Measuring the Baroreflex: Equivalency, Repeatability, and Its

Measuring the Baroreﬂex: Equivalency, Repeatability, and its Relationship with Sex, Age, and Orthostatic Tolerance

by Matthew Geoﬀrey Lloyd

B.Sc., Simon Fraser University, 2009

Thesis Submitted in Partial Fulﬁllment of the Requirements for the Degree of Doctor of Philosophy

in the Department of Biomedical Physiology and Kinesiology Faculty of Science

c Matthew Geoﬀrey Lloyd 2018 SIMON FRASER UNIVERSITY Summer 2018

All rights reserved. However, in accordance with the Copyright Act of Canada, this work may be reproduced without authorization under the conditions for “Fair Dealing.” Therefore, limited reproduction of this work for the purposes of private study, research, education, satire, parody, criticism, review and news reporting is likely to be in accordance with the law, particularly if cited appropriately. Approval

Name: Matthew Geoﬀrey Lloyd Degree: Doctor of Philosophy (Biomedical Physiology and Kinesiology) Title: Measuring the Baroreﬂex: Equivalency, Repeatability, and its Relationship with Sex, Age, and Orthostatic Tolerance Examining Committee: Chair: Will Cupples Professor

Victoria Claydon Senior Supervisor Associate Professor

Andrew Blaber Supervisor Professor

Michael Koehle Supervisor Associate Professor University of British Columbia

Damon Poburko Internal Examiner Assistant Professor Simon Fraser University

Nisha Charkoudian External Examiner Research Physiologist U.S. Army Research Institute of Environmental Medicine

Date Defended: 17 August 2018

ii Ethics Statement

iii Abstract

The lifetime prevalence of fainting is estimated at 20-40%. Fainting, or syncope, is caused by reduced cerebral perfusion, often secondary to a reduction in blood pressure. During standing, the movement of blood into the lower limbs stimulates the arterial baroreflex to increase cardiac output and constrict peripheral blood vessels to maintain blood pressure and blood flow to the brain. Syncope is more common in older adults and women; however, it is unclear whether this might relate to age- and/or sex-related changes in baroreflex function. Different methods of baroreflex stimulation and measurement may not be comparable, and the repeatability of these techniques is uncertain. In Chapter 3, I quantify the agreement and repeatability of different baroreflex testing methods. Cardiac baroreflex tests have higher agreement and repeatability than sympathetic vascular tests, highlighting the need for better methods of vascular baroreflex measurement. In Chapter 4, I examine the relationship between aging, sex and both the cardiac and vascular arms of the baroreflex using novel sympathetic vascular baroreflex measurement techniques. In Chapter 5, I examine the influence of carotid sinus massage (CSM) methodology on the cardiovascular responses, and recommend that CSM be applied to the area of maximal pulsatility when ultrasound-guidance is unavailable. Lastly, in Chapter 6, I refute the hypothesis that large responses to carotid baroreceptor stimulation in older adults are related to chronic denervation of the sternocleidomastoid muscles. Thus, the pathophysiology of hypersensitive responses to CSM remains unknown. This work demonstrates that careful consideration should be taken when selecting baroreflex measurement techniques. Our novel sympathetic vascular baroreflex testing methods are critical for a thorough description of baroreflex control of blood pressure in younger and older men and women. While the vascular responses to orthostasis are large in older individuals, the vascular responses to acute carotid baroreceptor stimulation with the neck collar decline with age. The standard clinical methodology of CSM may not accurately deliver mechanical stimulation to the carotid baroreceptors. We have shown that the previously reported association with sternocleidomastoid denervation is likely coincidental, not causal. Overall, this thesis reiterates the importance of sympathetic vascular baroreflex function in the maintenance of blood pressure.

iv Keywords: Baroreﬂex; Repeatability; Orthostatic Tolerance

v Dedication

This thesis is dedicated to my grandfather, Geoﬀ Wilkins, who was unable to see me through to the end. Buster, your love and support are still with me.

vi Acknowledgements

I have been extraordinarily fortunate in the support and guidance I have received over the last five years. It would be impossible to acknowledge, individually, each colleague, mentor, friend and family member who has enabled the completion of this dissertation. To each of you, I can never properly thank you for your support and patience. First and foremost, my wife, Kristina: your unwavering support has been completely amazing. There is no way to adequately express the magnitude of help and understanding you have given me. Throughout all the late nights, working weekends, dinners, parenting, and pregnancies, you have been an incredible partner. Thank you for everything you do. To Vic, thank you for allowing me the time to explore my roles and strengths as a scientist, and for being such a valuable mentor for so many years. The long talks about physiology, napkin diagrams, and deep discussions on statistics were essential to my scientific development. I will be forever grateful for the family-positive working environment you have established in the lab. I look forward to continuing our strong working relationship. To my family, especially my parents: you have set wonderful examples for your children, and showed us how to further our education. Sam, you paved the way and provided key advice at the right times. To Logan, thank you for reminding me that true excitement and discovery lies in trains, buses, and doughnuts on boys’ night. Thanks to all my lab mates and the many undergraduate students who worked with and alongside me. You all have been such a pleasure to work with. We had so much fun, and managed to do some science as well. Lastly, thank you to all of the research participants who volunteered their valuable time to further the scientific process. Your contribution is immensely valued.

vii Table of Contents

Approval ii

Ethics Statement iii

Abstract iv

Dedication vi

Acknowledgements vii

Table of Contents viii

List of Tables xiii

List of Figures xiv

List of Abbreviations xvi

1 Introduction 1 1.1 Arterial baroreceptors are crucial for short-term blood pressure control . . . 1 1.2 Abnormal Baroreflex Function May Result in Syncope ...... 3 1.3 There are Many Ways to Faint ...... 4 1.4 Assessment of Baroreflex Function is Challenging ...... 5 1.5 Non-Invasive Baroreflex Tests ...... 10 1.5.1 Static Baroreflex Testing Techniques ...... 10 1.5.1.1 Responses to Orthostatic Stress ...... 10 1.5.1.2 Neck Collar Technique ...... 11 1.5.1.3 Valsalva Maneuver ...... 12 1.5.1.4 Carotid Sinus Massage ...... 12 1.5.2 Dynamic Baroreflex Testing Techniques ...... 13 1.5.2.1 Sequence Method ...... 13 1.5.2.2 Spectral Analysis ...... 14 1.5.2.3 Alpha Method ...... 17 1.5.2.4 Transfer Function Method ...... 17

viii 1.6 The relative importance of the sympathetic and cardiac baroreflex arms is debated ...... 18 1.7 Risk of Syncope Increases with Age and Female Sex ...... 19 1.8 Baroreflex Function is Different for Men and Women, and Changes with Age 20 1.8.1 Cardiac Baroreflex Function ...... 20 1.8.1.1 Sex Differences and Aging ...... 20 1.8.1.2 Functional Cardiac Response to Baroreflex Stimulation . . 20 1.8.2 Sympathetic Baroreflex Function ...... 20 1.8.2.1 Sex Differences ...... 20 1.8.2.2 Changes with Age ...... 21 1.8.2.3 Functional Sympathetic Responses to Baroreflex Stimulation 21 1.9 Responses to Carotid Sinus Massage Increase with Age ...... 23 1.10 Thesis Outline ...... 23

2 General Methodology 25 2.1 Ethical Approval ...... 25 2.2 Cardiovascular Monitoring ...... 26 2.2.1 Calculation of Vascular Resistance Using Brachial Doppler . . . . . 26 2.2.2 Measurement of Stroke Volume and Total Peripheral Resistance with the Model Flow Method ...... 26 2.3 Carotid Sinus Massage ...... 26 2.3.1 Methodological Considerations ...... 27 2.4 Neck Collar Technique ...... 31 2.4.1 Methodological Considerations ...... 33 2.5 Orthostatic Stress Test ...... 36 2.6 Valsalva Maneuver ...... 37 2.7 Cross-Spectral Analysis and α-Index ...... 38 2.8 Sequence Analysis ...... 43

3 The Agreement and Repeatability of Common Non-Invasive Baroreﬂex Assessments 46 3.1 Abstract ...... 46 3.2 Introduction ...... 47 3.3 Methods ...... 48 3.3.1 Calculation of Outcome Measures ...... 48 3.3.1.1 Agreement ...... 48 3.3.1.2 Repeatability ...... 48 3.3.2 Calculation of Agreement ...... 49 3.3.3 Repeatability ...... 49 3.4 Results ...... 50

ix 3.4.1 Participants ...... 50 3.4.2 Responses to Baroreflex Tests ...... 50 3.4.3 Agreement ...... 50 3.4.3.1 Cardiac Baroreflex Measures ...... 50 3.4.3.2 Sympathetic Baroreflex Measures ...... 52 3.4.3.3 Conversion of Baroreflex Measures ...... 52 3.4.4 Repeatability ...... 74 3.5 Discussion ...... 77 3.5.1 Agreement ...... 77 3.5.2 Repeatability ...... 78 3.5.3 Limitations ...... 79 3.6 Conclusion ...... 79

4 Sex and Age Differences in Baroreflex Function: Insights from Neck Collar Stimulation and an Orthostatic Stress Test 80 4.1 Abstract ...... 80 4.2 Introduction ...... 81 4.3 Methods ...... 82 4.3.1 Data Analysis ...... 82 4.3.1.1 Standardization of Cardiovascular Responses to Carotid Barore- ceptor Stimulation with the Neck Collar ...... 82 4.3.1.2 Calculation of Cardiac Output, Total Peripheral Resistance 84 4.3.1.3 Calculation of Percent Contribution ...... 84 4.3.1.4 Calculation of the Timing of Maximal Vascular Responses During Orthostasis ...... 85 4.3.1.5 Model Averaging ...... 85 4.4 Results ...... 85 4.4.1 Cardiovascular Responses to Orthostatic Stress ...... 86 4.4.1.1 Cardiac Responses ...... 86 4.4.1.2 Sympathetic Responses ...... 87 4.4.1.3 Cardiac and Sympathetic Baroreflex Sensitivity ...... 90 4.4.2 Cardiovascular Responses to Carotid Baroreceptor Stimulation with the Neck Collar ...... 90 4.4.2.1 Cardiac and Sympathetic Baroreflex Curves ...... 90 4.4.2.2 Absolute Responses to Neck Suction and Pressure . . . . . 94 4.4.2.3 Comparison of Cardiac Output and Total Peripheral Resis- tance Responses ...... 94 4.4.3 Baroreflex Strategies for the Maintenance of Blood Pressure During Orthostasis ...... 94

x 4.5 Discussion ...... 98 4.5.1 Cardiovascular Responses to Orthostatic Stress ...... 101 4.5.2 Cardiovascular Responses to Carotid Baroreceptor Stimulation with the Neck Collar ...... 102 4.5.2.1 Cardiac and Sympathetic Baroreﬂex Sigmoid Curves . . . . 102 4.5.2.2 Absolute Cardiovascular Responses to Neck Suction and Pres- sure ...... 103 4.5.2.3 Relative Cardiovascular Responses to Neck Suction and Pres- sure ...... 103 4.5.3 Comparison of the Vascular Responses to Orthostasis and Neck Collar Stimulation ...... 104 4.5.4 Baroreﬂex Strategies for the Maintenance of Blood Pressure During Orthostasis ...... 105 4.5.5 Limitations ...... 106 4.5.6 Conclusion ...... 106

5 The Inﬂuence of Methodology on Cardiovascular Responses to Carotid Sinus Massage 108 5.1 Abstract ...... 108 5.2 Introduction ...... 109 5.3 Methods ...... 109 5.3.1 Healthy Controls ...... 109 5.3.2 Patients ...... 110 5.3.3 Data Analysis ...... 110 5.4 Results ...... 111 5.5 Discussion ...... 112 5.5.1 Anatomical Findings ...... 112 5.5.2 Responses to Carotid Sinus Massage ...... 118 5.6 Limitations ...... 118 5.7 Conclusion ...... 119

6 Carotid Sinus Hypersensitivity: Block of the Sternocleidomastoid Mus- cle does not Aﬀect Responses to Carotid Sinus Massage in Healthy Young Adults 120 6.1 Abstract ...... 120 6.2 Introduction ...... 121 6.3 Methods ...... 121 6.3.1 Ethical Approval ...... 121 6.3.2 Protocol ...... 122 6.3.3 Screening for Carotid Stenosis ...... 122

xi 6.3.4 Cardiovascular Monitoring ...... 122 6.3.5 Carotid Sinus Massage ...... 123 6.3.6 Muscle Activation ...... 123 6.3.7 Data Analyses ...... 123 6.3.8 Statistics ...... 124 6.4 Results ...... 124 6.4.1 Muscle Activity ...... 124 6.4.2 Carotid Sinus Massage ...... 124 6.5 Discussion ...... 128 6.5.1 Limitations ...... 131

7 Discussion 133 7.1 The Agreement and Repeatability of Non-Invasive Baroreflex Techniques . . 133 7.2 Sex and Age Differences in Baroreflex Function ...... 134 7.3 The Methodology of Carotid Sinus Massage ...... 137 7.4 The Pathophysiology of Carotid Sinus Hypersensitivity ...... 138 7.5 Conclusion ...... 139 7.6 Key Concepts ...... 139

Bibliography 141

xii List of Tables

Table 1.1 Dynamic baroreﬂex testing methods ...... 7 Table 1.2 Static baroreﬂex testing methods ...... 9

Table 2.1 Example carotid sinus massage protocol...... 28 Table 2.2 Example neck collar protocol...... 33

Table 3.1 Cardiovascular responses to supine baroreflex tests ...... 51 Table 3.2 Cardiovascular responses to HUT Baroreflex Tests ...... 51 Table 3.3 Agreement between supine cardiac baroreflex measures ...... 57 Table 3.4 Agreement between HUT cardiac baroreflex measures ...... 59 Table 3.5 Agreement between supine sympathetic baroreflex measures . . . . . 61 Table 3.6 Agreement between HUT sympathetic baroreflex measures ...... 63 Table 3.7 Supine cardiac baroreflex conversion table...... 65 Table 3.8 HUT cardiac baroreflex conversion table...... 67 Table 3.9 Supine sympathetic baroreflex conversion table...... 69 Table 3.10 HUT Sympathetic baroreflex conversion table...... 71 Table 3.11 Repeatability of cardiac baroreflex tests ...... 75 Table 3.12 Repeatability of sympathetic baroreflex tests ...... 76

Table 4.1 Tilt test participant demographics ...... 86 Table 4.2 Participant menstrual status, and medications ...... 86 Table 4.3 Cardiac and sympathetic baroreﬂex sensitivity during supine and tilt 87 Table 4.4 Neck collar sigmoid curve paremeters ...... 93

Table 5.1 Participant demographics ...... 112 Table 5.2 Responses to ultrasound-guided carotid sinus massage...... 113

Table 6.1 Cardiovascular responses to placebo and lidocaine injection ...... 126

xiii List of Figures

Figure 1.1 Anatomy of baroreflex pathways ...... 2 Figure 1.2 Blood pressure distribution during orthostasis ...... 4 Figure 1.3 Subtypes of syncope ...... 6 Figure 1.4 Baroreflex sequences with different lags ...... 14 Figure 1.5 Example power spectral densities using a fast-fourier transform and an autoregressive model ...... 16 Figure 1.6 Sympathetic baroreflex function vs. age and sex ...... 22

Figure 2.1 Example trace of carotid sinus massage ...... 29 Figure 2.2 Effect of breath holding on the cardiovascular responses to CSM . . 30 Figure 2.3 Effect of breath hold on the cardiovascular responses to HUT . . . 32 Figure 2.4 Example trace of cardiac and vascular neck collar responses . . . . 34 Figure 2.5 Temporal response patterns of neck collar stimulation ...... 35 Figure 2.6 Impact of timing of stimulus onset on cardiac baroreflex response . 36 Figure 2.7 Example tilt test ...... 39 Figure 2.8 Example valsalva maneuver ...... 40 Figure 2.9 Example traces of α-index and cross-spectral cardiac baroreflex measures ...... 41 Figure 2.10 Example traces of α-index and cross-spectral sympathetic vascular baroreflex measures ...... 42 Figure 2.11 Elimination of sequence overlap in spontaneous sequence analysis . 44 Figure 2.12 Cardiac and sympathetic baroreflex sequences in one participant during supine and head-up tilt ...... 45

Figure 3.1 Agreement and repeatability protocol figure ...... 54 Figure 3.2 Example Bland-Altman plot demonstrating proportional bias and heteroscedasticity ...... 55 Figure 3.3 Cardiac and sympathetic dynamic baroreflex prediction intervals for an individual ...... 72 Figure 3.4 Cardiac and sympathetic dynamic baroreflex prediction intervals for a sample ...... 73

Figure 4.1 Sex and age differences in baroreflex function protocol figure . . . . 83

xiv Figure 4.2 Cardiac responses to HUT plus LBNP...... 88 Figure 4.3 Sympathetic responses to HUT plus LBNP...... 89 Figure 4.4 Cardiac and sympathetic baroreﬂex sensitivity during supine, HUT, and the supine-HUT response ...... 91 Figure 4.5 Averaged cardiac and sympathetic baroreﬂex sigmoid curves for younger and older men and women ...... 92 Figure 4.6 Absolute cardiac and sympathetic responses to neck suction and pressure ...... 95 Figure 4.7 Percent contributions of cardiac output and total peripheral resistance during neck suction and pressure ...... 96 Figure 4.8 Model averaging results for age and sex ...... 97 Figure 4.9 Model averaging results for men and women ...... 98 Figure 4.10 Model averaging results for old and young ...... 99 Figure 4.11 Model averaging results for constrictors and non-constrictors . . . . 100

Figure 5.1 Carotid sinus massage protocol ﬁgure ...... 110 Figure 5.2 Anatomical location of the carotid bifurcation relative to cricoid cartilage and location of maximum pulsatility ...... 114 Figure 5.3 Cardiovascular responses to carotid sinus massage at the cricoid cartilage and carotid bifurcation...... 115 Figure 5.4 Cardiovascular responses to CSM vs. distance from carotid bifurcation.116 Figure 5.5 Treatment vs. sham CSM and non-responder range...... 117

Figure 6.1 Effect of lidocaine administration on sternocleidomastoid electromyography ...... 127 Figure 6.2 Effect of lidocaine and placebo injection on the supine cardiovascular responses to carotid sinus massage ...... 128 Figure 6.3 Effect of lidocaine and placebo injection on the tilted cardiovascular responses to carotid sinus massage ...... 129

xv List of Abbreviations

AIC ...... aikeke information criteria

AR ...... autoregressive

BBFV ...... brachial blood ﬂow velocity

BRS ...... baroreﬂex sensitivity

BRSα ...... alpha baroreﬂex sensitivity

BRSα1 ...... alternative alpha baroreﬂex sensitivity

cBRS ...... cardiac baroreﬂex sensitivity

cBRSdec ...... cBRS decrease

cBRSinc ...... cBRS increase

CO ...... cardiac output

CSH ...... carotid sinus hypersensitivity

CSM ...... carotid sinus massage

CSP ...... carotid sinus pressure

DAP ...... diastolic arterial pressure

ECG ...... electrocardiogram

EMG ...... electromyography

FFT ...... fast fourier transform

FVR ...... forearm vascular resistance

HF ...... high frequency

xvi HRV ...... heart rate variability

HUT ...... head-up tilt

LBNP ...... lower body negative pressure

LF ...... low frequency

LOA ...... limits of agreement

MAP ...... mean arterial pressure

MSNA ...... muscle sympathetic nerve activity

NTS ...... nucleus tractus solitarius

PI ...... prediction interval

POTS ...... postural orthostatic tachycardia syndrome

PRT ...... pressure recovery time

RMS ...... root mean squared

RRI ...... R-R interval

SAP ...... systolic arterial pressure sBRS ...... sympathetic baroreﬂex sensitivity

SD ...... standard deviation

SV ...... stroke volume

TPR ...... total peripheral resistance

VASIS ...... Vasovagal Syncope International Study

VLF ...... very low frequency

xvii Chapter 1

Introduction

1.1 Arterial baroreceptors are crucial for short-term blood pressure control

Peripheral arterial baroreceptors consist of spray nerve endings in the splanchnic circulation, as well as the carotid, aortic, and coronary arteries, and are responsive to stretch or distension of the vessels[21]. Afferent fibers in the artery walls are stimulated physiologically by alterations in transmural distending pressure associated with changes in arterial blood pressure. Neural signals are relayed via the vagus nerve for coronary and aortic baroreceptors, and via the glossopharyngeal nerve for the carotid baroreceptors, to the nucleus tractus solitarius (NTS)[21] (Figure 1.1). From the NTS, excitatory interneurons synapse onto efferent parasympathetic nerves in the “cardioinhibitory area” (the dorsal motor nucleus of the vagus and the nucleus ambiguus). These efferent parasympathetic nerves synapse on postganglionic nerves near the heart. The NTS also connects to the vasomotor area in the ventrolateral medulla via inhibitory interneurons. The vasomotor area is then connected to preganglionic sympathetic efferent nerves via the bulbospinal pathway. When mean arterial pressure (MAP) decreases, baroreceptor firing rate decreases pro- portionally to the magnitude of the MAP reduction. Due to the excitatory and inhibitory baroreflex circuitry in the brain, the baroreflex response to a step reduction in MAP elicits an increase in sympathetic neural activity to the heart and vasculature, coupled with reduced vagal activity to the heart. Blood pressure is therefore rectified through three main effects: 1) increased heart rate and contractility; 2) increased systemic vascular resistance via constriction of the arterioles; 3) active venoconstriction, predominately in the splanchnic vasculature. The resulting increase in cardiac output, total peripheral resistance, and effective circulating volume (respectively) causes a rise in MAP[21]. The baroreflex-mediated response to a rise in MAP involves the same mechanisms acting in reverse. This negative feedback loop is crucial in maintaining systemic blood pressure and adequate cerebral perfusion during perturbations in blood pressure such as during orthostatic stress.

1 Points of Access: CSM Neck Collars Tilt table + LBNP Valsalva Oxford Method

Tilt table + LBNP Valsalva Oxford Method

Figure 1.1: Anatomy of the afferent and efferent baroreflex pathways[21]. Note that the parasympathetic efferent fibers travel together via the vagus nerve. Also, sympathetic efferent fibers only affect the venules in the splanchnic vascular beds. On the left, points of access of different baroreflex assessment techniques are noted. Splanchnic baroreceptors are not shown, but will be accessed by the same methods as the aortic arch and coronary artery baroreceptors. Dynamic baroreflex testing methods will use the aggregate baroreflex responses from all baroreceptor locations. NTS = nucleus tractus solitarius, ACh = Acetyl- choline, IX = 9th cranial nerve (glossopharyngeal), X = 10th cranial nerve (Vagus), CSM = Carotid Sinus Massage, LBNP = Lower Body Negative Pressure, NE = norepinephrine, Glu = Glutamate. Reproduced with permission.

2 1.2 Abnormal Baroreﬂex Function May Result in Syncope

While the body is in a horizontal position, arterial pressure throughout the body is approximately equal, with small reductions in pressure at distal sites due to arterial resistance[21]. However, during orthostasis the distribution of arterial pressure changes markedly[73] (Fig- ure 1.2). Upon assuming an upright posture, approximately 400mL-600mL of blood immediately accumulates in the abdomen and lower limbs due to venous pooling, with a further 300 mL accumulating over several minutes due to increased capillary filtration via increased hydrostatic pressure[27]. This fluid shift results in reduced venous return and cardiac output, leading to unloading of the baroreceptors[21]. In particular, MAP at the level of the carotid baroreceptors is 15-18mmHg lower than at the level of the heart, and lower still at the level of the brain[73]. While the cerebral vasculature has multiple overlapping mechanisms to control cerebral blood flow over a range of pressure (termed cerebral autoregulation)[3, 143, 157, 52], cerebral blood flow starts to decline below about 60 mmHg cerebral arterial pressure[73]. Reducing cerebral flow to 50-60% of supine flow causes symptoms of presyncope including, but not limited to, dizziness, nausea, and visual and auditory disturbances[104, 133]. Further reductions in global cerebral blood flow, or full interruption of flow for at least 5-8 seconds, is sufficient to cause a loss of consciousness and syncope[200]. Syncope is therefore defined as a transient loss of consciousness of short duration, rapid onset, and spontaneous recovery, caused by global cerebral hypoperfusion[156]. Experienc- ing a loss of consciousness while upright typically involves a loss of postural tone and a fall, and can result in secondary injuries, including concussion or bone fractures. Furthermore, syncope during activities of daily living such as driving, cooking, or child-rearing can be extremely hazardous. Together, these factors likely contribute to the low reported quality of life in patients with recurrent syncope, which has been compared to those living with severe rheumatoid arthritis and chronic low back pain[175]. Studies examining the prevalence of syncope most commonly estimate prevalence at 20%[11], but range as high as 41%[103]. Syncope accounts for between one[23, 16] and three[122] percent of emergency department visits. The only national estimate of the cost of syncope-related health care comes from the U.S. Healthcare Utilization Project, which estimated the annual cost of patients with a primary discharge diagnosis of syncope as $2.4 billion. This was comparable to the annual hospitalization-related costs of asthma ($2.8 billion), human immunodeficiency virus ($2.2 billion), and chronic obstructive pulmonary disease ($2.2 billion)[176]. Therefore, a thorough and nuanced understanding of baroreflex function and syncope is paramount to the treatment and prevention of recurrent syncope.

3 Figure 1.2: The distribution of arterial and venous pressure during orthostasis [73]. Repro- duced with permission.

1.3 There are Many Ways to Faint

The central feature fainting (syncope) is a transient loss of consciousness[156]. However, the physiology of syncope is distinct from other disorders that cause a transient loss of consciousness (eg. epilepsy). Physiologically, all true faints are caused by a reduction in global cerebral perfusion sufficient to cause a loss of consciousness and loss of postural tone. Different mechanisms that cause cerebral hypoperfusion can be used to classify separate syncope subtypes (Figure 1.3). The most common type of syncope is neurally-mediated (reflex) syncope, in which blood pressure falls suddenly due to sudden reflex sympathetic withdrawal, resulting in a reduction in total peripheral resistance (TPR) (vasodepressor), cardiac output (CO) (cardioinhibitory), or both (mixed). The impetus for reflex vasodilation and/or bradycardia can include vasovagal syncope (mediated by emotional and/or orthostatic stress), carotid sinus syndrome (mechanical stimulation of hypersensitive carotid baroreceptors), or situational triggers such as a cough or sneeze[130]. Cardiac syncope refers to syncope caused by a reduction in cerebral perfusion secondary to cardiac insufficiency, most commonly caused by arrhythmia (e.g. Mobitz type 2 atrioven- tricular block). However, structural disease (e.g. hypertrophic cardiomyopathy) may also lead to reduced cardiac output and MAP, causing syncope.

4 Alternative to reductions in CO, syndromes impairing the ability to increase TPR during orthostatic stress may also lead to reduced MAP and syncope. In primary autonomic failure (1◦ ANF), diseases affecting the autonomic nervous system (e.g. Parkinson’s disease, Lewy body dementia) impair baroreflex-mediated vasoconstriction. In secondary autonomic failure (2◦ ANF), peripheral nerves are affected by secondary disease (e.g. diabetes, spinal cord injury) and vasoconstriction is again impaired. Poor vasoconstrictor function during orthostasis leads to an immediate and sustained reduction in MAP until syncope, termed orthostatic hypotension. Postural orthostatic tachycardia syndrome (POTS) refers to a spectrum of fainting syndromes characterized by orthostatic tachycardia. The excessive tachycardia during orthostasis may be secondary to peripheral neuropathy (neurogenic POTS), low blood volume (hypovolemic POTS), or a chronic hyperadgrenergic state[149]. The common thread running through most subtypes of syncope is alteration of baroreflex function. While the underlying physiology and nature of the baroreflex impairment varies widely, it is clear that accurate and precise measurement of baroreflex function is essential to the understanding and treatment of the various syncope disorders.

1.4 Assessment of Baroreﬂex Function is Challenging

There is unfortunately no gold standard test to assess both the cardiac and vascular arms of the baroreflex. Instead, a large number of baroreflex methods, broadly categorized as static and dynamic, each provide limited information on baroreflex function. Dynamic methods of baroreflex measurement rely on the non-invasive evaluation of spontaneously occurring oscillations in blood pressure and heart rate, and have traditionally only examined the cardiac arm of the baroreflex. Cardiac baroreflex sensitivity (cBRS) can be determined from data obtained while the participant is at rest. These methods are summarized in Table 1.1. Static baroreflex testing methods introduce a stress to the system, and measure the baroreflex-mediated response (Table 1.2). Of these methods, only the neck collar technique provides sufficient data to create full baroreflex response curves and allows for measurement of both the cardiac and vascular responses simultaneously (see Figure 2.4). Since its first description in 1969[166], the Oxford Method has served as the “gold standard” for cardiac baroreflex assessment. The technique involves sequential infusion of vasoactive drugs to alter blood pressure, typically sodium nitroprusside and phenylephrine. Heart rate responses are the main outcome measure for this technique, but due to the administration of vasoactive drugs the vascular responses cannot be measured. Furthermore, it requires the presence of a qualified physician, and restoration of blood pressure through phenylephrine infusion can be slow, increasing the risk of profound hypotension. Lastly, the method is invasive, which may alter autonomic responses during the procedure[123, 170].

5 Figure 1.3: Classification of syncope subtypes. ANF = autonomic nervous failure, BP = blood pressure. Impaired baroreflex function is a likely common factor in many subtypes of syncope. Modified from [130].

For these reasons, our study will not include the modiﬁed Oxford Method, and will instead focus on non-invasive baroreﬂex tests.

6 Outcome Test Description Measure(s) Sequence Method Progressive increases/decreases in blood pres- Sensitivity and sure that are followed by reﬂex increases/de- coherence. creases in R-R interval are used to compute the slope of the regression line relating the two variables[13]. Transfer Function The relationship between the R-R interval and Transfer function Method systolic arterial pressure spectral frequencies is gain, coherence, assessed using cross-spectral analysis. Sensi- and phase. tivity is computed at the frequency of peak coherence in the low frequency range (0.07- 0.15Hz*)[152, 144]. Alpha Method Baroreﬂex sensitivity is calculated as the square Sensitivity. root of the ratio between systolic arterial pressure and R-R interval spectral powers over a large frequency band[137].

