcataplexy and other factors. The increased frequency of transitions in hcrt-1 deficientpatients may cause increased sympathetic activity during sleep and thereby increased HRas seen in Study 2, or the increased HR could be caused by decreased parasympatheticactivity due to the lack of hypocretin signalling.The attenuated HRR to arousal is primarily predicted by hcrt-1 deficiency, and may beexplained by the already elevated HR and/or by impaired hypocretin signalling. Thisfinding suggests that the hypocretin system plays an important role in the modulationof cardiovascular function at rest.

3.3 Methods, data quality and validity

The studies included in this PhD dissertation describe methods for assessment of auto-nomic function. Study 1 described a method to assess the autonomic function based onHRR to arousal and LM during sleep. Same method was applied in Study 2 on narcolepsypatients, and Study 2 and Study 3 together assessed the influence of hypocretin loss onautonomic function.The method described in Study 4 is a more systematic and well known method to assessthe contributions of parasympathetic and sympathetic activity on the autonomic func-tion, and this study was partly performed to validate the method of Study 1 and 2, butalso to further evaluate on the autonomic dysfunction in PD patients as revealed by theresults in Study 1.

The HRR to arousal and LM during sleep provided a simple method to assess autonomicactivity. Similar methods have been described in other studies [20, 27, 30, 33, 84, 85],however none of these assessed the HRR by the AUC, but instead compared several pointsalong the curves. The AUC provided a simple and robust measure for the HRR and madethe comparison of various groups simple, unlike the many comparisons necessary with abeat-by-beat analysis. The AUC also allowed to do univariate and multivariate regressionanalyses that could take into account the influence of several confounder variables.In Study 1 and 2 the HRR to motor activation was assessed in association with LM andnot PLM, as have been done in other similar studies [20, 27, 85]. LM was chosen insteadof PLM to ensure a stable HR unaffected by other events before the onset of LM. TheHR could be slightly increased during PLM, which could give rise to an elevated baseline

44

and thus a false reduced HRR. Furthermore, in 18 of 22 control subjects in Study 2, notenough (or even none) PLM were available for a valid statistical analysis.Another advantage compared to the similar studies, is that the HRR was assessed in dif-ferent sleep stages (both non-REM 2 and REM), as opposed to only assessing the HRRin non-REM 2 [20, 27, 30, 33, 85]. Generally, the fact that HRR was measured duringsleep, ensured stable conditions.

In healthy subjects the HR is controlled by sympathetic and parasympathetic activ-ity in mutual balance. Although the HRR method presented a single value for assessingautonomic function during stable conditions, it may be difficult to assess whether theattenuated HRR is an expression of altered sympathetic or parasympathetic activity, ora combination of both. Study 4 revealed that mainly impaired sympathetic activity ex-plained the differences for PD/iRBD patients compared to controls, and as sympatheticactivity is less active during sleep compared to wake, autonomic testing during wake, maybe preferred in neurodegenerative diseases, which then would favor the HRV measures inwake rather than HRR to events during sleep.One disadvantage of the HRV measures is that although the Task Force of 1996 [90]provided the guidelines of HRV calculations, there are numerous of variables both in thetime domain, frequency domain and for the nonlinear parameters. Researchers have usedmany methods for analyzing these variables and reproducibility among these methods islimited [96]. HRV measures are also subject to intra-individual variation and due to theoften numerous variables included, Bonferroni correction or another method is requiredto counteract the problem of multiple comparisons.Although several studies have reported on the clinical and prognostic value of HRV ana-lysis, this technique has not been incorporated into clinical practice [96]. In a previousstudy [88], an arousal detection algorithm was presented, which could be directly linkedto the method of assessing HRR to arousals, and this would provide a complete toolfor assessing the autonomic changes in sleep. The method could be part of the PSGassessment, and provide a standard clinical test to assess the autonomic function andtogether with other electrophysiological biomarkers assess the possibility of an incipientneurodegenerative disease. However, this would require clear threshold limit values ofthe HRR, that have to be established through studies in larger patient populations.

Study 3 provided a simple method to assess the sleep state switching in narcolepsy. Onelimitation of this study is that sleep-wake regulation as determined by simple macro-sleep

45

scoring may be too simplistic a model to allow the evaluation of the sleep-wake/REM-NREM transition, because this may occur much faster and not in the 30-sec epochs usedin traditional scoring. This may in part explain why no significant association was foundwith non-REM transitions.To date, no reliable or validated sleep scoring model has been developed for scoringepochs shorter than 30 sec. Alternative methods may include a quantitative measureusing a sliding window with a shorter time interval including not only the conventionalsleep stages, but also transition zones between stages. One attempt using a state spaceanalysis technique has been performed in orexin knockout mice [25], where the variationsin the depth of sleep and the intensity of wakefulness were measured by EEG power inthe delta and theta bands. The orexin knockout mice had an overlap between wake andREM sleep and spent more time in transition regions causing the altered quality of sleepand wake.Variations within states and changes in transition zones may be a reason for the morefrequent changes in orexin knockout mice and hypocretin-deficient patients with narco-lepsy. Future evaluation of sleep transition and its relation to the hypocretinergic systemshould include analysis of transition states and shorter windows used in the sleep scoring.In Study 3, only nocturnal PSGs were considered, but daytime monitoring could be atleast as interesting, because patients with narcolepsy have sleep episodes during daytime.Another limitation is that interscorer variability was not considered and sleep scoringmay have been subject to individual scoring. This is relevant for all 4 studies in thisPhD thesis, as all studies included analysis in specific states (wake, non-REM or REM).Especially for patients with neurodegenerative diseases, sleep scoring may be challengingdue to altered EEG, EOG and EMG as a result of disease progression [24, 64, 77]. How-ever, all PSGs were scored by experienced technicians in accordance to the currently bestaccepted international criteria [40].

The number of included patients is limited in all 4 studies, and even though the find-ings in Study 1 were in accordance with the findings in Study 4, studies including largerpatient populations are needed to validate the HRR method as an autonomic test. Thehead-up tilt test, the Valsalva maneuver and deep breathing at 0.1 Hz are normally usedtests to assess autonomic function. Common for these tests are, that they measure forcedvariations in HR as opposed to the spontaneous variations measured in both Study 1, 2and 4.

46

Chapter 4

Conclusion and Perspectives

Study 1 and 4 showed that the SNS activity of iRBD patients was attenuated, especiallywhen the disease progressed to PD. In Study 4, no differences were seen in the HRVmeasures ascribed to parasympathetic activity, indicating that the cardiac PNS, emana-ting from the nucleus ambiguus, may be relatively preserved in patients with PD. Theprogressive reduction of sympathetic nervous activity is in line with the postganglionicsympathetic nervous dysfunction seen in early PD and may represent an early manifesta-tion of a neurodegenerative process involving brain stem areas, which is consistent withthe Braak hypothesis.Larger prospective studies are needed to clarify the exact connection between dysauto-nomia in iRBD and the risk and type of neurodegenerative development.

Study 2 showed for the first time that a reduced autonomic response to arousals inpatients with narcolepsy is primarily predicted by hcrt-1 deficiency in both REM andnon-REM sleep, independent of cataplexy and other factors. The result confirm thathcrt-1 deficiency affects the autonomic nervous system of patients with narcolepsy andthat the hypocretin system is important for proper HR modulation at rest.The hypocretin deficiency may also cause a low threshold to transition between stages,causing the fragmented sleep pattern and increased frequency of transition seen in pa-tients with narcolepsy. The increased frequency of transitions in hcrt-1 deficient patientsmay cause increased sympathetic activity during sleep and thereby increased HR as seenin Study 2, or the increased HR could be caused by decreased parasympathetic activitydue to the lack of hypocretin signalling.Future analysis could include comparisons of HRR from patients with narcolepsy withhcrt-1 deficiency and without cataplexy with that of patients with narcolepsy with nor-

47

mal hcrt-1 and with cataplexy.The evaluation of sleep transition and its relation to the hypocretinergic system shouldinclude analysis of transition states and shorter time windows used in the sleep scoringand also daytime monitoring could be of interest.

The HRR measured as the AUC provided a simple and robust measure for autonomicfunction and made the comparison of various groups simple, unlike the many comparisonsneeded when analyzing HRV.Further studies should seek to validate the HRR findings compared with other autonomictests such as forced variations in HR by deep breathing at 0.1 Hz, the Valsalva manoeuvreand HR change during active standing.A critical issue for planning for disease-modifying therapies focusing on iRBD is identi-fying one or more biomarkers. It will be necessary to evaluate the predictive value ofautonomic (dys-)function for predicting development of a neurodegenerative disease iniRBD. If the HRR could be used as a predictive biomarker, it would be easy to imple-ment it as a standard clinical test as part of a PSG, and would provide a noninvasive,inexpensive and practical tool for screening purposes.

48

Bibliography

[1] Akincioglu C., Unlu M., and Tunc T. Cardiac innervation and clinical correlates inidiopathic parkinson’s disease. Nuclear Medicine Communications, 24(3):267–271,2003.

[2] American Academy of Sleep Medicine. The international classification of sleep dis-orders, revised. Diagnostic and coding manual. American Academy of Sleep Medi-cine, 1 edition, 2001.

[3] Baharav A., Kotagal S., Gibbons V., Rubin B.K., Pratt G., Karin J., and Akselrod S.Fluctuations in autonomic nervous activity during sleep displayed by power spectrumanalysis of heart rate variability. Neurology, 45:1183–1187, 1995.

[4] Barbic F., Perego F., Canesi M., Gianni M., Biagiotti S., Costantino G., Pezzoli G.,Porta A., Malliani A., and Furlan R. Early abnormalities of vascular and cardiacautonomic control in parkinson’s disease without orthostatic hypotension. Hyper-tension, 49(1):120 – 126, 2007.

[5] Bastianini S., Silvani A., Berteotti C., Elghozi J.L., Franzini C., Lenzi P., Lo MartireV., and Zoccoli G. Sleep related changes in blood pressure in hypocretin-deficientnarcoleptic mice. SLEEP, 34:213–8, 2011.

[6] Boeve B.F. REM sleep behavior disorder: Updated review of the core features,the REM sleep behavior disorder-neurodegenerative disease association, evolvingconcepts, controversies, and future directions. Ann N Y Acad Sci., 1184:15–54,2010.

[7] Bonnet M.H., Carley D., Guilleminault C.G., Harper R., Haydes B., HirshkowitzM., Ktonas P., Keenan S., Roehrs T., Smith J., Walsh J., Weber S., and WestbrookP. EEG arousals: scoring rules and examples: a preliminary report from the sleepdisorders atlas task force of the american sleep disorders association. Sleep, 15:173–184, 1992.

49

[8] Bonnet M.H. and Arand D.L. Heart rate variability: sleep stage, time of night, andarousal influences. Electroencephalography and Clinical Neurophysiology, 102:390–396, 1997.

[9] Bonnet M.H. and Arand D.L. EEG arousal norms by age. Journal of Clinical SleepMedicine, 3(3):271–274, 2007.

[10] Bourgin P., Huitrón-Résendiz S., Spier A.D., Fabre V., Morte B., Criado J.R.,Sutcliffe J.G., Henriksen S.J., and de Lecea L. Hypocretin-1 modulates rapideye movement sleep through activation of locus coeruleus neurons. J Neurosci,20(20):7760–5, 2000.

[11] Braak H., Tredici K.D., Rüb U., de Vos R.A.I., Jansen Steur E.N.H., and Braak E.Staging of brain pathology related to sporadic parkinson’s disease. Neurobiology ofAging, 24(2):197 – 211, 2003.

[12] Braak H., Bohl J.R., Müller C.M., Rüb U., de Vos R.A., and Del Tredici K. Stanleyfahn lecture 2005: The staging procedure for the inclusion body pathology associatedwith sporadic parkinson’s disease reconsidered. Mov Disord., 21:2042–2051, 2006.

[13] Braak H. and Tredici K.D. Invited article: Nervous system pathology in sporadicparkinson disease. Neurology, 70:1916 – 1925, 2008.

[14] Brennan M., Palaniswami M., and Kamen. P. Do existing measures of poincaré plotgeometry reflect nonlinear features of heart rate variability? IEEE Trans BiomedEng., 48:1342–1347, 2001.

[15] Busek P., Vanková J., Opavský J., Salinger J., and Nevsímalová S. Spectral analysisof the heart rate variability in sleep. Physiol Res., 54:369–376, 2005.

[16] Camerlingo M., Aillon C., and Bottacchi E. Parasympathetic assessment in parkin-son’s disease. Adv. Neurol., 45:267–269, 1987.

[17] Chaudhuri K.R., Healy D.G., and Schapira A.H.V. Non-motor symptoms of par-kinson’s disease: diagnosis and management. The Lancet Neurology, 5(3):235 – 245,2006.

[18] Chemelli R.M., Willie J.T., and Sinton et al. C.M. Narcolepsy in orexin knockoutmice: molecular genetics of sleep regulation. Cell, 98:437–51, 1997.

[19] Ciriello J. and de Oliveira C.V.R. Cardiac effects of hypocretin-1 in nucleus am-biguus. Am J Physiol Regul Integr Comp Physiol, 284:R1611–20, 2003.

50

[20] Dauvilliers Y., Pennestri M.H., Whittom S., Lanfranchi P.A., and Montplaisir J.Y.Autonomic response to periodic leg movements during sleep in narcolepsy-cataplexy.Sleep, 34:219–23, 2011.

[21] de Lau L.M. and Breteler M.M. Epidemiology of parkinson’s disease. The LancetNeurology, 5(6):525–535, 2006.

[22] de Oliveira C.V.R, Rosas-Arellano M.P., Solano-Flores L.P., and Ciriello J. Cardi-ovascular effects of hypocretin-1 in nucleus of the solitary tract. Am J Physiol HeartCirc Physiol, 284:H1369–77, 2003.

[23] Dergacheva O., Philbin K., Batemana R., and Mendelowitza D. Hypocretin-1(orexin A) prevents the effects of hypoxia/hypercapnia and enhances the gabaergicpathway from the lateral paragigantocellular nucleus to cardiac vagal neurons in thenucleus ambiguus. Neuroscience, 175:18–23, 2011.

[24] Diederich N.J., Vaillant M., and Mancuso G. Progressive sleep destructuring inparkinsons’s disease. A polysomnographic study in 46 patients. Sleep Medicine,6:313–318, 2005.

[25] Diniz Behn C.G., Klerman E.B., Mochizuki T., Lin S., and Scammell T.E. Abnormalsleep/wake dynamics in orexin knockout mice. SLEEP, 33:297–306, 2010.

[26] Ergun U., Demirci M., Nurlu G., and Komürcü F. Power spectral analysis ofheart rate variability: normal values of subjects over 60 years old. Int J Neurosci.,118:1165–1173, 2008.

[27] Fantini M.L., Michaud M., Gosselin N., Lavigne G., and Montplaisir J. Periodicleg movements in REM sleep behavior disorder and related autonomic and EEGactivation. Neurology, 59:1889 – 1894, 2002.

[28] Ferini-Strambi L., Spera A., Oldani A., and et al. Autonomic function in narcolepsy:power spectrum analysis of heart rate variability. J Neurol, 244:252–5, 1997.

[29] Ferri R., Miano S., Bruni O., Vankova J., Nevsimalova S., Vandi S., Montagna P.,Ferini-Strambi L., and Plazzi G. NREM sleep alterations in narcolepsycataplexy.Clinical Neurophysiology, 116:2675 – 84, 2005.

[30] Freilich S., Goff E.A., Malaweera A.S.N., Twigg G.L., Simonds A.K., Mathias C.J.,and Morrell M.J. Sleep architecture and attenuated heart rate response to arousalfrom sleep in patients with autonomic failure. Sleep Medicine, 11(1):87 – 92, 2010.

51

[31] Friesen G.M., Jannett T.C., Jadallah M.A., Yates S.L., Quint S.R., and Nagle H.T.A comparison of the noise sensitivity of nine QRS detection algorithms. BiomedicalEngineering, IEEE Transactions on, 37(1):85 –98, jan. 1990.

[32] Fronczek R., Overeem S., Reijntjes R., Lammers G.J., van Dijk J.G., and Pijl H. In-creased heart rate variability but normal resting metabolic rate in hypocretin/orexin-deficient human narcolepsy. J Clin Sleep Med, 4:248–54, 2008.

[33] Gosselin N., Michaud M., Carrier J., Lavigne G., and Montplaisir J. Age differencein heart rate changes associated with micro-arousals in humans. Clinical Neuro-physiology, 113(9):1517 – 1521, 2002.

[34] Grimaldi D., Pierangeli G., Barletta G., and et al. Spectral analysis of heart ratevariability reveals an enhanced sympathetic activity in narcolepsy with cataplexy.Clinical Neurophysiology, 121:1142–7, 2010.

[35] Haapaniemi T.H., Pursiainen V., Korpelainen J.T., Huikuri H.V., and SotaniemiK.A. Ambulatory ECG and analysis of heart rate variability in parkinson’s disease.Journal of Neurology, Neurosurgery and Psychiatry, 70:305–310, 2001.

[36] Hadjileontiadis L.J., Panoulas K.I., Penzel T., and Panas S.M. Performance ofthree QRS detection algorithms during sleep: a comparative study. In Engineeringin Medicine and Biology Society, 2001. Proceedings of the 23rd Annual InternationalConference of the IEEE, volume 2, pages 1954 – 1957, 2001.

[37] Hadjiyannakis K., Ogilvie R.D., Alloway C.E., and Shapiro C. FFT analysis ofEEG during stage 2-to REM transitions in narcoleptic patients and normal sleepers.Electroencephalogr Clin Neurophysiol, 103:543–53, 1997.

[38] Haensch C.A., Lerch H., Jörg J., and Isenmann S. Cardiac denervation occurs inde-pendent of orthostatic hypotension and impaired heart rate variability in parkinson’sdisease. Parkinsonism Relat. Disord., 15:134–7, 2009.

[39] Hamilton P.S. and Tompkins W.J. Quantitative investigation of QRS detection rulesusing the MIT/BIH arrhythmia database. Biomedical Engineering, IEEE Transac-tions on, BME-33(12):1157 –1165, dec. 1986.

[40] Iber C., Ancoli-Israel S., Chesson A.L., and Quan S.F. The AASM Manual for theScoring of Sleep and Associated Events: Rules, Terminology and Technical Specific-ations. Wastchester, IL: American Academy of Sleep Medicine, 1 edition, 2007.

52

[41] Iranzo A., Molinuevo J.L., Santamaría J., Serradell M., Martí M.J., Valldeoriola F.,and Tolosa E. Rapid-eye-movement sleep behaviour disorder as an early marker fora neurodegenerative disorder: a descriptive study. The Lancet Neurology, 5(7):572 –577, 2006.

[42] Iranzo A., Santamaría J., Serradell M., Marti M.J., and Valldeoriola F. Sixty-fourpercent of patients with idiopathic REM sleep behavior disorder developed a neuro-logical disorder after a mean clinical follow-up of seven years. Sleep, 31(Suppl.):A280,2008.

[43] Jennum P., Zoetmulder M., Korbo L., and Kjellberg J. The health-related, social,and economic consequences of parkinsonism: a controlled national study. J Neurol.,258(8):1497–506, 2011.

[44] Kaitin K.I., Kilduff T.S., and Dement W.C. Sleep fragmentation in canine narco-lepsy. SLEEP, 9:116–9, 1986.