Table 1.1: Dynamic baroreflex testing methods. Coherence between two signals is similar in concept to a correlation coefficient, and gives an indication of the strength of the relationship at a specified frequency. Phase refers to the delay between oscillations in two signals. While similar conceptually to the notion of time delay, it is most often expressed in radians or degrees. Transfer function gain refers to the magnitude of change that occurs in the output signal (e.g. R-R interval) relative to the magnitude of change in the input signal (e.g. systolic arterial pressure) at a specified frequency. *Note that the low frequency range cited here is taken from the original publication - a frequency range of 0.04-0.15Hz is used in this thesis, see section 1.5.2.3, page 17.

7 Outcome Baroreceptors Test Description Measure(s) Strength(s) Limitation(s) Stimulated

Tilt Table Patient is passively tilted to Heart rate, vascu- Able to produce Does not quantify barore- All barore- 60-70 degrees to allow ve- lar resistance, or- presyncope in all flex sensitivity, but ceptors. nous pooling. Orthostatic thostatic tolerance. participants when baroreflex response to stress is often increased using combined with LBNP, orthostatic challenge. LBNP[147]. assesses baroreflex function to a real- world stimulus.

Valsalva Ma- Patient exhales forcefully Several indices of Test is cheap and re- May be diﬃcult for some All barore- neuver against a closed glottis at cardiac and sym- quires little time or populations to achieve ceptors. 40 mmHg for 20 seconds. pathetic baroreﬂex equipment. and maintain target Increased intrathoracic pres- function, including pressures. sure drives blood to the cBRS and sBRS. 8 extremities, causing an initial drop in blood pressure that recovers when the breath is released.

Neck Collar Variable air pressure is ap- cBRS and sBRS, Both vascular and Technique is difficult, re- Carotid plied to the neck, caus- several metrics of cardiac responses ob- quires specialized equip- barorecep- ing distention or compres- baroreflex sigmoid tained. Construction ment not commercially tors only. sion of the carotid sinus curves. of full baroreflex curves available, responses may and activation of the carotid for both variables are be subject to buffering by baroreceptors[36]. possible. other baroreceptors.

Continued on next page Outcome Baroreceptors Test Description Measure(s) Strength(s) Limitation(s) Stimulated

Carotid Si- A physician massages the pa- Absolute heart rate Test is cheap and re- Results are highly vari- Carotid nus Massage tient’s neck at the medial bor- and blood pressure quires little time or able, physiological signif- barorecep- der of the sternocleidomas- responses. equipment. icance questionable, pro- tors only. toid muscle at the level of vides only absolute car- the cricoid cartilage to acti- diac and blood pressure vate carotid sinus barorecep- responses, responses are tors. Massage is maintained buﬀered by other barore- for 5-10 seconds[130]. ceptors, usually a qualitative assessment (normal vs. abnormal).

Oxford Intravenous infusion of cBRS only. Considered the “gold Requires injection of va- All barore- Method sodium nitroprusside causes standard” baroreﬂex soactive drugs, provides ceptors. 9 vasodilation, reducing blood test, so normative data only cardiac responses, pressure. After 60 sec, and correlations with restoration of blood pres- phenylephrine is injected to other syndromes are sure can be slow. vasoconstrict and raise blood better studied. pressure[154].

Table 1.2: Static baroreﬂex testing methods. LBNP = lower body negative pressure. 1.5 Non-Invasive Baroreﬂex Tests

1.5.1 Static Baroreﬂex Testing Techniques

1.5.1.1 Responses to Orthostatic Stress

The most common stimulus for baroreflex regulation of blood pressure is orthostasis. As mentioned in section 1.2 (page 3), assumption of an upright posture involves redistribution of arterial blood in the body and activation of the baroreflex. However, in daily life the reduction in venous return is partially attenuated by activation of the skeletal muscle pump during standing and walking[21]. In order to isolate the effect of orthostasis on the cardiovascular system, passive orthostatic stress testing using head-up tilt (HUT) was introduced in the 1980s[87]. For patient groups unable to maintain an upright posture for an extended period of time (e.g. elderly, patients with spinal cord injury), negative pressure can be applied to the lower body to reduce venous return and simulate orthostatic stress[171]. Currently, orthostatic stress testing can involve any combination of the following: passive HUT, pharmacological provocation via vasodilatory drugs, passive HUT with lower body negative pressure (LBNP), or LBNP alone. Orthostatic stress testing has classically been hampered by low reproducibility, as well as low sensitivity and specificity to diagnose orthostatice intolerance[10]. For instance, passive HUT alone (typically for 20-60 minutes) often fails to evoke presyncope in patients, lower- ing the sensitivity of the test[81, 193]. Pharmacological provocations such as isoprenaline and sublingual nitrate are often used during a tilt test to increase the orthostatic challenge. However, this reduces the specificity of the test and produces high positive response rates following administration[86]. Use of vasoactive drugs also does not allow for measurement of the sympathetically-mediated vascular responses, which are arguably central to the baroreflex-mediated maintenance of blood pressure (see section 1.6, page 18). The application of negative pressure to the lower body reduces the venous pressure gradient to the heart and reduces venous return[171]. Its effects on central hypovolemia and baroreceptor unloading are comparable to HUT. However, the method has been criticized as a model of orthostatic stress due to differences between simulated orthostatic stress using LBNP and actual postural changes. These include (but are not limited to) differences in splanchnic blood flow[182], the height of the carotid baroreceptors relative to the heart[201], and activation of the vestibulosympathetic reflex[151, 206]. The development of a combined HUT plus graded LBNP protocol was a major advance in orthostatic stress testing, with high reproducibility (1.1±0.6 minutes), sensitivity (85%), and specificity (92%)[48, 71, 107]. The method has several advantages: 1) it is completely non-invasive, eliminating the risk of autonomic effects of venipuncture[170]; 2) nearly all participants reach presyncope, allowing for symptom recognition in patient populations and description of maximal cardiovascular responses and orthostatic tolerance,

10 deﬁned as the time from tilt to the onset of presyncope; 3) as no vasoactive drugs are used, sympathetically-mediated vascular responses can be assessed; 4) the high sensitivity and speciﬁcity allows the test to be used for diagnosis and the evaluation of treatments for orthostatic intolerance[147]. Accordingly, combined HUT plus graded LBNP to presyncope should be considered the “gold standard” for orthostatic stress testing[147].

1.5.1.2 Neck Collar Technique

The neck collar technique was first developed by Ernsting and Perry in 1957[51]. Their Plexiglas box sealed a pocket of air around the neck from the surrounding atmosphere, and air was removed from the box to decrease air pressure around the neck. This causes the carotid arteries to distend, and the carotid baroreceptors are stimulated. Many different iterations of collar construction have been developed, however early, rigid designs were limited to use in the participants for whom they had been designed[36]. The malleable lead neck collar after which our system is modeled was originally described by Eckberg in 1975[47], and is able to fit a wide range of participants due to the easily manipulated, malleable lead construction. There are a number of advantages to baroreceptor stimulation using the neck collar system. The method is completely non-invasive, so responses are unaffected by changes in autonomic function associated with venipuncture. Both positive and negative pressures can be applied to the carotid baroreceptors, allowing for a wide range of stimuli, and construction of full baroreflex sigmoid curves (Figure 2.4B and D). Separate aspects of the baroreflex sigmoid curve, such as centering point, operating point, saturation, and threshold can be calculated, providing a wealth of information on baroreflex function. Important parameters of baroreflex stimulation such as rate of pressure onset, intensity, and the duration of the stimulus can be tightly controlled. Lastly, vasoactive drugs are not used to stimulate baroreceptors, so both cardiac and vascular responses to carotid sinus stimulation can be quantified[36] (Figure 2.4A and C). Use of the neck collar technique has been restricted to specialized research labs for a number of reasons. Firstly, no commercial versions of a neck collar system exist, and the pressure/suction source, central controller, and collar itself need to be custom-fabricated. Secondly, achieving and maintaining collar pressures is technically challenging, particularly for high positive collar pressures. Lastly, the experimental protocol required to build full cardiac and sympathetic vascular baroreflex sigmoid curves is long and may be inappropriate for routine clinical use. The main physiological limitation of the technique is that it stimulates only the carotid baroreceptors, and responses are likely buffered during longer duration stimuli by other baroreceptor locations that are not stimulated.

11 1.5.1.3 Valsalva Maneuver

The Valsalva Maneuver was first described in 1704 by Antonio Valsalva. While originally devised to clear material from the middle ear, the maneuver has been used for decades as a routine clinical test for the evaluation of cardiovascular function[60]. A clinical Valsalva maneuver involves forcefully expiring to 40 mmHg for 15-20 seconds. This causes a series of blood pressure changes, characterized by four distinct phases, that elicit baroreflex- mediated heart rate and vascular responses. Phase I of the maneuver involves an initial rise in intrathoracic pressure upon initiation of the forced expiration, which corresponds with a transient increase in blood pressure. As intra-thoracic pressure rises, the venae cavae are occluded, and venous return drops, causing a fall in blood pressure during phase IIa. This reduction in blood pressure elicits a baroreflex-mediated increase in heart rate, as well as a burst of muscle sympathetic nerve activity (MSNA)[159]. Due to the rise in heart rate and vasoconstriction, blood pressure rises during phase IIb. Upon release of the forced expiratory pressure, blood pressure falls transiently during phase III before venous return is rapidly restored as the vena cava are decompressed. In phase IV, a blood pressure overshoot is commonly seen due to the rapid recovery of venous return[60] (Figure 2.8). Historically, clinical indices of the Valsalva maneuver have been limited to the cardiac baroreflex-mediated responses during phase IIa, and phases III and IV[108, 61, 138]. More recently, several indices of sympathetic baroreflex function have been developed with the help of microneurography[191, 159]. The Valsalva maneuver therefore provides a large number of numerical outcomes, and provides information on both cardiac and sympathetic vascular baroreflex function. The Valsalva maneuver is a quick and easy clinical maneuver to perform for most healthy individuals. However, the bulk of autonomic information available from the maneuver ne- cessitates beat-to-beat blood pressure and heart rate recording, which is not always available in the clinic. There are methodological differences in the literature regarding peak expiratory pressure and expiratory time that make comparisons between studies challenging. Lastly, holding an expiratory pressure of 40 mmHg for 15-20 seconds can be challenging for some healthy controls, and is certainly difficult in clinical populations, children, and the elderly. Interpretation of the autonomic responses to a sub-optimal Valsalva maneuver is also difficult.

1.5.1.4 Carotid Sinus Massage

CSM is a clinical diagnostic technique used to assess the carotid baroreflex in which digital stimulation of the carotid baroreceptors is employed to elicit baroreflex-mediated reductions in heart rate and blood pressure. While the test protocols have differed since its first description[96], current guidelines recommend 10 seconds of massage at the anterior margin of the sternocleidomastoid muscle at the level of the cricoid cartilage[130]. The test is

12 typically performed first with the patient supine, and then again while upright, usually on a tilt table[130]. However, many aspects of this technique vary between physicians, including the location of the massage on the neck, massage patterns, and the duration of massage. The approach is further complicated because the location of the carotid sinus varies substantially between individuals[148]. CSM is the diagnostic test used to confirm carotid sinus hypersensitivity (CSH)[130], a condition in which abnormally sensitive baroreflex responses result in hypotension and bradycardia, leading to fainting episodes (see section 1.9). An abnormal response to CSM is defined as an asystole lasting >3 seconds, and/or a fall in systolic blood pressure >50 mmHg[130].

1.5.2 Dynamic Baroreﬂex Testing Techniques

Heart rate and blood pressure continuously oscillate due to hormonal, mechanical (e.g. respiration), and neural mechanisms[183, 172]. In particular, heart rate and blood pressure are tightly controlled through a relative balance of sympathetic and parasympathetic nervous system outflow[119]. As these two variables are in a negative feedback loop with some time delay[106], baroreflex-mediated control of heart rate and blood pressure involves oscillations of both signals at 0.04-1.5 Hz[183]. These fluctuations are particularly pronounced during orthostasis, when venous pooling challenges the maintenance of MAP. Several methods exist that use these spontaneous, autonomically-mediated oscillations in this low-frequency range to quantify baroreflex function.

1.5.2.1 Sequence Method

Perhaps the simplest (mathematically) dynamic baroreflex testing technique is the sequence method, developed by Blaber et al in 1995[15]. Unlike the transfer function method and the alpha method, the sequence method does not require working in the frequency domain, and all calculations are performed in the time domain. After collecting beat-to-beat heart rate and systolic arterial pressure (SAP), progressive increases or decreases in at least three beats of systolic pressure that coincide with at least three beats of progressive reflex increases or decreases (respectively) in R-R interval (RRI) are identified[15]. Originally, Blaber et al considered sequences that occurred simultaneously, but also considered sequences with one and two beats of delay (termed "lag"). This raises the problem of overlapping sequences (Figure 1.4). Blaber et al account for this by first choosing the longest sequence, and for overlapping sequences with the same length, choosing the first sequence to begin. However, two overlapping sequences may have the same length and also start on the same beat, but with a different lag. Gulli et al (2005) address this problem by taking the sequence with the highest RRI ∼ SAP correlation coefficient. After eliminating sequence overlap, individual sequences are only accepted for final analysis if the RRI ∼ SAP correlation coefficient (R2)

13 is greater than 0.85[15]. Cardiac baroreﬂex sensitivity is then calculated as the average slope of all accepted baroreﬂex sequences.

Figure 1.4: Baroreflex sequences with different lags[15]. In A, a five-beat RRI:SAP sequence with lag zero could be counted as a four-beat sequence with lag 1, or a three-beat sequence of lag 2. In B, three-beat sequences exist at lags 0, 1, and 2. * = sequence chosen by the longest sequence (A) or the first to begin (B). Used with permission.

1.5.2.2 Spectral Analysis

While the sequence method analyses oscillations in SAP and RRI in the time domain, other dynamic methods of baroreflex measurement examine these variables by the frequencies at which they oscillate[119]. This involves examining the variability of SAP and RRI in the frequency domain. Following data collection, and removal of ectopic beats and artifact, data are resampled, typically at 4 Hz, in order to provide a time series with equally distanced time intervals. This is crucial for the next stage of data processing, which typically involves either a fast fourier transform (FFT) or autoregressive (AR) analyses[120]. FFT involves transformation of a time series into a frequency domain following division of the signal into its constituent x and y vectors. The process is fully reversible and involves no loss of information. By contrast, AR involves fitting of a model to the data. Following successful fitting of a model, a Fourier-like transform is performed to convert the signal to the frequency domain. AR is widely used due to its greater ability to outline distinct frequency bands with little noise

14 compared to FFT[183]. However, AR is problematic in that the “optimum” model order to produce a model that reflects true physiological processes can be difficult to determine[97]. Differences in model order selection may significantly impact results. In addition, the model chosen for the dataset may explain only part of the variance of the overall signal, which may bias the results[183]. There are typically three frequency regions of interest in the power spectrum density plot: very low frequency (VLF) of 0-0.04 Hz, low frequency (LF) of 0.04-0.15 Hz, and high frequency (HF) of 0.15-0.4 Hz (HF)[183, 6, 5, 79, 136, 145] (Figure 1.5).

15 16

Figure 1.5: Example power spectral densities using a fast-fourier transform and an autoregressive model. Oscillations in R-R intervals in the time domain (A) can be examined in the frequency domain using a fast-fourier transform (B) or an autoregressive model (C). Similarly, oscillations in systolic arterial pressure (D) can also be examined in the frequency domain with a fast-fourier transform (E) or autoregressive model (F). The frequency ranges are indicated here with dashed lines, with very low frequency range indicated with a white background, low frequency represented with light grey, and high frequency represented with dark grey. RRI = R-R interval, SAP = systolic arterial pressure. Oscillations in RRI and SAP in the HF range are believed to be due to mechanical influences of respiration on heart rate and thoracic vasculature, as well as the changes in cardiac output due to respiratory sinus arrhythmia[135, 139]. The VLF frequency band generally accounts for the greatest percentage power of the three frequencies. Oscillations of RRI in the VLF range are thought to be due to a combination of neural and hormonal influences, most prominently the parasympathetic nervous system as well as the renin-angiotensin system[184]. RRI VLF power has been previously shown to correlate with mortality following myocardial infarction[14, 98]. However, the physiological underpinnings of oscillations in the VLF range are less well understood. The LF band is perhaps more interesting from a physiological standpoint, as it appears to reflect the relative balance of sympathetic and parasympathetic nervous system outflow[135, 139]. Oscillations of LF RRI is generally accepted to reflect baroreflex function[4, 28]. These are thought to, in turn, be driven by sympathetically-mediated blood pressure oscillations in the LF band[64]. Thus the RRI and blood pressure LF bands provide information about baroreflex function.

1.5.2.3 Alpha Method

The alpha method was ﬁrst introduced in 1988 by Pagani et al[137]. First, the autoregressive power spectrum of SAP and RRI is calculated. Spontaneous baroreﬂex sensitivity is estimated by computing the square root of the ratio of SAP and RRI spectral power.

Pagani calculated two alpha indices: 1) αLF , calculated over a frequency band centered at

0.1 Hz, and 2) αHF , calculated over a frequency band centered at 0.25 Hz. The upper and lower limits of these frequency bands are not described in the original paper[137]. However, a later paper from the same group[115] deﬁnes the frequency bands as LF = 0.03-0.15, and HF = 0.15-0.35, citing the 1996 task force guidelines on heart rate variability (HRV)[183]. Thus, baroreﬂex sensitivity (BRS) is calculated via the alpha index as:

R 0.15 !0.5 0.03 (P owerRRI ) V agal α-Index = R 0.15 0.03 (P owerSAP )

1.5.2.4 Transfer Function Method

The transfer function method was originally described by Robbe et al in 1987[152]. It is a metric of BRS obtained using cross-spectral analysis, which refers to the assessment of the relationship between the RRI and SAP spectral frequencies (Figures 2.9 and 2.10). The coherence between these signals (similar in concept to a correlation coefficient) gives an indication of the strength of the relationship at the specified frequency[152]. For instance, a coherence of 0.5 means that 50% of the variance in the output signal is explained by the input, and is commonly used as a cutoff value. Signals with coherence values of less than 0.5 are commonly designated as ”poor“ coherence, and are not investigated further. The

17 main baroreflex-related outcome measure of the transfer function method is the transfer function gain, and is used as an index of BRS. The transfer function method has two major limitations. Firstly, the method is computa- tionally complicated, which has lead to the development of “black box” computer software where significant aspects of the calculations (use of autoregressive model or fast-fourier transform alone, model order, visual inspection and elimination of ectopic beats, etc.) are hidden. Alternatively, custom software is developed by individual research groups, making comparisons between studies questionable. Secondly, in participants with low coherence (due to deficient baroreflex function or an unloaded baroreflex, as during supine rest), transfer function gain cannot be calculated.

1.6 The relative importance of the sympathetic and cardiac baroreﬂex arms is debated

The relative importance of the sympathetic and cardiac baroreflex arms in the overall maintenance of MAP during orthostasis is still unclear. Proponents of the cardiac baroreflex responses argue that individuals with high orthostatic tolerance to LBNP have larger heart rate responses than those with lower orthostatice tolerance. Accordingly, orthostatic tolerance is reduced when the cardiac arm of the baroreflex is blocked[33]. The argument that the cardiac baroreflex is the major contributor to the overall maintenance of blood pressure has been criticized[201]. There are several lines of evidence indicating that the vascular responses contribute more to baroreflex-mediated blood pressure maintenance. For instance, if cardiac responses are indeed important, patients with cardiac transplants (and thus no autonomically-mediated changes in heart rate upon standing) should exhibit poor orthostatic tolerance. However, these patients have intact blood pressure control during orthostasis. Conversely, patients with POTS produce excessive cardiac responses upon standing but suffer from orthostatic presyncope[201]. Taken together, these two examples highlight that baroreflex-mediated cardiac responses are insufficient to support blood pressure during orthostasis. In support of sympathetic predominance, patients with significantly reduced ability to vasoconstrict due to sympathetic vasomotor lesions display profound orthostatic hypotension during orthostasis[201]. In healthy controls, there is a robust correlation between the large vascular resistance response and increased orthostatic tolerance during orthostatic stress using HUT with LBNP[30]. Much of the evidence in support of the importance of cardiac responses to orthostatic stress comes from studies using only LBNP, which may not be generalizable to true orthostasis. For instance, during LBNP the vestibulosympathetic reflex is not activated, which has been shown to have an additive interaction for the increase in MSNA (and therefore vasoconstriction) during orthostasis[151]. During HUT, increased splanchnic pool-

18 ing activates subdiaphragmatic baroreceptors that likely contribute to baroreflex-mediated vasoconstriction[43]. However, the effect of LBNP alone on splanchnic blood flow is in opposition to that during HUT, likely resulting in a reduced pressor response. Lastly, the carotid baroreceptors are kept at the same level as the heart during supine LBNP, and are likely not unloaded to the same degree as during standing[73] (Figure 1.2). The nature of baroreflex stimulation via LBNP alone, therefore, appears distinct from that of orthostasis, and likely underlies the disparate baroreflex responses to each stimulus. During orthostasis, stroke volume (SV) and CO are reduced[173], leaving TPR as the main mechanism for the maintenance of MAP. While a minimum cardiac response to orthostasis is necessary, the baroreflex control of resistance and splanchnic capacitance vessels appears to be the more important mechanism for the maintenance of MAP during orthostatic stress. Certainly if this important point is to be resolved the accurate and reproducible determination of both cardiac and vascular baroreflex control will be crucial.

1.7 Risk of Syncope Increases with Age and Female Sex

The population is aging rapidly, with 9% (660 million people) aged >65 years worldwide, and this figure is expected to continue to rise substantially[168, 169, 188]. The risk of syncope is high in older adults[130], with a three-fold higher risk of syncope in adults aged 70-79 years than adults aged 40-49 years[155]. This increase likely contributes to falls and fall-related injuries in older adults[83]. The impact of syncope and associated falls in older adults is severe - elderly inpatients who faint display a 13-33% one-year mortality and a 79% four-year mortality[187]. Although the pathophysiology of syncope in the elderly is complex and incompletely understood, abnormal baroreflex function, particularly during postural change, is implicated[167]. The risk of syncope in men increases approximately linearly, while in women the risk is bimodal[155], and young women are at increased risk of syncope compared to young men. The distribution of syncope subtypes also varies with age. For instance, young women are at increased risk for vasovagal syncope and POTS, while older individuals are more likely to experience orthostatic intolerance and cardiac syncope[130]. While the underlying causes of the different subtypes of syncope are complex and incompletely understood, baroreflex dysfunction is likely a common factor due to its central role in the regulation of orthostatic cardiovascular homeostasis. In addition, age- and sex-related changes in baroreflex function may be at least partly responsible for the varying mechanisms and susceptibility to syncope through the lifespan and between sexes.

19 1.8 Baroreﬂex Function is Diﬀerent for Men and Women, and Changes with Age

1.8.1 Cardiac Baroreﬂex Function

1.8.1.1 Sex Diﬀerences and Aging

Multiple lines of evidence indicate reduced cBRS with aging[128, 36, 77]. While the exact mechanisms are not known, age-related increases in arterial stiffening have been proposed to reduce baroreceptor stimulation during alterations in MAP[127]. Reduced cardiac efferent baroreflex function may also contribute to declining cBRS with age[128]. Resting cBRS is greater in men than women[101].

1.8.1.2 Functional Cardiac Response to Baroreﬂex Stimulation

Irrespective of age, men achieve a similar cBRS as women upon HUT. However, as their resting cBRS is higher, men mount a larger cBRS response to HUT than women[102, 9]. Baroreﬂex-mediated heart rate responses to orthostasis are larger in younger individuals vs. older individuals[56], and larger in men than women[158].

1.8.2 Sympathetic Baroreﬂex Function

1.8.2.1 Sex Diﬀerences

If baroreflex-mediated changes in vascular resistance during orthostasis are indeed the main mechanism of maintaining MAP, one should see differences in sympathetic neural baroreflex function between men and women, as women have lower orthostatic tolerance than men[147] and are at an increased risk of syncope[155]. Resting MSNA is greater in men than women[85], but with respect to the sympathetic neural response to orthostasis, there is evidence both for and against sex differences. During HUT, plasma norepinephrine[116] and MSNA burst frequency[58, 162] rises similarly in young men and women, reflecting similar neural responses to orthostasis. However, Shoemaker et al (2001) argue that young men display increased burst amplitude and decreased MSNA burst delay, which reflects an increase in sympathetic outflow via different central neural recruitment patterns[162] (Fig- ure 1.6B). Young men may therefore be better able to modulate sympathetic outflow to the vasculature in response to orthostasis. There is evidence from frequency domain analysis that suggests larger diastolic arterial pressure (DAP)-MSNA coherence in men compared to women at rest and during LBNP, indicating a closer relationship between MSNA and DAP in men[205]. Perhaps more persuasive is the evidence indicating sex differences in sympathetic vascular transduction, ie. the amount of vasoconstriction elicited by a given level of sympathetic nerve activity. Following the development of direct sympathetic nerve recordings

20 via MSNA, researchers noted a perplexing lack of correlation between resting sympathetic nerve activity and TPR[177, 164] in young men and women. In young men, MSNA displays a positive relationship with TPR and a negative relationship with CO, accounting for the poor relationship between MSNA and MAP. However, young women display neither relationship[75]. This is due to the enhanced β2-adrenergic vasodilatory response to noradrenaline release in young women, which oﬀsets the vasoconstrictor eﬀects of α1-adrenergic receptor activation[93, 74].

1.8.2.2 Changes with Age

Resting MSNA increases with age in both sexes[131] (Figure 1.6A). While the relationship between resting MSNA and TPR is not signiﬁcant in younger individuals, it becomes signiﬁcant in both sexes with age. In men, the negative relationship between resting CO and MSNA progressively declines throughout life, creating a positive relationship between resting MSNA and MAP[131]. In women, hormonal changes during menopause reduce the

β2-adrenergic vasodilatory response to noradrenaline release, again producing a positive relationship between MSNA and TPR[74]. The sympathetic neural response to orthostatic stress appears to be maintained with age[128, 44] (Figure 1.6C). However, vascular transduction declines in older men, likely due to reduced α-adrenergic responsiveness[22] (Figure 1.6D). Conversely, the reduction in

β2-adrenergic vasodilatory response to noradrenaline release in postmenopausal increases sympathetic vascular transduction[74, 22]. Vascular transduction in aging men and women appears to intersect at about age 55 years[22].

1.8.2.3 Functional Sympathetic Responses to Baroreﬂex Stimulation

In response to baroreflex activation via orthostasis, the rise in TPR during tilt is larger in young men vs. young women[55, 62, 34]. However, the TPR response to orthostasis also increases in both age groups, with older women exhibiting the largest vasoconstrictor response to orthostatic stress[56, 124]. This is in opposition to the change in vascular transduction seen in older men, but may reflect compensation for reduced cardiac baroreflex sensitivity[101]. Early investigations from the neck collar protocol (Section 1.5.1.2, page 11) indicate reduced maximum DAP responses to neck suction (a hypertensive stimulus) with age[70], indicating a reduced sympathetically-mediated vasoconstrictor response. However, they did not perform a sex comparison, likely due to the low number of older women in their sample. Kim et al (2011)[92] demonstrated reduced pressor responses to neck suction (a hypertensive stimulus) in young women compared to young men. However, both sexes responded similarly to neck pressure (hypotensive stimulus)[92].

21 Figure 1.6: Sympathetic baroreflex function changes with age and sex. A) Resting muscle sympathetic nerve activity (MSNA) rises with age in both sexes, but rises more in females[131]. B) The MSNA response to head-up tilt is not different for young men and women when assessing only the change in burst frequency. However, males increase the MSNA burst amplitude in response to tilt, while females do not[162]. C) The MSNA response to changes in diastolic arterial pressure induced by the modified Oxford technique is similar across ages, but shifts to higher values of diastolic pressure and MSNA with increasing age[45]. D) The effect of age on vascular transduction (the amount of vasoconstriction elicited by a burst of muscle sympathetic nerve activity) is different for men and women. Vascular transduction decreases with age in men, but increases with age in women, intersecting at about age 55 years. The slopes of the regression lines were statistically different (p < 0.001). Green circles indicate outliers that were removed from the analysis[22]. Reproduced with permission.

The MAP response to neck suction is similar between young and older women; however, older women exhibited a larger contribution of TPR to the response. During neck pressure, MAP responses were blunted in older women, with similar contributions from TPR for both young and older women[40]. The physiology underlying differences in responses to a hyper- vs. hypotensive stimulus is not clear. The body of work examining the changes in sympathetic baroreflex function with age and sex remains incomplete. To date, full vascular sympathetic baroreflex curves have not been constructed to compare sympathetic vascular baroreflex function between sexes and with age. Furthermore, our understanding of sympathetic vascular baroreflex function during orthostasis is limited, as continuous measures of sympathetic vascular baroreflex sensitivity have previously not been available.

22 1.9 Responses to Carotid Sinus Massage Increase with Age

The aggregated evidence for a decline in baroreflex-mediated cardiac responses with age were highlighted in section 1.8.1.1. In stark opposition to this, the responses to CSM increase with age, with a high percentage of healthy older adults (35% aged >70 years) exhibiting significant asystoles (>3000 msec)[96]. The reason for this discrepancy is not clear - isolated stimulation of the carotid baroreceptors does not appear to explain the phenomenon, as the cardiac responses to neck suction and pressure decline with age[70]. However, due to the proximity of the vagus nerve to the location of CSM[130], the large responses to CSM may relate to digital mechanical stimulation of the efferent vagus nerve during CSM. Several pathophysiologic mechanisms have been proposed to explain the large cardiovascular responses to CSM, including generalized autonomic dysfunction, reduced carotid sinus compliance, and sternocleidomastoid proprioceptive denervation[7]. However, the pathophysiology of large cardiovascular responses to CSM remains unclear. The question of the mechanism by which CSM alters heart rate (via stimulation of the carotid baroreceptors vs. stimulation of the vagus nerve), in addition to the opposing relationship with age, challenges the utility of CSM as a baroreflex testing technique. As much of the data regarding baroreflex function in older adults stems from this technique, its limitations also likely cloud our understanding of the effect of aging on baroreflex function.

1.10 Thesis Outline

This thesis investigates baroreflex function from a varied perspective, including characterization of the cardiac and sympathetic vascular arms of the baroreflex via different testing methods. The changes in cardiac and vascular baroreflex function with age and sex are explored, and set within the larger context of the risk of syncope in these subgroups. My first study (Chapter 3) examined the agreement and repeatability of commonly-used non-invasive baroreflex tests. This study was not hypothesis driven, but provides a variety of tools to aid the use and interpretation of baroreflex measurement. My second study (Chapter 4) examined sex and age differences in the baroreflex- mediated maintenance of blood pressure using simultaneous measures of cardiac and sympathetic vascular baroreflex function. I hypothesized that: 1) robust vascular responses to orthostatic stress will correspond with greater orthostatic tolerance; and 2) this relationship will vary based on different age and sex subgroups. My third study (Chapter 5) examined the effect of different methods of carotid sinus massage on cardiovascular responses to this stimulus. I hypothesized that the magnitude of responses would be larger with CSM performed at the ultrasound-guided location of the carotid bifurcation.

23 My last study (Chapter 6) investigated the pathophysiology of carotid sinus hypersensitivity. I hypothesized that sensory block of the sternocleidomastoid muscles would interrupt central integration of sensory and baroreﬂex aﬀerent information, increasing the responses to CSM. In Chapter 7, I discuss the overall implications of this body of work, including limitations and avenues for future research.

24 Chapter 2

General Methodology

Chapters 3-5 of this thesis share similar experimental protocols. Each of the following protocols was completed on a different day, and the time of day was standardized for each participant. Following familiarization to the laboratory and the experimental protocol, participants provided written informed consent and completed a brief medical history. Participants underwent baroreflex testing on three different testing dates, with the order randomized. We considered four static baroreflex methods (the variable pressure neck collar, the Valsalva maneuver, responses to an orthostatic stress test, and carotid sinus massage) and three dynamic baroreflex methods (the alpha-index, cross spectral analyses, and the spontaneous sequence method) derived from data collected during the baseline supine rest period of the orthostatic stress test and neck collar protocols. The Valsalva maneuver and orthostatic stress test occurred on the same day, while the carotid sinus massage and neck collar protocols were performed on separate days. Tests were performed at the same time of day to account for circadian fluctuations in baroreflex function[35]. Prior to each test day participants were asked to eat only a light meal, and to abstain from alcohol, caffeine, and strenuous exercise for 12 hours.

2.1 Ethical Approval

These experiments were approved by the Department Of Research Ethics at Simon Fraser University and conform to the principles outlined in the Declaration of Helsinki[203]. Prior to testing, participants provided written informed consent, and completed a brief medical history to conﬁrm they met our inclusion criteria. General exclusion criteria included: pregnancy; pre-existing cardiovascular or neurological disease; or use of medications with cardiovascular actions. Additional exclusion criteria applied to speciﬁc tests and are detailed elsewhere.

25 2.2 Cardiovascular Monitoring

On each test day we continuously monitored beat-to-beat cardiac (RRI: electrocardiogram (ECG); Lead II) and blood pressure responses using the Finometer Pro (Finapres Medi- cal Systems BV, Amsterdam, Netherlands). Finger cuff measurements were calibrated to brachial blood pressure using the return-to-flow calibration [66] prior to commencing testing, and throughout using the internal calibration (PhysiocalTM). Prior to data recording, the PhysiocalTM was turned off to prevent interruption of the waveform. The participants’ hand was kept at approximately heart height throughout testing, and a height correction unit was used to account for small changes in vertical height between the finger cuff and the heart [66]. Brachial artery blood flow velocity was measured using an 8 MHz probe positioned overlying the brachial artery and clamped at a constant angle throughout testing (Doppler-BoxTM, Compumedics, DWL, Singen Germany). The mean blood flow velocity was calculated as the area under the curve of the waveform signal. Data were sampled at 1000 Hz with an analog-to-digital converter (Powerlab 16/30, AD Instruments, Colorado Springs, CO), acquired using LabChart (version 10, AD Instru- ments), and stored for offline analysis. ECG data were low-pass filtered at 50 Hz to remove electrical noise. A median filter was applied to the brachial blood flow signal to remove artifacts. Beat-to-beat heart rate, SAP, and DAP were extracted from the beat-to-beat blood pressure waveform using cyclic peak detection algorithms in Labchart. MAP was calculated as MAP = DAP + 1/3 × (SAP − DAP ). Forearm vascular resistance (FVR) was calculated as MAP divided by mean brachial blood flow velocity (BBFV).

2.2.1 Calculation of Vascular Resistance Using Brachial Doppler

The measurement of FVR via brachial Doppler during HUT was first introduced in 2000[26]. Doppler ultrasound measures flow velocity, and the magnitude of the measured velocity is highly dependent on the angle of insonation. Thus it is essential that 1) the angle of insonation is less than 60◦, as the error associated with flow velocity measurements increases exponentially with the angle of insonation[186]; and 2) the angle of insonation does not change throughout the recording. As we measure only flow velocity, which is highly dependent on the size of the artery, and therefore the size of the individual, our calculations of FVR are expressed as percentage changes from supine rest.

2.2.2 Measurement of Stroke Volume and Total Peripheral Resistance with the Model Flow Method

2.3 Carotid Sinus Massage

The carotid arteries were screened for the presence of stenosis using ultrasound, and participants were excluded if >50% carotid stenosis was detected[112]. Examinations were

26 conducted using the GE Logiq I system (GE Healthcare, Chicago, Illinois, USA), with a 6.3 MHz linear transducer. The carotid arteries on both sides were evaluated in both transverse and longitudinal planes for the presence of visible significant narrowing on B-mode greyscale image and the presence of mosaic pattern on colour Doppler image. Examinations were performed with the standard presets for carotid ultrasound initially, with technician optimisation where necessary. The location of the carotid bifurcation was identified and marked on the skin. The locations of the maximal carotid artery pulsatility[90] and cricoid cartilage[25] were also identified. The distance between the gonion of the mandible and the locations of maximum pulsatility, cricoid cartilage, and ultrasound-guided bifurcation were recorded. After cardiovascular instrumentation and a 10-minute supine rest period, participants then underwent supine and tilted CSM in duplicate, on both sides of the neck at the ultrasound-guided level of the carotid bifurcation and at the level of the cricoid cartilage. A total of eight CSM were therefore completed in both supine and HUT positions (Table 2.1), with two additional sham procedures in each position in which the fingers were placed overlying the carotid sinus on each side of the neck, but no pressure was applied. To reduce the influence of respiratory sinus arrhythmia on the measured cardiac responses to CSM, participants performed a breath hold at normal end-expiration for 15 seconds, with CSM occurring during the last ten seconds of the breath hold. This breath hold length is long enough to limit respiratory sinus arrhythmia but is not long enough to provoke a significant chemoreflex response. Cardiac responses were calculated as both the maximum RRI elicited during the stimulus, and the maximum increase in RRI relative to the baseline RRI (the average RRI over the five seconds prior to massage onset). Systolic arterial pressure responses were calculated as the minimum SAP elicited during the stimulus relative to the baseline SAP. Vascular resistance responses were calculated as the minimum FVR during the stimulus, expressed as the percentage change from baseline (the average FVR over the 5 seconds prior to massage onset, Figure 2.1).

2.3.1 Methodological Considerations

We conducted CSM during breath hold at end-expiration to reduce the eﬀects of respiratory sinus arrhythmia. While this is common practice during examination of the cardiac responses to neck suction/pressure[36], clinical CSM is performed during normal breathing[25], and the diﬀerent breathing protocol could theoretically alter the cardiovascular responses to CSM. To examine this, we performed repeated CSMs during normal breathing and end-expiration in 12 young, healthy participants.

27 Table 2.1: Example of one carotid sinus massage (CSM) protocol. One minute of rest separated each massage to allow cardiovascular responses from the CSM to return to baseline. Following head-up tilt (HUT), we waited for at least two minutes before administering CSM to allow for the initial reduction in blood pressure to be rectified via baroreflex responses. Following CSM at the ultrasound and cricoid locations, two sham massages were performed during supine and HUT. During the sham procedure, participants performed a typical breath hold, and fingers were lightly placed overlying the carotid sinus, but CSM was not performed.

28 Figure 2.1: A representative example tracing from one individual showing cardiac, blood pressure, and forearm vascular resistance responses to carotid sinus massage. The period of massage is indicated by the dashed lines. Note the modest bradycardia and vasodilation induced by the massage. Responses were calculated as the appropriate minimum (Min) or maximum (Max) value (indicated) compared to the mean value for ﬁve seconds prior to the onset of massage (baseline). CSM = carotid sinus massage, ECG = electrocardiogram, RRI = R-R interval, BP = blood pressure, SAP = systolic arterial pressure, BBFV = brachial blood ﬂow velocity, FVR = forearm vascular resistance.

29 30

Figure 2.2: The effect of breath holding on the cardiovascular responses to CSM. A) There was no detectable difference between the change in RRI with CSM during normal breathing vs. breath hold in supine or tilted positions. B) The tilted SAP response to CSM during normal breathing was larger than the tilted SAP response with breath hold. During normal breathing SAP responses increased upon tilt, but did not increase with breath holding. C) The tilted FVR response to CSM during normal breathing was larger than the tilted FVR response with breath hold. During normal breathing, FVR responses were larger with tilting, but not during breath holding. RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance, NB = normal breathing, BH = breath hold. * = p < 0.05 between indicated conditions. During the tilted position only, the SAP and FVR responses to CSM were larger with normal breathing than during breath hold (Figure 2.2). However, the magnitude of this effect was small (<5mmHg for SAP and <10% for FVR), and is unlikely to affect further investigations. The SAP and FVR responses to CSM were larger during tilt for the normal breathing condition only. It is likely that oscillations in thoracic pressure during breathing caused oscillations in both SAP and FVR, which may have resulted in our detecting larger minimum values of both variables during CSM. We also performed CSM in the tilted position with repeated breath holds. As the fluctuations in thoracic pressure during respiration contributed to venous return[202], repeated interruption of normal breathing during orthostasis may have altered the baroreflex-mediated response to tilt and in turn altered the responses to CSM. To examine this, we performed two 10-minute tilts to 70◦ in 10 young, healthy participants. Between tilts, participants rested in the supine position for 10 minutes to allow return of cardiovascular homeostasis. During one of the tilts (order randomized), participants performed eight 15-second breath holds at one-minute intervals. Heart rate, SAP, SV, and CO were assessed during the breath holds and compared to time-matched periods during normal breathing. We found no difference in any outcome measure during the breath hold vs. normal breathing tilts (Figure 2.3). Therefore, the effect of breath holding on the responses to CSM during tilt are minimal.

2.4 Neck Collar Technique

Following instrumentation, participants rested in a supine position for 10 minutes. A malleable lead neck collar[47] was fitted to the participant so that a pocket of air anterior to the neck was isolated from atmospheric pressure. A rheostat device modified current flow to a pressure source to generate graded positive and negative pressures (-60, -40, -20, 0, 20, 40, and 60 mmHg). Pressure onset (300ms) and offset (650ms) in the collar were controlled via solenoid valves. To assess predominately vagally-mediated cardiac responses, participants were instructed to breath hold at the end of normal expiration for ten seconds to reduce respiratory sinus arrhythmia. Negative/positive pressure was applied during the last five seconds of breath hold, and responses were calculated as the maximum or minimum RRI during the stimulus for neck suction and pressure respectively minus the baseline RRI (the average RRI over the 5 seconds prior to stimulus onset) (Figure 2.4A). Sympathetic vascular responses are slower in onset[80, 76] and so were assessed during normal breathing and application of 20 seconds of neck suction/pressure. Sympathetic vascular responses were reported as the maximum increase or decrease in FVR for neck pressure or suction respectively, expressed as the percentage change in FVR during the

31 Figure 2.3: The effect of breath hold on the cardiovascular responses to head-up tilt. Car- diovascular variables were assessed during breath hold and time-matched sections of the normal breathing tilt, and averaged over all breath hold repetitions. There were no differences between breath hold and non-breath hold tilts for systolic arterial pressure (A), heart rate (B), stroke volume (C), or cardiac output (D). NB = normal breathing, BH = breath hold, SAP = systolic arterial pressure, HR = heart rate, SV = stroke volume, CO = cardiac output. stimulus relative to the baseline (five seconds preceding the onset of the stimulus) (Figure 2.4C). Cardiac and sympathetic vascular baroreflex time delay was calculated as the delay between the onset of neck collar pressure/suction and the onset of the directionally appropriate maximum/minimum response[146, 204, 40, 50] (Figure 2.4A and 2.4C).

32 Duration (s) Pressure (mmHg) Breathing Protocol

20 -20 Normal breathing 20 -40 Normal breathing 5 -60 Breath hold 5 0 Breath hold 20 0 Normal breathing 5 +20 Breath hold 5 +40 Breath hold 5 -20 Breath hold 5 +60 Breath hold 20 +60 Normal breathing 20 +40 Normal breathing 20 -60 Normal breathing 20 +20 Normal breathing 5 -40 Breath hold

Table 2.2: Example of one neck collar protocol. Each pulse was repeated twice before moving to the next pulse protocol, with at least one minute separating each pulse to allow washout of the cardiovascular responses.

Each pressure stimulus was applied in duplicate for both the cardiac and sympathetic protocols. The order of timing protocol and pressure was randomised (Table 2.2). Carotid sinus pressure (CSP) was calculated from MAP – neck collar pressure. Barore- flex sigmoid curves were fitted to the data using a non-linear least-squares algorithm (nlsLM function from the minpack.lm package, R version 3.4.2)[160]. Cardiac BRS was calculated as the maximum or minimum of the first derivative of the cardiac or sympathetic vascular sigmoid curve, respectively. Centering point was calculated as the CSP corresponding with the cardiac/sympathetic BRS. Saturation and threshold were calculated as the maximum/minimum of the second derivative of the baroreflex sigmoid curve (Figure 2.4B and 2.4D). Operating point, or the resting blood pressure at which the baroreflex ”operates,“ was calculated as the averaged baseline CSP from all neck collar repetitions, split by five- or twenty-second pulses.

2.4.1 Methodological Considerations

It is important to note that we did not time the onset of neck pressure/suction to the p wave, as has been previously performed[46]. Eckberg (1976) found that timing of the stimulus onset relative to the p wave of the next heart beat inﬂuenced the magnitude of

33 Figure 2.4: Example traces of cardiac and vascular neck collar responses. A) Vagally- mediated cardiac responses to neck pressure/section were elicited during 5 seconds of neck pressure/suction during a 10-second breath hold at end-expiration. Responses were calculated as the max/min RRI (for pressure/suction respectively) during the stimulus minus baseline RRI (5 seconds preceding the stimulus). B) Cardiac BRS was calculated as the max of the first derivative of the curve. Centering point was calculated as the CSP corresponding with the max BRS. Threshold and saturation points of the curve were calculated as the CSP corresponding with the max/min values of the second derivative of the sigmoid curve, respectively. Operating point was calculated as the averaged baseline CSP from all 5-second neck collar repetitions. C) Sympathetically-mediated vascular responses were elicited during 20 seconds of neck pressure/suction during normal breathing. D) Sympathetic BRS was calculated as the min of the first derivative of the sigmoid curve. Centering point was calculated as the CSP corresponding with the min BRS. Threshold and saturation points of the curve were calculated as the CSP corresponding with the min and max values of the second derivative of the sigmoid curve, respectively. Operating point was calculated as the averaged baseline CSP from all 20-second neck collar repetitions. RRI = R-R interval, FVR = forearm vascular resistance, NCP = neck collar pressure, BRS = baroreflex sensitivity, CSP = carotid sinus pressure, Thr = threshold, CP = centering point, Sat = saturation, Max = maximum, Min = minimum. Forearm vascular resistance was calculated as MAP divided by brachial blood flow velocity, and re-expressed as the percent change from baseline. the cardiac response[46] (Figure 2.5). However, his protocol used stimulus durations less than 1.5 sec. This shorter stimulus length may not capture the maximal cardiac response to

34 carotid baroreceptor stimulation, which may take up to two cardiac cycles to manifest[70]. With this in mind, we used a ﬁve-second stimulus length that has been used previously[36].

Figure 2.5: Temporal response patterns of neck collar stimulation at 0.58 sec and 1.13 sec stimuli lengths[46]. Notice that for stimulus durations < 1.5 seconds, the cardiac response magnitude depends on the timing of the impulse relative to the next p wave of the electrocardiograph. Used with permission.

It is unclear whether timing to the ECG inﬂuences the response magnitude with the longer stimulus length. Previous studies using this stimulus length also did not time the stimulus onset to the p wave[36]. In order to address this question, we performed repeated carotid sinus stimulations using a variable-air-pressure neck collar in seven participants. The onset of each stimulus was applied randomly and a wide spread of stimulus onset times were obtained. Figure 2.6 indicates the results from a linear regression analysis, which failed to identify a relationship between stimulus-intensity-corrected cardiac responses and timing onset to the next p wave. This is likely due to the stimulus duration lasting over several heartbeats, so Eckberg’s timing "hotspot" occurs within our 5-second pressure pulse (Figure 2.5).

35 Figure 2.6: Impact of timing of stimulus onset on cardiac baroreflex responses. Responses to a 5 s stimulus of -40 mmHg neck suction (A) and + 40mmHg neck pressure (B) applied with variable timing relative to the subsequent p wave can be seen in 7 healthy controls. Responses for each individual are denoted in the different colours. There was no significant relationship between timing of stimulus onset and magnitude of response. Solid lines denote the linear regression of the entire data set for each stimulus. Dashed lines show the 95 % confidence intervals.

2.5 Orthostatic Stress Test

Following instrumentation, participants performed an orthostatic stress test consisting of HUT combined with graded LBNP and continued until presyncope[147]. Participants rested in the supine position for 20 minutes before being head-up tilted to 70◦. After 20 minutes of HUT, graded LBNP was applied at -20, -40, and -60 mmHg for 10 minutes each. The test was stopped if any of the following criteria were met: SAP dropped below 80 mmHg; heart rate decreased to below 50 bpm or increased to higher than 170 bpm; the participant experienced symptoms of presyncope such as dizziness, warmth, and requested to stop; or the protocol was completed. Orthostatic tolerance was calculated as the time at presyncope minus the onset of HUT (Figure 2.7). Following HUT, cardiovascular data were averaged every 30 seconds. The maximum response for each parameter was calculated as the peak averaged value during the orthostatic stress minus the average from the last ﬁve minutes of supine ("Baseline" in Figure 2.7). The maximum FVR response was calculated as the maximum percentage increase from the supine average. Data during presyncope (deﬁned as thirty seconds preceding the minimum SAP preceding termination of the test) were excluded from determination of maximal responses.

36 2.6 Valsalva Maneuver

Following familiarization with the equipment, participants performed at least one practice Valsalva maneuver in the supine position. Participants were instructed to forcefully expire against a closed glottis to 40 mmHg for 20 seconds. The Valsalva strain was quantified with a digital pressure transducer attached to the expiratory tubing and participants had continuous visual feedback of the Valsalva strain achieved (Figure 2.8). Valsalva maneuvers were repeated up to four times, until two satisfactory repetitions were performed. A Valsalva was considered satisfactory if it was at least 15 seconds in duration with an average Valsalva pressure of at least 30 mmHg. Valsalva maneuvers were not repeated if the participant exhibited a phase 4 SAP overshoot response greater than 200 mmHg. Valsalva maneuvers were not included in the final analysis if the Valsalva pressure was less than 30 mmHg or if the maneuver was less than 15 seconds long. The average Valsalva pressure in our study was 40 mmHg (standard deviation (SD) 5 mmHg), and the average Valsalva duration was 20 seconds (SD 1 second). Baseline values for SAP and RRI were calculated as the average of 15 seconds prior to the onset of the maneuver. Three indices of cardiac baroreflex function were selected for analysis. The Valsalva ratio was calculated as the maximum RRI in phase 4 divided by the minimum RRI in phase 2b[108]. Cardiac baroreflex sensitivity was calculated during two phases of the Valsalva maneuver: 1) during decreasing SAP in phase 2a (cBRSdec)[61] and 2) during increasing

SAP in phase 4 (cBRSinc)[138]. In both phases, cBRS was calculated as the slope of the regression line between RRI and SAP (Figure 2.8). Sympathetic vascular baroreflex function was also quantified using three indices (Figure 2.8. Pressure recovery time (PRT) was calculated as the time from the minimum SAP in phase 3 to the first SAP beat exceeding baseline SAP during phase 4[191]. Standard alpha baroreflex sensitivity (BRSα) was calculated as the reduction in SAP during phase 3 divided by PRT:

∆SAP BRSα = PRT

where ∆SAP = SAPBaseline−SAP3Min . Alternative alpha baroreﬂex sensitivity (BRSα1) was calculated[159] as:

BRSα1 = SAPBaseline − SAP2aMin + 0.75 ∗ (SAP2bMax − SAP3Min )

37 2.7 Cross-Spectral Analysis and α-Index

Beat-to-beat SAP, RRI, MAP, and FVR ﬁles were extracted from the supine periods of the orthostatic stress test and neck collar protocols. Occasional ectopic beats were identiﬁed and corrected via linear interpolation of adjacent beats. Data were re-expressed at 4Hz via cubic spline interpolation, and transformed into the frequency domain via an autoregressive model (model order 10). The α-Index BRS was calculated[137] as:

R 0.15 !0.5 0.04 (P owerRRI ) Cardiac α-Index = R 0.15 0.04 (P owerSAP )

R 0.15 !0.5 0.04 (P owerFVR) Sympathetic α-Index = R 0.15 0.04 (P owerMAP ) Autoregressive cross-spectral analyses were performed for both the SAP:RRI (Figure 2.9A) and MAP:FVR (Figure 2.10A) time series using a model order of 15. Cardiovagal and sympathetic BRS were calculated as the transfer function gain corresponding with the peak coherence in the LF frequency band (0.04-0.15 Hz, Figures 2.9B and 2.10B)[183]. Transfer function gain values were only used if they corresponded with a coherence greater than 0.5, indicating a signiﬁcant relationship between the two variables.

38 Figure 2.7: Example trace from one participant during the orthostatic stress test. The test was stopped during -40 LBNP due to a reduction in SAP below 80 mmHg and the onset of symptoms of presyncope (indicated by the red asterisk). Cardiovascular measures during the minimum systolic arterial pressure associated with presyncope are shown in blue. Orthostatic tolerance was calculated as the time from the onset of HUT to presyncope. Maximal cardiovascular responses were calculated as the maximum response elicited during the orthostatic phase minus baseline (defined as the last five minutes of the supine period). Forearm vascular resistance (FVR) was calculated as FVR = MAP/BBF V , and was re- expressed as the percent change from the average FVR from baseline. BP = blood pressure, HUT = head-up tilt, LBNP = lower-body negative pressure, RRI = R-R interval, MAP = mean arterial pressure, BBFV = brachial blood flow velocity, FVR = forearm vascular resistance.

39 Figure 2.8: Valsalva maneuvers were performed for 20 seconds with a Valsalva pressure of 40 mmHg. Cardiac baroreflex measures included the Valsalva ratio, calculated as the maximum RRI during phase 4 divided by the minimum RRI during phase 2b. Cardiac baroreflex sensitivity (cBRS) was calculated by correlating the RRI responses to changes in systolic arterial pressure (SAP) during the fall in SAP during phase 2a (light grey boxes, cBRSdec), and during the rise in SAP in phase 4 (dark grey boxes, cBRSinc). Sympathetic baroreflex measures included pressure recovery time (PRT), calculated as the time from the minimum SAP in phase 3 until the first SAP exceeding baseline SAP (black horizontal dashed line). Sympathetic baroreflex sensitivity (BRSα) was calculated as BRSα = ∆SAP3/P RT . Alternative sympathetic baroreflex sensitivity (BRSα1) was calculated as BRSα1 = ∆SAP1 + 0.75 ∗ ∆SAP2. BP = blood pressure, RRI = R-R interval, SAP = systolic arterial pressure, ECG = electrocardiogram, VP = Valsalva pressure.

40 41

Figure 2.9: Example trace of cardiac (SAP:RRI) α-index and cross-spectral baroreﬂex measures in one participant. A) Example tracing of supine SAP and RRI in the time domain. B) Power spectral densities were calculated for both SAP and RRI using an autoregressive model with model order 10. The area under the curve in the low frequency range (0.04-0.15 Hz) was used to calculate the SAP:RRI α-index. Cross-spectral analyses were performed for SAP:RRI, and baroreﬂex sensitivity was determined from the transfer function gain of the relationship taken at the frequency corresponding with peak coherence in the low frequency range (0.04-0.15 Hz). SAP = systolic arterial pressure, RRI = R-R interval, AUC = area under the curve, Coh = coherence. 42

Figure 2.10: Example trace of sympathetic vascular (MAP:FVR) α-index and cross-spectral baroreﬂex measures in one participant. A) Example tracing of supine MAP and FVR in the time domain. B) Power spectral densities were calculated for both MAP and FVR using an autoregressive model with model order 10. The area under the curve in the low frequency range (0.04-0.15 Hz) was used to calculate the MAP:FVR α-index. Cross-spectral analyses were performed for MAP:FVR, and baroreﬂex sensitivity was determined from the transfer function gain of the relationship taken at the frequency corresponding with peak coherence in the low frequency range (0.04-0.15 Hz). MAP = mean arterial pressure, FVR = forearm vascular resistance, AUC = area under the curve, Coh = coherence. 2.8 Sequence Analysis

Sequence analysis was inspired by Blaber et al (1995)[15] and Gulli et al (2005)[68] and again conducted on the supine data recorded prior to the neck collar protocol, as well as the supine, HUT, and LBNP phases of the orthostatic stress test. Following data collection, we performed cardiac baroreflex sequence analyses (Figure 2.11). Beat-to-beat SAP and RRI were scanned to identify baroreflex sequences, defined as sequentially increased/decreased SAP by at least 1 mmHg with concomitant increased/decreased RRI by at least 4 ms in the same direction. To allow for the possibility of variable baroreflex delay between individuals, we determined baroreflex sequences over a range from 0-4 beats of delay[68]. If two or more baroreflex sequences overlapped, we first took the longest sequence (Figure 2.11A); second, if sequences with equal length overlapped, the sequence that started on the earliest SAP beat was chosen (Figure 2.11B); lastly, the sequence with the highest R2 was chosen if neither the longest nor “first to begin” criteria applied (Figure 2.11C). Sequences with an R2 < 0.85 were not considered valid baroreflex sequences. We also determined vascular resistance baroreflex sequence analyses. All FVR values were re-expressed as the percent change from the average FVR from the last five minutes of supine. The MAP:FVR sequence analysis was performed in a similar manner to the SAP:RRI analysis, except that baroreflex sequences were defined as sequential increases in MAP by at least 1mmHg accompanied by sequential decreases in FVR (and vice versa). As baroreflex-mediated vascular responses are slower in onset[80, 76], the beat lags considered were 4 to 16. Lastly, a minimum change in FVR for inclusion was 1% of the baseline supine FVR. Both cardiac and sympathetic baroreflex sensitivity were calculated from the mean slope of all the sequences (Figure 2.12). In this thesis, references to sympathetic baroreflex function (ie. sympathetic baroreflex sensitivity) will refer to the sensitivity of the sympathetically-mediated vascular responses to baroreceptor stimulation.

43 Figure 2.11: Examples of baroreflex sequence overlap. A) For overlapping baroreflex sequences of different lengths, the longest sequence was chosen (A1)[15]. B) For overlapping baroreflex sequences of the same length, the sequence starting on the earliest SAP beat was chosen (B3)[15]. C) For overlapping baroreflex sequences of the same length that started on the same beat, the sequence with the highest R2 was chosen (C2)[68]

44 Figure 2.12: Cardiac and sympathetic baroreflex sequences from one individual during supine and 70◦ HUT. Average baroreflex sensitivity (black) was calculated as the average slope from all available baroreflex sequences (grey). Cardiac baroreflex sensitivity decreases when moving from supine rest (A) to HUT (B). Conversely, sympathetic baroreflex sensitivity increases in magnitude when moving from supine (C) to HUT (D). RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance, MAP = mean arterial pressure, HUT = head-up tilt.

45 Chapter 3

The Agreement and Repeatability of Common Non-Invasive Baroreﬂex Assessments

3.1 Abstract

Arterial baroreflex responses serve to maintain arterial blood pressure through the regulation of efferent sympathetic and parasympathetic outflow to the heart and vasculature. Abnormal baroreflex function is implicated in a number of cardiovascular diseases, including orthostatic hypotension, sleep apnoea, and hypertension. Accordingly, a wide variety of non-invasive baroreflex testing methods have been developed for both research and routine clinical use. The large number of available non-invasive baroreflex assessments has created the need for evaluation of the agreement between these tests, to facilitate translation of clinical and research observations between assessments using different approaches. Prior comparisons of baroreflex tests have been hampered by the low numbers of different tests compared, reliance on correlation coefficients as indices of agreement, and lack of assessment of repeatability (test-retest reliability). In this study, we report agreement, repeatability, and conversion metrics for baroreflex indices derived from seven non-invasive baroreflex measurement techniques: the Valsalva maneuver; carotid sinus massage, variable pressure neck collar technique; cross-spectral analysis; α-index; spontaneous sequence analysis; and maximum responses to orthostatic stress. Data were collected from 63 participants (29 males) with a mean age of 39 years (range 19-84 years). Indices of cardiac baroreflex function had higher agreement and repeatability than sympathetic baroreflex indices. Dynamic baroreflex tests had higher agreement and repeatability than static tests. Due to the wide prediction intervals, conversion between baroreflex measures in an individual should be avoided; however, conversion of baroreflex measures for a sample provides reasonable prediction intervals.

46 Agreement and repeatability of common non-invasive baroreﬂex tests are variable and careful consideration should be taken during the planning and interpretation of studies involving baroreﬂex measurement.