[45] Kallio M., Haapaniemi T., Turkka J., Suominen K., Tolonen U., Sotaniemi K.,Heikkilä V.-P., and Myllylä V. Heart rate variability in patients with untreatedparkinson’s disease. European Journal of Neurology, 7(6):667 – 672, 2000.

[46] Kim J.W., Shin H.B., and Robinson P.A. Quantitative study of the sleep onsetperiod via detrended fluctuation analysis: normal vs. narcoleptic subjects. ClinNeurophysiol, 120:1245–51, 2009.

[47] Knudsen S., Gammeltoft S., and Jennum P.J. Rapid eye movement sleep behaviourdisorder in patients with narcolepsy is associated with hypocretin-1 deficiency. Brain,133:568 – 79, 2010.

[48] Kohler B.-U., Hennig C., and Orglmeister R. The principles of software QRSdetection. Engineering in Medicine and Biology Magazine, IEEE, 21(1):42 –57, jan.-feb. 2002.

[49] Kohlmeier K.A., Watanabe S., Tyler C.J., Burlet S., and Leonard C.S. Dual orexinactions on dorsal raphe and laterodorsal tegmentum neurons: noisy cation currentactivation and selective enhancement of ca2+ transients mediated by L-type calciumchannels. J Neurophysiol, 100:2265–81, 2008.

[50] Lanfranchi P.A., Fradette L., Gagnon J., Colombo R., and Montplaisir J. Cardiacautonomic regulation during sleep in idiopathic REM sleep behavior disorder. Sleep,30(8):1019 – 1025, 2007.

53

[51] Lindgren P., von Campenhausen S., Spottkeb E., Siebert U., and Dodel R. Cost ofparkinson’s disease in Europe. European Journal of Neurology, 12:S68–S73, 2005.

[52] Ludin S.M., Steiger U.H., and Ludin H.P. Autonomic disturbances and cardiovas-cular reflexes in idiopathic parkinson’s disease. Journal of Neurology, 235:10–15,1987.

[53] Maetzler W., Liepelt I., and Berg D. Progression of parkinson’s disease in theclinical phase: potential markers. The Lancet Neurology, 8:1158 – 1171, 2008.

[54] Marple S.L. Digital Spectral Analysis. Prentice-Hall International, 1987.

[55] Mesec A., Sega S., and Trost M. The deterioration of cardiovascular reflexes inparkinson’s disease. Acta Neurology Scandinavia, 100:296–299, 1999.

[56] Mignot E. Etiology and genetics of human narcolepsy. In Baumann C.R., Bas-setti C.L., and Scammell T.E., editors, Narcolepsy. Pathophysiology, Diagnosis, andTreatment, chapter 1, pages 3 – 17. Springer, New York Dordrecht Heidelberg Lon-don, 2011.

[57] Miyamoto T., Miyamoto M., Suzuki K., Nishibayashi M., Iwanami M., and Hirata K.123I-MIBG cardiac scintigraphy provides clues to the underlying neurodegenerativedisorder in idiopathic REM sleep behavior disorder. SLEEP, 31:717–723, 2008.

[58] Mochizuki T., Crocker A., McCormack S., Yanagisawa M., Sakurai T., and Scam-mell T.E. Behavioral state instability in orexin knock-out mice. J Neurosci, 24:6291–6300, 2004.

[59] Nishino S., Riehl J., Hong J., Kwan M., Reid M., and Mignot E. Is narcolepsya REM sleep disorder? analysis of sleep abnormalities in narcoleptic dobermans.Neurosci Res, 38:437–46, 2000.

[60] Nussbaum R.L. and Ellis C.E. Alzheimer’s disease and parkinson’s disease. TheNew England Journal of Medicine, 14(348):1356–1364, 2003.

[61] Oka H., Toyoda C., Yogo M., and Mochio S. Cardiovascular dysautonomia in denovo parkinson’s disease without orthostatic hypotension. Eur J Neurol., 18:286–292,2011.

[62] Pan J. and Tompkins W.J. A real-time QRS detection algorithm. BiomedicalEngineering, IEEE Transactions on, BME-32(3):230 –236, march 1985.

54

[63] Parkinson J. An essay on the shaking palsy. Whittingham and Roland, London,1817.

[64] Petit D., Gagnona J.-F., Fantinia M.L., Ferini-Strambi L., and Montplaisir J. Sleepand quantitative EEG in neurodegenerative disorders. Journal of PsychosomaticResearch, 56:487 – 496, 2004.

[65] Peyron C. Is narcolepsy a neurodegenerative disorder? In Baumann C.R., Bas-setti C.L., and Scammell T.E., editors, Narcolepsy. Pathophysiology, Diagnosis, andTreatment, chapter 1, pages 27 – 33. Springer, New York Dordrecht Heidelberg Lon-don, 2011.

[66] Peyron C., Tighe D.K., van den Pol A.N., and et al. Neurons containing hypocretin(orexin) project to multiple neuronal systems. J Neurosci, 18:9996–10015, 1998.

[67] Peyron C., Faraco J., Rogers W., Ripley B., Overeem S., Charnay Y., NevsimalovaS., Aldrich M., Reynolds D., Albin R., Li R., Hungs M., Pedrazzoli M., PadigaruM., Kucherlapati M., Fan J., Maki R., Lammers G.J., Bouras C., Kucherlapati R.,Nishino S., and Mignot E. A mutation in a case of early onset narcolepsy and ageneralized absence of hypocretin peptides in human narcoleptic brains. Nat Med,6:991 – 7, 2000.

[68] Pizza F., Moghadam K.K., Vandi S., Detto S., Poli F., Mignot E., Ferri R., andPlazzi G. Daytime continuous polysomnography predicts MSLT results in hyper-somnias of central origin. J Sleep Res, 22(1):32–40, 2013.

[69] Portet F., Hernández A.I., and Carrault G. Evaluation of real-time QRS detectionalgorithm in variable contexts. Medical and Biological Engineering and Computing,43(3):379–385, 2005.

[70] Postuma R.B. and Montplaisir J. Predicting parkinson’s disease - why, when, andhow? Parkinsonism and Related Disorders, 15(Supplement 3):S105 – S109, 2009.

[71] Postuma R.B., Gagnon J.F., Vendette M., Fantini M.L., Massicotte-Marquez J.,and Montplaisir J. Quantifying the risk of neurodegenerative disease in idiopathicREM sleep behavior disorder. Neurology, 72(15):1296–1300, 2009.

[72] Postuma R.B., Gagnon J.F, Vendette M., and Montplaisir J.Y. Markers of neuro-degeneration in idiopathic rapid eye movement sleep behaviour disorder and parkin-son’s disease. Brain, 139:3298–3307, 2009.

55

[73] Postuma R.B., Gagnon J.F., and Montplaisir J.Y. Clinical prediction of parkinson’sdisease: planning for the age of neuroprotection. Neurosurg Psychiatry., 81:1008–13,2010.

[74] Postuma R.B., Lanfranchi P.A., Blais H., Gagnon J.F., and Montplaisir J.Y. Car-diac autonomic dysfunction in idiopathic REM sleep behavior disorder. MovementDisorders, 25(14):2304–10, 2010.

[75] Postuma R.B., Montplaisir J.Y., Lanfranchi P., Blais H., Rompré S., Colombo R.,and Gagnon J.F. Cardiac autonomic denervation in parkinson’s disease is linked toREM sleep behavior disorder. Mov Disord., 26:1529–33, 2011.

[76] Samson W.K., Gosnell B., Chang J.K., Resch Z.T., and Murphy T.C. Cardiovascularregulatory actions of the hypocretins in brain. Brain Res, 831:248–53, 1999.

[77] Santamaria J., Högl B., Trenkwalder C., and Bliwise D. Scoring sleep in neurologicalpatients: the need for specific considerations. SLEEP, 34(10):1283–4, 2011.

[78] Saper C.B., Fuller P.M., Pedersen N.P., Lu J., and Scammell T.E. Sleep stateswitching. Neuron, 68:1023–1042, 2010.

[79] Saper C.B., Chou T.C., and Scammell T.E. The sleep switch: hypothalamic controlof sleep and wakefulness. Trends in Neurosciences, 24(12):726–731, 2001.

[80] Saper C.B., Scammell T.E., and Lu J. Hypothalamic regulation of sleep and circa-dian rhythms. Nature, 437:1257–1263, 2005.

[81] Schenck C.H., Bundlie S.R., and Mahowald M.W. Delayed emergence of a parkin-sonian disorder in 38 percentage of 29 older men initially diagnosed with idiopathicrapid eye movement sleep behavior disorder. Neurology, 46(2):388–393, 1996.

[82] Schenck C.H., Bundlie S.R., and Mahowald M.W. REM sleep behaviour disorder(RBD) delayed emergence of parkinsonism and/or dementia in 65 percent of oldermen initially diagnosed with idiopathic RBD, and an analysis of the maximum andminimum tonic and/or phasic electromyographic abnormalities found during REMsleep. Sleep, 26(Suppl.):A316, 2003.

[83] Schenck C.H., Bundlie S.R., and Mahowald M.W. Delayed emergence of a par-kinsonian disorder or dementia in 81 percentage of older males initially diagnosedwith idiopathic REM sleep behavior disorder (RBD): 16 year update on a previouslyreported series. Sleep Med, 2013. http://dx.doi.org/10.1016/j.sleep.2012.10.009.

56

[84] Sforza E., Jouny C., and Ibanez V. Cardiac activation during arousal in humans:further evidence for hierarchy in the arousal response. Clinical Neurophysiology,111:1611–1619, 2000.

[85] Sforza E., Jouny C., and Ibanez V. Time-dependent variation in cerebral andautonomic activity during periodic leg movements in sleep: implications for arousalmechanisms. Clinical Neurophysiology, 113:883 – 891, 2002.

[86] Shirasaka T., Nakazato M., Matsukura S., Takasaki M., and Kannan H. Sym-pathetic and cardiovascular actions of orexins in conscious rats. Am J Physiol,277:R1780–5, 1999.

[87] Shirasaka T., Kunitake T., Takasaki M., and Kannan H. Neuronal effects of orexins:relevant to sympathetic and cardiovascular functions. Regul Pept, 104:91–5, 2002.

[88] Sorensen G.L., Jennum P., Kempfner J., Zoetmulder M., and Sorensen H.B. AComputerized Algorithm for Arousal Detection in Healthy Adults and Patients withParkinson Disease. J Clin Neurophysiol., 29(1):58–64, 2012.

[89] Tarvainen M.P. and Niskanen J.P. Kubios HRV version 2.1 User’s guide. Universityof Finland, 2012.

[90] Task Force of The European Society of Cardiology, The North American Societyof Pacing, and Electrophysiology. Heart rate variability. standards of measurment,physiological interpretation, and clinical use. European Heart Journal, 17:354–381,1996.

[91] Thannickal T.C., Moore R.Y., Nienhuis R., Ramanathan L., Gulyani S., AldrichM., Cornford M., and Siegel J.M. Reduced number of hypocretin neurons in humannarcolepsy. Neurology, 27(3):469–474, 2000.

[92] Tomlinson C.L., Stowe R., Patel S., Rick C., Gray R., and Clarke C.E. Systematicreview of levodopa dose equivalency reporting in parkinson’s disease. Mov. Disord.,25:2649–2653, 2010.

[93] Valappil R.A., Black J.E., Broderick M.J., Carrillo O., Frenette E., Sullivan S.S.,Goldman S.M., Tanner C.M., and Langston J.W. Exploring the electrocardiogramas a potential tool to screen for premotor parkinson’s disease. Movement Disorders,25(14):2296–2303, 2010.

57

[94] Vaughn B.V., Quint S.R., Messenheimer J.A., and Robertson K.R. Heart periodvariability in sleep. Electroencephalography and Clinical Neurophysiology, 94(3):155– 162, 1995.

[95] Voss A., Heitmann A., Schroeder R., Peters A., and Perz S. Short-term heart ratevariability - age dependence in healthy subjects. Physiol Meas., 33:1289 – 1311,2012.

[96] Xhyheri B., Manfrini O., Mazzolini M., Pizzi C., and Bugiardini R. Heart ratevariability today. Prog Cardiovasc Dis., 55(3):321–331, 2012.

[97] Young J.K., Wu M., Manaye K.F., Kc P., Allard J.S., Mack S.O., and Haxhiu M.A.Orexin stimulates breathing via medullary and spinal pathways. J Appl Physiol,98:1387–95, 2005.

[98] Zoccolella S., Savarese M., Lamberti P., Manni R., Pacchetti C., and Logroscino G.Sleep disorders and the natural history of parkinson’s disease: The contribution ofepidemiological studies. Sleep Medicine Reviews, 15(1):41–50, 2011.

58

Appendix A

Paper 1

Title: Attenuated Heart Rate Response in REM Sleep Behavior Disorder and Parkin-son’s Disease

Status: Published in Movement Disorders

Citation: Sorensen G.L., Kempfner J., Zoetmulder M., Sorensen H.B., and Jennum P.Attenuated heart rate response in REM sleep behavior disorder and Parkinson’s disease.Mov Disord. 2012; 27(7):888-94

59

Attenuated Heart Rate Response in REM Sleep Behavior Disorder andParkinson’s Disease

Gertrud Laura Sorensen, MSc,1,2,3* Jacob Kempfner, MSc,1,2,3 Marielle Zoetmulder, MSc,2,3,4

Helge B. D. Sorensen, MSc, PhD,1 and Poul Jennum, MD, PhD2,3

1Department of Electrical Engineering, Technical University of Denmark, Kongens Lyngby, Denmark2Danish Center for Sleep Medicine, Glostrup University Hospital, Glostrup, Denmark

3Center for Healthy Aging, University of Copenhagen, Copenhagen, Denmark4Department of Neurology, Bispebjerg Hospital, Copenhagen, Denmark

ABSTRACT: The objective of this study was todetermine whether patients with Parkinson’s diseasewith and without rapid-eye-movement sleep behaviordisorder and patients with idiopathic rapid-eye-move-ment sleep behavior disorder have an attenuated heartrate response to arousals or to leg movements duringsleep compared with healthy controls. Fourteen and 16Parkinson’s patients with and without rapid-eye-move-ment sleep behavior disorder, respectively, 11 idiopathicrapid-eye-movement sleep behavior disorder patients,and 17 control subjects underwent 1 night of polysom-nography. The heart rate response associated witharousal or leg movement from all sleep stages was ana-lyzed from 10 heartbeats before the onset of the sleepevent to 15 heartbeats following onset of the sleepevent. The heart rate reponse to arousals was signifi-cantly lower in both parkinsonian groups comparedwith the control group and the idiopathic rapid-eye-movement sleep behavior disorder group. The heart

rate response to leg movement was significantly lowerin both Parkinson’s groups and in the idiopathic rapid-eye-movement sleep behavior disorder group comparedwith the control group. The heart rate response for theidiopathic rapid-eye-movement sleep behavior disordergroup was intermediate with respect to the control andthe parkinsonian groups. The attenuated heart rateresponse may be a manifestation of the autonomic defi-cits experienced in Parkinson’s disease. The idiopathicrapid-eye-movement sleep behavior disorder patientsnot only exhibited impaired motor symptoms but alsoincipient autonomic dysfunction, as revealed by theattenuated heart rate response. VC 2012 Movement Disor-der Society

Key Words: Parkinson’s disease; REM sleep behaviordisorder; arousal; sleep; heart rate response; autonomicdysfunction

Parkinson’s disease (PD) is a neurodegenerativedisease with bradykinesia, rigidity, rest tremor, andpostural instability.1 In addition to these motor symp-toms, PD patients have numerous abnormalities ofother systems, which may precede the motor signs.2,3

Autonomic dysfunction is experienced by 40%–70%of PD patients.2 The degeneration and dysfunctioninclude nuclei-mediating autonomic functions such asthe dorsal vagal nucleus, nucleus ambiguus, and othermedullary centers, which exert differential control onthe sympathetic preganglionic neurons via descendingpathways.2 Parasympathetic and sympathetic functionsmay be impaired.4

Several studies have examined cardiovascularresponses as markers of autonomic nervous system(ANS) dysfunction in PD patients.5–9 This study

------------------------------------------------------------Additional Supporting Information may be found in the online version ofthis article.

*Correspondence to: Gertrud Laura Sorensen, Danish Center for SleepMedicine, Department of Clinical Neurophysiology, Faculty of HealthSciences, University of Copenhagen, Glostrup Hospital, DK 2600Glostrup, Denmark; Gertrud_l@yahoo.dk

Funding agencies: The study is supported by an unrestricted grant fromCenter for Healthy Aging (CEHA), University of Copenhagen. The study isfurther part of the research in the Biomedical Signal ProcessingResearch Group, Technical University of Denmark (DTU). Neither theCEHA nor the DTU had any further role in the study design, collection,analysis, and interpretation of data, the writing of the report, or thedecision to submit the article for publication. None of the authors hasreported any conflict of interests.Relevant conflicts of interest/financial disclosures: Nothing to report.Full financial disclosures and author roles may be found in the onlineversion of this article.

Received: 16 June 2011; Revised: 22 February 2012; Accepted: 26March 2012Published online 3 May 2012 in Wiley Online Library(wileyonlinelibrary.com). DOI: 10.1002/mds.25012

R E S E A R C H A R T I C L E

888 Movement Disorders, Vol. 27, No. 7, 2012

focused on the cardiac response to arousals and legmovements (LMs) during sleep, which has not previ-ously been assessed in PD patients.Wakefulness and arousals are regulated by 2 ascend-

ing neuron networks in the brain, referred to as theascending reticular activation system (ARAS).10 Thefirst activating neuron network consists of the pedun-culopontine and laterodorsal tegmental nuclei in thebrain stem, which are active during wakefulness andrapid-eye-movement (REM) sleep.11 The second acti-vating neuron network consists of the locus coeruleus,dorsal end median Raphe nuclei, ventral periaquaduc-tal gray matter, and tuberomammilary nucleus, whichare active during wakefulness, slightly active duringnon-REM sleep, and inactive during REM sleep.11

Sleep is promoted by activation of the ventrolateralpreoptic nucleus in the basal forebrain, which actsthrough inhibition of hypocretin neurons and the neu-rons in the ARAS.11 Transition from sleep to wakeful-ness and vice versa is regulated by the sleep–wake flip-flop switch, which involves the mutual inhibition ofthe sleep-promoting nuclei and the wakefulness-pro-moting nuclei.12 This flip-flop ensures rapid sleep–wake transitions. Arousal from sleep involves activa-tion of the sympathetic nervous system, including spi-nal motor activations. Therefore, arousal from sleepincluding motor activations in sleep is associated withan increased heart rate (HR) and blood pressure inhealthy adults,13 although the increase in HR getslower with age.14 An intact sympathetic nervous sys-tem is a key component of the HR response (HRR)associated with LM and arousal from sleep.15

PD is a progressive neurodegenerative disease believedto commence in the medulla oblongata and progressingupward.16 Idiopathic REM sleep behavior disorder(iRBD) is a strong predictor of the later development ofPD, probably because of the involvement of the sublatero-dorsal (SLD) nucleus, which is involved in atonia duringREM sleep.3,17–21 Because of the early involvement of themedulla oblongata in iRBD and PD, we hypothesized thatANS deficiency should be present early in the diseasecourse and that therefore these patients will demonstratean attenuated HRR associated with arousal and LM.Because iRBD is reported to herald the onset of PD byyears, attenuation of HRR in these patients would supportour hypothesis.