3.2 Introduction

The arterial baroreflex is critical for the maintenance of blood pressure, especially during upright posture (orthostasis) where lower body venous pooling reduces the effective circulating blood volume and blood pressure tends to decrease[21]. Abnormal baroreflex function is widely implicated in numerous cardiovascular diseases including, but not restricted to, orthostatic hypotension, vasovagal syncope, POTS, sleep apnoea, hypertension, stroke, diabetes mellitus, and myocardial infarction[114, 84, 197, 39, 189, 165, 99, 180, 98]. Given the prominence of baroreflex responses in physiological and pathophysiological cardiovascular control, accurate, repeatable and translatable assessment of baroreflex responses is paramount. The Oxford Method[166] is often cited as the “gold standard" for baroreflex assessment, and non-invasive baroreflex assessment techniques have been compared against the Oxford Method to assess their validity[140, 195, 194, 118, 125, 20]. In this technique, infusions of vasodilator (nitric oxide donors) and vasoconstrictor (alpha-agonists) agents are given to induce changes in blood pressure while subsequent baroreflex-mediated changes in heart rate are evaluated[44]. However, the Oxford Method is an invasive protocol, which may alter baseline autonomic function[170]. It is also time-consuming, and is often inappropriate for research or routine clinical assessments. Accordingly, a wide variety of non-invasive methods[54, 179] have been developed for both research and routine clinical use. However, these methods have been shown previously to have low agreement with the Oxford Method, and may reflect different aspects of baroreflex function[111]. Another disadvantage of the Oxford Method is that it is not possible to evaluate sympathetic vascular baroreflex function in the presence of systemic vasodilator and vasoconstrictor drugs. This is important because the sympathetic vascular baroreflex responses are thought to be more important for the maintenance of blood pressure during orthostasis[30] and in the setting of hypertension[189], stroke[180], myocardial infarction[99, 98] and sleep apnoea[39]. Given the large number of non-invasive baroreflex assessments available, one area of research need is to evaluate how well these different measures agree with one another – to facilitate translation of clinical and research observations between assessments using different approaches. Some attempts to quantify the agreement/equivalency between non- invasive baroreflex assessments have previously been undertaken[61, 31, 111, 105, 192]. However, prior comparisons of baroreflex tests have been hampered by the low numbers of different tests compared, reliance on correlation coefficients as indices of agreement, and lack of assessment of repeatability (test-retest reliability). This is important because poor

47 repeatability of a method will affect its agreement with other methods, and is an important consideration when performing agreement studies[19]. In this study, we report agreement and repeatability of baroreflex indices derived from seven non-invasive baroreflex measurement techniques: the Valsalva maneuver; carotid sinus massage, variable pressure neck collar technique; cross-spectral analysis; α-index; spontaneous sequence analysis; and maximum responses to orthostatic stress (tilt test combined with lower-body negative pressure and continued until presyncope). From each method we report indices of both cardiovagal and sympathetic vascular baroreflex function, as well as conversion factors to enable comparison of responses between measures.

3.3 Methods

We examined the baroreﬂex responses to four static, and three dynamic baroreﬂex tests. The protocols and methods for these tests are in Section 2, page 25. The valsalva maneuver and orthostatic stress test took place on the same day, while CSM and neck collar protocols took place on separate days. Diagrams describing each protocol are in Figure 3.1. Results are reported as mean ± standard error unless otherwise stated.

3.3.1 Calculation of Outcome Measures

3.3.1.1 Agreement

For the calculation of baroreflex function for the agreement analysis, dynamic BRS was taken from the 20-minute supine period preceding HUT for comparisons with the Valsalva maneuver, CSM, and the maximum response to tilt. Comparison between dynamic baroreflex measures and the neck collar protocol used dynamic BRS from 10 minutes of supine rest preceding the neck collar protocol, and were thus taken from the same day of testing. Neck collar BRS for agreement analysis was calculated by fitting a sigmoid curve to all available responses. Responses to CSM occurring at the level of the ultrasound-guided carotid bifurcation were used for all analyses, and responses were averaged between repetitions and side (left and right) for the agreement analysis.

3.3.1.2 Repeatability

Dynamic BRS repeatability was assessed by calculating BRS for the first and second half of the 20-minute supine period preceding tilt. Analysis of the repeatability of BRS derived from the neck collars was calculated from sigmoid curves fitted to the first and then second repetitions of each pulse pressure. The repeatability of indices from utlrasound-guided CSM and Valsalva maneuvers were calculated using the first and second repetitions.

48 3.3.2 Calculation of Agreement

Agreement between baroreflex measures was first assessed using linear correlation. Propor- tional bias was then examined by correlating the mean differences between two variables with the average of the two variables, using a Bland-Altman plot[19]. If p < 0.05 for the regression, we concluded there was evidence for proportional bias. Regardless of the presence of proportional bias, we examined the presence of heteroscedasticity (measurement error proportional to the mean) by correlating the absolute value of the residuals from the Bland-Altman regression against the average of the two variables. Heteroscedasticity was concluded if p < 0.05. For comparisons with heteroscedasticity but no proportional bias, data were log-transformed before calculating limits of agreement. The limits of agreement were then back-transformed to reflect the absolute data[19]. When the mean difference or limits of agreement for two measurements varied with the mean of the two measurements, each was expressed as a function of the average between two measurements (A) (Figure 3.2).

3.3.3 Repeatability

There is currently no gold standard for reporting of repeatability. Consequently, we have selected several measures from the literature that comprehensively describe repeatability. The Intraclass Correlation Coeﬃcient (ICC) is the proportion of variance in a set of scores that is attributable to the “true score variance,” and is calculated using metrics from a repeated-measures ANOVA:

MS − MS ICC = B E MSB + (k − 1) × MSE

where MSB = between-subject mean square, MSE = mean squared error, and k = number of repetitions performed in each participant[163]. The Standard Error of Measurement (SEM) is calculated as:

√ SEM = SDBS × 1 − ICC

where SDBS = the between-subject standard deviation. The SEM is deﬁned as the standard deviation of errors of measurement that is associated with the test scores[196]. The Repeatability Coeﬃcient (RC) is calculated as:

√ RC = 1.96 × 2 × SDWS

where SDWS = within-subject standard deviation. Similar to the 95% limits of agreement, two readings by the same method will be within -RC:RC. This can be compared to the 95% LOA to estimate how much of the lack of agreement may be due to low repeatability[19]. The Minimum Diﬀerence to Detect (MDD) is calculated as:

49 √ MDD = 1.96 × 2 × SEM

It is the minimum diﬀerence between two measurements in one participant for the diﬀerence to be considered real[196].

3.4 Results

3.4.1 Participants

Data were obtained from 63 participants (29 males) with a mean age of 39 years (range 19-84 years). Participants had a mean height of 172 (SD 9cm) and weight of 71 (SD 3kg). The average resting supine SAP was 125 (SD 13mmHg) and DAP was 68 (SD 8mmHg), with a heart rate of 63 (SD 7mmHg).

3.4.2 Responses to Baroreﬂex Tests

Mean, maximum, and minimum responses to supine baroreﬂex tests are listed in Table 3.1. Responses to baroreﬂex tests performed during HUT are listed in Table 3.2. Where common units allow for direct comparisons, cardiac and sympathetic BRS were larger with dynamic measures compared to static measures.

3.4.3 Agreement

3.4.3.1 Cardiac Baroreﬂex Measures

Agreement between cardiac baroreflex measures during supine and head-up tilt is described in Table 3.3 and Table 3.4, respectively. Agreement was highest between dynamic cardiac baroreflex measures. Regarding static baroreflex measures, cardiac baroreflex indices from the Valsalva maneuver (Valsalva ratio, cBRS increase (cBRSinc), cBRS decrease (cBRSdec)) had good agreement with each other, as well as good agreement with static baroreflex measures. The magnitude of the heart rate response to orthostatic stress had good agreement with all three supine dynamic cardiac baroreflex measures and cBRSinc; however, agreement was reduced with the HUT dynamic cardiac baroreflex measures. Cardiac baroreflex sensitivity using the neck collar technique had good agreement with all three supine dynamic indices, and with cBRSinc. However, the correlation between neck collar cBRS and HUT dynamic cardiac baroreflex sensitivity was not significant. The cardiac baroreflex measure derived from CSM had poor agreement with all other cardiac baroreflex measures. The agreement between dynamic cardiac baroreflex measures is shown in Figure 3.3A-C.

50 Baroreﬂex Arm Test Mean SD Min Max n Cardiac Seq (ms.mmHg-1) 15.5abc 10.2 3.8 49.7 47 XPSD (ms.mmHg-1) 15.4abc 12.6 2.4 60.1 42 α-index (ms.mmHg-1) 13.2abc 8.8 3 38.6 48 NC (ms.mmHg-1) 6.5 4.4 0.6 19 45 CSM (ms) 87 82 -11 509 60 Valsalva Ratio 1.6 0.4 1.1 2.5 45 -1 cBRSinc (ms.mmHg ) 5 3.3 0.1 12.6 43 -1 cBRSdec (ms.mmHg ) 3.8 2.5 -0.1 9 40 Sympathetic Seq (%.mmHg-1) -4.3ad 1.5 -7.3 -1.5 48 XPSD (%.mmHg-1) -3.3a 1.2 -6.4 -1.5 25 α-index (%.mmHg-1) -4ad 1.2 -7.8 -1.6 48 NC (%.mmHg-1) -1.9 1 -5.4 -0.7 45 CSM (%) -14 9 -29 7.3 58 PRT (s) 2 1.4 0.6 6.2 40 BRSα (mmHg.s-1) 11.6 7 0.9 27.3 38 BRSα1 (mmHg) 30.8 16.8 8.4 85.3 43

Table 3.1: Responses to supine cardiac and sympathetic baroreflex tests. Note the low n with sympathetic XPSD is due to low coherence. a = significantly different from neck b c collar BRS within condition, = significantly different from cBRSinc, = significantly d different from cBRSdec, = significantly different from XPSD BRS, all p < 0.02. SD = standard deviation, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, cBRSinc = cardiac baroreflex sensitivity (increase), cBRSdec = cardiac baroreflex sensitivity (decrease), PRT = pressure recovery time, BRSα = sympathetic baroreflex sensitivity, BRSα1 = alternative sympathetic baroreflex sensitivity

Baroreﬂex Arm Test Mean SD Min Max n Cardiac Seq (ms.mmHg-1) 6.7* 3.2 2.5 16.4 48 XPSD (ms.mmHg-1) 6.3* 3.3 1.1 20.2 45 α-index (ms.mmHg-1) 6.7* 2.9 1.5 16.5 47 CSM (ms) 87 61 -9 260 59 Tilt Max ΔRRI (ms) -389 117 -674 -116 48 Sympathetic Seq (%.mmHg-1) -6.3* 1.9 -11.4 -3 46 XPSD (%.mmHg-1) -4.0* 1.7 -7.4 -1.6 26 α-index (%.mmHg-1) -4.7* 1.9 -8.5 -0.2 45 CSM (%) -16.5 13 -51.8 12.5 57 Tilt Max ΔFVR (%) 179 111 22 430 48

Table 3.2: Responses to cardiac and sympathetic baroreflex tests performed during head-up tilt. * = significantly different from supine test, all p < 0.03. SD = standard deviation, Seq = sequence analysis, XPSD = cross-spectral analysis, RRI = R-R interval, FVR = forearm vascular resistance.

51 3.4.3.2 Sympathetic Baroreﬂex Measures

Agreement between sympathetic baroreflex measures during supine and head-up tilt is described in Table 3.5 and 3.6, respectively. Agreement was highest between dynamic sympathetic baroreflex measures. Regarding static baroreflex measures, two sympathetic barore-

flex indices from the Valsalva maneuver (BRSα, BRSα1) had good agreement with each other. Sympathetic spontaneous sequence analysis had good agreement with PRT. None of the supine sympathetic baroreflex measures had good agreement with the vascular resistance response to orthostatic stress; however, sympathetic baroreflex sensitivity (sBRS) as measured by sequence analysis and α-index during HUT were correlated with the maximal FVR response during the orthostatic stress test. The agreement of dynamic sympathetic baroreflex measures is shown in Figure 3.3D-F.

3.4.3.3 Conversion of Baroreﬂex Measures

Metrics necessary for the numerical conversion of supine and HUT cardiac baroreflex measures are listed in Tables 3.7 and 3.8, respectively. Conversion metrics for supine and HUT sympathetic baroreflex measures are listed in Tables 3.9 and 3.10, respectively. As a working example, to convert a supine cBRS of 20 ms.mmHg-1 derived from sequence analysis to the α-index, first solve for the equation of a line:

y = Slope ∗ x + Int

y = 0.71 ∗ 20 + 1.6

y = 15.9 ms.mmHg−1

To calculate the 95% prediction interval (PI95%) for this estimate:

v u 2 ! u 1 1 (xh − x¯) PI = y ± t ∗ tMSE ∗ + + 95% (α/2,n−2) m n SSD

-1 Where xh is the original cBRS of 20 ms.mmHg , t(α/2,n−2) is the t-statistic for n-2 and alpha = 0.05, n is the sample size from our study, m is the size of the sample being converted, SSD is the sum of the squared diﬀerences, and x¯ is the mean response from our study. For calculation of a 95% prediction interval (PI) for one individual, m = 1. If converting cBRS for a sample of 30 participants, m = 30:

v u 2 ! u 1 1 (20 − 15) PI = 15.9 ± 2.01 ∗ t12 ∗ + + 95% 30 47 4769

−1 PI95% = 14.2 − 17.6ms.mmHg

52 Therefore, cBRS via the α-index is 15.9 ms.mmHg-1 (95% PI 14.2-17.6).

53 Figure 3.1: Schematic diagrams of the experimental protocols - each protocol was completed on a different day, with the order randomized. A) Experimental protocol for the orthostatic stress test. Following 20 minutes supine rest, participants were head-up tilted to 70 degrees for a further 20 minutes. Incremental lower-body negative pressure was applied for ten minute intervals at -20, -40, and -60mmHg. The test was terminated if any of the following criteria were met: systolic arterial dropped below 80 mmHg; heart rate decreased to below 50 bpm or increased to higher than 170 bpm; the participant experienced symptoms of presyncope such as dizziness, warmth, and requested to stop; or the protocol was completed. Each test was therefore stopped at a different time point. B) Neck collar experimental protocol. Following 10 minutes supine rest, graded positive and negative pressures (-60, -40, -20, 0, 20, 40, and 60 mmHg) were generated over the carotid sinus. Vagally-mediated cardiac responses were assessed during 5 seconds of neck pressure/suction applied for five seconds during breath hold. Sympathetically-mediated vascular responses were assessed during 20 seconds of neck suction/pressure during normal breathing. Each pressure/timing protocol was repeated twice, and were separated by at least 1 minute to allow return of cardiovascular homeostasis. C) Following a 10-minutes supine rest, participants underwent supine and tilted carotid sinus massage in duplicate, on both sides of the neck at the ultrasound-guided level of the carotid bifurcation and at the level of the cricoid cartilage. Two additional sham procedures were conducted in each position in which the fingers were placed overlying the carotid sinus on each side of the neck, but no pressure was applied. LBNP = lower-body negative pressure, US = ultrasound-guided location of the carotid bifurcation, Cr = cricoid cartilage, R = right, L = left.

54 Figure 3.2: Example Bland-Altman plot demonstrating the agreement between cardiac α- index BRS and cBRSinc. In this example, the relationship displays proportional bias, as the regression between the difference and the average is significant (p < 0.0001). The mean difference (μdiff) is therefore expressed as a function of the average (A): µdiff = −1.28 + 1.04 × A (solid line). Similarly, as the regression between the residuals of this regression correlated with the average (not shown, p = 0.007), heteroscedasticity was also present, and the 95% limits of agreement (95%LOA) were expressed as a function of the average (A) and the μdiff: 95%LOA = µdiff ± −3.14 + 0.43 × A (dashed lines). BRS = baroreflex sensitivity, cBRSinc = cardiac baroreflex sensitivity (increasing).

55 Para- Valsalva

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec

XPSD r[p] 0.88[<0.01]

μdiﬀ 2.04

LOA95% 9.87 n 41

α-index r[p] 0.91[<0.01] 0.96[<0.01]

μdiﬀ -0.72+0.25*A† -3.66+0.35*A†

LOA95% -1.76+0.47*A‡ -9+0.24*A‡ n 47 42

NC r[p] 0.6[<0.01] 0.4[0.02] 0.56[<0.01]

μdiﬀ -0.44+0.96*A† 1.21+0.63*A† 0.95+0.63*A†

LOA95% 10.88 2.98+0.85*A‡ 10.22 n 39 33 39

56 CSM r[p] -0.08[0.6] -0.14[0.41] -0.08[0.6] 0.11[0.52]

μdiﬀ 32.05-1.98*A† 33.41-2*A† 27.52-1.99*A† 11.32-1.97*A†

LOA95% 41.03 51.31 67.69+0.22*A‡ 17.13 n 41 36 42 39

Continued on next page Para- Valsalva

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec

Valsalva r[p] 0.2[0.19] 0.26[0.12] 0.24[0.12] 0.16[0.38] -0.23[0.16]

Ratio μdiﬀ -2.86+1.96*A† -2.93+1.97*A† -2.85+1.95*A† -2.85+1.91*A† -3.39+2*A†

LOA95% 1.37 1.37 1.35 1.4 1.45 n 43 39 44 34 38

cBRSinc r[p] 0.63[<0.01] 0.52[<0.01] 0.65[<0.01] 0.37[0.04] 0.01[0.94] 0.3[0.05]

μdiﬀ -0.94+1.1*A† -3.97+1.4*A† -1.28+1.04*A† 0.83 -9.52+1.98*A† 2.62-1.83*A†

LOA95% -2.32+0.45*A‡ -9.77+0.43*A‡ -3.14+0.43*A‡ 8.32 13.51 1.31 n 41 37 42 32 36 43

cBRSdec r[p] 0.39[0.01] 0.47[0.01] 0.43[0.01] 0.33[0.08] -0.02[0.93] 0.24[0.14] 0.36[0.02]

μdiﬀ -2.37+1.46*A† -3.48+1.55*A† -2.37+1.38*A† -1.44+0.75*A† -7.84+1.99*A† 2.63-1.8*A† 1.12

LOA95% 8.01 8.08 7.74 6.9 10.07 1.31 6.6 n 38 34 39 29 33 40 40

57 Tilt Max r[p] -0.37[0.01] -0.52[<0.01] -0.51[<0.01] 0.11[0.5] 0.2[0.21] -0.22[0.15] -0.54[<0.01] -0.24[0.14]

ΔRRI μdiﬀ 11.04-2.11*A† -4.85-2.19*A† 0.35-2.14*A† 17.41-1.98*A† 397.44-0.55*A† 2.69-2*A† -0.96-2.06*A† 4.14-2.02*A†

LOA95% 38.26 -11.93-0.2*A‡ 30.77 17.27 977.71+1.38*A‡ 1.38 11.05 9.64 n 47 42 48 38 42 44 42 39

Table 3.3: Agreement between supine cardiac baroreflex tests. μdiff = mean difference, LOA95% = 95% limits of agreement, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, cBRSinc = cardiac baroreflex sensitivity (increase), cardiac baroreflex sensitivity (decrease), RRI = R-R interval, A = average of two measurements in one individual. † = proportional bias, ‡ = heteroscedasticity. Bold text denotes significant correlations between indicated baroreflex measures. Para- Valsalva

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec

XPSD r[p] 0.7[<0.01]

μdiﬀ 0.33

LOA95% 3.52 n 40

α-index r[p] 0.78[<0.01] 0.96[<0.01]

μdiﬀ -0.26 -0.53

LOA95% 3.08 1.39 n 42 43

NC r[p] 0.29[0.1] 0.24[0.16] 0.27[0.12]

μdiﬀ 5.71-0.9*A† 4.61-0.74*A† 5.41-0.73*A†

LOA95% 14.04+0.65*A‡ 7.85 7.7 n 33 35 36

58 CSM r[p] -0.1[0.56] -0.24[0.15] -0.05[0.74] -0.06[0.71]

μdiﬀ 12.72-2.01*A† 14.4-2.03*A† 13.84-2*A† 12.64-2*A†

LOA95% 8.26 9.84 10.14 17.33 n 36 38 39 39

Continued on next page Para- Valsalva

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec

Valsalva r[p] -0.23[0.15] 0.05[0.76] -0.01[0.94] -0.15[0.37]

Ratio μdiﬀ -3.36+2.04*A† -2.85+1.9*A† -2.92+1.93*A† -3.4+2*A†

LOA95% 1.41 1.37 1.4 1.45 n 40 41 42 37

cBRSinc r[p] 0.29[0.07] 0.45[<0.01] 0.46[<0.01] -0.02[0.9]

μdiﬀ 4.85-0.75*A† 0.83 1.38 -10.26+1.99*A†

LOA95% 5.85 6.22 6.15 13.3 n 38 39 41 35

cBRSdec r[p] 0.47[<0.01] 0.45[0.01] 0.45[<0.01] -0.16[0.4]

μdiﬀ 1.85 2.04 2.52 -9.08+2.02*A†

LOA95% 4.66 5.4 5.32 10.05 n 35 36 38 32

59 Tilt Max r[p] -0.13[0.26] -0.36[0.02] -0.29[0.05] 0[0.98]

ΔRRI μdiﬀ 10.35-2.01*A† 5.95-2.03*A† 8.23-2.02*A† 325.77-1.05*A†

LOA95% 8.11 9.65 9.71 801.4+1.06*A‡ n 43 44 46 41

Table 3.4: Agreement between head-up tilt (HUT) cardiac baroreflex tests. Agreement between tests not involving a HUT component have been omitted from this table. μdiff = mean difference, LOA95% = 95% limits of agreement, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, cBRSinc = cardiac baroreflex sensitivity (increase), cardiac baroreflex sensitivity (decrease), RRI = R-R interval, A = average of two measurements in one individual. † = proportional bias, ‡ = heteroscedasticity. Bold text denotes significant correlations between indicated baroreflex measures. Para-

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1

XPSD r[p] 0.81[<0.01]

μdiﬀ -0.79

LOA95% 1.57 n 25

α-index r[p] 0.65[<0.01] 0.96[<0.01]

μdiﬀ -0.23 0.14-0.14*A†

LOA95% 2.24 0.73 n 48 25

NC r[p] 0.22[0.2] 0.-0.57[0.01] 0.44[0.01]

μdiﬀ -2.19 5.69 -2.53

LOA95% 2.37 4.31 2.67 n 34 22 34

60 CSM r[p] 0.14[0.38] 0.27[0.28] 0.25[0.11] 0.53[<0.01]

μdiﬀ -7.04-1.81*A† -5-1.78*A† -6.47-1.81*A† -1.67-1.74*A†

LOA95% 5.69 4.47 4.42 -4.1-0.3*A‡ n 40 18 40 38

Continued on next page Para-

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1

PRT r[p] 0.48[<0.01] 0.39[0.08] 0.29[0.07] 0.13[0.52] 0.32[0.08]

μdiﬀ -6.24 -5.18 -5.93 -3.79 -5.18+1.78*A†

LOA95% 2.77 2.62 2.93 2.98 -12.8+0.37*A‡ n 39 21 39 29 32

BRSα r[p] 0.16[0.33] 0.15[0.53] 0.13[0.46] 0.06[0.78] -0.14[0.46] -0.21[0.21]

μdiﬀ -9.67-1.72*A† -6.98-1.83*A† -8.56-1.83*A† -3.96-1.9*A† -26.13 4.07-2*A†

LOA95% 5.27 4.11 4.19 3.77 25.45 5.64 n 37 20 37 27 30 38

BRSα1 r[p] 0.34[0.03] 0.32[0.16] 0.24[0.13] 0.22[0.22] -0.04[0.83] 0.2[0.21] 0.7[<0.01]

μdiﬀ -10.56-1.86*A† -8.16-1.89*A† -9.24-1.91*A† -4.91-1.93*A† -36.5-1.17*A† 2.5-1.91*A† 0.73-0.97*A†

LOA95% 5.32 4.65 4.63 4.25 -89.9+0.9*A‡ 5.36 14.8 n 42 21 42 32 35 40 38

61 Tilt Max r[p] 0.2[0.07] -0.11[0.6] -0.01[0.93] 0.18[0.29] 0.1[0.55] -0.06[0.69] 0.2[0.19] 0.25[0.1]

ΔFVR μdiﬀ -9.5-1.99*A† -6.27-2*A† -8.04-2*A† -4.49-1.99*A† -35.21-1.93*A† 3.96-2*A† 16.93-1.93*A† 36.27-1.74*A†

LOA95% 5.63 4.66 4.8 4.13 35.73 4.93 26.06 60.71 n 48 25 48 38 40 39 37 42

Table 3.5: Agreement between supine sympathetic baroreflex tests. μdiff = mean difference, LOA95% = 95% limits of agreement, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSα= alpha baroreflex sensitivity, BRSα1 = alternative alpha adrenergic baroreflex sensitivity, FVR = forearm vascular resistance, A = average of two measurements in one individual. † = proportional bias, ‡ = heteroscedasticity. Bold text denotes significant correlations between indicated baroreflex measures. Para-

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1

XPSD r[p] 0.42[0.04]

μdiﬀ -2.6

LOA95% 3.99 n 25

α-index r[p] 0.5[<0.01] 0.81[<0.01]

μdiﬀ -1.67 0.82

LOA95% 3.67 2.31 n 44 26

NC r[p] -0.01[0.94] 0.06[0.79] 0.09[0.61]

μdiﬀ 0.01+1.06*A† 0.03+0.81*A† -0.01+0.88*A†

LOA95% 0.04-1.51*A‡ 0.07-1.6*A‡ -0.03-1.27*A‡ n 36 22 35

62 CSM r[p] -0.16[0.33] -0.26[0.24] -0.1[0.56] 0.03[0.87]

μdiﬀ -12.94-2.01*A† -9.33-2.07*A† -9.15-1.96*A† -3.38-1.96*A†

LOA95% 7.35 6.47 7.63 4.12 n 38 23 37 38

Continued on next page Para-

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1

PRT r[p] 0.17[0.3] 0.04[0.87] -0.13[0.43] -0.11[0.57]

μdiﬀ -6.56+0.7*A† -4.58+1.14*A† -5.12+1*A† -3.6+2*A†

LOA95% 3.74 3.25 -12.59+1.52*A‡ 4.64 n 38 21 37 31

BRSα r[p] -0.09[0.62] 0.41[0.06] 0.09[0.59] -0.21[0.28]

μdiﬀ -12.96-1.82*A† -10.45-1.49*A† -10.69-1.6*A† -27.02+1.07*A†

LOA95% 7.08 5.9 7.63 27.01 n 36 21 35 30

BRSα1 r[p] -0.07[0.65] 0.44[0.04] -0.02[0.92] -0.4[0.02]

μdiﬀ -12.8-1.98*A† -10.92-1.81*A† -9.99-1.96*A† -40.26-1.22*A†

LOA95% 7.36 6.11 7.77 44.33 n 41 22 40 34

63 Tilt Max r[p] -0.4[<0.01] -0.34[0.09] -0.49[<0.01] 0.03[0.84]

ΔFVR μdiﬀ -10.07-2.03*A† -6.23-2.02*A† -6.31-2.04*A† -42.08-1.94*A†

LOA95% 6.83 6.28 6.52 47.63 n 46 26 45 40

Table 3.6: Agreement between head-up tilt (HUT) sympathetic baroreflex tests. Agreement between tests not involving a HUT component have been omitted from this table. μdiff = mean difference, LOA95% = 95% limits of agreement, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSα= alpha baroreflex sensitivity, BRSα1 = alternative alpha adrenergic baroreflex sensitivity, FVR = forearm vascular resistance, A = average of two measurements in one individual. † = proportional bias, ‡ = heteroscedasticity. Bold text denotes significant correlations between indicated baroreflex measures. Para- Valsalva Tilt Max

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec ΔRRI

Seq Slope; Int 0.87; 3.9 1.2; 0.88 1.4; 9.1 -0.01; 17 5.2; 6.6 1.8; 6.4 1.5; 9.4 -0.034; 2.4 MSE 24 19 73 112 82 53 78 92 n; x¯ 41; 14 47; 13 39; 6.7 41; 89 43; 1.6 41; 5 38; 3.7 47; -382 SSD 4459 2957 770 294306 5 442 233 557576

XPSD Slope; Int 0.9; -0.37 1.4; -3.5 0.63; 10 -0.02; 19 8.5; 1.5 2; 5.2 2.3; 6.5 -0.055; -5.5 MSE 25 13 45 173 142 113 129 118 n; x¯ 41; 16 42; 14 33; 6.6 36; 94 39; 1.6 37; 5.1 34; 3.9 42; -382 SSD 4329 3232 665 286036 5 358 221 596067

α-index Slope; Int 0.71; 1.6 0.68; 3.5 0.94; 8 -0.0088; 15 5.7; 4 1.7; 4.7 1.5; 7.4 -0.038; -1.5 MSE 12 6.3 39 84 70 44 64 59 n; x¯ 47; 15 42; 15 39; 6.7 42; 87 44; 1.6 42; 5 39; 3.8 48; -389 SSD 4769 6508 770 299555 6 444 242 640536

64 NC Slope; Int 0.26; 2 0.25; 3 0.34; 1.9 0.0052; 5.9 1.7; 3.6 0.43; 3.9 0.57; 4.1 0.0042; 8.2 MSE 13 18 14 20 17 15 16 20 n; x¯ 39; 18 33; 14 39; 14 39; 88 34; 1.6 32; 5.3 29; 3.9 38; -392 SSD 4192 1652 2139 298040 4 364 170 547686

CSM Slope; Int -0.69; 100 -0.98; 110 -0.78; 98 2.1; 75 -55; 177 0.23; 77 -0.32; 76 0.14; 143 MSE 7493 8244 7438 7966 7854 3374 2574 7199 n; x¯ 41; 17 36; 17 42; 14 39; 6.4 38; 1.6 36; 5.1 33; 4.1 42; -394 SSD 4390 5997 3379 743 5 418 212 587666