Patients and Methods

Subjects

PD and iRBD patients were identified from thepatient database at the Danish Center for Sleep Medi-cine, Department of Clinical Neurophysiology,Glostrup University Hospital, Denmark. All patientswere evaluated by neurological examination and

underwent PSG. Furthermore, PD patients also had aDAT scan to confirm the PD diagnosis.Exclusion criteria were additional neurological, psy-

chiatric, or cardiovascular disorders, alcohol abuse,head trauma, dementia, and an apnea-hypopnea index>5. No subjects were taking any agents known toaffect the sympathetic system.It was not possible to withdraw patients from their

Parkinson’s medication, but in order to rule out anyinfluence of these drugs, patients treated with intraspi-nal medication, duodopa, or intercerebral stimulationwere excluded. For the patients who were included,the total levodopa equivalent dose (LED) was calcu-lated as described previously22 in order to check forany correlations between the results and the medica-tion. All PD patients were categorized based onwhether they had symptoms of RBD, as it was ofinterest to know whether these 2 PD types expresseddifferences in the HRR because this might indicatedifferent disease progression in the 2 groups. RBDwere diagnosed according to the criteria of the Inter-national Classification of Sleep Disorders (ICSD-2).23

iRBD patients had no other disorders explaining theRBD. All patients were further assessed with theHoehn and Yahr (HY) scale and the Unified ParkinsonDisease Rating Scale (UPDRS).In total, 14 PD patients with RBD, 16 PD patients

without RBD, and 11 iRBD patients fulfilled the inclu-sion criteria of the study. Seventeen healthy controls,matched to the patient group by age, sex, andbody mass index (BMI), were also recruited tothe study. All included subjects gave their signedinformed consent. The study received ethical approval(HA20070120).Demographic data of the included subjects are given

in Table 1.

Polysomnographic Recordings

All subjects underwent 1 night of PSG recorded ei-ther under ambulant conditions or at the Danish Cen-ter for Sleep Medicine. All recordings included aminimum of 6 electroencephalography (EEG) leads(F3-A2, F4-A1, C3-A2, C4-A1, O1-A2, O2-A1), andinpatient recordings included EEG electrodes accord-ing to the 10-20 system, all in accordance with thegold standard.24 Furthermore, surface electromyogra-phy (EMG) of the left and right anterior tibialismuscles and the submental muscle, electrocardio-graphy (ECG), and vertical and horizontal electro-oculography (EOG) were included in all recordings.Impedances were kept below 10 kX. The amplifierwas different for the ambulant and inpatient record-ings, although this only influenced frequencies above60 Hz, which is irrelevant in the context of this study.Sleep stages and arousals from sleep were manually

scored according to standard criteria.24,25 LM were

H E A R T R A T E R E S P O N S E I N P D / i R B D P A T I E N T S

Movement Disorders, Vol. 27, No. 7, 2012 889

quantified from the surface EMG in accordance withstandard criteria.24

Data Analysis

The HRR to arousals or LM was estimated by cal-culating the change in RR intervals in the ECG signaland determining the area under the curve (AUC) forthe HR change (HRC) from 10 beats before to 15beats after onset of the events. Data were analyzed bythe following steps.

Detection of QRS Complexes and R Waves

For each sleep event (arousal or LM), QRS com-plexes were automatically detected in the ECG signalusing the Pan-Tompkins QRS detector.26 In eachdetected QRS complex, the R wave was determined asthe largest peak.

Validation of QRS Detector andCalculation of RR Intervals

Arousals and LMs associated with noise-contami-nated ECG signals were excluded from the calculationof the HRR because detection of QRS complexes inthese ECG signal segments was impossible to validate.RR intervals were calculated as the interbeat inter-

val between successive R waves by subtracting thesample positions, Ri, of succeeding R waves and divid-ing by the sample frequency, fs,:

RRi ¼ Riþ1 � Ri

fs; i ¼ f�10;�9; . . . ; 14; 15gnf0g: (1)

where a negative i and a positive i denote RR intervalsbefore and after onset of the event, respectively

Calculation of HRR Curves

For every included sleep event, the HR was analyzedfrom 10 heartbeats before to 15 heartbeats followingthe onset of the event. The identified RR intervalswere converted to HR in beats/minute according tothe following equation:

HRi ¼ 1

RRi� 60; i ¼ f�10;�9; . . . ; 14; 15gnf0g: (2)

Previous studies of HRR to arousal or motor activ-ity had shown that the HR began to rise 2 heartbeatsbefore the event.13,27 Therefore, a baseline value ofHR was calculated by averaging the first 7 of the 10beats before the event. To determine the HRC, thisbaseline value was subtracted from all 25 HR values(from 10 beats before to 15 beats after the event):

HRCi ¼ HRi �X�4

j¼�10

HRj

7; i

¼ f�10;�9; . . . ; 14; 15gnf0g: (3)

For each subject, the values of Equation (3) werecalculated for all events, and the average HRC wascalculated as the mean of the HRC related to allevents, calculated separately for arousal and LM. Acurve was created for each patient, displaying the av-erage HRC for each of the 25 heartbeats. By averagingindividual curves, the HRC associated with arousalsand LM was obtained for each group. Such curveswere calculated for 2 cases: events from non–REM 2sleep and events from REM sleep.

Calculation of HRR

The AUCs for the HRC curves were calculated foreach event type for each subject to obtain a measureof the HRR. Student t tests were performed to com-pare HRR/AUC between groups, and Pearson Productmoment correlations were calculated to evaluate therelationship of HRR with age, BMI, LED, HY, andUPDRS scores within groups.

PSG Variables

The following PSG variables were measured: sleeplatency (SL), total sleep time (TST), sleep efficiency(SE), time from sleep onset to first epoch of REMsleep (REM SL), percentage of stages 1, 2, 3 non-REM and REM sleep, arousal index (AI; arousals perhour of sleep), and mean duration of arousals. Toevaluate between-group differences, a log transforma-tion was performed for each variable and 2-sided ttests were calculated for each variable for each of the3 patient groups compared with the control group. ABonferroni correction was performed to address theproblem of multiple comparisons.

Results

Polysomnographic Variables

Sleep parameters are presented in Table 2. Onlythose of the PD patients without RBD were signifi-cantly different from those of the control group. TSTand SL were both significantly lower (P < .05), but

TABLE 1. Demographic data of subjects includedin the study

Controls

PD with

RBD

PD

without

RBD iRBD

Total number 17 14 16 11Female/male 7/8 3/11 8/8 2/9Age (y) 62.4 6 9.7 63.4 6 6.7 62.0 6 7.3 60.2 6 7.8BMI (kg/m2) 23.4 6 3.1 26.3 6 3.2 25.7 6 5.1 24.9 6 2.9UPDRS score — 23 6 11 23 6 10 1 6 1HY score — 1.6 6 0.8 1.8 6 0.8 0.0 6 0.0LED score — 623 6 360 680 6 552 —

Age, BMI, LED, UPDRS, and HY scores are given as mean 6 standarddeviation (SD).

S O R E N S E N E T A L .

890 Movement Disorders, Vol. 27, No. 7, 2012

there were no significant differences between any ofthe other sleep parameters given in Table 2.

Heart Rate Response to Arousals

Figure 1 shows the HRC associated with arousalsfrom non–REM 2 sleep. The curves for the PD, iRBD,and control groups all showed an acceleration of HRafter the onset of arousal. However, the HRRs for thePD groups were significantly lower than both the con-trol (P < .01) and iRBD (P < .02) groups. There wasno significant difference in the HRR of the iRBDgroup compared with the control group or betweenthe 2 PD groups.Figure 2 shows the HRC associated with arousals

from REM sleep. Two of 17 control subjects did nothave sufficient arousals during REM sleep to meritinclusion in this study. Again, all groups showed anacceleration of HR after the onset of arousal. TheHRRs for the PD groups were significantly lower than

those of the control (P < .01) and iRBD (P < .05)groups. There was no significant difference in theHRR of the iRBD group compared with that of thecontrol group or between the 2 PD groups.No significant correlations were observed between

HRR to arousals and age, BMI, LED, HY, or UPDRSwithin any of the groups. Baseline values were similarfor all groups in Figures 1 and 2. All HRR values aregiven in Table 2.

Heart Rate Response to Leg Movements

Figure 3 shows the HRC associated with LM duringnon–REM 2 sleep. As with arousals, there was anincrease in HR as a response to LM. The HRRs weresignificantly lower for both PD groups (P < .02) andfor the iRBD group (P ¼ .03) relative to controls.There was no significant difference in the HRRbetween the iRBD group and the PD groups orbetween the 2 PD groups.

TABLE 2. Sleep parameters and HRR values

Controls PD with RBD PD without RBD iRBD P valuea P valueb P valuec

TST (min) 418.0 6 48.3 372.9 6 74.1 348.5 6 59.9 449 6 79.9 NS <.05 NSSL (min) 9.8 6 8.2 13.3 6 22.9 3.7 6 3.5 8.0 6 7.1 NS <.05 NSREM SL (min) 100.4 6 50.6 139.9 6 87.4 124.4 6 85.8 109.3 6 53.4 NS NS NSSE (%) 88.2 6 7.6 80.8 6 14.1 81.5 6 6.5 84.1 6 14.3 NS NS NSStage 1 sleep (%) 10.5 6 7.8 10.4 6 6.8 9.6 6 7.1 9.3 6 5.1 NS NS NSStage 2 sleep (%) 46.7 6 11.1 48.0 6 16.5 51.7 6 17.4 48.7 6 7.4 NS NS NSStage 3 sleep (%) 20.9 6 9.6 23.9 6 18.9 23.5 6 14.3 21.1 6 4.9 NS NS NSREM sleep (%) 22.0 6 4.0 17.7 6 10.5 15.2 6 9.4 22.2 6 6.2 NS NS NSAI (events/hour) 14.5 6 6.0 12.1 6 5.4 10.5 6 4.1 14.4 6 6.4 NS NS NSArousal duration (s) 9.8 6 2.4 10.0 6 1.7 10.9 6 3.2 9.1 6 2.0 NS NS NSHRR to arousals in NREM 2 91.9 6 10.3 40.5 6 4.8 48.8 6 8.0 66.9 6 9.9 <.01 <.01 NSHRR to arousals in REM 106.0 6 14.1 24.1 6 7.8 24.8 6 17.5 69.5 6 16.5 <.01 <.01 NSHRR to LM in NREM 2 73.5 6 14.8 25.5 6 4.7 35.5 6 6.0 32.0 6 8.3 <.01 .01 .03HRR to LM in REM 61.9 6 18.2 10.8 6 5.7 15.6 6 9.7 18.4 6 10.3 .02 .04 .05

Sleep parameters are expressed as mean 6 SD, and P values are calculated from 2-sided t tests after log transformation. HRR values are expressed as mean6 SEM, and P values are calculated from 2-sided t tests.aComparison of PD with RBD and control groups.bComparison of PD without RBD and control groups.cComparison of iRBD patients and control group. NS, not significant.

FIG. 1. HRC associated with arousals from non–REM 2 sleep stages.Vertical lines indicate standard error of the mean (SEM).

FIG. 2. HRC associated with arousals from REM sleep stages. Verti-cal lines indicate SEM.

H E A R T R A T E R E S P O N S E I N P D / i R B D P A T I E N T S

Movement Disorders, Vol. 27, No. 7, 2012 891

Figure 4 shows the HRC associated with LM fromREM sleep. Two of 17 control subjects and 2 of 16PD patients without RBD did not have enough LMduring REM sleep to warrant inclusion. The HRRwas significantly lower in both PD groups (P < .05)and in the iRBD group (P ¼ .05) compared with thecontrols. There was no significant difference in theHRR between the iRBD group and the PD groups orbetween the 2 PD groups.No significant correlations were observed between

HRR to LM and age, BMI, LED, HY, or UPDRSwithin any of the groups. Baseline values were similarfor all groups in Figures 3 and 4. All HRR values aregiven in Table 2.

Discussion

PD patients have an attenuated HRR to arousalsand LM compared with healthy controls. Because PDpatients are known to have autonomic dysfunction,this attenuated HRR was expected. For the baselineHR, calculated from the prearousal period, there wasno significant difference between any of the groups.However, some of the graphs indicated that theincrease in HR started earlier for the control group,indicating that the cardiac response to arousal or LMis not only attenuated but also delayed in PD patients.The HRR to LM for the iRBD patients was signifi-

cantly lower than in the control group. Furthermore,the curves for the iRBD group were intermediatebetween those of the control and PD groups.The correlation of the HRR with LED verified that

the attenuated HRR was not caused by medication.Furthermore, no iRBD patients took any parkinsonianmedication, and yet they showed an attenuated HRRto LM.Ferri et al28 suggested that sleep EEG, HR, and

motor activity appear to be modulated by a complex,dynamically interacting system of cortical and subcort-

ical mechanisms, each of which influences the others.It is therefore not surprising that PD patients have alower HRR to LM and arousals because PD patientsare known to undergo a degenerative process in thecortical and subcortical areas.29

The attenuated HRR is consistent with the reducedHRR associated with periodic LM (PLM) experiencedby RBD patients.27 Because RBD often precedes theclinical manifestation of PD,21 the change in auto-nomic function may represent an early manifestationof a neurodegenerative process involving brain stemareas, which is consistent with the Braak hypothesis.16

We suggest that the attenuated HRR worsens fromthe onset of iRBD to the onset of PD. If the attenuatedHRR is an early manifestation of a neurodegenerativeprocess, this could be an early marker of the disease.It remains unknown, however, which patients withiRBD will develop PD or another synucleinopathy.Although some studies suggest that combinations ofpremotor symptoms like iRBD and anosmia predictPD,3 it remains uncertain whether iRBD with dysauto-nomia presents an increased risk of further neurodege-nerative development.30 It will be necessary to followiRBD patients in a prospective study to answer thesequestions. We have provided 2 scatter plots of theresults for the iRBD patients in the supplementary sec-tion, which revealed that the iRBD patients could bedivided in 2 groups. One group has a normal HRR toboth arousal and LM, and the other group has anattenuated HRR to arousal and LM, Following theseiRBD patients in a prospective study to see if these 2groups predict who will develop a neurodegenerativedisease will answer some of our questions.There were no differences in the HRR for PD with

RBD compared with PD without RBD. If this hadbeen the case, this would indicate that brain stemareas were not involved to the same extent and thusthat disease progression was different in the 2 groups.However, PSG variables showed that only the results

FIG. 3. HRC associated with LM from non–REM 2 sleep stages. Verti-cal lines indicate SEM.

FIG. 4. HRC associated with LM from REM sleep stages. Verticallines indicate SEM.

S O R E N S E N E T A L .

892 Movement Disorders, Vol. 27, No. 7, 2012

of the PD without RBD were different from controls,which implies that sleep may be affected more in thispatient group.Previous studies have established the involvement of

the autonomic system. In a study using 123I-MIBGcardiac scintigraphy, a reduced heart/mediastinum (H/M) ratio was found for iRBD, PD, and dementia ofLewy bodies patients relative to controls. However,normal values were found in multiple system atrophyand progressive supranuclear palsy.31 R–R variabili-ty32 and orthostatic testing33 during wakefulness wereaffected in iRBD and PD with/without RBD comparedwith controls, although the differences between thegroups were not significant. These data confirm ourfindings that the presence of iRBD or PD with orwithout RBD is the main predictor of autonomicdysfunction.Studies of the HRR to arousal in healthy subjects

have found that tachycardia starts approximately 2beats before arousal onset and is followed by brady-cardia before the HR returns to baseline.13 The brady-cardia was not observed in our study. In the study bySforza et al,13 the bradycardia occurred approximately10 beats after arousal onset and reached baseline at15 beats after arousal onset. However, the mean ageof the subjects in this study was 28.8 years, and com-pared with the mean age of 62.4 years of the controlgroup, tachycardia and bradycardia would beexpected to be of significantly lower amplitude, asGosselin et al demonstrated in middle-aged subjects.14

Furthermore, no constraints were placed on the maxi-mum duration of arousals used in the analysis. For allgroups, the mean duration of arousal was around 10seconds, with a minimum duration of 3 seconds, asdefined in the American Academy of Sleep Medicinemanual.24 Sforza et al13 included arousals down to aminimum duration of 1.5 seconds. This could explainwhy, in the latter study, the HRR reached baselinevalues faster after arousal onset. The HRR for thecontrol subjects of this study reached the baselinevalue between 20 and 30 heartbeats after arousalonset. Because we wished to investigate the immediateHRR to arousal, we displayed the graphs only up to15 heartbeats after onset.We chose to assess the HRR in association with LM

and not PLM, as Fantini et al27 had done, in order toensure that the HR was stable and unaffected by otherevents before the onset of LM. The HR could beslightly increased during PLM, which could give riseto an elevated baseline and thus a reduced change inHR.The analysis of sleep parameters showed significant

differences in TST and SL for the PD group withoutRBD compared with the control group. Other studieshave shown lower TST and a lower percentage ofstage 3 sleep in PD patients.1,23,34 The percentage of

stage 3 sleep was not significantly reduced in the PDpatients in this study. The AI index was not increasedfor any of the diseased groups compared with the con-trol group, although PD and iRBD patients are knownto have disrupted sleep with frequent arousals. How-ever, all groups had a lower AI than had previouslybeen noted in groups of similar age.14,35

The time course analysis over 25 beats, unlike theaverage values for 3 time periods presented by Freilichet al,15 enabled the display of the whole postarousal/post-LM period and provided a more detailed descrip-tion of the shape of the HRR to arousal/LM. TheAUC provided a simple measure for the HRR andmade the comparison of various groups simple, unlikethe many comparisons necessary with a beat-by-beatanalysis. Furthermore, this study also assessed theHRR to arousal and LM in REM sleep, rather thanonly stage 2 non-REM sleep.The attenuated HRR in patients with iRBD and PD

with/without RBD may be a manifestation of the au-tonomic deficits experienced in these patients andcould indicate a rather significant impact on multiplesystems in the brain including the ARAS and the ANS.The iRBD patients not only exhibited impaired motorfunction with REM sleep without atonia but also in-cipient autonomic dysfunction, revealed in the attenu-ated HRR to LM. Further studies should seek tovalidate the findings compared with other autonomictests, and it will be necessary to evaluate the predictivevalue of autonomic (dys)function for predicting devel-opment of a neurodegenerative disease in iRBD.

References1. Maetzler W, Liepelt I, Berg D. Progression of Parkinson’s disease

in the clinical phase: potential markers. Lancet Neurol. 2008;8:1158–1171.

2. Chaudhuri KR, Healy DG, Schapira AHV. Non-motor symptomsof Parkinson’s disease: diagnosis and management. Lancet Neurol.2006;5:235–245.

3. Postuma RB, Montplaisir J. Predicting Parkinson’s disease–why,when, and how?. Parkinsonism Relat Disord. 2009;15(Suppl 3):S105–S109.

4. Akincioglu C, Unlu M, Tunc T. Cardiac innervation and clinicalcorrelates in idiopathic Parkinson’s disease. Nucl Med Commun.2003;24:267–271.

5. Barbic F, Perego F, Canesi M, et al. Early abnormalities of vascu-lar and cardiac autonomic control in Parkinson’s disease withoutorthostatic hypotension. Hypertension. 2007;49:120–126.

6. Haapaniemi TH, Pursiainen V, Korpelainen JT, Huikuri HV, Sota-niemi KA. Ambulatory ECG and analysis of heart rate variability inParkinson’s disease. J Neurol Neurosurg Psychiatry. 2001;70:305–310.

7. Kallio M, Haapaniemi T, Turkka J, et al. Heart rate variability inpatients with untreated Parkinson’s disease. Eur J Neurol. 2000;7:667–672.