Valsalva Slope; Int 0.008; 1.5 0.0077; 1.5 0.01; 1.5 0.014; 1.5 -0.001; 1.7 0.034; 1.4 0.035; 1.5 -0.00066; 1.3 Ratio MSE 0.13 0.13 0.12 0.14 0.14 0.12 0.13 0.13 n; x¯ 43; 15 39; 15 44; 13 34; 6.3 38; 88 43; 5 40; 3.8 44; -387 SSD 3513 5619 3141 547 298823 447 243 618645

Continued on next page Para- Valsalva Tilt Max

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec ΔRRI

cBRSinc Slope; Int 0.23; 1.6 0.13; 3 0.25; 1.7 0.31; 3.4 0.00083; 5.1 2.7; 0.66 0.48; 3.1 -0.015; -0.7 MSE 6.8 7.4 6.4 11 12 9.9 9.9 7.9 n; x¯ 41; 15 37; 16 42; 13 32; 6.2 36; 79 43; 1.6 40; 3.8 42; -391 SSD 3426 5454 3047 509 114750 6 243 598887

cBRSdec Slope; Int 0.1; 2.2 0.096; 2.4 0.12; 2.2 0.2; 2.7 -0.0008; 4.1 1.7; 1.2 0.27; 2.5 -0.0049; 1.9 MSE 5.5 5.4 5.3 5.6 6.8 6 5.6 6.1 n; x¯ 38; 15 34; 15 39; 13 29; 6.3 33; 74 40; 1.6 40; 4.9 39; -385 SSD 3339 5270 2888 488 79807 5 432 576315

Tilt Max Slope; Int -4; -321 -5; -305 -6.7; -300 3.1; -413 0.28; -418 -73; -270 -20; -292 -12; -340 ΔRRI MSE 10708 10812 10369 15017 14124 14018 10680 14674 n; x¯ 47; 15 42; 15 48; 13 38; 6.6 42; 87 44; 1.6 42; 5 39; 3.8 SSD 4769 6508 3617 735 299555 6 444 242 65 Table 3.7: Metrics for conversion of supine cardiac baroreflex tests. Int = intercept, MSE = mean squared error, n = sample size, x¯ = mean of the x variable, SSD = sum of the squared differences, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, cBRSinc = cardiac baroreflex sensitivity (increase), cardiac baroreflex sensitivity (decrease), RRI = R-R interval. Bold text denotes significant correlations between indicated baroreflex measures. Para- Valsalva Tilt Max

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec ΔRRI

Seq Slope; Int 0.61; 2.5 0.66; 1.8 0.16; 5.3 -0.0032; 6.4 -1.3; 8 0.18; 4.9 0.38; 4.3 -0.0024; 5 MSE 2.4 1.8 3.9 4.6 4.2 3.9 3.3 4.4 n; x¯ 40; 5.6 42; 6.2 33; 5.9 36; 91 40; 1.6 38; 4.9 35; 3.9 43; -375 SSD 231 251 472 158906 5 433 213 500052

XPSD Slope; Int 0.8; 0.88 0.97; -0.31 0.15; 5.4 -0.0093; 7.3 0.37; 5.4 0.36; 4.1 0.48; 4.1 -0.0084; 2.8 MSE 3.1 0.51 6.9 6.4 7.1 5.7 5.9 6.1 n; x¯ 40; 6 43; 6.6 35; 6.7 38; 92 41; 1.6 39; 5.1 36; 3.9 44; -387 SSD 176 287 644 156708 5 418 230 542734

α-index Slope; Int 0.91; 0.78 0.96; 0.77 0.16; 5.8 -0.0023; 7.1 -0.089; 6.6 0.35; 4.6 0.46; 4.6 -0.0067; 3.9 MSE 2.5 0.5 6.8 6.9 7 5.4 5.6 6.2 n; x¯ 42; 5.9 43; 6.1 36; 6.3 39; 86 42; 1.6 41; 5 38; 3.8 46; -387 SSD 181 288 658 146683 5 443 240 585892

66 NC Slope; Int 0.55; 2.5 0.4; 4.2 0.43; 3.3 -0.0042; 6.7 MSE 14 18 18 20 n; x¯ 33; 6.2 35; 6.4 36; 6.9 39; 85 SSD 133 243 248 150242

CSM Slope; Int -3.2; 110 -6; 131 -1.3; 95 -0.86; 90 -27; 135 -0.42; 89 -3.8; 98 -0.0022; 88 MSE 4626 4111 3953 4046 4569 4406 4076 4245 n; x¯ 36; 6.1 38; 6.4 39; 6.9 39; 6.4 37; 1.6 35; 5.3 32; 4.1 41; -400 SSD 156 242 254 743 5 392 207 530225

Valsalva Slope; Int -0.04; 1.8 0.0067; 1.6 -0.0017; 1.6 -0.00084; 1.7 Ratio MSE 0.13 0.13 0.13 0.14 n; x¯ 40; 5.8 41; 6 42; 6.5 37; 91 SSD 170 278 282 163671

Continued on next page Para- Valsalva Tilt Max

Test meter Seq XPSD α-index NC CSM Ratio cBRSinc cBRSdec ΔRRI

cBRSinc Slope; Int 0.49; 2.1 0.57; 1.8 0.59; 1.2 -0.0011; 5.4 MSE 11 9 9 12 n; x¯ 38; 5.7 39; 6 41; 6.4 35; 87 SSD 154 264 265 145482

cBRSdec Slope; Int 0.57; 0.58 0.43; 1.3 0.43; 1 -0.0063; 4.6 MSE 5.1 5.4 5.3 6.7 n; x¯ 35; 5.7 36; 5.9 38; 6.3 32; 82 SSD 140 254 254 125304

Tilt Max Slope; Int -6.6; -335 -15; -294 -13; -303 -0.007; -399 ΔRRI MSE 12002 11244 12159 13595 n; x¯ 43; 5.9 44; 6 46; 6.5 41; 89 SSD 182 294 299 165559 67 Table 3.8: Metrics for conversion of baroreflex tests performed during head-up tilt (HUT). Int = intercept, MSE = mean squared error, n = sample size, x¯ = mean of the x variable, SSD = sum of the squared differences, Seq = sequence analysis, XPSD = cross- spectral analysis, NC = neck collar, CSM = carotid sinus massage, cBRSinc = cardiac baroreflex sensitivity (increase), cardiac baroreflex sensitivity (decrease), RRI = R-R interval. Bold text denotes significant correlations between indicated baroreflex measures. Para- Tilt Max

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1 ΔFVR

Seq Slope; Int 0.92; -1.1 0.78; -1.1 0.18; -3.7 0.023; -4.1 0.56; -5.4 0.034; -4.7 0.03; -5.3 0.0027; -4.7 MSE 0.66 1.3 0.72 2.3 1.7 2.1 2 2.1 n; x¯ 25; -3.3 48; -4 34; -1.9 40; -15 39; 1.9 37; 12 42; 31 48; 179 SSD 34 71 38 3361 60 1728 11736 579359

XPSD Slope; Int 0.71; -0.4 0.83; -0.01 -0.81; 2.3 0.039; -2.9 0.44; -4.1 0.023; -3.4 0.024; -4 -0.0011; -3.1 MSE 0.51 0.12 1.5 1.5 1.5 1.2 1.6 1.5 n; x¯ 25; -4.1 25; -4 22; -1.9 18; -15 21; 1.9 20; 12 21; 32 25; 206 SSD 44 45 22 1281 26 972 5879 349368

α-index Slope; Int 0.54; -1.7 1.1; -0.32 0.56; -3.4 0.033; -3.6 0.29; -4.6 0.02; -4.2 0.017; -4.6 -0.00014; -4 MSE 0.88 0.16 1.7 1.4 1.5 1.3 1.5 1.5 n; x¯ 48; -4.3 25; -3.3 34; -2.1 40; -15 39; 1.9 37; 12 42; 31 48; 179 SSD 102 34 41 3361 60 1728 11736 579359

68 NC Slope; Int 0.28; -0.76 -0.39; -0.36 0.34; -0.47 0.06; -1 0.12; -2.1 0.0075; -1.8 0.014; -2.4 0.0017; -2.2 MSE 1.1 0.74 1 0.86 1.3 0.99 1.3 1.1 n; x¯ 34; -4.1 22; 3.8 34; -4.6 38; -15 29; 1.9 27; 11 32; 29 38; 179 SSD 24 45 66 3337 39 1446 9636 471264

CSM Slope; Int 0.87; -11 1.9; -8.5 1.9; -6.8 4.6; -6 2.4; -19 -0.18; -12 -0.02; -15 0.0093; -16 MSE 87 74 83 67 86 95 91 88 n; x¯ 40; -4.4 18; -3.5 40; -4.1 38; -1.9 32; 1.9 30; 12 35; 30 40; 170 SSD 90 26 58 43 51 1597 10792 367131

PRT Slope; Int 0.41; 3.6 0.35; 3 0.29; 3.1 0.13; 2.1 0.042; 2.5 -0.043; 2.5 0.016; 1.5 -0.0008; 2 MSE 1.3 1.2 1.5 1.4 1.5 2 2 1.6 n; x¯ 39; -4.4 21; -3.3 39; -4.1 29; -1.9 32; -15 38; 12 40; 31 39; 170 SSD 83 33 59 35 2856 1798 11798 396073

Continued on next page Para- Tilt Max

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1 ΔFVR

BRSα Slope; Int 0.8; 15 0.96; 15 0.78; 15 0.44; 12 -0.11; 10 -1; 14 0.28; 2.6 0.015; 9.3 MSE 48 53 49 58 56 48 25 47 n; x¯ 37; -4.3 20; -3.1 37; -3.9 27; -1.7 30; -14 38; 2 38; 32 37; 171 SSD 74 23 44 25 2714 76 11022 391931

BRSα1 Slope; Int 3.8; 48 4.2; 46 3.2; 44 3.5; 36 -0.07; 29 2.5; 26 1.7; 12 0.043; 24 MSE 260 278 277 305 327 298 155 274 n; x¯ 42; -4.3 21; -3.3 42; -4 32; -2 35; -15 40; 2 38; 12 42; 168 SSD 91 33 63 40 3000 79 1798 399921

Tilt Max Slope; Int 15; 243 -11; 168 -1.2; 174 19; 215 1; 185 -5.3; 180 3.3; 132 1.5; 122 ΔFVR MSE 12088 15002 12593 12681 9570 10660 10657 9354 n; x¯ 48; -4.3 25; -3.3 48; -4 38; -1.9 40; -15 39; 1.9 37; 12 42; 31 SSD 102 34 71 42 3361 60 1728 11736 69 Table 3.9: Metrics for conversion of sympathetic baroreflex measures performed while supine. Int = intercept, MSE = mean squared error, n = sample size, x¯ = mean of the x variable, SSD = sum of the squared differences, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSα= alpha baroreflex sensitivity, BRSα1 = alternative alpha adrenergic baroreflex sensitivity, FVR = forearm vascular resistance. Bold text denotes significant correlations between indicated baroreflex measures. Para- Tilt Max

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1 ΔFVR

Seq Slope; Int 0.52; -4.5 0.51; -3.9 -0.022; -6.4 -0.025; -6.9 0.26; -6.8 -0.022; -6 -0.0081; -6 -0.0079; -4.9 MSE 3.7 2.8 4 3.5 3.6 3.4 3.6 3.1 n; x¯ 25; -3.9 44; -4.6 36; -1.9 38; -18 38; 1.8 36; 12 41; 31 46; 168 SSD 66 146 42 5639 57 1705 11625 418824

XPSD Slope; Int 0.34; -1.7 0.69; -0.69 0.093; -4.1 -0.032; -4.9 0.075; -4.2 0.1; -5.4 0.042; -5.5 -0.0055; -3.1 MSE 2.4 1 2.9 2.7 3.6 3 2.8 2.6 n; x¯ 25; -6.5 26; -4.8 22; -1.8 23; -22 21; 1.8 21; 13 22; 34 26; 172 SSD 103 99 23 3919 18 1164 7569 281623

α-index Slope; Int 0.49; -1.6 0.95; -1 0.15; -4.7 -0.016; -5.1 -0.21; -4.3 0.028; -4.9 -0.0019; -4.6 -0.0095; -3.1 MSE 2.6 1.4 3.5 3.9 4.3 4.5 4.1 2.8 n; x¯ 44; -6.3 26; -4 35; -1.9 37; -19 37; 1.9 35; 12 40; 31 45; 173 SSD 154 72 41 5337 60 1698 11713 422333

70 NC Slope; Int -0.0068; -2 0.038; -1.7 0.054; -1.7 0.0023; -1.8 MSE 1.2 1.2 1.2 1.1 n; x¯ 36; -6.3 22; -4.3 35; -5 38; -18 SSD 135 58 115 5746

CSM Slope; Int -1.1; -25 -2.1; -31 -0.61; -22 0.32; -18 -1.1; -16 -0.33; -15 -0.26; -9.7 0.0037; -19 MSE 153 174 151 160 144 142 121 154 n; x¯ 38; -6.4 23; -4.2 37; -4.8 38; -1.9 31; 1.8 30; 12 34; 30 40; 186 SSD 130 61 137 41 42 1632 10347 485564

PRT Slope; Int 0.11; 2.5 0.019; 1.9 -0.083; 1.5 -0.011; 1.6 MSE 1.5 0.94 1.7 1.4 n; x¯ 38; -6.3 21; -4.1 37; -4.7 31; -18 SSD 135 69 154 4225

Continued on next page Para- Tilt Max

Test meter Seq XPSD α-index NC CSM PRT BRSα BRSα1 ΔFVR

BRSα Slope; Int -0.33; 9.9 1.7; 20 0.32; 13 -0.13; 9.6 MSE 50 51 51 56 n; x¯ 36; -6.2 21; -4.1 35; -4.6 30; -19 SSD 118 69 149 4163

BRSα1 Slope; Int -0.66; 27 4.6; 53 -0.14; 30 -0.6; 20 MSE 296 306 308 272 n; x¯ 41; -6.3 22; -4.1 40; -4.7 34; -18 SSD 141 69 155 4581

Tilt Max Slope; Int -21; 38 -21; 87 -25; 53 0.3; 191 ΔFVR MSE 7965 10364 7455 12764 n; x¯ 46; -6.3 26; -4 45; -4.7 40; -18 SSD 161 72 157 5876 71 Table 3.10: Metrics for conversion of sympathetic baroreflex measures performed during HUT. Int = intercept, MSE = mean squared error, n = sample size, x¯ = mean of the x variable, SSD = sum of the squared differences, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSα= alpha baroreflex sensitivity, BRSα1 = alternative alpha adrenergic baroreflex sensitivity, FVR = forearm vascular resistance. Bold text denotes significant correlations between indicated baroreflex measures. 72

Figure 3.3: Prediction intervals for dynamic cardiac and vascular BRS indices for one individual at various confidence levels. Agreement between dynamic cardiac (A-C) and sympathetic vascular (D-F) baroreflex tests were highest compared to static tests. Note the presence of proportional bias with the cardiac α-index (A and C) and sympathetic α-index vs. Seq (D). Dashed line indicates the line of identity (perfect agreement). XPSD = cross-spectral analysis, Seq = spontaneous sequence, BRS = baroreflex sensitivity. 73

Figure 3.4: Prediction intervals for dynamic cardiac and vascular BRS indices for various sample sizes at 95% confidence. Agreement between dynamic cardiac (A-C) and sympathetic vascular (D-F) baroreflex tests were highest among all baroreflex tests. Note the presence of proportional bias with the cardiac α-index (A and C) and sympathetic α-index vs. Seq (D). Dashed line indicates the line of identity (perfect agreement). XPSD = cross-spectral analysis, Seq = spontaneous sequence, BRS = baroreflex sensitivity. 3.4.4 Repeatability

The repeatability of cardiac baroreflex measures is listed in Table 3.11. Repeatability was highest with dynamic baroreflex measures and those derived from the Valsalva maneuver. The repeatability of sympathetic baroreflex measures is listed in Table 3.12. Repeatabil- ity was again highest with dynamic baroreflex measures and those derived from the Valsalva maneuver. Repeatability for the sympathetic measures appeared lower overall than for the cardiac baroreflex measures.

74 2 Position Test n Side μdiﬀ p ICC SEM MDD RC CV R Supine Seq 48 0.42 0.46 0.95 3.92 10.85 7.64 10.6 0.92 XPSD 37 -1.39 0.04 0.91 4.34 12.02 8.19 16.98 0.83 α-index 47 -0.02 0.97 0.91 3.64 10.08 7.05 12.83 0.84 NC 26 0.53 0.56 0.42 5.85 16.21 9.55 33.18 0.28 CSM 57 L 20.04 0.04 0.57 73.4 203.45 146.79 86.83 0.33 58 R 2.89 0.85 0.56 117.96 326.97 227.67 115.26 0.35 Valsalva Ratio 34 -0.02 0.62 0.84 0.21 0.59 0.41 6.97 0.74 cBRSinc 30 -0.84 0.05 0.56 2.38 6.59 4.71 41.73 0.32 cBRSdec 29 -0.8 0.12 0.75 2.83 7.83 5.39 37.2 0.6 HUT Seq 48 0.29 0.07 0.94 1.08 3 2.17 9.68 0.89 XPSD 43 0.2 0.37 0.9 1.48 4.09 2.79 11.99 0.82 α-index 46 0.32 0.04 0.94 1.02 2.82 2.07 9.67 0.89 CSM 51 L 2.73 0.82 0.27 85.87 238.03 164.78 78.96 0.07

75 52 R 18.24 0.27 0.39 119.78 332.02 230.94 70.63 0.17

Table 3.11: Repeatability of cardiac baroreflex tests. Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSa = alpha baroreflex sensitivity, μdiff = mean difference of two repetitions, p = p-value for a paired t-test comparing the first and second repetitions, ICC = intraclass correlation coefficient, SEM = standard error of measurement, MDD = minimum difference to detect, RC = repeatability coefficient, CV = coefficient of variation. Bold denotes high repeatability (ICC > 0.7). Position Test n Side Mean Diff p ICC SEM MDD RC CV R2 Supine Seq 48 0 0.99 0.56 1.56 4.32 3.02 19.1 0.32 XPSD 18 0.15 0.4 0.72 1.02 2.84 1.55 15.11 0.66 α-index 41 -0.02 0.86 0.78 0.86 2.37 1.66 13.72 0.62 NC 33 -0.18 0.44 0.49 1.5 4.16 2.68 33.31 0.26 CSM 48 L -0.48 0.81 0.34 14.66 40.63 27.4 111.5 0.12 48 R 1.08 0.6 0.43 15.24 42.24 27.94 118.06 0.21 PRT 26 0 1 0.79 0.84 2.32 1.47 27.54 0.69 BRSα 23 -0.47 0.63 0.79 5.18 14.36 8.98 31.59 0.67 BRSα1 31 5.76 0 0.78 11.23 31.13 22.82 26.09 0.62 HUT Seq 48 -0.52 0.14 0.8 2.39 6.63 4.74 17.93 0.65 XPSD 21 0.44 0.06 0.77 1.22 3.39 2.12 14.92 0.57 α-index 37 0.17 0.57 0.78 1.83 5.08 3.56 19.22 0.62 CSM 39 L 4.19 0.14 0.34 19.54 54.16 35.69 107.06 0.13

76 38 R -0.58 0.82 0.59 18.16 50.33 31.66 232.28 0.41

Table 3.12: Repeatability of sympathetic baroreflex tests. SD = standard deviation, Seq = sequence analysis, XPSD = cross-spectral analysis, NC = neck collar, CSM = carotid sinus massage, PRT = pressure recovery time, BRSα = sympathetic baroreflex sensitivity, BRSα1 = alternative sympathetic baroreflex sensitivity, μdiff = mean difference of two repetitions, p = p-value for a paired t-test comparing the first and second repetitions, ICC = intraclass correlation coefficient, SEM = standard error of measurement, MDD = minimum difference to detect, RC = repeatability coefficient, CV = coefficient of variation. Bold denotes high repeatability (ICC > 0.7). 3.5 Discussion

In this paper, we present a variety of analytics to aid researchers and clinicians in the use and interpretation of the various baroreflex measurement options. The numerous methods available to researchers studying the arterial baroreflex makes the interpretation of the literature on baroreflex function as a whole difficult. The descriptions of mean responses, variance, agreement, repeatability, and metrics for conversion between baroreflex measurement techniques are valuable tools for future investigations into baroreflex function. Quantification of ”good“ or ”bad“ agreement is difficult, as acceptable agreement is dependent on the intended use, between-subject variance, and units of measurement. In this chapter, we provide bland-altman mean differences and limits of agreement to aid researchers in the quantification and interpretation of agreement of various baroreflex tests. In order to provide a relative framework for test comparisons, my discussion of agreement will rely on the pearson r values from linear correlation analyses. While not a robust metric of agreement in itself[18], this will allow for qualitative comparisons of agreement between tests. Similarly, while we have presented several metrics of repeatability, to allow for between-test comparisons, my discussion will focus largely on the intraclass correlation coefficients.

3.5.1 Agreement

This study is the first to compare such a large number of baroreflex tests for both equivalency and repeatability. Among all baroreflex tests, the dynamic tests displayed the highest equivalency, likely reflecting the similar theoretical underpinnings of these tests. Cardiac baroreflex tests had higher equivalency than sympathetic baroreflex tests. As we included healthy adults with no history of cardiovascular disease, the range of responses to some tests was low, and likely impacted our ability to detect significant relationships. For example, the relationship between PRT and BRSα is very clear with a range of PRT exceeding 10 seconds[159]. In our cohort, no participants displayed a PRT greater than 6 seconds, and the smaller range likely contributes to the poor relationship observed. The poor equivalency of two popular clinical metrics, the Valsalva ratio and CSM, is surprising given their widespread clinical use. As with PRT, they do not appear to precisely describe baroreflex function in healthy individuals, and are likely more useful as metrics of pathology. In particular, a large number of healthy participants do not exhibit responses to CSM that are significantly different from sham massage (see Figure 5.5 in Chapter 5, page 117) and will therefore have low agreement with other measures. The dynamic sympathetic baroreflex tests described here (sequence analysis, cross- spectral analysis, and α-index) are, to our knowledge, novel, and have yet to be validated. In this study, we found no relationship between the supine sBRS and the functional maximum FVR response during a tilt test to presyncope. While dynamic cBRS in the supine position

77 is a reliable indicator of its response upon tilt, the supine sBRS is not. However, the sBRS during HUT correlated with the maximum FVR response during the tilt test (Seq: r = 0.4, n = 46, p = 0.01; α-index: r = 0.49, n = 45, p < 0.01, Table 3.6). Due to the large drop in n due to poor coherence, the correlation between XPSD sBRS during HUT and the maximum FVR response during tilt did not quite achieve statistical significance (r = 0.34, n = 26, p = 0.09). Additionally, sequence sBRS correlated with two previously validated sympathetic measures from the Valsalva maneuver, PRT and BRSα1. We therefore believe that these techniques are useful metrics of sympathetic baroreflex function, and provide similar advantages to dynamic cardiac baroreflex tests. The agreement between cardiac baroreflex measures was, on average, greater than that of the sympathetic vascular baroreflex measures. Calculation of the cardiac responses to baroreflex stimulation involves direct measurement baroreflex modulation of RRI. Con- versely, our measurements of the vascular responses were based on changes in brachial blood flow velocity (eg. sequence analysis sBRS) or blood pressure (eg. PRT), and were thus indirect measures of changes in vascular resistance. The associated reduction in measurement precision likely (in part) underlies the reduced agreement between indices of vascular baroreflex function. We have reported agreement between baroreflex measures using Bland-Altman analysis (Tables 3.3, 3.4, 3.5, and 3.6). While this gives a useful, quantitative measure of agreement, this paper is also intended as a resource to other researchers who wish to convert baroreflex values between different tests. To this end, we have calculated single-individual prediction intervals for dynamic cardiac and sympathetic baroreflex tests (Figure 3.3). At 95% confidence, converting BRS values between tests on one individual is likely not feasible due to the wide prediction intervals. However, for researchers comparing BRS values of a sample, conversion of baroreflex tests gives reasonable prediction intervals (Figure 3.4). Necessary metrics for conversion of supine and tilted cardiac baroreflex tests are listed in Tables 3.7 and 3.8, respectively. Metrics for conversion of supine and tilted sympathetic baroreflex tests are listed in Tables 3.9 and 3.10, respectively.

3.5.2 Repeatability

The repeatability of baroreflex tests are described in Tables 3.11 and 3.12. Repeatability was generally higher with dynamic baroreflex tests, and with metrics from the Valsalva maneuver. The repeatability of the BRS measures from the neck collar protocol were lower than expected. We feel that this is due to extra sources of variance inherent in the technique: maintenance of target pressures is difficult and achieved pressures vary between repetitions; during the calculation of BRS, sigmoid curves are fitted to the raw data, adding another source of variance. Accordingly, use of this technique to construct full baroreflex sigmoid

78 curves should involve repetitions of each target pressure as performed during calculation of BRS for the agreement section of this study.

3.5.3 Limitations

While the time of day was standardized between participants, for practical reasons and to minimize participant burden, the neck collar protocol, carotid sinus massage, and head- up tilt protocols occurred on different days. This was done deliberately to reduce the possibility of tests interacting with one another. However, comparison of tests performed on different days have added variance of day-to-day fluctuations in baroreflex function[113]. Indeed, the agreement between tests performed on the same day (the Valsalva maneuver, orthostatic stress test, and dynamic baroreflex tests) was higher than those performed on different days (carotid sinus massage, neck collar protocol, and orthostatic stress test). This predominately affects comparisons with the neck collar protocol, where agreement with metrics from the Valsalva maneuver and the maximal response during the orthostatic stress test is likely underestimated. We do not believe day-to-day variation in baroreflex function affected agreement with CSM, as most participants in our study did not display a response significantly different from sham massage (see Figure 5.5, page 117). Logically, agreement with a non-response will be low. We did not assess repeatability of the tilt test to presyncope; however, it has been assessed previously and was found to be highly repeatable (between-test variance of ±1.1min).

3.6 Conclusion

The large number of baroreflex assessment techniques currently available allows for baroreflex measurement in a wide variety of methodological situations. However, agreement between techniques varies widely and comparisons should be made with caution. Indices of cardiac baroreflex function displayed higher agreement than sympathetic baroreflex measures. In this chapter we provide tools to aid in the comparison of baroreflex techniques and consolidation of the baroreflex literature as a whole.

79 Chapter 4

Sex and Age Diﬀerences in Baroreﬂex Function: Insights from Neck Collar Stimulation and an Orthostatic Stress Test

4.1 Abstract

The arterial baroreflex is critical for the regulation of blood pressure, particularly during orthostasis. Rapid changes in blood pressure are rectified via baroreflex-mediated cardiac and vascular responses, but the relative importance of each response is still debated. Recent evidence indicates substantial age- and sex-related differences in vascular baroreflex function, suggesting that the relative contributions of the cardiac and vascular arms of the baroreflex may be subgroup-dependent. We hypothesized that: 1) robust vascular responses to orthostatic stress will correspond with greater orthostatic tolerance; and 2) this relationship will be stronger in men, and in older participants. We recruited 49 participants (26 females, aged 19-84 years). Cardiac and sympathetic vascular baroreflex function was assessed with the variable pressure neck collar, and a maximal orthostatic stress test to presyncope. The maximum heart rate and cBRS responses during the orthostatic stress test were larger in younger participants (Δheart rate: Younger = 47±3 bpm, Older = 35±4 bpm, p = 0.0095; ΔcBRS: Younger = -14±1.6 ms.mmHg-1, Older = -3.5±1.9 ms.mmHg-1, p = 0.0003). The absolute (younger ΔRRI = -118±14 ms, older ΔRRI = -37±18 ms, p = 0.001) and relative (older-younger: -20±6 %, p = 0.01) cardiac responses to simulated hypotension with the neck collar also declined with age. The maximum FVR response during orthostasis was smaller in younger women compared to all other groups (all p < 0.03). Younger participants mounted a smaller ΔsBRS to HUT (Younger = -1.5 SE 0.4

80 %.mmHg-1, Older = -3.5 SE 0.5 %.mmHg-1, p = 0.0015). Men also mounted a larger ΔsBRS to HUT than women (Males = -3.5 SE 0.4 %.mmHg-1, Females = -1.5 SE 0.4 %.mmHg-1, p = 0.0016). Younger females did not display an increase in sBRS upon HUT (ΔsBRS = 0.5±0.5 %.mmHg-1, p = 0.95). Conversely, the absolute vascular responses to simulated hypotension with the neck collar were larger in younger participants (younger ΔFVR = +83±9 %, older ΔFVR = +45±12 %, p = 0.01). The effect of aging on baroreflex function is different for men and women, suggesting different strategies for baroreflex control of blood pressure. However, significant intra-group variance exists, which may indicate that grouping by sex is not the best method to stratify baroreflex function. Orthostatic stress and discrete stimulation of the carotid baroreceptors demonstrate different age-related changes in vascular baroreflex function.

4.2 Introduction

Rapid changes in blood pressure are buffered by the arterial baroreflex through sympathetically-mediated vascular resistance responses and predominately vagally-mediated changes in heart rate[121, 21]. Baroreflex function is commonly quantified by its sensitivity, or the magnitude of the cardiac or vascular response elicited divided by the magnitude of the change in blood pressure that evoked the reflex[166, 15]. BRS therefore reflects the short- term responsive capacity of the cardiovascular system to perturbations in blood pressure, such as during orthostatic stress. However, previous work has shown that the sensitivity of the cardiac component of the arterial baroreflex does not play a major role in orthostatic tolerance[37, 67], or the ability to maintain blood pressure during orthostatic stress. This has led to the hypothesis that the vascular responses are the main mechanism behind the maintenance of blood pressure during orthostasis[48, 73]. It has been shown that in young adults an impairment in the ability to increase vascular resistance responses to baroreflex simulation when upright is associated with poor orthostatic tolerance[38, 30]. However, there continues to be debate regarding the relative contribution of the cardiac and vascular resistance responses to orthostasis[33, 201]. To our knowledge, no study to date has examined the relationship between the sensitivity of the vascular arm of the baroreflex and orthostatic tolerance. Sympathetic control of vascular tone and MAP appears to be different for men and women. Resting MSNA is correlated with TPR but inversely correlated with CO in young men, while young women display neither relationship, ostensibly due to increased beta- adrenergic vasodilation[75]. Thus resting MSNA is not correlated with MAP in either sex. However, with age this relationship becomes significant in both sexes[75]. Young men have larger TPR responses to tilt than young women[55], which may be mediated by different sympathetic neural recruitment strategies[162] and increased vascular transduction[93, 22]. Young men also mount a greater relative vascular response to neck suction (a hypertensive

81 stimulus), but not neck pressure (a hypotensive stimulus), compared to young women[92]. The effect of age on sympathetic vascular function does not appear to be the same for both sexes[22, 124, 74], as older individuals appear to mount a larger TPR response during orthostasis compared to younger individuals[56, 102], with older women mounting larger TPR responses than older men[56, 124]. The substantial variability in sympathetic baroreflex function in these subgroups may indicate that the extent of sympathetic predominance in the regulation of blood pressure varies between individuals. Certainly the it appears that those who are able to mount a robust baroreflex-mediated vasoconstriction are better able to regulate their blood pressure, but whether this holds true equally for different subgroups has not been established. We hypothesized that: 1) robust vascular responses to orthostatic stress will correspond with greater orthostatic tolerance; and 2) this relationship will be stronger in men, and in older participants.

4.3 Methods

We examined the cardiac and sympathetic responses to two static baroreflex tests: the orthostatic stress test, and the neck collar protocol. Methodology for the orthostatic stress test is detailed in Section 2.5, page 36. The neck collar methods are described in Section 2.4, page 31, and quantification of cBRS and sBRS via sequence analysis is described in Section 2.8, page 43. The experimental protocols for this chapter are outlined in Figure 4.1. The orthostatic stress test and neck collar protocols were performed on different days, with the order randomized.

4.3.1 Data Analysis

For paired analyses, mixed linear models were used with participant as a random block; for unpaired analyses, simple linear models were constructed. Residual vs. ﬁtted and q-q plots were examined to ensure normality of the residuals. Signiﬁcance was assumed when p < 0.05. Results are reported as mean ± standard error unless otherwise stated.

4.3.1.1 Standardization of Cardiovascular Responses to Carotid Baroreceptor Stimulation with the Neck Collar

The neck collar protocol involves delivery of distinct pressures to the carotid sinus. However, achieving and maintaining collar pressures can be diﬃcult, and there is often error associated with achieving a target pressure. As the cardiovascular responses are proportional to the neck collar pressure delivered (calculated as the mean pressure), comparison of the absolute responses should involve ﬁrst normalizing the response to the pressure delivered in order to reduce the error associated with the pressure delivery. For the comparison of the

82 Figure 4.1: Schematic diagrams of the experimental protocols - each protocol was completed on a different day, with the order randomized. A) Experimental protocol for the orthostatic stress test. Following 20 minutes supine rest, participants were head-up tilted to 70 degrees for a further 20 minutes. Incremental lower-body negative pressure was applied for ten minute intervals at -20, -40, and -60mmHg. The test was terminated if any of the following criteria were met: systolic arterial dropped below 80 mmHg; heart rate decreased to below 50 bpm or increased to higher than 170 bpm; the participant experienced symptoms of presyncope such as dizziness, warmth, and requested to stop; or the protocol was completed. Each test was therefore stopped at a different time point. B) Neck collar experimental protocol. Following 10 minutes supine rest, graded positive and negative pressures (-60, -40, -20, 0, 20, 40, and 60 mmHg) were generated over the carotid sinus. Vagally-mediated cardiac responses were assessed during 5 seconds of neck pressure/suction applied for five seconds during breath hold. Sympathetically-mediated vascular responses were assessed during 20 seconds of neck suction/pressure during normal breathing. Each pressure/timing protocol was repeated twice, and were separated by at least 1 minute to allow return of cardiovascular homeostasis. LBNP = lower-body negative pressure. absolute cardiovascular responses to neck suction/pressure, we therefore standardized the cardiovascular responses to neck pressure and suction via the following formula:

T arget P ressure Response = ∗ Response Standardized Achieved P ressure Actual In Chapter 3, we describe sBRS using three dynamic techniques: the α-index, cross- spectral analysis, and spontaneous sequence analysis. One important consideration for these techniques is that brief signal interruptions are common with brachial Doppler. For

83 calculation of baroreflex methods using spectral methods (α-index and cross-spectral analysis), these interruptions can be smoothed using linear interpolation, but interpolating longer signal interruptions (>5 beats) reduces the validity of the resulting metric of sBRS. Fur- thermore, we found that low coherence using cross-spectral analysis hampered the utility of the technique in the supine position. Despite these caveats, agreement between the different dynamic tests was high. We ultimately chose to extend our analyses to incorporate spontaneous sequence analysis as: 1) it is more robust to signal interruptions; and 2) it displayed reasonable agreement with two previously validated sympathetic baroreflex measures, PRT and BRSα1.

4.3.1.2 Calculation of Cardiac Output, Total Peripheral Resistance

We calculated CO and TPR using the Modelflow method. Briefly, a nonlinear three-element model of arterial input impedence is used to calculate flow through the aorta based on the arterial wave form, and demographic information from the participant (age, sex, height, weight). This technique has been validated previously[198, 82], and has also been validated in a previous study using the neck collar[92]. Following beat-to-beat monitoring of CO and MAP, TPR can also be calculated using Ohm’s law.

4.3.1.3 Calculation of Percent Contribution

The calculation of the percent contribution of CO and TPR during the cardiovascular response to neck suction/pressure was inspired by Kim et al (2011)[92]. First, the maximal changes in CO and TPR were identiﬁed. The predicted change in MAP was then calculated using the individual change in CO or TPR while assuming no response in the other variable.

Therefore, the predicted MAP response using the CO response (MAPCO) was calculated as:

MAPCO = COP eak Response ∗ TPRBaseline

Similarly, the predicted MAP response using the TPR response (MAPTPR) was calculated as:

MAPTPR = TPRP eak Response ∗ COBaseline

The percent contributions of CO and TPR were then calculated as:

MAPCO CO%Contribution = ∗ 100 MAPCO + MAPTPR

MAPTPR TPR%Contribution = ∗ 100 MAPCO + MAPTPR

84 4.3.1.4 Calculation of the Timing of Maximal Vascular Responses During Or- thostasis

The timing of the maximal vascular response during orthostasis (HUT and all achieved levels of LBNP) was calculated by ﬁrst standardizing the length of the orthostatic phase for each individual from 0-100% of orthostatic tolerance. The timing of the maximal vascular response during orthostasis was then expressed as a percent of orthostatic tolerance.

4.3.1.5 Model Averaging

To explore subgroups within our data set that may display different strategies to maintain blood pressure during orthostatic stress, we applied an automated model selection tool based on aikeke information criteria (AIC)[8]. Model selection was performed using the glmulti package (version 1.0.7) in R (version 3.4.3) and RStudio (version 1.1.423). In short, linear models were constructed from a candidate set of predictive variables. All possible combinations of predictive variables were used to construct a set of linear models. Models were assigned an AIC value based on: 1) how well the model fit the data, and 2) the complexity of the model (ie. how many terms were included in the model). Simple models that explained the data well were ranked highly, while more complex models were penalized to avoid over-fitting, and were ranked lower. Based on these criteria, the posterior probability that each model is the “true” model was calculated. Models that summed to 95% posterior probability were considered for further analysis, and posterior probability was re-calculated using only these top models. Variable importance was calculated as the sum of posterior probabilities of the models in which the variable appeared. Beta coefficients were also estimated as the weighted average beta coefficient of a variable over the top models. Coefficients were weighted to the posterior probability of each model. For models in which the variable did not appear, the beta coefficient was considered to be zero.

4.4 Results

Data were collected in 49 participants (26 females) over a wide age range (19-84 years). Demographics information is in Table 4.1. Participants were split by sex, as well as age into “older” and “younger” groups based on an age cutoff of 45 years, which appropriately divided women into pre- and post-menopausal groups. Older participants were shorter than younger participants (-7±2cm, p = 0.002). Men were taller (+8±2cm, p = 0.0002) and heavier (+14±2 kg, p < 0.0001) than women. There were no significant differences in resting DAP, heart rate, or orthostatic tolerance. SAP in the older participants was not quite statistically significantly different from the younger participants (p = 0.058).

85 Participants’ medications, menopausal status, and menstrual cycle day are presented in Table 4.2. Age Height Weight BP HR OT Sex Group n Age (y) (cm) (kg) (mmHg) (bpm) (min) Female Older 10 60 (47-69) 162±7‡ 64±7 131±15 / 67±9 60±6 27±7 Female Younger 16 24 (19-38) 171±6 64±7 124±8 / 70±8 66±10 22±10† Male Older 10 68 (59-84) 173±7*‡ 75±10* 130±20 / 68±9 61±9 26±8 Male Younger 13 26 (20-32) 177±8* 80±9* 122±9 / 68±4 63±7 30±9

Table 4.1: Demographics information for the four age groups. BP = blood pressure, HR = heart rate, OT = orthostatic tolerance. * = significant main effect of Sex (p<0.05), ‡ = significant main effect of Age Group (p<0.05), † = p = 0.06 compared to young men. Data are expressed as mean ± standard deviation.

Menstrual status Post-menopausal 10 Birth control 5 No birth control 11 Menstrual cycle day (neck collar protocol) 11 (SD 11) Menstrual cycle day (tilt protocol) 15 (SD 9) Medication No medication 37 Synthroid 3 SSRIs 3 Aspirin 4 Tylenol 1 NSAID 2 Antihistamine 2

Table 4.2: Participants’ menopause status, birth control, and medications. SSRIs = selective seratonin re-uptake inhibitor, NSAIDs = non-steroidal anti-inﬂammatories

4.4.1 Cardiovascular Responses to Orthostatic Stress

4.4.1.1 Cardiac Responses

The maximum heart rate response during the orthostatic stress test was larger in younger participants (Younger = 47±3 bpm, Older = 35±4 bpm, p = 0.0095, Figure 4.2A). Younger participants exhibited a larger resting cBRS than older participants (+14±3 ms.mmhg-1, p < 0.0001, Table 4.3). Younger participants also mounted a larger cBRS response to HUT (Younger = -14±1.6 ms.mmHg-1, Older = -3.5±1.9 ms.mmHg-1, p = 0.0003, Figure 4.2B). The maximum heart rate response during the orthostatic stress test was correlated with orthostatic tolerance in all participants (R2 = 0.15, p = 0.0051); however, a subgroup analysis revealed this to be only true of females (R2 = 0.41, p = 0.0004, Figure 4.2C).

86 Maximum heart rate responses correlated with orthostatic tolerance in participants who experienced a Vasovagal Syncope International Study (VASIS)[24] Type 3 (vasodepressor) faint (R2 = 0.33, p = 0.0014) but did not quite achieve statistical signiﬁcance for those who experienced a VASIS Type 1 (mixed vasodepressor and cardio-inhibitory) faint (p = 0.08, Figure 4.2D).

4.4.1.2 Sympathetic Responses

The maximum FVR response during the orthostatic stress test was smaller in younger women compared to all other groups (p < 0.03, Figure 4.3A). Younger participants exhibited a larger resting sBRS than older participants (+2±0.3, p < 0.0001, Table 4.3), and mounted a smaller ΔsBRS to HUT (Younger = -1.5 SE 0.4 %.mmHg-1, Older = -3.5 SE 0.5 %.mmHg-1, p = 0.0015). Men also mounted a larger ΔsBRS to HUT than women (Males = -3.5 SE 0.4 %.mmHg-1, Females = -1.5 SE 0.4 %.mmHg-1, p = 0.0016, Figure 4.3B). Younger females did not display an increase in sBRS upon HUT (ΔsBRS = 0.5±0.5 %.mmHg-1, p = 0.95). The maximum FVR response during the orthostatic stress test was correlated with orthostatic tolerance in all participants (R2 = 0.16, p = 0.0051); however, a subgroup analysis revealed this to be only true of younger females (R2 = 0.59, p = 0.0005, Figure 4.3C). Maximum FVR responses did not correlate with orthostatic tolerance in participants who experienced a VASIS Type 3 faint (p = 0.3), but almost achieved statistical signiﬁcance in those with a VASIS Type 1 faint (p = 0.06, Figure 4.3D).

cBRS sBRS Position Age Group Sex (ms.mmHg-1) (%.mmHg-1) Supine Older Female 8.6±0.8‡ -3±0.3‡ Older Male 7.8±1.1‡ -3.2±0.4‡ Younger Female 23.6±3 -5±0.3 Younger Male 19.8±4 -5.2±0.3 HUT Older Female 5±0.5Φ‡ -5.4±0.6Φ* Older Male 4.3±0.4Φ‡ -7.7±1.1Φ Younger Female 7.5±0.7Φ -5.5±0.3* Younger Male 8.8±1.1Φ -7.6±0.4Φ

Table 4.3: Cardiac and sympathetic baroreflex sensitivity during supine and tilt. Data are expressed as mean ± standard error. * = significant main effect of sex within position, ‡ = significant main effect of age group within position, φ = significantly different from the supine position within age group and sex. cBRS = cardiac baroreflex sensitivity, sBRS = sympathetic baroreflex sensitivity.

87 Figure 4.2: Cardiac responses to tilt. A) Maximum heart rate response for younger and older men and women. Younger participants displayed larger maximum heart rate responses to tilt (‡ = main effect of age, p = 0.01). B) Cardiac baroreflex response to head-up tilt for younger and older men and women. Younger participants displayed a larger change in cBRS when moving from supine to tilt (‡ = main effect of age, p < 0.001). C) The maximum heart rate response during tilt was correlated with orthostatic tolerance; however, this relationship was driven by the female participants. D) Maximum heart rate response during tilt was correlated with orthostatic tolerance in those participants who experience a Type 3 (vasodepressor)[24] faint at presyncope. Horizontal dashed lines represent group mean, horizontal solid lines represent 95% confidence interval. HR = Heart Rate, SAP = systolic arterial pressure, RRI = R-R interval, BRS = baroreflex sensitivity, OT = orthostatic tolerance, VASIS = Vasovagal Syncope International Study.

88 Figure 4.3: Vascular responses to tilt. A) Maximum vascular response for younger and older men and women. Younger women displayed smaller maximum vascular responses to tilt († = different from other groups, p = 0.01 ). B) Vascular baroreflex response to head-up tilt for younger and older men and women. Younger participants displayed a larger change in sBRS when moving from supine to tilt (‡ = main effect of age, p = 0.002). Female participants displayed a smaller change in sBRS when moving from supine to tilt (* = main effect of sex, p < 0.001). C) The maximum vascular response during tilt was correlated with orthostatic tolerance; however, this relationship was driven by the younger female participants. D) The maximum FVR response during orthostasis was not correlated with with orthostatic tolerance for either VASIS type 3 (vasodepressor)[24], and did not quite achieve statistical significance in those meeting VASIS type 1 (mixed) criteria. Horizontal dashed lines represent group mean, horizontal solid lines represent 95% confidence interval. FVR = forearm vascular resistance, MAP = mean arterial pressure, RRI = R-R interval, BRS = baroreflex sensitivity, OT = orthostatic tolerance, VASIS = Vasovagal Syncope International Study.

89 4.4.1.3 Cardiac and Sympathetic Baroreflex Sensitivity cBRS and sBRS were compared to the maximum cardiac and sympathetic responses during orthostasis, respectively (Figure 4.4). Supine cBRS was significantly correlated with the maximum RRI response during orthostasis (R2 = 0.14, p = 0.01) but HUT cBRS was not (R2 = 0.03, p = 0.26). However, the HUT-supine cBRS response correlated with the maximum RRI response during orthostasis (R2 = 0.22, p = 0.001). Supine sBRS did not correlate with the maximum FVR response during orthostasis (R2 = 0.07, p = 0.07), but the relationship was significant with HUT sBRS (R2 = 0.21, p = 0.001). The HUT-supine ΔsBRS correlated best with the maximum FVR response during orthostasis (R2 = 0.34, p < 0.001).

4.4.2 Cardiovascular Responses to Carotid Baroreceptor Stimulation with the Neck Collar

4.4.2.1 Cardiac and Sympathetic Baroreﬂex Curves

Averaged cardiac and sympathetic baroreflex curves for men and women split by age group are shown in Figure 4.5. Cardiac baroreflex curves were right-shifted to higher blood pressures compared to sympathetic baroreflex curves for all groups (Table 4.4). Younger participants had larger cBRS compared to older participants (ΔcBRS +2.4 SE 1.3 ms.mmHg-1, p = 0.049). In younger females, the position of the operating point relative to the cardiac centering point was significantly left-shifted compared to younger males (-17 SE 6 mmHg, p = 0.04). Younger participants had larger sBRS compared to older participants (ΔsBRS -0.7 SE 0.3%.mmHg-1, p = 0.03). In younger participants, the sympathetic threshold and centering point was significantly left-shifted compared to older participants (Δthreshold: -19±7 mmHg, p = 0.0075; Δcentering point: -14±6 mmHg, p = 0.01).

90 Figure 4.4: Cardiac and sympathetic baroreflex sensitivity during supine, HUT, and the supine-HUT BRS response. cBRS correlates with the maximum RRI response during orthostasis during supine (A) but not HUT (B). The HUT-supine cBRS response correlates best with the maximum RRI response during orthostasis (C). sBRS in the supine position (D) does not correlate with the maximum FVR response during orthostasis. The relationship becomes significant during HUT (E). The HUT-supine sBRS response correlates best with the maximum FVR response during orthostasis (F). RRI = R-R interval, FVR = forearm vascular resistance, cBRS = cardiac baroreflex sensitivity, sBRS = sympathetic baroreflex sensitivity. Outliers are circled in black and were removed from the correlation analysis.

91 Figure 4.5: Averaged cardiac and sympathetic baroreflex sigmoid curves for younger and older men and women. Points on the left indicate baroreflex threshold, points on the right indicate baroreflex saturation. Baroreflex curves were fitted to five averaged points using the threshold, centering point, saturation, and minimum/maximum asymptotes of the baroreflex sigmoids from each group. RRI = R-R interval, FVR = forearm vascular resistance, CSP = carotid sinus pressure.

92 Baroreflex Arm Parameter Younger Older Males Females Males Females Cardiac BRS (ms.mmHg-1) 6.3±1.0‡ 8.6±1.3‡ 3.8±0.8 6.4±1.8 Threshold (mmHg) 64.3±3.1Φ 79.8±4.1Φ 81.4±8.0Φ 74.0±12.8Φ Centering Point (mmHg) 84.2±4.3Φ 100.6±3.7Φ 100.8±7.2Φ 90.1±9.5Φ Saturation (mmHg) 104.8±7.2Φ 122.6±5.3Φ 121.2±7.8Φ 107.3±9.4Φ OP-CP (mmHg) 2.5±4.2†Φ -14.4±3.2Φ -8.5±5.6Φ -0.9±8.2Φ Operating Range (mmHg) 40.5±6.3 42.8±6.3 39.8±7.8 33.3±9.3 Sympathetic BRS (%.mmHg-1) -2.4±0.4‡ -2.0±0.2‡ -1.4±0.2 -1.6±0.1 Threshold (mmHg) 31.8±3.3‡ 46.1±5.4‡ 55.5±8.5 58.6±10.4 Centering Point (mmHg) 57.2±4.2‡ 68.5±4.7‡ 78.1±6.5 76.0±8.0 Saturation (mmHg) 85.1±7.2 93.2±4.7 101.7±5.7 94.1±7.1 OP-CP (mmHg) 28.9±4.2 16.6±4.4 13.9±4.2 14.3±7.1 Operating Range (mmHg) 53.3±6.1 47.1±5.7 46±7.9 35.5±8.3 93 Table 4.4: Neck collar sigmoid curve parameters. ‡ = significant main effect of age group, Φ = significant main effect of baroreflex arm, † = significantly different from young females. Note that due to the difference in units, cardiac baroreflex sensitivity was not compared to sympathetic baroreflex sensitivity. Operating range was calculated as saturation-threshold. BRS = baroreflex sensitivity, OP = operating point, CP = centering point. Data are expressed as mean ± standard error 4.4.2.2 Absolute Responses to Neck Suction and Pressure

The cardiac responses to a hypertensive stimulus of -60mmHg neck suction were larger in younger participants (younger ΔRRI = 214±21 ms, older ΔRRI = 120±27 ms, p = 0.009), and in women (males = 118±24 ms, females ΔRRI = 215±25 ms, p = 0.007, Figure 4.6A). Cardiac responses to a hypotensive stimulus of +40mmHg were larger in younger participants (younger ΔRRI = -118±14 ms, older ΔRRI = -37±18 ms, p = 0.001). Sympathetic responses to a hypertensive stimulus of -60mmHg and a hypotensive stimulus of +40mmHg were larger in younger participants (-60mmHg: younger ΔFVR = -20±3 %, older ΔFVR = -12±3 %, p = 0.05; +40mmHg: younger ΔFVR = +83±9 %, older ΔFVR = +45±12 %, p = 0.01, Figure 4.6B). The cardiac responses to orthostasis were greater than those to +40mmHg neck pressure in all groups (+310±20 ms, p < 0.0001). The sympathetic vascular responses to orthostasis were greater than the vascular responses to orthostasis in young men (127±27, p = 0.0004), older men (167±35%, p = 0.0003), older women (179± 34%, p < 0.0001), but not young women (7±25%, p = 0.99).

4.4.2.3 Comparison of Cardiac Output and Total Peripheral Resistance Re- sponses

The percent contribution of CO and TPR to a theoretical maximum MAP response to neck suction and neck pressure are shown in Figure 4.7. All groups except young women responded to neck pressure (a hypotensive stimulus) with a larger contribution of TPR than CO (all p < 0.002). Conversely, all groups except older males responded to neck suction (a hypertensive stimulus) with a larger contribution of CO than TPR (all p < 0.02). The relative contributions of CO and TPR were diﬀerent between negative and positive pressures for all groups (p < 0.0001), where neck pressure elicited a predominant TPR response and neck suction elicited a predominant CO response. Older participants had a larger relative contribution of TPR to neck pressure than younger participants (+20±6 %, p = 0.01).

4.4.3 Baroreﬂex Strategies for the Maintenance of Blood Pressure During Orthostasis

We performed model averaging using cardiac and sympathetic variables from the baseline period (defined as the last five minutes of supine), and the maximal response during tilt. Cardiac variables included CO, SV, heart rate, cBRS, and SAP. Sympathetic variables included DAP, TPR, FVR, and sBRS. Model selection and averaging was performed separately for older and younger females, and older and younger males (Figure 4.8). None of the measured variables reached high importance, defined as >0.8. The magnitude of the decrease in SV was a moderately important (defined as 0.6-0.8) predictor of orthostatic tolerance for young males and older

94 Figure 4.6: Absolute cardiac and sympathetic responses to simulated carotid hypotension (+40 mmHg neck pressure) and hypertension (-60 mmHg neck suction). ‡ = main eﬀect of age, p < 0.05. * = main eﬀect of sex, p < 0.05. RRI = R-R interval, FVR = forearm vascular resistance. females. Supine cBRS was also a moderately important predictor of orthostatic tolerance for young females. When the data were split by sex, supine sBRS and the HUT-supine ΔsBRS were moderately and highly important for men, respectively (Figure 4.9). Important variables in women were the maximal heart rate response during tilt, as well as the supine sBRS. Important variables for younger participants were largely cardiac in nature (maximal heart rate response during orthostasis, supine cBRS and the HUT-supine cBRS response),

95 Figure 4.7: Percent contributions of cardiac output and total peripheral resistance during neck suction (a hypertensive stimulus) and pressure (a hypotensive stimulus). All positive and negative pressures were averaged, as no significant differences were found within the positive or negative pressures. ‡ = significantly different between cardiac output and total peripheral resistance within condition, † = significantly different from younger males and females during positive pressure, * = significantly different between negative and positive pressures within condition. but also included supine sBRS. In older participants, the only variable to reach moderate importance was the HUT-supine cBRS response (Figure 4.10). Model averaging was also performed on subsets of the data based on the maximal FVR response during orthostatic stress. Participants who had a maximal FVR response greater than 125% were considered “constrictors,” and those with FVR responses smaller than 125% were considered “non-constrictors” (Figure 4.11). This value was chosen as it appears to reflect an inflection point in the relationship between the maximal FVR response during orthostasis, and orthostatic toleranc (Figure 4.3). Predominately sympathetically-mediated variables (baseline DAP, supine sBRS, and HUT-supine ΔsBRS) were highly important in

96 the constrictor group, while non-constrictors relied primarily on cardiac variables (maximal heart rate response during orthostasis, supine cBRS, and HUT-supine ΔcBRS). Important variables for these two groups were also ranked as important variables for the group as a whole.

Figure 4.8: Results of model averaging for the prediction of orthostatic tolerance, split by sex and age group. A small orthostatic decrease in stroke volume was a moderately important predictor of higher orthostatic tolerance for older women and younger men. Larger supine cBRS was a moderately important predictor of greater orthostatic tolerance for younger women. BL = baseline (the last five minutes preceding head-up tilt), Δ= the maximal change during orthostasis, CO = cardiac output, SV = stroke volume, HR = heart rate, cBRS = cardiac baroreflex sensitivity, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, TPR = total peripheral resistance, FVR = forearm vascular resistance, sBRS = sympathetic baroreflex sensitivity. The horizontal dark grey bar represents high variable importance (80-100), while the horizontal light grey bar represents moderate variable importance (60-80).

97 Figure 4.9: Results of model averaging for the prediction of orthostatic tolerance, split by sex, and for the group as a whole (“All”). Important variables predicting greater orthostatic tolerance in females included greater maximal increases in heart rate during orthostasis, and smaller supine sBRS. In men, the only highly important variable predictive of greater orthostatic tolerance was smaller HUT-supine ΔsBRS responses, however smaller supine sBRS was a moderately important predictor of increased orthostatic tolerance. BL = baseline (the last five minutes preceding head-up tilt), Δ= the maximal change during orthostasis, CO = cardiac output, SV = stroke volume, HR = heart rate, cBRS = cardiac baroreflex sensitivity, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, TPR = total peripheral resistance, FVR = forearm vascular resistance, sBRS = sympathetic baroreflex sensitivity. The horizontal dark grey bar represents high variable importance (80-100), while the horizontal light grey bar represents moderate variable importance (60-80).

4.5 Discussion

In this study, we performed cardiac and sympathetic baroreﬂex testing using the neck collar technique and an orthostatic stress test. We have utilized a novel method of dynamic vascular sBRS measurement to simultaneously describe cardiac and vascular function during orthostatic stress. Using the neck collar technique, we have constructed full sympathetic and cardiac baroreﬂex sigmoid curves, and demonstrated that sympathetic sigmoid curves

98 Figure 4.10: Results of model averaging for the prediction of orthostatic tolerance, split by age group, and for the group as a whole (“All”). In younger participants, larger maximum heart rate responses, larger supine cBRS, and smaller HUT-supine ΔcBRS responses were highly important predictors of greater orthostatic tolerance. In older participants, smaller HUT-supine ΔcBRS responses were moderately predictive of greater orthostatic tolerance. BL = baseline (the last five minutes preceding head-up tilt), Δ= the maximal change during orthostasis, CO = cardiac output, SV = stroke volume, HR = heart rate, cBRS = cardiac baroreflex sensitivity, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, TPR = total peripheral resistance, FVR = forearm vascular resistance, sBRS = sympathetic baroreflex sensitivity. The horizontal dark grey bar represents high variable importance (80- 100), while the horizontal light grey bar represents moderate variable importance (60-80). are left-shifted compared to cardiac baroreflex curves. The vascular responses to orthostasis appear to reflect known sex differences in vascular function, while the vascular responses to simulated hypotension using the neck collar do not. This suggests that previously reported sex differences in baroreflex function may be dependent on the time course of the baroreflex stimulus (with different responses to acute baroreflex stimulation vs. longer baroreflex stimuli, eg. orthostatic stress), but may also reflect methodological differences between orthostatic stress and the neck collar technique. Lastly, we have identified distinct strategies

99 Figure 4.11: Results of model averaging for the prediction of orthostatic tolerance, split by constrictors (maximal FVR response during orthostasis > 125%) vs. non-constrictors (maximal FVR response during orthostasis < 125%), and for the group as a whole (“All”). Im- portant variables for constrictors were primarily sympathetic in nature (larger supine DAP, smaller supine sBRS and HUT-supine ΔsBRS responses), while important non-constrictor variables were primarily cardiac (larger maximum increases in heart rate, larger supine cBRS, and smaller HUT-supine ΔcBRS responses). BL = baseline (the last five minutes preceding head-up tilt), Δ= the maximal change during orthostasis, CO = cardiac output, SV = stroke volume, HR = heart rate, cBRS = cardiac baroreflex sensitivity, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, TPR = total peripheral resistance, FVR = forearm vascular resistance, sBRS = sympathetic baroreflex sensitivity. The horizontal dark grey bar represents high variable importance (80-100), while the horizontal light grey bar represents moderate variable importance (60-80). for the baroreflex regulation of blood pressure based on the maximal FVR response during orthostasis.