8. Ludin SM, Steiger UH, Ludin HP. Autonomic disturbances andcardiovascular reflexes in idiopathic Parkinson’s disease. J Neurol.1987;235:10–15.

9. Mesec A, Sega S, Trost M. The deterioration of cardiovascular reflexesin Parkinson’s disease. Acta Neurol Scand. 1999;100:296–299.

10. Saper CB, Scammel TE, LU J. Hypothalamic regulation of sleepand circadian rhythms. Nature. 2005;437:1257–1263.

11. Saper CB, Fuller PM, Pedersen NP, LU J, Scammell TE. Sleep stateswitching. Neuron. 2010;68:1023–1042.

H E A R T R A T E R E S P O N S E I N P D / i R B D P A T I E N T S

Movement Disorders, Vol. 27, No. 7, 2012 893

12. Saper CB, Chou TC, Scammell TE. The sleep switch: hypothalamiccontrol of sleep and wakefulness. Trends Neurosci. 2001;24:726–731.

13. Sforza E, Jouny C, Ibanez V. Cardiac activation during arousal inhumans: further evidence for hierarchy in the arousal response.Clin Neurophysiol. 2000;111:1611–1619.

14. Gosselin N, Michaud M, Carrier J, Lavigne G, Montplaisir J. Agedifference in heart rate changes associated with micro-arousals inhumans. Clin. Neurophysiol. 2002;113:1517–1521.

15. Freilich S, Goff EA, Malaweera ASN, Twigg GL, Simonds AK,Mathias CJ, Morrell MJ. Sleep architecture and attenuated heartrate response to arousal from sleep in patients with autonomic fail-ure. Sleep Med. 2010;11:87–92.

16. Braak H, Tredici KD, Rub U, de Vos RAI, Jansen Steur ENH,Braak E. Staging of brain pathology related to sporadic Parkinson’sdisease. Neurobiol Aging. 2003;24:197–211.

17. Schenck CH, Bundlie SR, Mahowald MW. Delayed emergence ofa parkinsonian disorder in 38 percentage of 29 older men initiallydiagnosed with idiopathic rapid eye movement sleep behavior dis-order. Neurology. 1996;46:388–393.

18. Schenck CH, Bundlie SR, Mahowald MW. REM sleep behaviourdisorder (RBD) delayed emergence of parkinsonism and/or demen-tia in 65 percent of older men initially diagnosed with idiopathicRBD, and an analysis of the maximum and minimum tonic and/orphasic electromyographic abnormalities found during REM sleep.Sleep. 2003;26(Suppl):A316.

19. Iranzo A, Molinuevo JL, Santamarıa J, Serradell M, Martı MJ,Valldeoriola F, Tolosa E. Rapid-eye-movement sleep behaviour dis-order as an early marker for a neurodegenerative disorder: a de-scriptive study. Lancet Neurol. 2006;5:572–577.

20. Iranzo A, Santamarıa J, Serradell M, Marti MJ, Valldeoriola F.Sixty-four percent of patients with idiopathic REM sleep behaviordisorder developed a neurological disorder after a mean clinicalfollow-up of seven years. Sleep. 2008;31(Suppl):A280.

21. Postuma RB, Gagnon JF, Vendette M, Fantini ML, Massicotte-Marquez J, Montplaisir J. Quantifying the risk of neurodegenera-tive disease in idiopathic REM sleep behavior disorder. Neurology.2009;72:1296–1300.

22. Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke CE. Sys-tematic review of levodopa dose equivalency reporting in Parkin-son’s disease. Mov Disord. 2010;25:2649–2653.

23. American Academy of Sleep Medicine. The International Classifica-tion of Sleep Disorders, Revised.Diagnostic and Coding Manual. 1sted. Westchester, IL: American Academy of Sleep Medicine; 2001.

24. Iber C, Ancoli-Israel S, Chesson AL, Quan SF. The AASM Manualfor the Scoring of Sleep and Associated Events: Rules, Terminologyand Technical Specifications. 1st ed. Westchester, IL: AmericanAcademy of Sleep Medicine; 2007.

25. Bonnet MH, Carley D, Guilleminault CG, et al. EEG arousals:scoring rules and examples: a preliminary report from the SleepDisorders Atlas Task Force of the American Sleep Disorders Asso-ciation. Sleep. 1992;15:173–184.

26. Pan J, Tompkins WJ. A real-time QRS detection algorithm. IEEETrans Biomed Eng. 1985;32:230–236.

27. Fantini ML, Michaud M, Gosselin N, Lavigne G, Montplaisir J. Per-iodic leg movements in REM sleep behavior disorder and related au-tonomic and EEG activation. Neurology. 2002;59:1889–1894.

28. Ferri R, Zucconi M, Rundo F, Spruyt K, Manconi M, Ferini-Strambi L. Heart rate and spectral EEG changes accompanyingperiodic and non-periodic leg movements during sleep. Clin Neuro-physiol. 2007;118:438–448.

29. Yanagisawa N. Natural history of Parkinson’s disease: From dopa-mine to multiple system involvement. Parkinsonism Relat Disord.2006;12:S40–S46.

30. Postuma RB, Gagnon JF, Montplaisir J. Clinical prediction of Par-kinson’s disease: planning for the age of neuroprotection. J NeurolNeurosurg Psychiatry. 2010;81:1008–1013.

31. Miyamoto T, Miyamoto M, Suzuki K, Nishibayashi M, IwanamiM, Hirata K. 123I-MIBG cardiac scintigraphy provides clues to theunderlying neurodegenerative disorder in idiopathic REM sleepbehavior disorder. Sleep. 2008;31:717–723.

32. Postuma RB, Montplaisir J, Lanfranchi P, et al. Cardiac autonomicdenervation in Parkinson’s disease is linked to REM sleep behaviordisorder. Mov Disord. 2011;26:1529–1533.

33. Postuma RB, Gagnon JF, Vendette M, Montplaisir JY. Markers ofneurodegeneration in idiopathic rapid eye movement sleep behav-iour disorder and Parkinson’s disease. Brain. 2009;132:3298–3307.

34. Petit D, Gagnona JF, Fantinia ML, Ferini-Strambi L, MontplaisirJ. Sleep and quantitative EEG in neurodegenerative disorders. JPsychosomatic Res. 2004;56:487–496.

35. Bonnet MH, Arand DL. EEG arousal aorms by age. J Clin SleepMed. 2007;3:271–274.

S O R E N S E N E T A L .

894 Movement Disorders, Vol. 27, No. 7, 2012

Appendix B

Paper 2

Title: Attenuated Heart Rate Response is Associated with Hypocretin Deficiency in Pa-tients with Narcolepsy

Status: Published in SLEEP

Citation: Sorensen G.L., Knudsen S., Petersen E.R., Kempfner J., Gammeltoft S.,Sorensen H.B.D., Jennum P. Attenuated heart rate response is associated with hypocre-tin deficiency in patients with narcolepsy. SLEEP 2013;36(1):91-98.

67

SLEEP, Vol. 36, No. 1, 2013 91 Heart Rate Response in Narcolepsy—Sorensen et al

INTRODUCTIONNarcolepsy is a neurologic disease affecting one in 2,000

individuals. It is characterized by excessive daytime sleepi-ness and early onset of rapid eye movement (REM) sleep. The disease is closely associated with low levels of hypocretin-1 (hcrt-1) in the cerebrospinal fluid (CSF), especially in the two-thirds of patients with cataplexy (muscle weakness triggered by emotions).1 The neuropeptides hcrt-1 and hcrt-2 are produced and co-released by neurons in the lateral posterior hypothala-mus of the brain, which project widely within the central ner-vous system (CNS).2 Animal studies have suggested that hcrt-1 may affect neuronal circuits that control the cardiovascular sys-tem,3-6 as exemplified by intrathecal hypocretin administration resulting in increased blood pressure (BP) and heart rate (HR) mediated by enhancement of the sympathetic output.4,6

Two smaller human studies of patients with hcrt-1-deficient narcolepsy with cataplexy (NC) found contradictory results of sympathetic tone during daytime wakefulness.7,8 However, dur-ing sleep, reduced HR and sympathetic tone were found in NC patients of unknown hcrt-1 status,9,10 supporting the results of the animal studies.

HEART RATE RESPONSE AND HYPOCRETIN DEFICIENCY IN NARCOLEPSYhttp://dx.doi.org/10.5665/sleep.2308

Attenuated Heart Rate Response is Associated with Hypocretin Deficiency in Patients with NarcolepsyGertrud Laura Sorensen, MScE1-3; Stine Knudsen, MD, PhD1; Eva Rosa Petersen, MD1; Jacob Kempfner, MScE1-3; Steen Gammeltoft, MD, Dr. Med Sci.4; Helge Bjarup Dissing Sorensen, MScE, PhD3; Poul Jennum, MD, Dr. Med. Sci.1,2

1Danish Center for Sleep Medicine, Glostrup University Hospital, Glostrup Hospital, Denmark; 2Center for Healthy Aging, University of Copenhagen, Copenhagen, Denmark; 3Department of Electrical Engineering, Technical University of Denmark, Kongens Lyngby, Denmark; 4Department of Diagnostics, Glostrup University Hospital, Denmark

A commentary on this article appears in this issue on page 9.

Submitted for publication January, 2012Submitted in final revised form July, 2012Accepted for publication August, 2012Address correspondence to: Gertrud Laura Sorensen, Danish Center for Sleep Medicine, Glostrup University Hospital, DK-2600 Glostrup, Den-mark; Tel: +45 61662828; Fax +45 45880117; E-mail: gertrud_l@yahoo.dk

Study Objective: Several studies have suggested that hypocretin-1 may influence the cerebral control of the cardiovascular system. We analyzed whether hypocretin-1 deficiency in narcolepsy patients may result in a reduced heart rate response. Design: We analyzed the heart rate response during various sleep stages from a 1-night polysomnography in patients with narcolepsy and healthy controls. The narcolepsy group was subdivided by the presence of +/- cataplexy and +/- hypocretin-1 deficiency.Setting: Sleep laboratory studies conducted from 2001-2011.Participants: In total 67 narcolepsy patients and 22 control subjects were included in the study. Cataplexy was present in 46 patients and hypo-cretin-1 deficiency in 38 patients.Interventions: None.Measurements and Results: All patients with narcolepsy had a significantly reduced heart rate response associated with arousals and leg move-ments (P < 0.05). Heart rate response associated with arousals was significantly lower in the hypocretin-1 deficiency and cataplexy groups com-pared with patients with normal hypocretin-1 levels (P < 0.04) and patients without cataplexy (P < 0.04). Only hypocretin-1 deficiency significantly predicted the heart rate response associated with arousals in both REM and non-REM in a multivariate linear regression.Conclusions: Our results show that autonomic dysfunction is part of the narcoleptic phenotype, and that hypocretin-1 deficiency is the primary predictor of this dysfunction. This finding suggests that the hypocretin system participates in the modulation of cardiovascular function at rest.Keywords: Arousal, autonomic dysfunction, heart rate response, hypocretin-1, narcolepsyCitation: Sorensen GL; Knudsen S; Petersen ER; Kempfner J; Gammeltoft S; Sorensen HBD; Jennum P. Attenuated heart rate response is as-sociated with hypocretin deficiency in patients with narcolepsy. SLEEP 2013;36(1):91-98.

We speculated that hcrt-1 deficiency is the main predictor/cause of autonomic dysfunction during sleep in patients with narcolepsy. We therefore measured the HR response at night in REM sleep and non-REM sleep in patients with narcolepsy +/- hcrt-1 deficiency and +/- cataplexy compared with healthy, age-matched controls.

METHOD

SubjectsOver a 10-year period (2001-2011), patients with narco-

lepsy seen at the Danish Center for Sleep Medicine, Glos-trup University Hospital, were identified and consecutively included after ethical approval (KA03119) and written in-formed consent had been obtained. All patients fulfilled the International Classification of Sleep Disorders (ICSD-2) criteria for narcolepsy11 and were evaluated by neurologic examination, determination of routine blood characteristics, polysomnography (PSG), the multiple sleep latency test, and determination of CSF hcrt-1. The hypocretin measurement protocol and definition of hcrt-1 deficiency used were as pre-viously published.1 Exclusion criteria were: additional neuro-logic, psychiatric, and cardiovascular disorders, and an apnea index > 5/h. All patients (except three with severe cataplexy) were free of antidepressants and stimulants 7-14 days before inclusion; 44 of 67 were completely drug-naïve (i.e., they had never taken medication).

In total, 81 patients fulfilled the inclusion criteria, although because HR response is known to be age-related,12 14 patients were excluded to enable accurate age-matching of the narcolepsy

SLEEP, Vol. 36, No. 1, 2013 92 Heart Rate Response in Narcolepsy—Sorensen et al

groups. Of the excluded group, 12 of 14 had cataplexy and 10 of 13 had hcrt-1 deficiency. Thus, 67 patients with narcolepsy were included in the study, along with 22 healthy controls matched for age, sex, and body mass index (BMI). The control subjects were recruited by advertising for normal volunteers. None of the control subjects had any history of neurologic or sleep disorders, and their PSGs were evaluated by a neurologist as being normal.

Cataplexy was present in 46 of 67 patients and hcrt-1 defi-ciency was present in 38 of 61 patients. Six patients (five of six with cataplexy) refused lumbar puncture. Normal hcrt-1 levels were present in nine of 46 patients with cataplexy, and hcrt-1 deficiency was present in five of 21 patients without cataplexy.

Demographic data of patient and control subjects are sum-marized in Table 1.

PSG RecordingsAll subjects underwent 1 night of PSG recording consisting

of electroencephalography (C3-A2, C4-A1), vertical and hori-zontal electrooculography (EOG), surface electromyography (EMG) of the submentalis and anterior muscles, electrocardi-ography (ECG), nasal air flow, thoracic respiratory effort, and oxygen saturation. Subjects were instructed not to consume any caffeinated or alcoholic drinks for at least 6 hours before the recordings were made.

Sleep stages, leg movements (LM), and arousals from sleep were manually scored according to standard criteria.13,14 The raw sleep data, hypnograms, and sleep events were extracted from Nervus® (V5.5, Cephalon DK, Nørresundby, Denmark) or Somnologica Studio® (V5.1, Embla, Broomfield, CO USA), using the built-in export data tool, and imported to the Mat-lab® analysis program (R2009b, The MathWorks, Natick, MA, USA) for further processing.

Data AnalysisThe HR response was estimated as the area under the curve

(AUC) describing the HR change associated with arousals and the HR change associated with isolated LM (separated from arousals). For each sleep event (arousal or LM), a signal seg-ment was extracted from the ECG signal and QRS complexes were automatically detected using a self-implemented version of the Pan-Tompkins QRS-detector15 in Matlab. The intervals between QRS complexes (RR intervals) were calculated as the interbeat interval between successive R waves.

Before the HR was calculated, a manual noise analysis ex-cluded any ECG signal segments contaminated with too much noise, because the detection of QRS complexes (and R waves) in these signal segments was impossible to validate. Therefore,

not all scored arousals and LMs were included in the calcula-tion of the HR response.

For every selected sleep event, the HR was analyzed for 10 heartbeats before and for 15 heartbeats after the onset of the event. Previous studies of HR response to arousal or motor ac-tivity had shown that the HR started to rise two heartbeats before the event.16,17 Therefore, a baseline value of HR was calculated by averaging the first seven beats of the 10 before the event. To determine the change in HR, this baseline value was subtracted from every HR from 10 beats before to 15 beats after the event.

For each subject, the change in HR was calculated for every event (separately for arousals and LM), and the average HR change was calculated from the mean of the 25 heartbeats re-lated to each event. Thus, a curve was created displaying the HR change for each beat before and after the onset of the event, and the HR response was calculated as the AUC. By averag-ing individual HR responses, the HR response associated with arousals (HRRA) and the HR response associated with LM (HRRL) were obtained for each group. This was calculated both for events from all non-REM 2 sleep stages and events solely from REM sleep stages. Analysis of HR in non-REM 1 and 3 sleep stages was not possible in most subjects due to the insufficient amount of arousals/LM, and these were therefore omitted from this study.

Statistical AnalysisTo compare the baseline HR (the average HR of the first

seven of the 10 beats before event onset) and the HR response between groups, t-tests were performed. Univariate and back-ward stepwise multivariate linear regression were performed to determine the relationship between HR response and age, sex, BMI, disease-duration, disease-onset, +/- previous anticata-plexy medication, +/- cataplexy, and +/- hcrt-1 status (Table 2). The +/- anticataplexy medication indicates whether patients were pausing with medication or had never been taking medi-cation before inclusion (drug-naïve).

PSG variables were measured for each subject. These in-cluded: sleep latency (SL), total sleep time (TST), sleep ef-ficiency (SE), time from sleep onset to first epoch of REM sleep (REM SL), and percentage of stages 1, 2, and 3 non-REM and REM sleep. To evaluate between-group differences, each variable was log-transformed and two-sided t-tests were calculated for each variable for each of the three patient groups compared with the control group (Table 3). A Bonferroni cor-rection was performed to take into account the problem of multiple comparisons.

A significance level of 0.05 is used in all statistical tests.

Table 1—Demographic data of study patientsa

Controls NC NwC +hcrt-1 -hcrt-1Number 22 46 21 23 38Female/Male 16/6 20/26 9/12 10/13 22/16Age (year) 32.2 ± 8.4 31.9 ± 7.9 32.2 ± 8.9 31.4 ± 9.3 32.5 ± 7.9BMI (kg/m2) 24.2 ± 4.3 26.3 ± 6.0 25.4 ± 4.1 25.4 ± 5.0 24.9 ± 4.0

aAge and BMI are given as the mean ± the standard deviation (SD). BMI, body mass index; +hcrt-1 = narcolepsy patients with normal hcrt-1. -hcrt-1 = narcolepsy patients with hcrt-1 deficiency; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy.

SLEEP, Vol. 36, No. 1, 2013 93 Heart Rate Response in Narcolepsy—Sorensen et al

RESULTS

PSG VariablesThe sleep parameters are presented in Table 3. All narcolep-

sy groups had a lower percentage stage 2 non-REM sleep com-pared with the control group. NC and narcolepsy with hcrt-1 deficiency groups had a higher percentage stage 1 sleep com-pared with the control group, patients with narcolepsy without cataplexy (NwC), and those with normal hcrt-1. Furthermore, the REM SL was significantly lower for all patients with narco-lepsy than for the controls, except for narcolepsy with normal hcrt-1. SL was only significantly lower for the narcolepsy pa-tients with hcrt-1 deficiency.

Heart Rate Response to ArousalsFigures 1 and 2 show the mean HRRA from non-REM sleep

stage 2 and REM sleep, respectively. The vertical lines indicate the standard error of the mean (SEM). The HR after arousal onset was accelerated in a similar way for patients with nar-colepsy and controls, although to a lesser degree for the nar-colepsy groups. The HRRA was significantly reduced for NC (P < 0.02) and for patients with narcolepsy with hcrt-1 defi-ciency (P < 0.01) compared with the controls in both REM and non-REM 2 sleep. Furthermore, in REM and non-REM 2 sleep, patients with cataplexy and patients with hcrt-1 deficiency had significantly lower HRRA than did those with NwC (P < 0.04) and narcolepsy with normal hcrt-1 (P < 0.04). Likewise, the

Table 3—Comparison of PSG variablesa

Controls NC NwC +hcrt-1 -hcrt-1 P valueTST (min) 429.9 ± 52.1 392.8 ± 77.7 427.0 ± 48.7 415.1 ± 69.9 399.5 ± 77.4 NSSL (min) 26.2 ± 33.2 9.6 ± 19.0 7.4 ± 10.9 10.9 ± 14.0 8.4 ± 19.4 < 0.05e

REM SL (min) 123.8 ± 98.0 81.1 ± 153.4 57.8 ± 34.2 102.8 ± 204.0 53.5 ± 54.0 < 0.05b,c,e

SE (%) 87.0 ± 16.5 86.6 ± 11.7 92.6 ± 5.0 87.2 ± 15.6 88.6 ± 7.5 NSStage 1 sleep (%) 5.3 ± 3.2 8.9 ± 6.4 4.9 ± 4.5 4.1 ± 3.1 9.6 ± 6.7 < 0.05b,e-g

Stage 2 sleep (%) 50.3 ± 7.0 39.4 ± 13.3 43.2 ± 10.7 39.3 ± 13.3 39.9 ± 11.9 < 0.05b-e

Stage 3 sleep (%) 23.0 ± 8.0 29.7 ± 14.3 29.2 ± 9.4 34.4 ± 13.1 26.6 ± 11.9 NSREM sleep (%) 22.2 ± 4.1 22.8 ± 5.9 23.4 ± 5.5 22.0 ± 5.4 23.7 ± 5.5 NSArousal per hr 8.8 ± 5.2 9.6 ± 6.8 5.8 ± 2.9 6.3 ± 2.8 9.7 ± 7.2 NSLM per hr 9.7 ± 9.7 22.2 ± 21.3 13.3 ± 9.8 14.1 ± 13.3 23.6 ± 21.5 <0.05b,e

aData are expressed as mean ± SD and the P values are calculated from two-sided t-tests after log-transformation. bNC vs. controls. cNwC vs. controls. d+hcrt-1 vs. controls. e-hcrt-1 vs. controls. fNC vs. NwC. g+hcrt-1 vs. -hcrt-1. +hcrt-1, narcolepsy patients with normal hcrt-1; -hcrt-1, narcolepsy patients with hcrt-1 deficiency; LM, leg movement; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy; NS, not significant; REM, rapid eye movement; SE, sleep efficiency; SL, sleep latency; TST, total sleep time.

Table 2—Univariate and multivariate linear regression analysis of HR response and demographic variables

Univariate analysis Multivariate analysisCoefficient (95% CI) P value Coefficient (95% CI) P value

HR response to arousal from NREM 2 sleepAge, per year -0.989 (-2.313 – 0.335) 0.14Sex, male -6.269 (-28.41 – 15.87) 0.57BMI (kg/m2) -0.325 (-2.297 – 1.646) 0.74Disease duration, per year -0.924 (-2.130 – 0.281) 0.13Disease onset, per year 0.205 (-1.115 – 1.525) 0.75Previous anticataplexy medication -1.345 (-21.81 – 19.12) 0.89Cataplexy, yes -13.846 (-37.58 – 9.891) 0.28Low HCRT-1, yes 0.071 (0.013 – 0.129) 0.01 0.072 (0.011 – 0.132) 0.02

HR response to arousal from REM sleepAge, per year -0.351 (-1.937 – 1.234) 0.65Sex, male -2.425 (-28.54 – 23.69) 0.85BMI (kg/m2) -1.837 (-4.101 – 0.427) 0.11Disease duration, per year -1.109 (-2.545 – 0.328) 0.12 -1.590 (-2.928 – -0.253) 0.02Disease onset, per year 0.910 (-0.599 – 2.419) 0.23Previous anticataplexy medication -14.782 (-37.78 – 8.217) 0.20Cataplexy, yes -45.543 (-71.23 – -19.85) < 0.01 -35.65 (-63.90 – -7.405 0.01Low HCRT-1, yes 0.071 (0.0134 – 0.130) < 0.001 0.091 (0.025 – 0.157) < 0.01

BMI, body mass index; CI, confidence interval; HCRT, hypocretin; HR, heart rate; NREM, nonrapid eye movement; REM, rapid eye movement.

SLEEP, Vol. 36, No. 1, 2013 94 Heart Rate Response in Narcolepsy—Sorensen et al

Figure 1—Heart rate response associated with arousals from non-rapid eye movement 2 sleep stages. +hcrt-1, patients with narcolepsy with normal hcrt-1; -hcrt-1, patients with narcolepsy with hcrt-1 deficiency; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy; RR-intervals, intervals between successive QRS complexes.

Figure 2—Heart rate response associated with arousals from rapid eye movement (REM) sleep stages. +hcrt-1, patients with narcolepsy with normal hcrt-1; -hcrt-1, patients with narcolepsy with hcrt-1 deficiency; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy; RR-intervals, intervals between successive QRS complexes.

SLEEP, Vol. 36, No. 1, 2013 95 Heart Rate Response in Narcolepsy—Sorensen et al

HRRA for the NwC patients and the patients with narcolepsy with normal hcrt-1 had a lower, although non-significant HRRA compared with controls (P = 0.10 and P = 0.09, respectively), placing them between the control group and NC and narcolepsy with hcrt-1 deficiency groups. The baseline HR was significant-ly (P < 0.05) elevated for all narcolepsy groups except NwC patients compared with the controls. There was no difference in the baseline HR within the narcolepsy group.

Table 2 shows the results for the univariate and multivariate backward stepwise linear regression analysis. In the univari-ate model, hcrt-1 deficiency was the only predictor of HRRA in non-REM 2 sleep stages (P = 0.01), whereas both cataplexy (P < 0.01) and hcrt-1 deficiency (P < 0.001) were predictors of the HRRA in REM sleep stages. In the multivariate analysis, the HRRA in REM sleep was predicted by disease duration (P = 0.02), hcrt-1 deficiency (P < 0.01), and cataplexy status (P = 0.01), but only by hcrt-1 deficiency in non-REM 2 sleep (P = 0.02).

Heart Rate Response to LMFigures 3 and 4 show the HRRL during non-REM 2 sleep

and REM sleep, respectively. The vertical lines indicate the SEM. Four of 67 patients with narcolepsy (three of four with cataplexy and two of three with hcrt-1 deficiency; CSF not available in one of four) were excluded from further analysis due to noisy EMG signals. Furthermore, eight of 22 controls and 10 of 63 patients (seven with and three without cataplexy) with narcolepsy were excluded from the REM analysis due to insufficient numbers of LM during REM sleep.

With respect to HRRA, patients with narcolepsy and con-trols both had an increased HRRL, but to a significantly lesser degree in the narcolepsy groups (P < 0.05). However, unlike the HRRA, the HRRL did not differ within the narcolepsy groups, when the cataplexy and hcrt-1 groups were analyzed separately. This was confirmed by the univariate and multivariate linear regression analysis, in which none of the parameters was sig-nificant (data not shown). The baseline HR was significantly (P < 0.05) elevated for all narcolepsy groups, except the NwC group, compared with the controls. There was no difference in baseline HR within the narcolepsy groups.

DISCUSSIONThis study shows for the first time that a reduced autonom-

ic response to arousals in patients with narcolepsy is primar-ily predicted by hcrt-1 deficiency in both REM and non-REM sleep, independent of cataplexy and other factors. The data con-firm that hcrt-1 deficiency has a great effect on the autonomic nervous system of patients with narcolepsy and that the hypo-cretin system is important for proper HR modulation at rest.

Cataplexy and hcrt-1 status both significantly predicted HRRA in REM sleep, although only hcrt-1 status remained significant in non-REM 2 sleep. None of the possible biasing factors (age, sex, BMI, disease duration, disease onset, anti-cata-plexy medication) predicted HRRA in both non-REM and REM.

Our results are in accordance with the findings of reduced HR response associated with periodic limb movement (PLM) with and without arousals during non-REM 2 sleep in 14 HLA-DQB1*0602-positive (and presumed hcrt-1-deficient) NC pa-

Figure 3—Heart rate response associated with leg movement (LM) from nonrapid eye movement (non-REM) 2 sleep stages. +hcrt-1, patients with narcolepsy with normal hcrt-1; -hcrt-1, patients with narcolepsy with hcrt-1 deficiency; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy; RR-intervals, intervals between successive QRS complexes.

SLEEP, Vol. 36, No. 1, 2013 96 Heart Rate Response in Narcolepsy—Sorensen et al

tients.9 The Dauvilliers study design, which is comparable to ours, analyzed the HR response beat by beat only for PLM with and without arousals during non-REM 2 sleep, and found a significantly lower amplitude of tachycardia and bradycardia after onset of PLM, although the hcrt-1 level of the patients was not known, but they were presumed hcrt-1 deficient.9 In our study we used LM instead of PLM to ensure that the HR was stable and unaffected by other events before the onset of LM. Furthermore, in 18/22 control subjects, not enough (or even none) PLM were available for a valid statistical analysais. In addition, in some of the patients with narcolepsy not enough PLM were available.

Although the HRRL did not differ within the narcolepsy groups, the HRRL was significantly reduced compared with the control group both in REM and non-REM 2 sleep. A previous study has found hypocretin deficiency to be independently associ-ated with the prevalence of REM sleep behavior disorder (RBD) during REM sleep in narcolepsy.18 Therefore, the LM in REM sleep may reflect underlying RBD in the narcolepsy patients.

Three studies have investigated the heart rate variability (HRV) in patients with narcolepsy using power spectrum anal-ysis (PSA) and obtained very different results.7,8,10 Fronczek et al.7 found a higher power in all frequency bands in 15 hcrt-de-ficient NC patients. In contrast, normal cardiovascular changes during head-up tilt tests, the Valsalva manoeuvre, deep breath-ing, isometric handgrip, and cold face tests, but an increased sympathetic drive on HR at supine rest was found in 10 NC patients (nine of 10 with hcrt-1 deficiency),8 and an increase in HRV and sympathetic activity using PSA was found in NC

patients of unknown hcrt-1 status,10 indicating normal and in-creased sympathetic tones, respectively. The latter study also reported an increased low/high frequency ratio during wakeful-ness before sleep, suggesting an altered circadian autonomic function in patients with narcolepsy patients. Although the study by Ferini-Strambi et al.10 was performed with sleeping patients with NC, the results are difficult to compare with ours because of the different study designs.

The great variability of these results may be due to the differ-ent study designs (day/night and measurement methods), small populations, and inadequate knowledge of the hcrt-1 level. Furthermore, as hypothesized by Grimaldi et al.,8 patients with narcolepsy may be highly vigilant during the daytime, when they are continuously fighting against sleepiness, which might imply stronger daytime sympathetic activation and thereby ex-plain the different results obtained in previous studies. A single autonomic test during wakefulness may be insufficient in pa-tients with narcolepsy.

Our results of intermediate low HRRA in narcolepsy with normal hcrt-1 indicate that this group could resemble a partially hcrt-1-deficient phenotype, as found in a postmortem study.19

Hypocretin neurons and receptors have been proposed as providing a link between central mechanisms that regulate arousal and sleep–wakefulness states and central control of au-tonomic functions,20 including parasympathetic cardiac activ-ity.21 Hypocretin fibers have been found in nuclei that are well known to be involved in cardiovascular regulation, and it has been shown that hcrt-1 diminishes parasympathetic cardiac ac-tivity and thereby contributes to HR acceleration.21 Also, hypo-

Figure 4—Heart rate response associated with leg movement (LM) from rapid eye movement sleep stages. +hcrt-1, patients with narcolepsy with normal hcrt-1; -hcrt-1, patients with narcolepsy with hcrt-1 deficiency; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy; RR-intervals, intervals between successive QRS complexes.

SLEEP, Vol. 36, No. 1, 2013 97 Heart Rate Response in Narcolepsy—Sorensen et al

cretin acting on neurons in the hypothalamic paraventricular nucleus is known to increase cardiovascular response.22

Chronic lack of hypocretin signalling may have consequenc-es for cardiovascular function, which may be revealed in the HR and BP.23 However, hcrt-1-deficient patients with narcolepsy in our study not only had an elevated baseline HR but also failed to increase their HR in response to arousals or LM. This may be explained by two separate mechanisms: patients with hypocre-tin deficiency had increased arousability, sleep stage shift and daytime somnolence, which may affect autonomic activity; and the hypocretin system may directly influence autonomic activi-ty on its own. The attenuated HR response may be explained by the already elevated HR in these patients, their impaired hypo-cretin signalling, or by both factors. Hcrt-1 fibers have been found in the lateral paragigantocellular nucleus, which elicits an inhibitory pathway to preganglionic cardiac vagal neurons.21 The hcrt-1 deficiency may cause this pathway to be impaired and parasympathetic activity would explain the attenuated HR response. This further supports the hypothesis that the normal hcrt-1 group could be a partially hcrt-1-deficient phenotype. These patients therefore also show an attenuated HR response due to impaired hypocretin signalling. To date, there have been few post-mortem studies of hypocretin-deficient patients,19,24 and currently there is no evidence of neurodegeneration of sys-tems other than the hypocretin neurons.

Consequently, we believe that the changes in autonomic function observed here reflect stage-dependent dynamic chang-es in autonomic activity, which could explain the differences between the findings of our study and those of others evaluating autonomic activity during wakefulness. This is in contrast to the abnormalities observed in Parkinson’s Disease (PD) patients who also show attenuated HRRA/HRRL,25 but this is probably due to the destruction of neurons and nuclei in multiple brain areas involving the autonomic systems.26

The limitation of the study is that the autonomic test used herein has not previously been validated by comparison with other autonomic tests. HR in wakefulness and sleep is con-trolled by sympathetic and parasympathetic activity in mutual balance. The autonomic function is primarily tested with the head-up tilt test, the Valsalva maneuver, deep breathing, etc. During arousal there is likely an activation of the sympathetic system due to HR increase. Our study was performed during sleep, ensuring that natural conditions were present for studying neural modulation of the cardiovascular system. Future studies imply comparison of arousal and LM HR activation with other autonomic tests.

The AUC used in the current study is a simple, robust mea-sure of HR response that allows a straightforward comparison of various groups instead of a beat-by-beat analysis with many comparisons. The AUC also allowed us to do univariate and multivariate regression analyses that took into account the in-fluence of several confounder variables.

Future analysis could include comparisons of HR response from patients with narcolepsy with hcrt-1 deficiency and with-out cataplexy with that of patients with narcolepsy with normal hcrt-1 and with cataplexy.

We conclude that patients with narcolepsy present autonom-ic deficits from arousals and locomotor activity during REM and non-REM 2 sleep, which indicates that there is a stage-

dependent effect on the autonomic system. The attenuated HR response was greatest for patients with narcolepsy with hcrt-1 deficiency, which suggests that the hypocretinergic system plays an important role in the autonomic tone and modulation at rest.

DISCLOSURE STATEMENTThis was not an industry supported study. The authors have

indicated no financial conflicts of interest.

REFERENCES1. Knudsen S, Jennum P, Alving J, Sheikh SP, Gammeltoft S. Validation of

the ICSD-2 Criteria for CSF hypocretin-1 measurements in the diagnosis of narcolepsy in the Danish population. Sleep 2010;33:169-76.

2. Peyron C, Tighe DK, van den Pol AN, et al. Neurons containing hypocretin (orexin) project to multiple neuronal systems. J Neurosci 1998;18:9996-10015.

3. Ciriello J, de Oliveira CVR. Cardiac effects of hypocretin-1 in nucleus ambiguus. Am J Physiol Regul Integr Comp Physiol 2003;284:R1611-20.

4. de Oliveira CVR, Rosas-Arellano MP, Solano-Flores LP, Ciriello J. Car-diovascular effects of hypocretin-1 in nucleus of the solitary tract. Am J Physiol Heart Circ Physiol 2003; 284:H1369–77.

5. Samson WK, Gosnell B, Chang JK, Resch ZT, Murphy TC. Cardiovas-cular regulatory actions of the hypocretins in brain. Brain Res 1999; 831:248–53.

6. Shirasaka T, Nakazato M, Matsukura S, Takasaki M, Kannan H. Sym-pathetic and cardiovascular actions of orexins in conscious rats. Am J Physiol 1999;277:R1780–5.

7. Fronczek R, Overeem S, Reijntjes R, Lammers GJ, van Dijk JG, Pijl H. Increased heart rate variability but normal resting metabolic rate in hypocretin/orexin-deficient human narcolepsy. J Clin Sleep Med 2008;4:248-54.

8. Grimaldi D, Pierangeli G, Barletta G, et al. Spectral analysis of heart rate variability reveals an enhanced sympathetic activity in narcolepsy with cataplexy. Clin Neurophys 2010;121:1142-7.

9. Dauvilliers Y, Pennestri MH, Whittom S, Lanfranchi PA, Montplaisir JY. Autonomic response to periodic leg movements during sleep in narcolep-sy-cataplexy. Sleep 2011;34:219-23.

10. Ferini-Strambi L, Spera A, Oldani A, et al. Autonomic function in nar-colepsy: power spectrum analysis of heart rate variability. J Neurol 1997;244:252-5.

11. The International Classification of Sleep Disorders (ICSD)-Diagnostic and Coding Manual. 2nd ed. Westchester, IL: American Academy of Sleep Medicine, 2005.

12. Gosselin N, Michaud M, Carrier J, Lavigne G, Montplaisir J. Age differ-ence in heart rate changes associated with micro-arousals in humans. Clin Neurophys 2002;113:1517-21.

13. Bonnet MH, Carley D, Guilleminault CG, et al. EEG arousals: scor-ing rules and examples: a preliminary report from the Sleep Disorders Atlas Task Force of the American Sleep Disorders Association. Sleep 1992;15:173-84.

14. Iber C, Ancoli-Israel S, Chesson AL, Quan SF. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Techni-cal Specifications. 1st ed. Westchester, IL: American Academy of Sleep Medicine, 2007.

15. Pan J, Tompkins WJ. A real-time QRS detection algorithm. IEEE Trans Biomed Eng 1985;32:230-6.

16. Fantini ML, Michaud M, Gosselin N, Lavigne G, Montplaisir J. Periodic leg movements in REM sleep behavior disorder and related autonomic and EEG activation. Neurology 2002; 59:1889- 94.

17. Sforza E, Jouny C, Ibanez V. Cardiac activation during arousal in humans: further evidence for hierarchy in the arousal response. Clin Neurophysiol 2000; 111:1611-9.

18. Knudsen S, Gammeltoft S, Jennum PJ. Rapid eye movement sleep behav-iour disorder in patients with narcolepsy is associated with hypocretin-1 deficiency. Brain 2010; 133:568-79.

19. Thannickal TC, Nienhuis R, Siegel JM. Localized loss of hypocretin (orexin) cells in narcolepsy without cataplexy. Sleep 2009;32:993-8.

20. Young JK, Wu M, Manaye KF, et al. Orexin stimulates breathing via med-ullary and spinal pathways. J Appl Physiol 2005;98:1387-95.

SLEEP, Vol. 36, No. 1, 2013 98 Heart Rate Response in Narcolepsy—Sorensen et al

21. Dergacheva O, Philbin K, Batemana R, Mendelowitza D. Hypocretin-1 (orexin A) prevents the effects of hypoxia/hypercapnia and enhances the gabaergic pathway from the lateral paragigantocellular nucleus to cardiac vagal neurons in the nucleus ambiguus. Neuroscience 2011;175:18–23.

22. Shirasaka T, Kunitake T, Takasaki M, Kannan H. Neuronal effects of orexins: relevant to sympathetic and cardiovascular functions. Regul Pept 2002;104:91-5.