100 4.5.1 Cardiovascular Responses to Orthostatic Stress

The heart rate and ΔcBRS responses to orthostasis were greater in younger individuals, in keeping with well-established reductions in cBRS with aging[128, 36, 77]. The significant correlation between the maximum heart rate response to orthostasis in women, but not men, is surprising, because the cardiac/vascular support of the blood pressure response to carotid baroreceptor unloading with the neck collar is similar between young men and women[92] (Figures 4.6 and 4.7). Furthermore, previous data as well as our study indicate that with age, older women rely predominately on the vascular responses to baroreceptor unloading[40] (Figure 4.7). That these two groups function similarly here, and distinctly from younger males, may hint at different baroreflex strategies for the maintenance of blood pressure during orthostasis between men and women despite similar acute responses to baroreceptor stimulation. Indeed, it has been suggested that MSNA has a lesser influence on beat-to-beat changes in MAP in older women compared to older men[190]. The smaller FVR responses to orthostatic stress that we observed in young women are broadly consistent with historical data, which indicate young women mount smaller TPR responses to tilt compared to young men[55, 62], perhaps due to increased activation of vasodilatory beta-adrenergic receptors[93, 74]. However, contrary to previous reports[56], we did not see increased maximal vascular responses to orthostatic stress in older participants compared to younger males. In this study, 6/20 (30%, 3 men) of our older participants requested to stop the test before reaching our numerical stopping criteria (defined as a reduction in SAP below 80 mmHg, or a decrease in heart rate below 50 bpm, or an increase in heart rate higher than 170 bpm) compared to only 3/29 (10%, 2 men) younger participants. Thus, we likely underestimated the maximal FVR response and orthostatic tolerance in the older participants, and this likely impacted our ability to detect age-related differences in the vascular responses during orthostasis. Despite similar maximum FVR responses in men and older women, we detected significant sex and age effects on the ΔsBRS to HUT, with older men displaying the largest ΔsBRS. This is also surprising as older males display reduced vascular responses to MSNA bursts[190] and reduced vascular transduction compared to younger males[22]. However, young women were the only group that did not display a significant increase in the magnitude of sBRS upon HUT, which is consistent with previously demonstrated reduced vascular transduction in young women[22] (Table 4.3). The maximal FVR response during orthostatic stress was correlated with orthostatic tolerance for the group as a whole; however, this was primarily driven by the young women. Given the small maximal FVR responses in this group, this suggests that vascular responses may be more important for participants with lower maximal FVR responses. A “minimum” level of vasoconstriction may be necessary for normal orthostatic tolerance, with further vasoconstriction providing diminished returns. The maximal vascular resistance response

101 to orthostasis occurred at 85% (SD 12%) of orthostatic tolerance, and was not different between older and younger men and women (all p > 0.8). The timing of the maximal vascular resistance response was also not significantly correlated with orthostatic tolerance (p = 0.18). We examined the relationships between the maximal cardiac and maximal vascular responses during orthostasis to orthostatic tolerance for participants who experienced a type 1 (mixed) or type 3 (vasodepressor) faint. The maximal heart rate responses to orthostasis correlated with orthostatic tolerance in the type 3 group, but did not quite reach statistical significance (p = 0.08) in participants who experienced a type 1 faint. The difference between the two appears to be primarily driven by the lower n of the type 1 group, as most participants in this study experienced a type 3 faint. The relationship between the maximum FVR response and orthostasis did not quite reach statistical significant in the type 1 group (p = 0.06), but may hint that participants who experience a type 1 faint rely more on the FVR response during orthostasis. To our knowledge, this is the first study to simultaneously characterize cardiac and sympathetic vascular baroreflex function during orthostatic stress. Supine cBRS and sBRS are poorer predictors of the maximal RRI and FVR response during orthostasis than the magnitude of the change in BRS with tilting, and extrapolation of the supine measurements to baroreflex-mediated function during orthostasis should be interpreted with caution. The HUT-supine changes in cBRS and sBRS appear to be much more reliable indicators of the cardiovascular responses during orthostasis (Figure 4.4). Future use of these dynamic measures as metrics of baroreflex function should report the BRS response to an orthostatic challenge.

4.5.2 Cardiovascular Responses to Carotid Baroreceptor Stimulation with the Neck Collar

4.5.2.1 Cardiac and Sympathetic Baroreﬂex Sigmoid Curves

This is the first study to construct full cardiac and sympathetic baroreflex curves in participants over a wide age range. We have demonstrated that sympathetic baroreflex curves are left-shifted compared to cardiac baroreflex curves. This supports previous findings that baroreflex-mediated sympathetic vascular responses to hypotension are more important for the control of blood pressure during orthostasis[201], as the cardiac responses to hypotension reach the lower plateau of the baroreflex sigmoid at higher values of CSP. We have also shown the sympathetic threshold and centering point are right-shifted in older participants (Figure 4.5, Table 4.4). While it is well-established that age-related hypertension is accompanied by baroreflex resetting to higher blood pressure[94], our cohort of healthy older participants did not have significantly elevated blood pressure compared to our younger group. The difference was only just non-significant (p = 0.057), and it

102 is therefore possible that our cohort of older participants exhibited larger SAP (+7± 3.8 mmHg). Regardless, the magnitude of the diﬀerence in older participants (Δthreshold = +18±6.6 mmHg, Δcentering point = +14±5.7 mmHg) exceeded that of the SAP in older participants. This observed shift in centering point and threshold is therefore likely a phenomenon of aging, rather than hypertension per se.

4.5.2.2 Absolute Cardiovascular Responses to Neck Suction and Pressure

In our study, females had a greater cardiac response to -60mmHg neck suction than men, while we detected no effect of sex at +40mmHg. These results are consistent with previous data using the neck collar technique[92]. The peak FVR responses to neck suction and pressure were similar between young men and women (Figure 4.6). This is at odds with a previous study which reported attenuated vascular responses to neck suction in young women compared to young men[92]. The same group has also shown reduced vascular responses to neck suction, but not neck pressure, in younger women compared to older women[40]. Conversely, we detected significantly attenuated FVR responses to neck pressure and suction in older men and women. This is likely due to reduced α-adrenergic responsiveness in older individuals [190, 174]. The reasons for these discrepancies are not clear. We quantified vascular resistance responses to neck suction/pressure as the nadir/peak FVR response, while vascular responses have previously been quantified as the total vascular conductance during the peak/nadir MAP response[40]. Thus, the relative timing of the TPR response to the maximum MAP response, which has been shown previously to occur on different beats[68], may influence the outcome. We also quantified vascular responses using FVR, while previous studies have used TPR[92] and leg vascular conductance[40]. The vascular responses to orthostasis have previously been shown to be larger in the leg vs. the forearm in younger men vs. older men[65]. Lastly, we measured responses during a continuous 20 second bout of neck suction/pressure, whereas previous studies have used a 5-second pulse. The latency of the maximal vascular resistance response is approximately 10 seconds in healthy controls[68]. Furthermore, upon withdrawal of neck pressure to ambient pressure, MSNA rapidly drops and undershoots pre-stimulus levels, which may also reduce the magnitude of the slow-onset vascular response[80]. Our longer stimulus length, therefore, likely captures slow maximal vascular responses, and does not cause MSNA undershoot prior to the maximal vascular response.

4.5.2.3 Relative Cardiovascular Responses to Neck Suction and Pressure

Given the large diﬀerences in the magnitude of cardiovascular responses to carotid stimulation between younger and older men and women, to examine the relative contribution of cardiac vs. vascular responses we calculated the percent contribution of the CO and TPR

103 responses to the overall MAP response. All groups displayed larger relative TPR responses to simulated carotid hypotension with neck collar pressure compared to carotid hypertension. The vasodilatory response in a supine individual with little baseline vasosonstrictor tone is likely limited as the vessel is already near maximum dilation, with little vasodilatory reserve. Conversely, the increase in vagal tone during neck suction likely has less of a floor effect, as vagal tone can increase until significant asystoles occur[2]. Accordingly, the acute sympathetic vascular responses appear to be more important for the maintenance of blood pressure during hypotensive stimuli. However, the magnitude of this sympathetic vascular predominance was larger in older participants, indicating an age-related shift in the baroreflex strategy. This relationship has been shown previously in older women[40] and is consistent with increased vascular transduction that occurs with menopause[22, 74]. However, older men also tended toward larger relative vascular responses with aging despite reduced vascular conductance[22] and similar MSNA responses to neck pressure[45]. This may be compensatory for reduced cardiac vagal tone and blunted cBRS with aging[128, 36, 77].

4.5.3 Comparison of the Vascular Responses to Orthostasis and Neck Collar Stimulation

In this study, young females exhibited an attenuated FVR response during orthostasis (Figure 4.3A) compared to all other groups. In contrast, the FVR response to simulated hypotension with the neck collar was similar between young men and women, and the younger group exhibited larger responses than the older participants (Figure 4.6B). A direct comparison between the ΔFVR during tilt and the ΔFVR during +40mmHg neck pressure demonstrated that all groups except for young women displayed a larger ΔFVR response to orthostasis. The reasons for these discrepancies are not clear. There is a grow- ing body of evidence indicating reduced vascular transduction in young women[93, 74, 22]. While the maximum FVR response during orthostasis appears to be consistent with this, one would also expect attenuated FVR responses to neck pressure. Our observation of large FVR responses to carotid stimulation in younger women is therefore perplexing. A previous study examining the TPR responses to single bursts of MSNA found no difference in the TPR response to a MSNA burst between young men and women[190]. Consistent with our data, this study also found reduced TPR responses in older men and women[190]. One key methodological similarity between this study and ours is the short timeline over which the vascular responses were assessed (<25 seconds). Conversely, the experimental methodology examining vascular transduction typically involves longer timescales (30 seconds - 5 minutes) due to injection of vasoactive drugs[53, 93, 74]. Collectively, this suggests that acute baroreflex-mediated vasoconstriction is not impaired in younger women, but may become reduced with sustained baroreceptor unloading. Another possible mechanism underlying the different cardiovascular responses is the baroreceptor regions activated by the two tests. Neck pressure stimulates only the carotid

104 baroreceptors, while orthostasis likely stimulates both the carotid and splanchnic baroreceptors [182] when blood pressure is well-regulated, as well as the cardiac and aortic baroreceptors[21] as blood pressure falls. Activation of splanchnic baroreceptors has been shown to augment the pressor response to carotid baroreceptor stimulation[43]. Further- more, young women have been shown to experience increased pelvic pooling during orthostasis compared to men[199]. However, greater splanchnic pooling should cause increased vasoconstriction in women, contrary to the results seen here.

4.5.4 Baroreﬂex Strategies for the Maintenance of Blood Pressure During Orthostasis

Despite identifying significant age- and sex-related differences in the cardiovascular responses to orthostatic stress, significant intra-group variability remains in both the maximal cardiovascular responses to orthostasis, and orthostatic tolerance. While we believe our data indicate a predominance of vascular support of blood pressure during orthostasis, the possibility exists that subgroups may employ different strategies for the maintenance of blood pressure. We investigated this using model selection and model averaging. Due to the known differences in cBRS and vascular transduction between men and women and with aging[22], we first performed a subgroup analysis on younger and older men and women (Figure 4.8). None of the candidate variables included in the selection algorithm reached high variable importance, so we performed model selection for the group split by men and women (Figure 4.9). In women, the maximum heart rate response was identified as an important predictor of orthostatic tolerance, while in men the HUT-supine ΔsBRS was an important predictor. This is in agreement with our current data indicating that the maximal heart rate response during orthostasis correlates with orthostatic tolerance in women only, and larger ΔsBRS during HUT in men. However, supine sBRS was also identified as highly important variable in women, and a moderately important variable in men. This is surprising, given its poor relationship with the maximum FVR response during orthostasis, but indicates that participants with higher baroreflex-mediated sympathetic vascular activity at rest exhibited lower orthostatic tolerance, perhaps related to reduced baroreflex reserve during an orthostatic challenge[57]. Subgroup analysis for age group revealed that important predictors of orthostatic tolerance in younger participants are predominately cardiac, although supine sBRS was again moderately important. However, in the older group no variables were highly important, and the only moderately important variable was the HUT-supine ΔcBRS. This suggests that grouping older participants is not a useful method of characterizing baroreflex strategies during orthostasis. Alternatively, we also performed a subgroup analysis based on the maximum FVR response during orthostasis. Distinct baroreflex strategies were the most clear with these subgroups, where “constrictors” rely primarily on sympathetic vascular variables for orthostatic tolerance, while “non-constrictors” rely primarily on cardiac variables. The

105 variables that were identified as important for these subgroups also reflected the variables that were identified for the group as a whole, suggesting that this method of subgrouping is a useful method of characterizing baroreflex strategies for the maintenance of blood pressure during orthostasis.

4.5.5 Limitations

The neck collar protocol was performed on a separate day to the orthostatic stress test. This was done deliberately to avoid test interactions and to reduce participant burden. However, added variance due to day-to-day changes in baroreflex function[113] should be considered when interpreting our results. We did not measure cardiovascular fitness in our participants, which has known effects on orthostatic tolerance, with very highly trained athletes paradoxically displaying lower orthostatic tolerance than moderately trained athletes.[72, 109]. However, resting heart rate was not different between young and old men and women, indicating similar levels of cardiovascular fitness in these healthy participants. Prior research into the physiology underlying differences in sympathetic neural control of vascular resistance has primarily focused on the role of norepinephrine, and the relative sensitivities of alpha and beta receptors in the vasculature[74], and we have therefore focused our discussion on this mechanism. It should be noted that sympathetic nerves also release neuropeptide Y and adenosine triphosphate in addition to norepinephrine[21], and these may contribute to the observed sex differences in baroreflex control of vascular resistance. We measured baroreflex-mediated changes in vascular function at the forearm. While changes in vascular resistance in the forearm are correlated with those from other vascular beds (ie. legs, splanchnic vessels), we did not directly measure changes in resistance in these other vascular beds, which may have influenced our results. Our model averaging subgroup analysis for younger and older men and women (Fig- ure 4.8) did not reveal any variables that were highly important in predicting orthostatic tolerance. The low number of participants in each of the groups may have contributed to our inability to identify clear patterns of baroreflex control of blood pressure. However, our subgroups analyses stratifying by age group, sex, and level of vasoconstriction were less affected by limited sample size.

4.5.6 Conclusion

Simultaneous measurement of cardiac and sympathetic vascular baroreflex function provides valuable insight into the baroreflex control of blood pressure. Sympathetic baroreflex sigmoid curves are left-shifted compared to cardiac baroreflex sigmoid curves, suggesting that sympathetic vascular responses may be more important for the baroreflex regulation of blood pressure during hypotensive stimuli such as orthostatic stress. Comparison of barore-

106 flex responses to orthostasis and neck suction/pressure reveal important nuances in the baroreflex-mediated cardiac and vascular responses. In young women, vascular responses to orthostasis are blunted despite the robust responses to discrete stimulation of the carotid baroreceptors. This suggests that other factors associated with orthostatic stress impair vascular baroreflex responses in young women. While the vascular responses during orthostasis are indeed predictive of orthostatic tolerance, this appears to be subgroup dependent, with non-constricting individuals relying more on the cardiac responses to orthostatic stress. This has significant implications for the management of patients with recurrent syncope, as patients with small vascular responses to orthostasis may benefit from large cardiac responses. For instance, patients with POTS are often treated with beta-blockers, which may be counter-productive for those patients who are unable to mount a sufficient vasoconstriction response to orthostasis[1]. Taken together, our evidence suggests that different baroreflex strategies exist for the maintenance of blood pressure during orthostasis.

107 Chapter 5

The Inﬂuence of Methodology on Cardiovascular Responses to Carotid Sinus Massage

5.1 Abstract

The arterial baroreflex is crucial for short-term blood pressure control – impaired baroreflex function predisposes affected individuals to orthostatic hypotension, fainting and falling. Baroreceptors in the carotid arteries are central to the maintenance of blood pressure during changes in body position. In elderly patients with syncope, 26-60% display a hypersensitive response during CSM that is considered pathological. However, up to 35% of healthy older adults also display a hypersensitive response, which questions the use of this test clinically. We hypothesize that CSM exhibits poor repeatability that is impacted by variable methodology. We measured cardiac and vascular responses to CSM in 51 healthy controls and 17 patients at both the clinically recommended level of the cricoid cartilage and the ultrasound- guided location of the carotid sinus, on both sides of the neck. CSM was performed both in supine and upright postures. Sham massages were also performed. The locations of the cricoid cartilage, maximal carotid arterial pulsation, and the carotid sinus were measured. The current recommendation for application of CSM at the cricoid cartilage substantially underestimates the location of the carotid sinus by -4±0.2 cm (p < 0.0001), and reduces the magnitude of the cardiovascular responses in patients in the supine position (maximum RRI: -236±53 ms, p = 0.0002, ΔRRI: -231±56 ms, p = 0.0005, and ΔSAP: -5±1.5 ms, p = 0.03). The location of maximal pulsation of the carotid artery provided the best anatomical landmark for the location of the carotid baroreceptors (mean difference = -0.9 cm, 95% limits of agreement (LOA) = 2 cm). A large percentage of healthy older adults do not display responses to CSM that are significantly different from sham massage.

108 Performing CSM at the location of maximal pulsatility approximates the carotid sinus well, and may be used by physicians who are unable to perform ultrasound-guided CSM. A large percentage of healthy controls do not exhibit cardiovascular responses to CSM that are signiﬁcantly diﬀerent from sham massage.

5.2 Introduction

CSH is a phenomenon in which external stimulation of the carotid sinus produces profound sinus asystole (greater than 3 seconds) and/or reduced SAP (greater than 50 mmHg)[96]. It affects up to 35% of healthy older adults and 26-60% of elderly patients with syncope[130], and is reported to contribute to the high incidence of syncope and falls in this population[155, 83]. The diagnostic test for CSH, CSM, involves external digital stimulation of the carotid sinus to distort the vessel walls and stimulate the carotid sinus baroreceptors[130]. The anatomical location and duration of the massage varies in the literature[25, 161], which may affect the cardiovascular responses to the test. Standard anatomical landmarks such as the cricoid cartilage[25] are often used to guide CSM, but may be inappropriate as the location of the carotid sinus varies substantially between individuals[148]. The relationship between some anatomical landmarks and the carotid sinus has been assessed previously, and the area of maximal carotid pulsatility was found to best approximate the location of the carotid sinus[49]. Performing CSM at the correct anatomical location of the carotid sinus, determined using ultrasound, enhances the cardiac responses to CSM[69]; however, the effect of massage location on the vasodepressor response, as well as the interaction between massage technique and CSM performed during HUT has not been assessed. This is essential, as current guidelines suggest performing CSM during HUT in addition to supine massage to better elicit the pressor response[25, 161]. Patients with a history of unexplained syncope who have a positive response to CSM are considered for an implanted pacemaker[130]. Thus, the downstream effect of a false positive response is non-trivial for the patient, and reliability of this diagnostic test is essential. We hypothesized that ultrasound-guided CSM performed over the carotid sinus will increase the cardiovascular responses to CSM, and improve the diagnostic utility of this technique.

5.3 Methods

5.3.1 Healthy Controls

We examined the cardiovascular responses to carotid sinus massage performed over the ultrasound-guided location of the carotid sinus, as well as the cricoid cartilage, on the right and left sides of the neck. Massages were completed while supine and during HUT. The methodology for CSM in healthy controls is outlined in Section 2.3 (page 26). The experimental protocol for patients and controls is described in Figure 5.1.

109 Figure 5.1: The carotid sinus massage experimental protocol. Following a 10-minute supine rest, participants underwent supine and tilted carotid sinus massage in duplicate, on both sides of the neck at the ultrasound-guided level of the carotid bifurcation and at the level of the cricoid cartilage. Two additional sham procedures were conducted in each position in which the ﬁngers were placed overlying the carotid sinus on each side of the neck, but no pressure was applied. US = ultrasound-guided location of the carotid bifurcation, Cr = cricoid cartilage, R = right, L = left. * Patients did not undergo sham massage. ** Patients underwent only one repetition of each massage during head-up tilt, in order to reduce the length of the tilted period.

5.3.2 Patients

Patients were recruited from Hôpital du Sacrè-Coeur in Montreal, Canada. Participants with structurally normal hearts and normal ECG who had been referred for tilt table testing and clinically indicated carotid sinus massage for the diagnosis of syncope were recruited. Participants with audible carotid bruits or significant (>50%) carotid stenosis were excluded from the study. Patients underwent a similar protocol as the healthy controls with three modifications: 1) as the patients had a history of spontaneous syncope, only four CSMs (once each at the ultrasound-guided bifurcation and cricoid cartilage on the left and right sides) were performed during HUT; 2) to reduce the length of the overall protocol, sham massages were not performed; and 3) BBFV was not assessed with Doppler ultrasound, so FVR was not able to be calculated. On the test day, participants fasted, avoided strenuous exercise and caffeine for 12 hours prior to the tests, and refrained from medication affecting blood pressure or heart rate for 24 hours prior to the test. Tests were always conducted in the morning.

5.3.3 Data Analysis

The method for quantification of the responses to CSM are outlined in Chapter 2. For paired analyses, mixed linear models were used with participant as a random block; for unpaired analyses, simple linear models were constructed. Residual vs. fitted and q-q plots were examined to ensure normality of the residuals. Significance was assumed when p < 0.05.

110 For calculation of the range of values observed during the sham massage, the “non- responder” range was calculated using the mean and SD of the sham responses:

Non − responder Range = ResponseMean ± 1.96 × SD

Results are reported as mean ± standard error unless otherwise stated.

5.4 Results

Data were collected in 51 healthy controls and 17 patients. Participant demographics are in Table 5.1. Patients had lower SAP (-22± 4 mmHg, p < 0.0001) and higher heart rate (+7± 2 bpm, p = 0.0005) than controls. The anatomical locations of the carotid bifurcation, location of maximal carotid pulsatility, and cricoid cartilage relative to the gonion of the mandible are shown in Figure 5.2. The location of the cricoid cartilage was signiﬁcantly lower than the carotid bifurcation and maximal pulsatility on both the left (-4± 0.2cm, p<0.0001 for both maximal pulsatility and carotid bifurcation) and right (-4± 0.2cm, p < 0.0001) sides. Responses to ultrasound-guided CSM are listed in Table 5.2. The maximum RRI during CSM was smaller during HUT for both patients and controls. The RRI response during CSM was larger in patients in the supine position compared to patients during HUT, and controls. The SAP response during CSM was smaller in patients during HUT than patients in the supine position, and controls in both supine and HUT positions. The DAP response to CSM was smaller in patients during HUT compared to controls in both supine and HUT positions. No participant had responses to CSM that were diagnostic of CSH (maximum RRI > 3000ms or reduction in SAP of >50 mmHg). Cardiovascular responses to CSM at the ultrasound-guided bifurcation and the cricoid cartilage are shown in Figure 5.3. Ultrasound-guided CSM elicited larger maximum RRI (+236±53 ms, p = 0.0002), ΔRRI (+231±56 ms, p = 0.0005), and ΔSAP (+5±1.5 ms, p = 0.03) responses in patients in the supine position, compared to responses to CSM at the cricoid cartilage. Ultrasound-guided CSM produced larger ΔDAP (+2 SE 0.7, p = 0.0027) in both patients and controls in the supine position. Controls exhibited larger ΔRRI (+97±31, p = 0.003), ΔSAP (-7±2, p = 0.003), and ΔDAP (-5±1.3, p = 0.0001) responses to CSM than patients during HUT. The cardiovascular responses to CSM at the cricoid cartilage were compared to the distance of the cricoid cartilage from the carotid bifurcation in the controls (Figure 5.4). After correcting for outliers, none of the cardiovascular responses were correlated with the distance from the carotid bifurcation. Responses to ultrasound-guided CSM were compared with age in healthy controls and patients. The SAP responses to CSM increased with age in both groups, however the eﬀect size was small (-0.1±0.04 mmHg/year, p = 0.01). The ΔRRI, and ΔDAP responses to

111 CSM were not significantly correlated with age (p = 0.76 and p = 0.62, respectively). The maximum RRI during CSM did not quite significantly correlate with age, and the effect size was also small (2±1 ms/year, p = 0.058). Responses to ultrasound-guided CSM were compared to sham massage in healthy controls (Figure 5.5). None of the measured cardiovascular responses (maximum RRI, ΔRRI, ΔSAP, ΔFVR) were significantly different from sham massage. Using the variance of the responses to sham massage, normal "non-responder" ranges were calculated for all variables and superimposed over the responses to ultrasound-guided responses.

Height Weight SAP DAP HR Group n Age (cm) (kg) (mmHg) (mmHg) (bpm) Controls 51 (23M) 37 (19-84) 173±8 71±11 127±14 69±9 59±6 Patients 17 (8M) 56 (44-82) 170±11 75±17 104±15† 70±13 66±9†

Table 5.1: Demographic information for patients and controls. Data are presented as mean ± standard deviation. BP = blood pressure, HR = heart rate. † = signiﬁcantly diﬀerent from controls (p < 0.0005).

5.5 Discussion

In this study, we demonstrate that the region of maximal pulsatility best approximates the location of the carotid sinus. Carotid sinus massage performed at the ultrasound- guided carotid bifurcation increases both the cardiac and vasodepressor responses to CSM in patients, but not controls. Blood pressure responses to CSM were larger during HUT in healthy controls, but not in patients. In healthy controls, responses to CSM were small and indistinguishable from sham massage. This likely underlies the poor relationship between responses to CSM and the distance from the carotid bifurcation that CSM is applied.

5.5.1 Anatomical Findings

We have confirmed previous reports from Finucane et al (2016) that the location of the maximal carotid pulsatility approximates the location of the carotid sinus[49]. The bias and 95% limits of agreement between maximal pulsatility and carotid bifurcation in this previous study (mean difference ± 95% LOA = [Left] -0.6±3cm, [Right] -0.8±2.5cm) were broadly comparable to those found here (mean difference ± 95% LOA = [Left] -1.3±2.1cm, [Right] 0.4±2cm). However, they calculated agreement between the cricoid cartilage and the ultrasound-guided bifurcation as (mean difference ± 95% LOA) -1.3±3cm (Left), and -1.7±3cm (Right). In our study, the agreement between the cricoid cartilage and the ultrasound-guided bifurcation was (mean difference ± 95% LOA) -4.2±2.7cm for both the right and left sides. The reasons for these differences in measurement between the two

112 Variable Position Group Response Max RRI (ms) Supine Controls 1177±21.6 Patients 1217±96.2 HUT Controls 945±17.1* Patients 867±11.6* ∆ RRI (ms) Supine Controls 154±20.7 Patients 297±95.1acd HUT Controls 161±16.2 Patients 47±11.1 ∆ SAP (mmHg) Supine Controls -11±0.9 Patients -10±2.6 HUT Controls -13±1.4 Patients -4±1.3abc ∆ DAP (mmHg) Supine Controls -7±0.5 Patients -6±1.9 HUT Controls -8±1 Patients -2±0.8ac ∆ FVR (%) Supine Controls -22±1.7 HUT Controls -28±2.8

Table 5.2: Responses to ultrasound-guided carotid sinus massage in patients and controls. Data are presented as mean ± standard error. RRI = R-R interval, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, FVR = forearm vascular resistance, HUT = head-up tilt. * = significant main effect of position, a = significantly different from supine controls, b = significantly different from supine patients, c = significantly different from HUT controls, d = significantly different from HUT patients.

113 Figure 5.2: Anatomical location of the carotid bifurcation relative to cricoid cartilage and location of maximum pulsatility for the left (A) and right (B) sides of the neck in healthy controls. On both sides, the cricoid cartilage was significantly lower than the ultrasound- guided bifurcation and the location of maximal pulsatility. In C and D, Bland-Altman analysis shows good agreement between the ultrasound and maximal pulsatility locations with no systematic or proportional bias, and 95% limits of agreement of ±2.1 cm (95% confidence interval ±0.6 cm) for the left and ±2 cm (95% confidence interval ± 0.5 cm) right side. Solid lines indicate the mean difference, dashed lines indicate the 95% limits of agreement, and dotted lines indicate the 95% confidence interval for the limits of agreement. US = ultrasound-guided location of the carotid bifurcation, MP = location of maximum carotid artery pulsatility, Cr = location of the cricoid cartilage. * = p < 0.0001. studies is not clear, but has significant implications for studies examining the cardiovascular responses to CSM at the level of the cricoid cartilage. The largest difference between the carotid bifurcation and cricoid cartilage found in our study was 7.5 cm. Applying CSM

114 Figure 5.3: Cardiovascular responses to carotid sinus massage at the cricoid cartilage and carotid bifurcation, identified using ultrasound. RRI = R-R interval, SAP = systolic arterial pressure, DAP = diastolic arterial pressure, Cr = cricoid, US = ultrasound. ‡ = significant differences during Tukey post-hoc testing, † = significant main effect of group, * = significant main effect of location. so far below the carotid sinus is clearly unlikely to elicit a significant baroreflex-mediated response.

115 Figure 5.4: Cardiovascular responses to carotid sinus massage performed at the level of the cricoid cartilage were compared to the distance of the cricoid cartilage from the carotid bifurcation. There were no significant relationships between distance from the carotid bifurcation and A) the maximum RRI during massage, B) the maximal change in RRI during massage, or C) the maximal change in SAP during massage. D) the maximal change in DAP during massage was correlated with the distance from the carotid bifurcation during HUT only; however, when one outlier response was removed (indicated by the red circle), the relationship was no longer significant. Additional outliers were tested but were not removed from the analysis as they had no effect on the significance of the correlation. HUT = head-up tilt, RRI = R-R interval, SAP = systolic arterial pressure, DAP = diastolic arterial pressure.

116 Figure 5.5: Cardiovascular responses to sham and treatment CSM in healthy controls in the supine position. There were no signiﬁcant diﬀerences between sham and treatment massages for A) the maximum RRI elicited during massage, B) the change in RRI during massage, C) the maximal SAP response, or D) the maximum FVR response. HUT = head-up tilt, RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance, CSM = carotid sinus massage.

The area of maximal carotid pulsatility had good agreement with the ultrasound-guided carotid bifurcation, and suggests that physicians unable to use ultrasound to locate the carotid sinus can use the maximal pulsatility as a reasonable substitute, particularly when typical CSM methodology involves a dynamic stimulus spanning a 2-4cm range[49].

117 5.5.2 Responses to Carotid Sinus Massage

We have confirmed that the cardiovascular responses to ultrasound-guided CSM are greater than those elicited with CSM at the cricoid cartilage (Figure 5.3). However, this was only true in the patient group, and only true in the supine position. With respect to responses from the healthy control group, we hypothesized two underlying mechanisms. First, as the cricoid cartilage ranged between 0.5 cm above, and 7.5 cm below the carotid bifurcation in our study, large responses in participants with co-local cricoid cartilage and carotid bifurcations may have clouded our ability to detect differences in responses to CSM. To examine this, we correlated the distance of the cricoid cartilage from the carotid bifurcation against the cardiovascular responses to CSM, and found no relationship after correcting for outliers. We then assessed whether our population of healthy controls exhibited responses to CSM that were significantly different from sham massage (Figure 5.5). No statistical differences were detected between sham and actual CSM, indicating that the majority of our population of healthy controls, even the older adults, do not respond to CSM. Our inability to detect location-based differences in responses, then, is unsurprising. The vasodepressor response to CSM did not increase during HUT in either our patients or healthy controls. Contrary to other reports in the literature[95, 142], after HUT, ΔSAP was unchanged in the controls, and was smaller in the patients (Table 5.2). In fact, the ΔSAP and ΔDAP responses to CSM during HUT were smaller in the patients than controls. This is perplexing, particularly as our cohort of controls did not exhibit responses to CSM that were significantly different from sham. It may be that the increase in sympathetic nervous system activity during HUT masked the modest increase in vagal activity elicited during CSM in this cohort of participants with small CSM responses.