23. Bastianini S, Silvani A, Berteotti C, et al. Sleep related changes in blood pressure in hypocretin-deficient narcoleptic mice. Sleep 2011;34:213-8.

24. Peyron C, Faraco J, Rogers W, et al. A mutation in a case of early onset narcolepsy and a generalized absence of hypocretin peptides in human narcoleptic brains. Nat Med 2000; 6:991-7.

25. Sorensen GL, Kempfner J, Zoetmulder M, Sorensen HB, Jennum P. At-tenuated heart rate response in REM sleep behavior disorder and Parkin-son’s disease. Mov Disord 2012;27:888-94.

26. Braak H, Tredici KD, Rb U, de Vos RAI, Jansen Steur ENH, Braak E. Staging of brain pathology related to sporadic Parkinson’s disease. Neu-robiol Aging 2003;24:197-211.

Appendix C

Paper 3

Title: Sleep Transitions in Hypocretin-Deficient Narcolepsy

Status: Published in SLEEP

Citation: Sorensen G.L., Knudsen S., Jennum P. Sleep Transitions in Hypocretin-Deficient Narcolepsy. SLEEP 2013;36(8):1173-1177.

77

SLEEP, Vol. 36, No. 8, 2013 1173 Sleep Transitions in HCRT-Deficient Narcolepsy—Sorensen et al

INTRODUCTIONNarcolepsy with cataplexy (NC) or narcolepsy without cata-

plexy (NwC) is a neurological sleep disorder found in one in 2,000 individuals. It is characterized by instability of sleep-wake regulation (daytime sleepiness, short sleep latency, disrupted nocturnal sleep with awakenings) and instability of rapid eye movement (REM) sleep regulation and motor tonus regulation, causing hypnagogic hallucinations, sleep paralysis, cataplexy, and REM sleep behavior disorder.1,2 The major pathophysiol-ogy is loss of the sleep-wake pattern and motor tonus regulat-ing hypocretinergic neurons in the hypothalamus,3,4 resulting in low or undetectable levels of cerebrospinal fluid hypocretin-1 (hcrt-1) in most patients with NC, but in only a small proportion of patients with NwC.5,6 The cause of hypocretin neuron loss is most probably autoimmunological, mainly based on a very tight association with specific HLA-types,7,8 whereas the cause of NC/NwC in patients with normal hcrt-1 levels remains unclear.

Wakefulness and arousals are believed to be regulated mainly by two ascending neuron networks in the brain, referred to as the ascending reticular activation system (ARAS).9 The first ac-tivating neuron network consists of the pedunculopontine and

SLEEP TRANSITIONS IN HYPOCRETIN-DEFICIENT NARCOLEPSYhttp://dx.doi.org/10.5665/sleep.2880

Sleep Transitions in Hypocretin-Deficient NarcolepsyGertrud Laura Sorensen, MScE; Stine Knudsen, MD, PhD; Poul Jennum, Prof., MD, Dr. Med. Sci.Danish Center for Sleep Medicine, Department of Clinical Neurophysiology, Center for Healthy Aging, Faculty of Health Sciences, University of Copenhagen, Glostrup Hospital, Copenhagen, Denmark

A commentary on this article appears in this issue on page 1123.

Submitted for publication September, 2012Submitted in final revised from January, 2013Accepted for publication January, 2013Address correspondence to: Poul Jennum, Dr Med Sci, Professor, Chief Physician, Danish Center for Sleep Medicine, Department of Clinical Neurophysiology, Faculty of Health Sciences, University of Copenhagen, Glostrup Hospital, DK 2600 Glostrup, Denmark; Tel: 45 43232512; Fax: 45 43233933; E-mail: poul.jennum@regionh.dk

Study Objectives: Narcolepsy is characterized by instability of sleep-wake, tonus, and rapid eye movement (REM) sleep regulation. It is associ-ated with severe hypothalamic hypocretin deficiency, especially in patients with cataplexy (loss of tonus). As the hypocretin neurons coordinate and stabilize the brain’s sleep-wake pattern, tonus, and REM flip-flop neuronal centers in animal models, we set out to determine whether hypocretin deficiency and/or cataplexy predicts the unstable sleep-wake and REM sleep pattern of the human phenotype.Design: We measured the frequency of transitions in patients with narcolepsy between sleep-wake states and to/from REM and NREM sleep stages. Patients were subdivided by the presence of +/- cataplexy and +/- hypocretin-1 deficiency.Setting: Sleep laboratory studies conducted from 2001-2011.Patients: In total 63 narcolepsy patients were included in the study. Cataplexy was present in 43 of 63 patients and hypocretin-1 deficiency was present in 37 of 57 patients.Measurements and Results: Hypocretin-deficient patients with narcolepsy had a significantly higher frequency of sleep-wake transitions (P = 0.014) and of transitions to/from REM sleep (P = 0.044) than patients with normal levels of hypocretin-1. Patients with cataplexy had a significantly higher frequency of sleep-wake transitions (P = 0.002) than those without cataplexy. A multivariate analysis showed that transitions to/from REM sleep were predicted mainly by hypocretin-1 deficiency (P = 0.011), whereas sleep-wake transitions were predicted mainly by cataplexy (P = 0.001).Conclusions: In human narcolepsy, hypocretin deficiency and cataplexy are both associated with signs of destabilized sleep-wake and REM sleep control, indicating that the disorder may serve as a human model for the sleep-wake and REM sleep flip-flop switches.Keywords: Flip-flop, hypocretin, narcolepsy, sleep transitionCitation: Sorensen GL; Knudsen S; Jennum P. Sleep transitions in hypocretin-deficient narcolepsy. SLEEP 2013;36(8):1173-1177.

laterodorsal tegmental nuclei in the brainstem, which are active during wakefulness and REM sleep.10 The second activating neuron network consists of the locus coeruleus, dorsal end me-dian raphe nuclei, ventral periaqueductal gray matter, and tu-beromammillary nucleus, which are active during wakefulness, slightly active during nonrapid eye movement (NREM) sleep, and inactive during REM sleep.10 Sleep is promoted by activa-tion of the ventrolateral preoptic nucleus in the basal forebrain, which acts by inhibiting hypocretin neurons and the neurons in ARAS.10 The currently prevailing theory is that transition from sleep to wakefulness and vice versa is regulated by a sleep-wake flip-flop switch, involving the mutual inhibition of the sleep pro-moting nuclei and the wakefulness-promoting nuclei.11

The REM-NREM flip-flop switch works through another mu-tually inhibitory system, in which gamma-aminobutyric acidergic (GABAergic) neurons in the sublaterodorsal region fire during REM sleep and inhibit GABAergic neurons in the ventrolateral periaqueductal gray matter and the adjacent lateral pontine teg-mentum that in turn fire during NREM states to inhibit REM sleep.10 The hypocretin neurons play a central role in the coordi-nation and stabilization of the sleep-wake and REM-NREM flip-flop switches.11 Consequently, hypocretin deficiency may be the cause of the observed instability of the sleep-wake pattern of the human narcoleptic phenotype. We therefore propose that instabil-ity in these systems due to hypocretin loss may lead to sleep-wake and NREM-REM transition instabilities. For this reason, narco-lepsy may serve as a human model for the flip-flop switches.

METHODS

PatientsOver a 10-y period (2001-2011), patients with narcolepsy

seen at the Danish Center for Sleep Medicine, Glostrup Univer-

SLEEP, Vol. 36, No. 8, 2013 1174 Sleep Transitions in HCRT-Deficient Narcolepsy—Sorensen et al

sity Hospital, were identified and consecutively included after ethical approval (KA03119) and written informed consent had been obtained. All patients fulfilled the International Classifica-tion of Sleep Disorders (ICSD-2) criteria for narcolepsy,12 and were evaluated by neurological examination, determination of routine blood characteristics, polysomnography (PSG), the mul-tiple sleep latency test (MSLT), and determination of hcrt-1 in the cerebrospinal fluid (CSF). The hypocretin measurement pro-tocol and definition of hcrt-1 deficiency used were as previously published.6 Exclusion criteria were: additional neurological and psychiatric disorders, an apnea index > 5/h, and a periodic leg movements during sleep index > 4/h. All patients (except three with severe cataplexy) were free of antidepressants and stimu-lants 7 to 14 days before inclusion; 44 of 63 were completely drug-naïve (i.e., they had never taken medication).

A total of 63 patients fulfilled the inclusion criteria. Cata-plexy was diagnosed as previously described2 and was present in 43 of 63 patients, and hcrt-1 deficiency was present in 37 patients. Six patients (five of six with cataplexy) refused lum-bar puncture. Normal hcrt-1 levels were present in eight of 43 patients with cataplexy, and hcrt-1 deficiency was present in seven of 20 patients without cataplexy. All patients have been included in a previous publication by Sorensen et al.13 and 34 of 63 patients have been previously featured in publications by Knudsen et al.2,6 Demographic data of the patients are summa-rized in Table 1. Table 2 gives an overview of the cataplexy and hcrt-1 status for the patients in the different groups.

PSG RecordingsAll patients underwent 1 night of PSG recording consisting

of electroencephalography (C3-A2, C4-A1), vertical and hori-zontal electrooculography (EOG), surface electromyography (EMG) of the submentalis and anterior muscles, electrocardi-ography (ECG), nasal airflow, thoracic respiratory effort, and oxygen saturation. Patients were instructed not to consume any caffeinated or alcoholic drinks for at least 6 h before the re-cordings were made. Sleep stages and events were manually

scored for epochs of 30 sec according to standard criteria by experienced PSG technicians supervised by PJ14,15 and without knowledge of the final diagnosis. The hypnograms and sleep events were extracted from Nervus® (V5.5, Cephalon DK, Nørresundby, Denmark) or Somnologica Studio® (V5.1, Em-bla, Broomfield, CO, USA), using the built-in export data tool, and imported into the Matlab® analysis program (R2009b, The MathWorks, Natick, MA, USA) for further processing.

Analysis of TransitionsThe frequency of transitions in the hypnograms for each pa-

tient was measured as the number of transitions per h of sleep of the investigated sleep stage. Analyzed transitions included were transitions between wake and sleep, transitions to/from REM sleep (hereafter noted as REM-NREM transitions), transitions to/from NREM stage 1, 2, and 3, and transitions between all sleep stages. Wilcoxon rank-sum tests were performed to compare the frequencies of transitions between different narcolepsy groups.

A backward-stepwise multivariate linear regression was per-formed to determine the relationship between transitions and age, sex, body mass index, disease duration, disease onset, +/- previous anticataplexy medication, +/- cataplexy, and +/- hcrt-1 status. The +/- anticataplexy medication indicates whether patients were pausing with previous medication or had never taken medication before inclusion (drug-naïve).

PSG variables, including sleep latency (SL), total sleep time (TST), sleep efficiency (SE), time from sleep onset to first ep-och of REM sleep (REM SL), and percentage of stages 1, 2, and 3 NREM and REM sleep, were also measured. Wilcoxon rank-sum tests were performed to evaluate between-group differenc-es (results shown in Table 3). A significance level of P < 0.05 was chosen for all statistical tests.

RESULTSThe frequencies of sleep-wake transitions are illustrated in

the box plots in Figure 1 and means and standard deviations (SDs) of each narcolepsy group are given in Table 4. Patients

Table 1—Demographic data of patients

NC NwC Normal hcrt-1 Low hcrt-1Number 43 20 20 37Female/male 20/23 12/8 14/6 16/21Age (y) 35.3 ± 12.8 35.4 ± 10.6 33.9 ± 9.9 36.2 ± 12.8BMI (kg/m2) 26.8 ± 6.2 26.0 ± 5.0 25.2 ± 5.2 26.4 ± 4.3

Age and BMI are given as the mean ± standard deviation. BMI, body mass index; hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy.

Table 2—Overview of cataplexy and hcrt-1 status for the included patients

Normal hcrt-1 Low hcrt-1 Unknown hcrt-1 level TotalNC 8 30 5 43NwC 12 7 1 20Total 20 37 6 63

hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy.

SLEEP, Vol. 36, No. 8, 2013 1175 Sleep Transitions in HCRT-Defi cient Narcolepsy—Sorensen et al

with narcolepsy and low hcrt-1 level had a signifi cantly higher frequency of sleep-wake transitions than patients with narco-lepsy and normal hcrt-1 level (P = 0.014). Patients with NC also had a signifi cantly higher frequency of sleep-wake transitions than those with NwC (P = 0.002).

The frequencies of REM-NREM transitions are illustrated in the box plots in Figure 2 and the means and SDs are given

in Table 4. Patients with narcolepsy and low hcrt-1 level had a signifi cantly higher frequency of REM-NREM transitions than patients with narcolepsy and normal hcrt-1 level (P = 0.044). Patients with NC also had a trend for more REM-NREM tran-sitions than those with NwC, although this difference was not statistically signifi cant.

Table 3—Comparison of PSG variables

NC NwC Normal hcrt-1 Low hcrt-1 PTST (min) 393.0 ± 78.0 418.2 ± 60.5 417.5 ± 54.2 399.8 ± 76.9 NSSL (min) 9.8 ± 19.0 5.2 ± 4.0 10.0 ± 12.2 8.0 ± 19.7 NSREM SL (min) 79.6 ± 157.4 62.2 ± 38.0 118.4 ± 216.5 49.0 ± 55.2 < 0.04b

SE (%) 86.4 ± 12.1 92.3 ± 8.1 88.8 ± 15.8 89.1 ± 7.7 NSStage 1 sleep (%) 9.3 ± 6.6 4.8 ± 4.4 3.5 ± 2.0 9.9 ± 6.7 < 0.01a,b

Stage 2 sleep (%) 39.6 ± 12.5 44.3 ± 10.6 41.7 ± 12.0 40.1 ± 11.9 NSStage 3 sleep (%) 28.1 ± 13.2 28.8 ± 9.9 32.9 ± 12.5 25.2 ± 11.3 NSREM sleep (%) 23.8 ± 5.8 22.9 ± 4.3 21.8 ± 4.3 24.5 ± 5.5 NS

Data are expressed as mean ± standard deviation and P values are those corresponding to Wilcoxon rank-sum tests. aNC versus NwC. bNormal hcrt-1 versus low hcrt-1. hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NS, not signifi cant; NwC, narcolepsy without cataplexy; PSG, polysomnography; REM, rapid eye movement; SE, sleep effi ciency; SL, sleep latency; TST, total sleep time.

Table 4—Frequency of sleep transitions (mean ± standard deviation)

NC NwC Normal hcrt-1 Low hcrt-1Sleep-wake transitions 6.3 ± 2.5 4.1 ± 1.8 4.4 ± 2.0 6.0 ± 2.2

P 0.002 0.014

REM-NREM transitions 9.0 ± 3.3 7.9 ± 2.4 7.3 ± 1.7 9.2 ± 3.2P 0.289 0.044

P values are given for comparisons between NC and NwC and between narcolepsy with normal hcrt-1 level and narcolepsy with low hcrt-1 level using Wilcoxon rank-sum tests. hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NREM, nonrapid eye movement; NwC, narcolepsy without cataplexy; REM, rapid eye movement.

Figure 1—Frequency of sleep-wake transitions illustrated in box plots for each patient group. Each box plot shows the fi fth, 25th, 50th, 75th, and 95th percentile. Signifi cant P values are shown. Hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy.

P = 0.002 P = 0.014

Figure 2—Frequency of rapid eye movement-nonrapid eye movement transitions illustrated in box plots for each patient group. Each box plot shows the fi fth, 25th, 50th, 75th, and 95th percentile. Signifi cant P values are shown. Hcrt, hypocretin-1; NC, narcolepsy with cataplexy; NwC, narcolepsy without cataplexy.

P = 0.044

SLEEP, Vol. 36, No. 8, 2013 1176 Sleep Transitions in HCRT-Deficient Narcolepsy—Sorensen et al

For the other analyzed transitions (to/from NREM stage 1, 2, and 3, and transitions between all sleep stages), there was no significance between any of the groups (results not shown).

In the multivariate analysis, the number of REM-NREM transitions were predicted only by hcrt-1 deficiency (P = 0.011), whereas the number of sleep-wake transitions were predicted only by cataplexy (P = 0.001).

The values for the PSG variables are displayed in Table 3. For patients with narcolepsy and hcrt-1 deficiency, the REM SL was significantly lower (P < 0.04), and the percentage of NREM 1 sleep was significantly higher (P < 0.01) than in pa-tients with narcolepsy without hcrt-1 deficiency. The percent-age of NREM 1 sleep was significantly higher in patients with NC than in those with NwC (P < 0.01).

DISCUSSIONThis is the first study to show that the frequency of sleep-

wake transitions and REM-NREM transitions are associated with cataplexy and hypocretin deficiency in patients with narcolepsy. The study documents that sleep-wake and REM-NREM transitions are more frequent in patients with NC and hypocretin-deficient narcolepsy. These results are consistent with the flip-flop model of transitions proposed by Saper et al.,11 and are consistent with a human model in which hypocretin enforces general state stability.

Hypocretin neurons in the lateral hypothalamus reinforce the activity in the wake-promoting projections arising from neurons in the upper brainstem, whereas sleep is promoted by activa-tion of the ventrolateral preoptic nucleus in the basal forebrain, which acts by inhibiting hypocretin neurons and the neurons in ARAS.10 The hypocretin neurons are therefore believed to play an important role in sustaining wakefulness and in stabilizing the flip-flop switch. However, other important arousal-produc-ing neurons are present in the posterior lateral hypothalamus, which helps maintain wakefulness when hypocretin neurons are deficient. The hypocretin neurons may help to sustain ac-tivity in aminergic and cholinergic arousal regions, and in the absence of hypocretin, these arousal regions may have reduced activity, resulting in inappropriately low thresholds to transition into NREM sleep.16 In patients with narcolepsy and hypocretin deficiency, one of the most important components in the flip-flop switch is absent, making the switch unstable and entailing frequent sleep/wake switches, as seen in this study.

Hypocretin neurons are also known to suppress REM sleep by reinforcing the activity of the monoaminergic neurons in the locus coeruleus and dorsal raphe nucleus,17,18 which in turn activate REM sleep-suppressing neurons and inhibit REM on neurons. It can be hypothesized that loss of hypocretin neurons therefore also enables more frequent transitions into REM sleep and makes the REM-NREM switch unstable. In our study we detected that patients with narcolepsy and hcrt-1 deficiency had more REM-NREM transitions than those without hcrt-1 defi-ciency, and this higher frequency of REM-NREM transitions was primarily predicted by hcrt-1 deficiency, independent of cataplexy and other factors, supporting the theory that hypocre-tin destabilizes the sleep stage switch.

Quantitative electroencephalographic studies suggest chang-es in frequencies in NREM 2 sleep19 as well as sleep onset20 in patients with NC. Another study has shown that the occur-

rence of multiple spontaneous daytime sleep onset REM peri-ods clearly identified patients with narcolepsy.21 Animal studies support that sleep-wake transition is under control by the hypo-cretinergic system, which facilitates wakefulness and REM-NREM cyclicity. Studies with hypocretin-deficient knockout mice and animals with narcolepsy have shown sleep stage dysregulations22 and considerably more transitions between all stages.16,23-25 As hypocretin is believed to stabilize wake and sleep, the hypocretin deficiency in patients with narcolepsy may cause a low threshold to transition between stages, causing the fragmented sleep pattern seen in patients with narcolepsy. This supports our study, showing that hypocretin-deficient patients with narcolepsy may present sleep stage instability as com-pared with non-hypocretin-deficient patients with narcolepsy.

One limitation of this study is that sleep-wake regulation as determined by simple macro-sleep scoring may be too simplistic a model to allow the evaluation of the sleep-wake/REM-NREM transition, because this may occur much faster and not in the 30-sec epochs used in traditional scoring. This may in part explain why we did not find a significant association with NREM tran-sitions. However, to date, no reliable or validated sleep scoring model has been developed for scoring epochs shorter than 30 sec. Alternative methods may include a quantitative measure using a sliding window with a shorter time interval including not only the conventional sleep stages, but also transition zones between stages. One attempt using a state space analysis tech-nique has been performed in orexin knockout mice,25 where the variations in the depth of sleep and the intensity of wakefulness were measured by electroencephalographic power in the delta and theta bands. The orexin knockout mice had an overlap be-tween wake and REM sleep and spent more time in transition regions causing the altered quality of sleep and wake. Varia-tions within states and changes in transition zones may be a reason for the more frequent changes in orexin knockout mice and hypocretin-deficient patients with narcolepsy. We suggest that future evaluation of sleep transition and its relation to the hypocretinergic system should include analysis of transition states and shorter windows used in the sleep scoring.

Another limitation is that interscorer variability was not consid-ered and sleep scoring may have been subject to individual scor-ing. Furthermore, only nocturnal PSGs are considered, whereas daytime monitoring could be at least as interesting, because pa-tients with narcolepsy have sleep episodes during daytime.

In conclusion, the current study shows that patients with narcolepsy and hypocretin deficiency or cataplexy have more frequent sleep-wake and REM-NREM transitions, which is consistent with the destabilization of the flip-flop switches that regulate sleep-wake and REM-NREM transitions due to the loss of hypocretin neurons. The elevated frequency of REM-NREM transitions is primarily predicted by hcrt-1 deficiency, independent of cataplexy and other factors.

ACKNOWLEDGMENTGertrud Laura Sorensen and Stine Knudsen contributed

equally to this work.

DISCLOSURE STATEMENTThis was not an industry supported study. The authors have

indicated no financial conflicts of interest.

SLEEP, Vol. 36, No. 8, 2013 1177 Sleep Transitions in HCRT-Deficient Narcolepsy—Sorensen et al

REFERENCES1. Ferri R, Miano S, Bruni O, et al. NREM sleep alterations in narcolepsy/

cataplexy. Clin Neurophysiol 2005;116:2675-84.2. Knudsen S, Gammeltoft S, Jennum PJ. Rapid eye movement sleep behav-

iour disorder in patients with narcolepsy is associated with hypocretin-1 deficiency. Brain 2010;133:568-79.

3. Peyron C, Faraco J, Rogers W, et al. A mutation in a case of early onset narcolepsy and a generalized absence of hypocretin peptides in human narcoleptic brains. Nat Med 2000;6:991-7.

4. Thannickal TC, Moore RY, Nienhuis R, et al. Reduced number of hypo-cretin neurons in human narcolepsy. Neuron 2000;27:469-74.

5. Mignot E, Lammers GJ, Ripley B, et al. The role of cerebrospinal fluid hypocretin measurement in the diagnosis of narcolepsy and other hyper-somnias. Arch Neurol 2002;59:1553-62.

6. Knudsen S, Jennum PJ, Alving J, Sheikh SP, Gammeltoft S. Validation of the ICSD-2 criteria for CSF hypocretin-1 measurements in the diagnosis of narcolepsy in the Danish population. Sleep 2010; 33:169-76.

7. Kornum BR, Kawashima M, Faraco J, et al. Common variants in P2RY11 are associated with narcolepsy. Nat Genet 2011;43:66-71.

8. Hor H, Kutalik Z, Dauvilliers Y, et al. Genome-wide association study identifies new HLA class II haplotypes strongly protective against narco-lepsy. Nat Genet 2010;42:786-9.

9. Saper CB, Scammel TE, Lu J. Hypothalamic regulation of sleep and cir-cadian rhythms. Nature 2005; 437:1257-63.

10. Saper CB, Fuller PM, Pedersen NP, Lu J, Scamell TE. Sleep state switch-ing. Neuron 2010;68:1023-1042.

11. Saper CB, Chou TC, Scamell TE. The sleep switch: hypothalamic control of sleep and wakefulness. Trends Neurosci 2001;24:726-31.

12. The International Classification of Sleep Disorders (ICSD)-diagnostic and coding manual, 2nd edition. Westchester, IL: American Academy of Sleep Medicine, 2005.

13. Sorensen GL, Knudsen S, Pedersen ER, et al. Attenuated heart rate re-sponse is associated with hypocretin deficiency in patients with narco-lepsy. Sleep 2013;36:91-8.

14. Bonnet MH, Carley D, Guilleminault CG, et al. EEG arousals: scor-ing rules and examples: a preliminary report from the Sleep Disorders Atlas Task Force of the American Sleep Disorders Association. Sleep 1992;15:173-84.

15. Iber C, Ancoli-Israel S, Chesson AL, Quan SF. The AASM manual for the scoring of sleep and associated events: rules, terminology and technical specifications. Westchester, IL: American Academy of Sleep Medicine, 2007.

16. Mochizuki T, Crocker A, McCormack S, Yanagisawa M, Sakurai T, Scammell TE. Behavioral state instability in orexin knock-out mice. J Neurosci 2004;24:6291-6300.

17. Bourgin P, Huitrón-Résendiz S, Spier AD, et al. Hypocretin-1 modulates rapid eye movement sleep through activation of locus coeruleus neurons. J Neurosci 2000;20:7760-5.

18. Kohlmeier KA, Watanabe S, Tyler CJ, Burlet S, Leonard CS. Dual orexin actions on dorsal raphe and laterodorsal tegmentum neurons: noisy cation current activation and selective enhancement of Ca2+ transients mediated by L-type calcium channels. J Neurophysiol 2008;100:2265-81.

19. Hadjiyannakis K, Ogilvie RD, Alloway CE, Shapiro C. FFT analysis of EEG during stage 2-to REM transitions in narcoleptic patients and normal sleepers. Electroencephalogr Clin Neurophysiol 1997;103:543-53.

20. Kim JW, Shin HB, Robinson PA. Quantitative study of the sleep onset period via detrended fluctuation analysis: normal vs. narcoleptic subjects. Clin Neurophysiol 2009;120:1245-51.

21. Pizza F, Moghadam KK, Vandi S, et al. Daytime continuous polysomnog-raphy predicts MSLT results in hypersomnias of central origin. J Sleep Res 2013;22:32-40.

22. Chemelli RM, Willie JT, Sinton CM, et al. Narcolepsy in orexin knockout mice: molecular genetics of sleep regulation. Cell 1999;98:437-51.

23. Nishino S, Riehl J, Hong J, Kwan M, Reid M, Mignot E. Is narcolepsy a REM sleep disorder? Analysis of sleep abnormalities in narcoleptic Do-bermans. Neurosci Res 2000;38:437-46.

24. Kaitin KI, Kilduff TS, Dement WC. Sleep fragmentation in canine narco-lepsy. Sleep 1986;9:116-9.

25. Diniz Behn CG, Klerman EB, Mochizuki T, Lin S, Scammell TE. Abnormal sleep/wake dynamics in orexin knockout mice. Sleep 2010;33:297-306.

Appendix D

Paper 4

Title: Reduced sympathetic activity in idiopathic Rapid-eye-movement sleep BehaviorDisorder and Parkinson’s disease.

Status: Published in Autonomic Neuroscience: Basic and Clinical

Citation: Sorensen G.L., Mehlsen J., and Jennum P. Reduced sympathetic activity inidiopathic rapid-eye-movement sleep behavior disorder and Parkinson’s disease. Auton.Neurosci. 2013;179(1):138-141.

83

Autonomic Neuroscience: Basic and Clinical 179 (2013) 138–141

Contents lists available at ScienceDirect

Autonomic Neuroscience: Basic and Clinical

j ourna l homepage: www.e lsev ie r .com/ locate /autneu

Reduced sympathetic activity in idiopathic rapid-eye-movement sleep behaviordisorder and Parkinson's disease☆

Gertrud Laura Sorensen a, Jesper Mehlsen b, Poul Jennum a,⁎a Danish Center for Sleep Medicine, Department of Clinical Neurophysiology, Faculty of Health Sciences, University of Copenhagen, Glostrup Hospital, Copenhagen, Denmarkb Coordinating Research Centre, University of Copenhagen, Frederiksberg Hospital, Frederiksberg, Denmark

☆ Financial disclosure: The study is supported by an unfor Healthy Aging (CEHA), University of Copenhagen. CEstudy design, collection, analysis and interpretation of dathedecision to submit the paper for publication.None of thof interests.⁎ Corresponding author at: Danish Center for Sleep M

Neurophysiology, Faculty of Health Sciences, University ofDK 2600 Glostrup, Denmark. Tel.: +45 43232512; fax: +4

E-mail address: poul.joergen.jennum@regionh.dk (P. J

1566-0702/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.autneu.2013.08.067

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 21 May 2013Received in revised form 25 July 2013Accepted 13 August 2013

Keywords:Heart rate variabilityParkinson's diseaseiRBDSleepAutonomic dysfunction

Background: More than 50% of patients with idiopathic REM sleep behavior disorder (iRBD) will developParkinson's disease or Lewy body dementia. In a previous study, we found attenuated heart rate responses iniRBD and Parkinson's disease patients during sleep. The current study aimed to evaluate heart rate variability fur-ther in order to identify possible changes in these components during wakefulness and sleep in patients withiRBD and Parkinson's disease.Methods: We evaluated heart rate variability in 5-minute electrocardiography segments from wakefulness, andnon-REM and REM sleep in 11 iRBD patients and 23 Parkinson's disease patients, and compared these with 10control subjects.Results and conclusions: Patients with iRBD had attenuated sympathetic nervous system activity compared withcontrols and this was more pronounced in patients with Parkinson's disease. The cardiac parasympathetic ner-

vous system seems to be relativelywell preserved in patients with iRBD and Parkinson's disease. The progressivereduction of sympathetic nervous activity is in line with the postganglionic sympathetic nervous dysfunctionseen in early Parkinson's disease.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Lewy bodies are found in Parkinson's disease (PD), in dementias withLewy bodies, in multiple system atrophy, and in REM sleep-associatedbehavior disorder (RBD) (Boeve et al., 2007; Jellinger, 2004; Uchiyamaet al., 1995). These conditions are considered to be different phenotypesof the same family of disorders and often showaffection of the autonomicnervous system. Lewy bodies may also be found in the sinoatrial nodalganglion, myocardium, and paravertebral sympathetic ganglia, andthese inclusions are considered by some to precede the affection of thecentral nervous system. It has been shown that patients with idiopathicRBD (iRBD) are at a high risk of later development. It is likely – but notyet proven – that comorbid conditions, e.g., autonomic dysfunction,may add to this risk.

In 1996, a task force composed of members of the European Societyof Cardiology and the North American Society of Pacing and Electro-physiology drew up the guidelines necessary for comparing different

restricted grant from the CenterHA had no further role in theta, the writing of the paper, nore authors reported any conflicts

edicine, Department of ClinicalCopenhagen, Glostrup Hospital,5 43233933.ennum).

ghts reserved.

assessment models of heart rate variability (HRV) (Task Force, 1996).These guidelines have proved to be the most valuable non-invasivetest for assessing autonomic nervous system function (Xhyheri et al.,2012). However, due to great intra-individual variance the tests donot work individually and are recommended only for describing groupsof patients. In a previous study, we found attenuated heart rate re-sponses in iRBD and PD patients during sleep (Sorensen et al., 2012).However, it is not clear whether this expresses suppression of sympa-thetic or parasympathetic activities or combinations of the two. Theaim of the current study was therefore to evaluate HRV further inorder to identify possible changes in these components during bothwakefulness and sleep in patientswith iRBD and PD, especially address-ing whether the observed changes are mainly due to sympathetic and/or parasympathetic abnormalities.

2. Materials & methods

2.1. Subjects

This studywas carried out as a secondary analysis based on data froma previous study that evaluated the heart rate response during sleep(Sorensen et al., 2012). Subjects included were patients with PD andiRBD who had undergone a polysomnograph (PSG) recording at theDanish Center for SleepMedicine (Glostrup, Denmark). Exclusion criteriawere additional neurological, psychiatric or cardiovascular disorders,alcohol abuse, head trauma, dementia, and an apnoea–hypopnoea

139G.L. Sorensen et al. / Autonomic Neuroscience: Basic and Clinical 179 (2013) 138–141

index ≥5. None of the subjects were taking agents known to affect theautonomic system such as antihypertensive drugs. It was not possibleto withdraw the patients from their Parkinson medication, but in orderto reduce influence of these drugs, the patients treated with intraspinalmedication, duodopa or intercerebral stimulation were excluded. RBDpatients were diagnosed according to the criteria of the InternationalClassification of Sleep Disorders (American Academy of Sleep Medicine,2005). iRBDpatients had no other disorders to explain their RBD. Diseaseseveritywas assessed for all patientswith theHoehn and Yahr (HY) scaleand the Unified Parkinson Disease Rating Scale (UPDRS).

In the initial study, 14 PD patients with RBD, 16 PD patients withoutRBD and 11 iRBD patients fulfilled the inclusion criteria. Seventeenhealthy controls, matched to the patient group by age, gender andbody mass index (BMI), were also included. The study was ethicallyapproved (HA20070120) and all included subjects gave their signed in-formed consent. However, 4 PD patients with RBD and 3 PD patientswithout RBDhad toomany nocturnal events to permit the identificationof stationary signals, as described below, and so these patients were ex-cluded from the study. Furthermore, 7 control subjects were excludeddue to insufficient sampling frequency. Demographic data of the includ-ed subjects are summarized in Table 1.

2.2. Methods

All subjects underwent one night of PSG recording consisting ofelectroencephalography (C3–A2, C4–A1), vertical and horizontal elec-trooculography (EOG), and surface electromyography (EMG) of thesubmentalis and tibialis anterior muscles and electrocardiography(ECG). Subjects were instructed not to consume any caffeinated oralcoholic drinks for at least 6 h before the recordings were made.

Sleep stages and events weremanually scored for 30-second epochsaccording to standard criteria (Iber et al., 2007) by experienced poly-somnographic technicians supervised by PJ, without knowledge of thefinal diagnosis. The hypnograms and sleep events were extracted fromNervus® (V5.5, Cephalon DK, Nørresundby, Denmark), using the built-in export data tool, and imported into theMATLAB® analytical program(R2009b, The MathWorks, Natick, MA, USA) for further processing.

Autonomic activity was analyzed using different quantitative mea-sures of HRV according to the guidelines given by the European Societyof Cardiology and the North American Society of Pacing and Electro-physiology (Task Force, 1996).

Five-minute continuous ECG segments were selected fromwakeful-ness, stage 2NREMand REM sleep during the night. The 5-minutewakewas selected from the pre-sleep period and the 5-minute NREM 2 wasselected from the beginning of the night. Where possible, the 5-minuteREM sleep was selected from the last REM period, as this tended to bethe longest period of REM in most subjects.

The segments included stationary signals (without microarousals,leg movements, complex movements during REM sleep, or apnoeas)recorded with a lead II derivation using torso electrode placement(Iber et al., 2007) with a sampling frequency of 512 Hz. HRV was ana-lyzed with respect to time domain, frequency domain, and nonlinearmethods using Kubios HRV version 2.1 software. Time domain variablesincluded the mean value of normal RR intervals (NN), the standard de-viation of normal RR intervals (SDNN), the root mean square of succes-sive RR interval differences (RMSSDs), the number of successive

Table 1Demographic data of subjects included in the study expressed as means (and standarddeviations).

Controls PD with RBD PD without RBD iRBD

Total number 10 10 13 11Female/male 5/5 3/7 6/7 2/9Age (years) 59.0 (9.7) 62.5 (6.9) 60.8 (6.7) 60.2 (7.7)UPDRS – 18.8 (7.1) 20.6 (5.1) 0.8 (1.0)HY – 1.4 (0.7) 0.9 (1.6) 0.0 (0)

intervals differing by more than 50 ms (NN50) and the correspondingrelative amount (pNN50).

The frequency domain variables were calculated by autoregressivemodeling (Marple, 1987) and included absolute powers of very low fre-quency (VLF, 0–0.04 Hz), low frequency (LF, 0.04–0.15 Hz) and highfrequency (HF, 0.15–0.4 Hz) bands, LF and HF band powers in normal-ized units (LFnu and HFnu) and the LF/HF power ratio.

The nonlinear properties of HRV were analyzed by Poincaré plots,the Shannon entropy index and detrended fluctuation analysis (DFA).An ellipse was fitted to the Poincaré plots, and the standard deviationof the points perpendicular to the line-of-identity (SD1) and the stan-dard deviation along the line-of-identity (SD2) were calculated. In theDFA, two correlation measures (α1 and α2) were calculated as de-scribed in theKubios HRVUser's Guide (Tarvainen andNiskanen, 2012).

The time domain, frequency domain and nonlinear parameters werecalculated as described in the Kubios HRV User's Guide (Tarvainen andNiskanen, 2012), which are equivalent to the descriptions given by theTask Force of 1996 (Task Force, 1996). All HRV variables were calculatedindividually and, subsequently, between-group differences were evalu-ated with Wilcoxon rank-sum tests. A Bonferroni correction was ap-plied to take into account the problem of multiple comparisons,however in order not to be too conservative with the Bonferroni correc-tion and allow false negative results, the variableswere grouped in eightindependent groups, which resulted in a significance level of 0.05/8 =0.00625 for all statistical tests.

3. Results

All HRV variables are shown in Table 2 for the control group andiRBD, PD with RBD and PD without RBD patients for the three stages(wakefulness, NREM, REM). Comparisons are made for each patientgroup with the control group.

For the control group the values from the time domain analysis werecomparable duringwakefulness and REM,while they tended to be largerduring NREM. In the frequency domain analysis using normalized units,LF and HFwere similar during wakefulness and REM, whereas LF tendedto be lower and HF higher during NREM. Short-term variation reflectingparasympathetic activity tended to be more pronounced during NREMthan in wakefulness and REM, with higher levels of NN, RMSSD, NN50,HFnu, and SD1. The pattern for long-term variationwas less clear in com-parisons of the wakefulness, NREM and REM stages.

For the iRBD patients, only the VLF component in the wakefulnessstage was significantly different from the control group. None of thevariables differed significantly between these groups during REM andNREM sleep.

For the PD patients, SDNN, VLF, LF and SD2 were significantlydifferent from the control group in the wakefulness stage. None of thevariables differed significantly in the REM and NREM stages for the PDpatients without RBD compared with the control group. For the PD pa-tients with RBD, only SD2 differed significantly from the control groupin the REM stage, while none of the HRV variables were significantly dif-ferent in the NREM stage.

Across the spectrum of clinical presentation there was a pattern ofprogressive reduction in VLF during wakefulness from controls(158 ms2), through iRBD (47 ms2) and PD without RBD (30 ms2), toPD with RBD (27 ms2). A similar pattern was noted for the long-termvariation expressed by LF and SD2.

4. Discussion

In this study, we have examinedmeasures of autonomic nervous ac-tivity during various sleep stages in patients at risk of developing PD andthose with clinically overt PD. The salient finding is that of a pattern ofprogressive reduction in HRV parameters associated with sympatheticnervous activity in linewith the notion that postganglionic sympatheticnervous dysfunction is an early component in the development of PD.

Table 2Measures of heart rate variability in A) wake, B) NREM 2 and C) REM sleep for controlsubjects, iRBD patients, PD with RBD patients and PD without RBD patients.

Controls iRBD PD without RBD PD with RBD

A)Wake

NN (ms) 872 (126) 936 (161) 881 (121) 1005 (150)SDNN (ms) 28 (6) 20 (8) 14.9 (4.5)⁎ 17.8 (7.4)⁎

RMSSD (ms) 19.5 (6.9) 15.7 (6.9) 15.7 (7.3) 18.8 (8.9)NN50 (count) 8.7 (8.2) 2.36 (3.32) 2.8 (4.6) 7.2 (9.5)pNN50 2.9 (3.3) 0.79 (1.06) 0.9 (1.6) 3.1 (4.1)VLF (ms2) 158 (90) 47 (34)⁎ 30 (28)⁎ 27 (26)⁎

LF (ms2) 492 (235) 293 (251) 101 (86)⁎ 146 (130)⁎

HF (ms2) 150 (105) 88 (63) 88 (66) 117 (98)LF (nu) 76 (14) 76 (15) 55 (23) 56 (14)HF (nu) 24 (14) 24 (15) 45 (23) 43 (14)LF/HF 5.3 (4.9) 5.4 (5.0) 2.0 (1.9) 1.5 (0.7)SD1 (ms) 13.8 (4.9) 11.1 (4.9) 11.1 (5.1) 13.4 (6.3)SD2 (ms2) 37.3 (8.7) 26.4 (10.2) 17.7 (4.8)⁎ 21.0 (8.7)⁎

Shannon Entropy 3.15 (0.44) 3.02 (0.21) 2.87 (0.19) 2.7 (0.3)DFA, alpha1 1.34 (0.29) 1.36 (0.27) 0.98 (0.25) 1.00 (0.21)DFA, alpha2 0.57 (0.17) 0.47 (0.26) 0.49 (0.11) 0.39 (0.15)

B)NREM 2

NN (ms) 979 (160) 1066 (164) 981 (135) 1114 (199)SDNN (ms) 29.3 (8.5) 28 (8) 22.0 (6.0) 23.4 (9.3)RMSSD (ms) 23.5 (8.4) 25.1 (5.9) 21.5 (8.2) 21.4 (5.9)NN50 (count) 18.4 (17.2) 15.6 (15.3) 10.9 (13.5) 8.4 (6.4)pNN50 6.6 (6.6) 5.5 (5.1) 4.0 (5.1) 3.4 (2.6)VLF (ms2) 148 (144) 122 (193) 96 (92) 85 (66)LF (ms2) 550 (432) 454 (420) 259 (213) 409 (416)HF (ms2) 240 (163) 266 (179) 174 (89) 114 (55)LF (nu) 67 (18) 54 (28) 56 (21) 68 (21)HF (nu) 33 (18) 46 (28) 44 (21) 32 (21)LF/HF 3.9 (4.8) 2.7 (3.3) 1.96 (1.67) 4.2 (4.3)SD1 (ms) 16.6 (5.9) 17.8 (4.2) 15.2 (5.8) 15.1 (4.2)SD2 (ms2) 37.5 (11.6) 34.7 (12.6) 26.7 (8.0) 29.2 (13.2)Shannon Entropy 2.94 (0.27) 2.96 (0.23) 2.87 (0.27) 2.80 (0.08)DFA, alpha1 1.19 (0.23) 1.03 (0.34) 1.01 (0.28) 1.08 (0.24)DFA, alpha2 0.43 (0.10) 0.33 (0.15) 0.43 (0.16) 0.38 (0.16)

C)REM

NN (ms) 887 (132) 954 (155) 934 (112) 1006 (128)SDNN (ms) 27.5 (7.2) 21 (8) 21.7 (4.2) 17.5 (5.6)RMSSD (ms) 20.8 (4.8) 15.5 (4.5) 20.5 (6.6) 16.1 (6.9)NN50 (count) 10.1 (8.2) 2.5 (5.3) 8.0 (11.3) 3.3 (8.7)pNN50 3.0 (2.4) 0.9 (1.8) 2.7 (3.7) 1.4 (3.7)VLF (ms2) 131 (83) 131 (123) 76 (52) 67 (72)LF (ms2) 527 (334) 250 (190) 217 (141) 191 (144)HF (ms2) 126 (51) 79 (38) 158 (106) 78 (74)LF (nu) 76 (15) 70 (16) 57 (17) 71 (12)HF (nu) 24 (15) 30 (16) 43 (17) 29 (11)LF/HF 4.7 (3.3) 3.4 (2.6) 1.76 (1.43) 3.2 (2.2)SD1 (ms) 14.7 (3.4) 11.0 (3.2) 14.5 (4.6) 11.4 (4.9)SD2 (ms2) 35.8 (10.0) 27.2 (11.5) 26.7 (5.5) 21.7 (7.0)⁎

Shannon Entropy 3.2 (0.3) 2.91 (0.16) 2.84 (0.27) 2.87 (0.28)DFA, alpha1 1.30 (0.21) 1.27 (0.27) 1.09 (0.25) 1.15 (0.20)DFA, alpha2 0.48 (0.12) 0.51 (0.14) 0.44 (0.13) 0.44 (0.17)

Values are presented asmean (and standard deviation). SDNN = standard deviation of allRR-intervals; VLF = very low frequency; LF = low frequency; HF = high frequency;SD1 = instantaneous beat-to-beat RR-interval variability; SD2 = long-term continuousRR-interval variability.⁎ Values that are significant (p b 0.00625) different from the control group.

140 G.L. Sorensen et al. / Autonomic Neuroscience: Basic and Clinical 179 (2013) 138–141

TheHRV values found in our control groupwere similar to the valuesreported by others for similar age groups (Valappil et al., 2010; Vosset al., 2012). Looking at the various sleep stages in the control subjects,we found that the HRV patterns were comparable during wakefulnessand REM sleep, whereas there was a pattern of increased parasympa-thetic relative to sympathetic activity during NREM sleep. These find-ings are in accordance with those of other studies analyzing sleepstage differenceswith respect to spectral and time domain variables de-scribing theHRV (Baharav et al., 1995; Bonnet andArand, 1997; Vaughnet al., 1995). Vaughn et al. found clear sleep stage differences in the HRV

from spectral analysis that were consistent with increased parasympa-thetic nervous system activity during NREM stage 2, decreasedparasympathetic and sympathetic nervous system activities duringthe NREM stage 3 and increased sympathetic activity during the REMsleep stage (Vaughn et al., 1995).

In our limited group of patients, we were only able to demonstratesignificant differences with respect to the control subjects duringwake-fulness, while no significant differences were seen in NREM sleep, andonly the PD patients with RBD had a significant difference for a singlevariable in REM. The sympathetic nervous system is more active duringwakefulness than during the NREM and REM sleep stages, making itmore likely thatwe could demonstrate a reduction in sympathetic activ-ity during the wakefulness stage.

The iRBD patients only differed significantly from the control groupwith respect to the VLF component. As this component reflects slowregulatorymechanisms (Busek et al., 2005) and is related to sympathet-ic activity (Ergun et al., 2008), this indicates that the sympathetic ner-vous system activity of iRBD patients was attenuated. The finding ofreduced activity in the slow components of HRV is in line with the factthat in diseases associated with Lewy bodies, the vulnerable neuronsare those that are long, thin and unmyelinated and therefore slow-conducting (Braak et al., 2006). Lanfranchi et al. investigated the HRVin NREM and REM sleep and found that cardiac and respiratory re-sponses were absent from subjects with iRBD (Lanfranchi et al., 2007).This implies that iRBD patients lack the possibility of activating the sym-pathetic nervous system, which explains why, in this study, we found asignificant difference in the VLF component in the wakefulness stage.We found no significant differences for the HF component, which isknown to reflect parasympathetic activity (Xhyheri et al., 2012), andthe attenuated heart rate response identified in our earlier study(Sorensen et al., 2012) is therefore mainly due to involvement of thesympathetic nervous system, but does not exclude the possibility thatthe parasympathetic nervous system is involved. In this context, it is im-portant to differentiate between “slow” parasympathetic activity medi-ated through thin unmyelinated fibers originating from neurons in thedorsal vagal motor nucleus and terminating primarily in organs belowthe diaphragm, and “fast” parasympathetic activity mediated throughlarge myelinated nerve fibers originating in the nucleus ambiguus andterminating primarily in the heart. In the post-mortem PD cases studiedto date, it is the unmyelinated vagal preganglionic projecting neuronsconnecting the central nervous system to the enteric nervous systemthat contain α-synuclein immunoreactive inclusions (Braak and DelTredici, 2008). In terms of HRV, the fast components (RMSSD, HF, SD1)are commonly ascribed to parasympathetic activity without stressingthe fact that this only holds true for the myelinated fibers terminatingin the heart. The importance of distinguishing between the parasympa-thetic activitymediated through thin unmyelinatedfibers and thickmy-elinatedfibers is reflected in the study byOka et al. (2011), who showedthat early-stage PD patients feature changes in HRV that are compatiblewith sympathetic but not parasympathetic nervous system dysfunctionin spite of the well known disturbances in intestinal motility and blad-der function associated with early PD.

For the PD patients, the LF component was found to be significantlylower than for the control subjects. The LF component is generallythought to reflect both sympathetic and parasympathetic activities(Ergun et al., 2008) and the difference between the PD and control sub-jects is probably due to a more pronounced dysfunction of the sympa-thetic nervous system than in the iRBD patients.

The SDNNwas significantly lower in the PD groups than in the con-trol group. SDNN is amarker of the total variance of HRV and reflects alllong-term components responsible for variability and correlates withVLF (Xhyheri et al., 2012), and therefore this result corresponds withthe results for LF and VLF and is indicative of a dysfunction of the sym-pathetic nervous system.

Of the nonlinear variables, only SD2 was significantly lower for thePD patients than the control subjects, which is also consistent with the

141G.L. Sorensen et al. / Autonomic Neuroscience: Basic and Clinical 179 (2013) 138–141

other results, since SD2 is believed to describe long-term variability andis related to the time domain measure SDNN (Brennan et al., 2001).

NREM sleep is characterized by relative autonomic stability, a dom-inance of parasympathetic influences to the heart, and a reduction insympathetic efferent vasomotor tone (Lanfranchi et al., 2007), whileREM sleep is a state of autonomic instability, dominated by remarkablefluctuations between parasympathetic and sympathetic influences(Lanfranchi et al., 2007) and, probably, activation of both nervous sys-tems (Busek et al., 2005). Therefore, if mainly the sympathetic nervoussystem is affected in iRBD and PD and unmyelinated parasympatheticnerve fibers in PD, this would explain why the HRV variables reflectingthe sympathetic nervous system differ in PD and iRBD compared withthe control subjects during wakefulness, rather than in NREM andREM sleep. Valappil et al. (2010) foundno differences in theVLF compo-nent in the wake but in both the LF and HF components when compar-ing iRBD patients with controls. Only 11 iRBD patients were included inthe Valappil study, which is also the case in our study. Studies with larg-er patient groups are recommended to clarify autonomic findings in theHRV measures.

According to the Braak staging study (Braak et al., 2003) the progres-sion of PD starts in the medulla oblongata, which includes the cardiaccenter and controls autonomic functions. Braak et al. found that bothparasympathetic structures (dorsal motor nucleus of vagus) and sympa-thetic structures (postganglionic neurons) demonstrate deposition ofsynuclein at early stages of PD pathology (Braak and Del Tredici, 2008).This is supported by the progressive reduction in VLF and the long-term variations expressed by LF and SD2 from controls to iRBD and toPD found in our study. Furthermore, other studies have shown that thebaroreceptor reflex in early-stage PD without orthostatic hypotensionis not severely impaired (Camerlingo et al., 1987; Haensch et al., 2009;Ludin et al., 1987), implying that the cardiac parasympathetic nervoussystem emanating from the nucleus ambiguus may be relatively pre-served in patients with PD.

From the medulla oblongata, the Braak staging system gradually as-cends to the more rostral structures implicating dysfunction in thesublaterodorsal nucleus and pre-locus coeruleus structures, whichcould lead to RBD (Boeve, 2010). This temporal sequence could indicatethat sympathetic nervous system dysfunction precedes iRBD, which isfollowed by PD or other diseases associated with Lewy body patholo-gies. However, we cannot conclude that iRBD with dysautonomia pre-sents an increased risk of further neurodegenerative development;furthermore, even if we could, we would not be able to predict thetype of neurodegenerative disease an iRBD patient would develop.

Postuma et al. (2011) found no evidence that baseline cardiac auto-nomic dysfunction can predict the risk of neurodegenerative diseaseand that autonomic dysfunction is linked with RBD independent of as-sociated PD or Lewy body dementia. Larger prospective studies areneeded to clarify the exact connection between dysautonomia in iRBDand the risk and type of neurodegenerative development. Furthermore,measurements of HRV based on the spontaneous variability in heartrate are known to vary greatly within individuals, which could blurthe outcome of small cohort studies. Future studies could include forcedvariations in heart rate by deep breathing at 0.1 Hz, theValsalvamaneu-ver and heart rate change during active standing, since these parame-ters reflect the capacity of the autonomic system rather than theactivity and are subject to smaller intra-individual variation.

In conclusion, we found that the sympathetic nervous system activityof iRBD patients was attenuated, especially when the disease progressedto PD. No differences were seen in the HRV variables ascribed to

parasympathetic activity, indicating that the cardiac parasympatheticnervous system, emanating from the nucleus ambiguus, may be rela-tively well preserved in patients with PD. The progressive reduction ofsympathetic nervous activity is in line with the postganglionic sympa-thetic nervous dysfunction seen in early PD.

References

American Academy of Sleep Medicine, 2005. The International Classification of SleepDisorders (ICSD) — Diagnostic and Coding Manual, 2nd ed. American Academy ofSleep Medicine, Westchester, IL.

Baharav, A., Kotagal, S., Gibbons, V., Rubin, B.K., Pratt, G., Karin, J., Akselrod, S., 1995. Fluc-tuations in autonomic nervous activity during sleep displayed by power spectrumanalysis of heart rate variability. Neurology 45, 1183–1187.

Boeve, B.F., 2010. REM sleep behavior disorder: updated review of the core features, theREM sleep behavior disorder–neurodegenerative disease association, evolving con-cepts, controversies, and future directions. Ann. N. Y. Acad. Sci. 1184, 15–54.

Boeve, B.F., Dickson,D.W., Olson, E.J., et al., 2007. Insights into REM sleep behavior disorderpathophysiology in brainstem-predominant Lewy body disease. Sleep Med. 8, 60–64.

Bonnet, M.H., Arand, D.L., 1997. Heart rate variability: sleep stage, time of night, andarousal influences. Electroencephalogr. Clin. Neurophysiol. 102, 390–396.

Braak, H., Del Tredici, K., 2008. Invited article: nervous system pathology in sporadicParkinson disease. Neurology 70, 1916–1925.

Braak, H., Del Tredici, K., Rüb, U., de Vos, R.A., Jansen Steur, E.N., Braak, E., 2003. Staging ofbrain pathology related to sporadic Parkinson's disease. Neurobiol. Aging 24, 197–211.

Braak, H., Bohl, J.R., Müller, C.M., Rüb, U., de Vos, R.A., Del Tredici, K., 2006. Stanley FahnLecture 2005: the staging procedure for the inclusion body pathology associatedwith sporadic Parkinson's disease reconsidered. Mov. Disord. 21, 2042–2051.

Brennan, M., Palaniswami, M., Kamen, P., 2001. Do existing measures of Poincaré plot ge-ometry reflect nonlinear features of heart rate variability? IEEE Trans. Biomed. Eng.48, 1342–1347.

Busek, P., Vanková, J., Opavský, J., Salinger, J., Nevsímalová, S., 2005. Spectral analysis ofthe heart rate variability in sleep. Physiol. Res. 54, 369–376.

Camerlingo, M., Aillon, C., Bottacchi, E., 1987. Parasympathetic assessment in Parkinson'sdisease. Adv. Neurol. 45, 267–269.

Ergun, U., Demirci,M., Nurlu, G., Komürcü, F., 2008. Power spectral analysis of heart rate var-iability: normal values of subjects over 60 years old. Int. J. Neurosci. 118, 1165–1173.

Haensch, C.A., Lerch, H., Jörg, J., Isenmann, S., 2009. Cardiac denervation occurs indepen-dent of orthostatic hypotension and impaired heart rate variability in Parkinson's dis-ease. Parkinsonism Relat. Disord. 15, 134–137.

Iber, C., Ancoli-Israel, S., Chesson, A.L., Quan, S.F., 2007. The AASMManual for the Scoringof Sleep and Associated Events: Rules, Terminology and Technical Specifications.American Academy of Sleep Medicine, Westchester, IL.

Jellinger, K.A., 2004. Lewy Body-related alpha-synucleinopathy in the aged human brain.J. Neural Transm. 111, 1219–1235.

Lanfranchi, P.A., Fradette, L., Gagnon, J.F., Colombo, R., Montplaisir, J., 2007. Cardiac auto-nomic regulation during sleep in idiopathic REM sleep behavior disorder. Sleep 30,1019–1025.

Ludin, S.M., Steiger, U.H., Ludin, H.P., 1987. Autonomic disturbances and cardiovascularreflexes in idiopathic Parkinson's disease. J. Neurobiol. 235, 10–15.

Marple, S.L., 1987. Digital Spectral Analysis. Prentice-Hall International.Oka, H., Toyoda, C., Yogo, M., Mochio, S., 2011. Cardiovascular dysautonomia in de novo

Parkinson's disease without orthostatic hypotension. Eur. J. Neurol. 18, 286–292.Postuma, R.B., Lanfranchi, P.A., Blais, H., Gagnon, J.F., Montplaisir, J.Y., 2011. Cardiac auto-

nomic dysfunction in idiopathic REM sleep behavior disorder. Mov. Disord. 25,2304–2310.

Sorensen, G.L., Kempfner, J., Zoetmulder, M., Sorensen, H.B., Jennum, P., 2012. Attenuatedheart rate response in REM sleep behavior disorder and Parkinson's disease. Mov.Disord. 27, 888–894.

Tarvainen, M.P., Niskanen, J.P., 2012. Kubios HRV Version 2.1 User's Guide. University ofFinland.

Task Force of the European Society of Cardiology, The North American Society of Pacingand Electrophysiology, 1996. Heart rate variability: standards of measurement, phys-iological interpretation and clinical use. Circulation 93, 1043–1065.

Uchiyama, M., Isse, K., Tanaka, K., et al., 1995. Incidental Lewy body disease in a patientwith REM sleep behavior disorder. Neurology 45, 709–712.

Valappil, R.A., Black, J.E., Broderick, M.J., Carrillo, O., Frenette, E., Sullivan, S.S., Goldman,S.M., Tanner, C.M., Langston, J.W., 2010. Exploring the electrocardiogram as a poten-tial tool to screen for premotor Parkinson's disease. Mov. Disord. 25, 2296–2303.

Vaughn, B.V., Quint, S.R., Messenheimer, J.A., Robertson, K.R., 1995. Heart period variabil-ity in sleep. Electroencephalogr. Clin. Neurophysiol. 94, 155–162.

Voss, A., Heitmann, A., Schroeder, R., Peters, A., Perz, S., 2012. Short-term heart ratevariability — age dependence in healthy subjects. Physiol. Meas. 33, 1289–1311.

Xhyheri, B., Manfrini, O., Mazzolini, M., Pizzi, C., Bugiardini, R., 2012. Heart rate variabilitytoday. Prog. Cardiovasc. Dis. 55, 321–331.