5.6 Limitations

Despite recruiting 17 healthy older adults and 17 patients, none of the participants in our study had a clinically significant response to CSM, classified as asystole greater than 3 seconds and/or a reduction in SAP greater than 50 mmHg. This is contrary to previously reported prevalence of CSH as between 4% and 35% in healthy older adults aged 40-80[96], which was the approximate age range of our older controls and patients. We have no explanation for the small responses in our older cohort, and this unfortunately makes extrapolation of our results difficult. While we saw clear differences in the cardiovascular responses to ultrasound-guided CSM in the patient group, this was only true in the supine position, and the magnitude of the differences were small (ΔRRI = +236± 53 ms, p = 0.0002; ΔSAP = -5± 2 mmHg, p = 0.03; ΔDAP = -2± 0.7mmHg, p = 0.003) and likely not clinically meaningful. While we believe the magnitude of the differences would be clinically meaningful in a population of patients with CSH, our data do not directly support

118 this. Future studies should expand the current methodology to patients with documented CSH.

5.7 Conclusion

We have shown that the current recommendations of CSM being performed at the level of the cricoid cartilage underestimates the location of the carotid sinus, and likely impacts the magnitude of the responses to CSM. Performing CSM at the location of maximal pulsatility approximates the carotid sinus well, and may be used by physicians who are unable to perform ultrasound-guided CSM. Contrary to current recommendations, the blood pressure responses to CSM did not increase when performed during HUT. If these results extend to patients with CSH, it suggests that repeating CSM during HUT may not provide added diagnostic benefit. A large percentage of healthy controls do not exhibit cardiovascular responses to CSM that are significantly different from sham massage.

119 Chapter 6

Carotid Sinus Hypersensitivity: Block of the Sternocleidomastoid Muscle does not Aﬀect Responses to Carotid Sinus Massage in Healthy Young Adults

6.1 Abstract

The arterial baroreflex is crucial for short-term blood pressure control – abnormal baroreflex function predisposes to syncope and falling. Hypersensitive responses to carotid baroreflex stimulation using CSM are common in older adults and may be associated with syncope. The pathophysiology of this hypersensitivity is unknown, but chronic denervation of the sternocleidomastoid muscles is common in elderly patients with CSH, and is proposed to interfere with normal integration of afferent firing from the carotid baroreceptors with proprioceptive feedback from the sternocleidomastoids, producing large responses to CSM. We hypothesized that simulation of sternocleidomastoid “denervation” using pharmacological blockade would increase cardiovascular responses to CSM. Thirteen participants received supine and tilted CSM prior to intramuscular injections (6-8ml distributed over 4 sites) of 2% lidocaine hydrochloride, and 0.9% saline (placebo) in contralateral sternocleidomastoid muscles. Muscle activation was recorded with electromyography (EMG) during maximal unilateral sternocleidomastoid contraction both pre- and post-injection. Supine and tilted CSM were repeated following injections and responses compared to pre-injection.

120 Following lidocaine injection, the muscle activation fell to 23±0.04% of the pre-injection value (p<0.001), conﬁrming neural block of the sternocleidomastoid muscles. Cardiac (RRI), FVR, and SAP responses to CSM did not increase after lidocaine injection in either supine or tilted positions (supine: ∆RRI -72±31ms, ∆SAP +2±1mmHg, ∆FVR +4±4%; tilted: ∆RRI -20±13ms, ∆SAP +2±mmHg, ∆FVR +2±4%; all p>0.05). Neural block of the sternocleidomastoid muscles does not increase cardiovascular responses to CSM. The pathophysiology of CSH remains unknown.

6.2 Introduction

Stretch receptors in the carotid arteries are central to the baroreflex response to changes in blood pressure during changes in body position. Up to 35% of healthy older adults and 26-60% of elderly patients with syncope display abnormally large responses when the carotid baroreceptors are stimulated using CSM, a simple clinical test whereby manual massage of the skin overlying the carotid sinus is proposed to stimulate the underlying arterial baroreceptors[130]. The high prevalence of CSH in the elderly, and its reported association with syncope[130], makes it an obvious candidate for consideration with respect to the increased risk of syncope and falling in older adults. Despite its high prevalence in the elderly, the pathophysiology of CSH remains unknown, and several possible mechanisms have been proposed[7]. One leading idea postulates that, in healthy individuals, stimulation of the sternocleidomastoid muscles near the carotid baroreceptors (during neck turning or carotid sinus massage), as well as the carotid baroreceptors themselves, leads to central integration of the two signals as “external” stretching of the sinus, and thus does not elicit a baroreflex response[185]. Loss of this central integration due to sternocleidomastoid denervation might lead to hypersensitive responses during daily activities, such as neck turning, as well as during CSM[17]. This led to the first description of sternocleidomastoid electromyographic abnormalities in elderly patients with CSH[17]. It was suggested that chronic denervation of the sternocleidomastoid muscles with aging interrupts normal integration of neck muscle proprioception and carotid baroreceptor information, leading to excessive responses to CSM and an increased risk of syncope. We aimed to test this theory, hypothesizing that pharmacological block of the sternocleidomastoid muscles in healthy individuals will interrupt central integration of proprioceptive and carotid baroreceptor information, and lead to hypersensitive responses to CSM.

6.3 Methods

6.3.1 Ethical Approval

This study was approved by the Oﬃce Of Research Ethics at Simon Fraser University and conforms to the principles outlined in the Declaration of Helsinki[203]. Prior to testing,

121 participants provided written informed consent, and completed a brief medical history to conﬁrm they met our inclusion criteria. Participants were excluded from the study if any of the following criteria were met: pregnancy; history of neck surgery, neck injury, or currently symptomatic neck pain; known sensitivities or allergies to medications containing lidocaine; pre-existing cardiovascular or neurological disease; use of medications containing lidocaine or with cardiovascular actions.

6.3.2 Protocol

The carotid arteries were screened for the presence of stenosis using ultrasound, and the ultrasound-guided location of the carotid bifurcation was determined and marked on the skin. After a 10-minute supine rest period, participants then underwent ﬁrst supine and then tilted CSM in duplicate, on both sides of the neck. Muscle activation (EMG) of the sternocleidomastoid muscles during standardised maximum voluntary contraction was recorded. Participants then received intramuscular injections using a 25G 1” needle (6-8ml distributed over 4 sites) of 2% lidocaine hydrochloride, and 0.9% saline (placebo) in contralateral sternocleidomastoid muscles (determined randomly). The total dose of lidocaine therefore did not exceed 400 mg, well below the maximum dose of 600 mg over 12 hours. After a 10-minute rest period (to allow time for optimal block from the lidocaine injections and the recovery of any cardiovascular responses to any perceived discomfort acutely associated with the injections), EMG recordings were repeated following injection to quantify the extent of block of muscle activity with lidocaine. Supine and tilted CSM were then repeated to determine the impact on cardiovascular responses obtained. Data were collected and analysed in a double-blind fashion.

6.3.3 Screening for Carotid Stenosis

Participants’ carotid arteries were screened for the presence of significant stenosis, and participants were excluded from further testing if arterial narrowing greater than 50% was detected by the study physician. Examinations were conducted using the GE Logiq I system (GE Healthcare, Chicago, Illinois, USA), with a 6.3 MHz linear transducer. The carotid arteries on both sides were evaluated in both transverse and longitudinal planes for the presence of visible significant narrowing on B-mode greyscale image and the presence of mosaic pattern on colour Doppler image. Examinations were performed with the standard presets for carotid ultrasound initially, with technician optimisation where necessary. The location of the carotid bifurcation was identified and marked on the skin.

6.3.4 Cardiovascular Monitoring

Beat-to-beat cardiac (RRI: ECG; Lead II) and blood pressure responses to CSM were recorded using a Finometer Pro (Finapres Medical Systems BV, Amsterdam, Netherlands).

122 Finger cuff measurements were calibrated to brachial blood pressure using the return-to-flow calibration[66] prior to commencing testing, and throughout using the internal calibration (PhysiocalTM). Immediately prior to CSM, the PhysiocalTM was turned off to prevent interruption of the waveform. The participants’ hand was kept at approximately heart height throughout testing, and a height correction unit was used to account for small changes in vertical height between the finger cuff and the heart[66]. Brachial artery blood flow velocity was measured using an 8 MHz probe positioned overlying the brachial artery and clamped at a constant angle throughout testing (Doppler-BoxTM, Compumedics, DWL, Singen Germany). FVR was calculated as MAP/BBFV.

6.3.5 Carotid Sinus Massage

CSM was performed for 10 seconds at the ultrasound-guided location of the carotid bifurcation. Participants were instructed to hold their breath at the end of normal expiration for 15 seconds; CSM occurred during the last 10 seconds of the breath hold in order to minimize the potential confounding inﬂuence of respiration on the responses obtained. Two CSM were performed and responses averaged on each side of the neck with the participant supine, and again during 70-degree head-up tilt. The pressure and technique of the massage stimulus was standardised by having the same investigator perform all massages. A representative example tracing showing responses to CSM can be seen in Figure 2.1.

6.3.6 Muscle Activation

Muscle activation was measured with EMG. Surface EMG signals (bipolar Na/NaCl elec- trodes) were recorded from the sternocleidomastoid muscles halfway between the mastoid process and sternum. Signals were recorded at a frequency of 2000Hz and amplified with a bandwidth of 10-500Hz (Biovision, Wehrheim, Germany). Prior to electrode placement the area was shaved, abraded and the skin cleaned with alcohol. Maximal unilateral sternocleidomastoid contraction was performed with the participant supine, and the head rotated. Participants were instructed to lift their head off the table and maximally push against a standardised opposing force for seven seconds, with EMG recorded during the last five seconds of contraction. Two repetitions were performed on each side.

6.3.7 Data Analyses

Data recorded from the side of the neck that received lidocaine injection were considered “lidocaine” and from the side that received saline injection were considered “placebo.” Method- ology for the calculation of the cardiovascular responses to CSM are presented in Chapter 2. Responses of the two repetitions for each side were averaged within each condition (lidocaine and placebo) and position (supine and tilt). Responses to CSM were classiﬁed as

123 hypersensitive if there was an asystole lasting >3 sec, and/or a fall in systolic blood pressure >50 mmHg[130]. Sternocleidomastoid muscle activity was quantiﬁed as the root mean squared (RMS) of the EMG signals from the last 5 seconds of each contraction. The postinjection RMS EMG was normalized to the maximum preinjection levels. RMS EMG for the two post-injection contractions on one side were averaged and expressed relative to the baseline.

6.3.8 Statistics

Data were analyzed using R (3.2.3), Rstudio (0.99.902) and SigmaPlot Version 12. Paired T-tests were used to compare changes in EMG activity with placebo and lidocaine injections. Repeated-measures ANOVA with factors for treatment (lidocaine vs. placebo), condition (supine vs. upright) and time (pre-injection vs. post-injection) were used to compare cardiovascular responses. Residual versus fitted plots and q-q plots were generated to confirm normality of the residuals, and standard deviations of each group were compared to confirm the assumption of equal variance. Holm-Sidak post hoc tests were used to analyze interaction terms, with p < 0.05 set as the cut-off for significance. Results are reported as mean ± standard error unless otherwise stated.

6.4 Results

Thirteen participants (aged 28±1 years; height 173±2 cm; weight 71±3 kg; 8 male) were recruited for the study. One participant experienced a mild vasovagal reaction to the injections, and completed only two CSM during head-tilt following the injections, one on each side of the neck.

6.4.1 Muscle Activity

Following injection of lidocaine the muscle (RMS) activity fell to 23±0.04% of the maximum pre-injection level, compared to 67±0.05% for placebo (p<0.0001, Figure 6.1).

6.4.2 Carotid Sinus Massage

Baseline responses to CSM prior to injection were small and not significantly different between placebo or lidocaine conditions in either the supine or tilted position (Table 6.1). Resting supine RRI, SAP and FVR prior to CSM were not significantly different between pre- and post-injection conditions. The maximum RRI during CSM while supine was greater than while tilted in both placebo and lidocaine conditions, pre- and post-injection (p<0.001). However, the change in RRI (maximum prolongation during CSM compared to the mean RRI during the five seconds before CSM) was not significantly different between supine and tilt conditions

124 either pre- or post-injection, for either placebo or lidocaine. The changes in SAP and FVR were also not significantly different between supine and tilt conditions either pre- or post-injection, for either placebo or lidocaine. In both placebo and lidocaine conditions, there was a significant main effect of injection independent of receipt of placebo or lidocaine injections, whereby the change in RRI during the supine phase was smaller post-injection than pre-injection (p<0.05). However, this was not statistically significant with post-hoc comparisons within each injection condition (p=0.07 for both placebo and lidocaine conditions respectively). There was no effect of injection on responses during tilted conditions. Our primary interest was in whether the responses to CSM were increased following injection of lidocaine. Therefore, we expressed the cardiovascular responses to CSM post- injection relative to pre-injection for both the lidocaine and placebo drug administration during supine (Figure 6.2) and tilted conditions (Figure 6.3). The post-injection changes in cardiovascular responses were not significantly different from zero in either supine (Figure 6.2) or tilted (Figure 6.3) positions, during either placebo or lidocaine conditions; there was no effect of injection of lidocaine or placebo on cardiovascular responses to carotid sinus massage. Hypersensitive responses were not observed in any participant in any condition.

125 Supine HUT

Pre-Injection Post-Injection Pre-Injection Post-Injection

Lidocaine Placebo Lidocaine Placebo Lidocaine Placebo Lidocaine Placebo

Max RRI (ms) 1075±33 1090±40 1030±44 1023±50 864±49* 895±53* 840±48* 859±47*

126 ∆ RRI (ms) +111±37 +103±32 +41±9† +39±14† +67±12 +104±25 +67±10 +64±16 ∆ SAP (mmHg) -5±1 -7±1 -3±1 -5±1 -5±1 -6±2 -3±1 -6±2 ∆ FVR (%) -18±3 -23±3 -16±4 -16±3 -15±4 -16±4 -13±4 -15±4

Table 6.1: Cardiovascular responses to placebo and lidocaine injection in the supine and tilted conditions. Responses are expressed as the largest response during carotid sinus massage, relative to the pre-massage baseline. RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance, HUT = head-up tilt. Significant differences (p<0.05) between corresponding values during supine and HUT are denoted by *; significant (p<0.05) main effect of injection in the supine position (post-injection vs pre-injection, independent of lidocaine or placebo condition) is denoted by †. 127

Figure 6.1: Effect of lidocaine administration on sternocleidomastoid electromyography (EMG). (A) Drawing showing the anatomical locations of the key structures in the neck. Carotid sinus massage was performed at the location of the carotid bifurcation determined using ultrasound (white circle). Lidocaine or placebo injections (2ml per site) were conducted on contralateral sternocleidomastoid muscles at four injection sites encompassing the full length of the muscle (yellow stars). (B) Representative EMG recorded from one participant before and after lidocaine and placebo injections. Traces show EMG recordings for the last 5 seconds of a 7-second maximal voluntary contraction. There was a near-abolition in EMG following lidocaine injection. EMG is expressed as percent of maximum RMS during pre-injection contraction. (B) Group data showing RMS EMG after injection with lidocaine and placebo (expressed as percent of maximum pre-injection contractions). The horizontal dashed lines represent group means, with solid horizontal lines representing the 95% confidence interval; * denotes statistical significance (p<0.0001). Figure 6.2: Effect of lidocaine and placebo injection on the supine cardiovascular responses to carotid sinus massage (CSM). Responses are expressed as the change in response to CSM in the post-injection condition compared to the pre-injection condition for: (A) maximum RRI prolongation; (B) change in RRI; (C) change in SAP; and (D) change in FVR. Hori- zontal dashed lines represent group means, solid horizontal lines represent 95% confidence interval. RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance.

6.5 Discussion

We showed that neural block of the sternocleidomastoid muscles does not increase cardiovascular responses to CSM in healthy young controls. This is contrary to the hypothesis[17] that chronic denervation of the sternocleidomastoid muscles with aging interrupts normal integration of neck muscle proprioception and carotid baroreceptor information, leading to excessive responses to CSM and an increased risk of syncope. Our results suggest that the previously reported association between sternocleidomastoid denervation and CSH in older adults[17] may be coincidental, rather than causal, and highlight the need for further research to identify the pathophysiological mechanisms underlying CSH. We are conﬁdent that we achieved an eﬀective block of the sternocleidomastoid muscles with lidocaine based on the known short onset of action (3-5 minutes) and relatively long

128 Figure 6.3: Effect of lidocaine and placebo injection on the tilted cardiovascular responses to carotid sinus massage (CSM). Responses are expressed as the change in response to CSM in the post-injection condition compared to the pre-injection condition for: (A) maximum RRI prolongation; (B) change in RRI; (C) change in SAP; and (D) change in FVR. Horizontal dashed lines represent group means, solid horizontal lines represent 95% confidence interval. RRI = R-R interval, SAP = systolic arterial pressure, FVR = forearm vascular resistance. half-life (1.5-2 hours) of lidocaine, as well as the near-abolition of sternocleidomastoid muscle activity that we observed following injection. Given that afferent fibres are blocked with lower doses of lidocaine than motor fibres[59], we consider the afferent input from the muscle to have been lost during the lidocaine condition; therefore, we infer that our negative findings are not a consequence of inadequate block. Rather than the proposed enhancement of cardiovascular responses to CSM, we actually saw a small but statistically significant decrease in RRI responses to CSM following injection. This small reduction was independent of the placebo or lidocaine conditions and may reflect the impact of the volume of injected fluid within the muscle on massage stimulus transmission to the carotid sinus. In healthy young adults, we observed larger cardiac responses to CSM when performed supine than when tilted, independent of the lidocaine or placebo condition. This is of interest because the current clinical guidelines[161] suggest that tilted CSM may be of

129 benefit in patients with recurrent syncope and/or suspected CSH because of the potential to unmask pathologically large vasodepressor responses in susceptible individuals when CSM is combined with orthostatic stress. This was not the case in our studies, presumably because the baroreflex engagement associated with normal orthostatic control in young healthy individuals (and resultant increases in sympathetic stimulation of the heart and vasculature) blunts the magnitude of the evoked bradycardia and vasodilation during CSM. If this observation extends to older adults, and those with CSH, it would suggest that orthostatic CSM may not be a necessary addition to autonomic function testing in these individuals. Our results do not support the prior observation[17] of sternocleidomastoid denervation contributing to CSH in older adults. This discrepancy may partly reflect the different methodology employed in the previous study, whereby sternocleidomastoid raw EMG was qualified based on physician observation as “normal” or “abnormal – moderate or severe”, and not subjected to quantitative evaluation. The method of standardisation of the stimulus for contraction of the sternocleidomastoid muscles was also not reported in the previous study so it is unclear whether abnormal EMG could be influenced by reduced contractile effort. Similarly, the CSM stimulus in the previous study was not standardised, and again responses were qualified as “normal, doubtful or CSH”. Where responses were discrepant between sides or there was disconnect between cardiac and blood pressure responses to CSM, the most abnormal response was considered in the correlative analyses. Nevertheless, despite these methodological differences, in the previous study abnormal EMG were observed in 57% of individuals and CSH in 27% of individuals, with strong concordance (p<0.00001) between the two[17]. However, correlation does not necessarily imply causality. It may be that these results reflect the known associations between aging and sarcopenia and age- related muscle weakness (reflected in reduced EMG activity)[29], and between aging and carotid sinus hypersensitivity[130], rather than an interaction between sternocleidomastoid muscle denervation and CSH per se. We have refuted the proposed potential role of sternocleidomastoid denervation in the pathophysiology of CSH. While this adds an interesting new piece to this physiological puz- zle, the mechanism underlying CSH remains elusive. Currently, two additional hypotheses have attempted to explain the phenomenon of CSH. The first posits that with aging, atherosclerotic buildup on the carotid artery vessel walls results in increased vessel stiffness, reduced barororeceptor activity, and upregulation of central gain with consequent overshoot responses[134]. While patients with CSH do exhibit increased arterial stiffening[117], efforts to investigate the mechanism of increased central gain have been unsuccessful[88, 141]. The second hypothesis argues that CSH is a symptom of more widespread autonomic dysregulation, including increased baroreflex gain and increased resting sympathetic tone, as measured by resting heart rate variability[181]. The pathophysiology of this dysregulation is unknown, but may involve altered function of central pathways involved in the

130 baroreflex. For example, CSH is common in patients with dementia with Lewy bodies, which points to central white matter lesions as contributors to the large heart rate and systolic blood pressure responses to CSM[89]. In patients with CSH, increased tau accumulation in baroreflex-associated nuclei has been observed[126], but cause and effect in terms of CSH has yet to be demonstrated.

6.5.1 Limitations

We saw a small but statistically significant reduction in the EMG signals following the placebo injections. We assume this reflects either modest fatigue during repeated EMG testing, or the presence of the volume of the injected saline within the muscle belly either slightly impairing contraction or providing impedance to the EMG signals recorded. How- ever, we do not believe this influences our overarching conclusion. The small reduction in EMG signals in the placebo condition was far less than the near-abolition of the EMG signals with lidocaine injections. Furthermore, the same volume of lidocaine was injected into the contralateral muscle, and comparisons between placebo and lidocaine conditions therefore control for the potential influence of injected volume on EMG activity. Lastly, the main implication of the small decrease in EMG activity in the saline condition would be hypothesised to be a small enhancement of the responses to CSM; this was not the case, further supporting our findings that neural block of the sternocleidomastoid muscles does not increase cardiovascular responses to CSM. While we were careful to ensure delivery of lidocaine into the muscle belly of the sternocleidomastoid, lidocaine may have spilled over to the vagus nerve or adjacent carotid baroreceptors, or may have migrated to the circulation. To ensure lidocaine was not injected into the circulation, prior to injection, a small syringe draw was performed to ensure no blood was present, indicating the needle was not in a vessel. Symptoms of intravenous lidocaine injection (dizziness, nausea, fainting, or seizure) were monitored, and no related symptoms were observed. If lidocaine spillover to the vagus had occurred, the resting heart rate following injection would be expected to decrease - this was not the case (pre-injection RRI = 953±47 ms, post-injection RRI = 990±47 ms, p = 0.15). We had a relatively small sample size and of course this has the potential to influence the statistical power. However, perhaps in part because of the highly standardised conditions, (ultrasound-guided carotid sinus massage; standardised massage procedure and pressure applied; breath holding to exclude respiratory sinus arrhythmia and respiratory-induced changes in blood pressure) we had very low within-subject variance. Accordingly, we were powered to detect a difference in the RRI response following injection of just 60ms, and for blood pressure and vascular responses of only 8mmHg and 7% respectively. We do not think the lack of enhancement of cardiovascular responses to CSM with sternocleidomastoid blockade can be explained by low statistical power.

131 We conducted our analyses in a cohort of young healthy participants, who had small baseline responses to CSM with no evidence of hypersensitive responses; this phenomenon is well-documented[96]. We chose to conduct our assessments in young individuals with small responses to CSM because we were concerned that the larger responses documented in older adults[130, 78, 96, 178] might create a ceiling effect, and that it would not be possible to enhance responses further with the lidocaine intervention in this population, leading to a “false-negative” result. Accordingly, we are confident that the lack of increase in response to CSM during lidocaine administration was not due to a ceiling effect. However, given the known effect of aging on baroreflex responses to CSM, it is possible that responses in healthy older adults to the intervention might be different (smaller) than those reported here in young individuals. Finally, we considered the potential role of only the sternocleidomastoid muscles in the pathophysiology of CSH, and did not consider that normal afferent feedback from other neck muscles might compensate for the imposed pharmacological “denervation” of the sternocleidomastoid muscle in this study. We do not believe this invalidates our findings for two reasons: firstly, we did not provide an afferent stimulus to any other muscle groups during our CSM protocol so they presumably did not ameliorate the impact of our intervention; secondly, prior research has shown that only reductions in EMG of the sternocleidomastoid muscles, and not other neck muscles (such as styloglossus and upper trapezius) were associated with CSH.

132 Chapter 7

Discussion

In this thesis, I describe the agreement, repeatability, and age- and sex-related differences in arterial baroreflex function. Accurate and repeatable measures of baroreflex function are fundamental to better understand the physiology of the wide array of diseases associated with altered baroreflex function[114, 84, 197, 39, 189, 165, 99, 180]. In particular, the prevalence of disorders associated with baroreflex function such as hypertension[189], stroke[180], myocardial infarction[99, 98] and sleep apnoea[39] all rise with age, suggesting aging-related changes in baroreflex function. Hypertension[110] and syncope[155, 130] also display strong associations with sex. However, our understanding of age- and sex-related differences in baroreflex function has been hampered by the wide range of baroreflex measurement techniques, their different underlying mechanisms, and uncertain agreement.

7.1 The Agreement and Repeatability of Non-Invasive Barore- ﬂex Techniques

In Chapter 3 I present the agreement and repeatability of commonly-used, non-invasive baroreflex tests. In general, cardiac baroreflex tests displayed better agreement and repeatability than sympathetic baroreflex tests. Dynamic baroreflex tests displayed better agreement and repeatability than static baroreflex tests, although the repeatability of cardiac and sympathetic metrics from the Valsalva maneuver was also high. Two commonly-used clinical baroreflex measures, CSM and the Valsalva ratio, had very poor agreement with other baroreflex measures. With CSM, responses in our cohort were not different from sham massage, and likely negatively impacted agreement. However, the

Valsalva ratio had poor agreement with all other cardiac metrics except cBRSinc. This is perplexing, as the valsalva ratio is purported to reflect cardiovagal function [132]. The measure has been criticized as the cardiac responses are largely dependent on the blood pressure fluctuations during the Valsalva maneuver, which are not accounted for with the Valsalva ratio[60]. The poor agreement of these two tests with other baroreflex tests questions their

133 utility as measures of baroreflex function in healthy participants. Future studies attempting to measure baroreflex function should select baroreflex tests that display reasonable agreement with other tests. This will ensure that conversion of BRS values is possible, allowing for more robust comparisons between studies. We have presented metrics that enable researchers and clinicians to convert BRS between tests, providing this conversion uses the mean of a sample. An obvious use of these metrics is to convert values of BRS that have been identified as important cut-off values to stratify risk or survival of certain patient groups. For instance, a cBRS of 3 ms.mmHg-1 has been identified as an important value to stratify outcomes for heart failure patients. While originally identified using the modified Oxford Method[98], this value has also been replicated using cross-spectral analysis[63] and the α-index[150]. Using the same value for both dynamic methods is likely appropriate given our demonstrated high level of agreement between these two tests. Using our conversion metrics, we can calculate similar cutoff val- -1 ues for cBRSinc (2.5, 95% PI 1.1-3.8 ms.mmHg ), and with cBRSdec (3.1, 95% PI 1.8-4.3 ms.mmHg-1). These values are consistent with empirically-derived values from the Valsalva maneuver[153]. Broadly, the results from this chapter indicate a need for careful consideration when interpreting the baroreflex literature, as the agreement and repeatability of non-invasive baroreflex tests vary widely. Furthermore, it highlights the need for better methods of sympathetic baroreflex measurement techniques. Our novel dynamic methods of sympathetic vascular function displayed high agreement and repeatability, and can be used alongside dynamic baroreflex measures. Lastly, the poor agreement of currently available sympathetic baroreflex tests may also indicate that sympathetic baroreflex function is difficult to quantify during steady state measurements, and may be more useful when expressed as a change from baseline. For example, supine sBRS correlates poorly (R2 = 0.07, p = 0.07) with the maximal FVR response during orthostasis, while the HUT-supine ΔsBRS is a more robust predictor of the maximal FVR response during orthostasis (R2 = 0.34, p < 0.0001, Figure 4.4, page 91).

7.2 Sex and Age Diﬀerences in Baroreﬂex Function

Recent investigations into sex- and age-related diﬀerences in sympathetic vascular function[75, 22] have highlighted the need for techniques that are able to directly measure vascular function in a variety of experimental conditions, in particular during orthostasis. A number of new sympathetic measures have recently been developed that have improved our ability to measure sympathetic vascular function in autonomic failure using the Valsalva maneuver[159, 191], and in healthy controls using MSNA[22]. However, indices from the Valsalva maneuver do not allow continuous measurement of sympathetic function, and measuring sympathetic function with MSNA during orthostasis is technically challenging. In

134 Chapter 3, I present novel dynamic sympathetic vascular baroreflex measurements. This represents a significant advance in baroreflex measurement, as we are now able to continuously and simultaneously measure sympathetic vascular and cardiac baroreflex function during orthostasis using brachial Doppler ultrasound. Short-term signal interruptions with brachial Doppler are common due to participant movement during orthostatic stress testing. This movement artifact, as well as physiological artifacts such as ectopic beats, are a significant problem for baroreflex techniques based on spectral analysis, as they introduce false frequency components into the power spectrum[12]. In this thesis, both movement artifacts and physiological artifacts were removed via linear interpolation. While this is a well-established approach for ectopic beats that only affect two consecutive beats[12], applying linear interpolation over a larger number of consecutive beats may negatively influence the validity of the results, and therefore reduce the utility of the technique. Furthermore, we originally applied cross-spectral techniques commonly used with cardiac baroreflex function to describe the relationship between MAP and FVR. However, we found that low coherence using cross-spectral analysis hampered the utility of the technique. We ultimately chose to use spontaneous sequence analysis as: 1) it is more robust to signal interruptions, and 2) it displayed reasonable agreement with two previously validated sympathetic baroreflex measures, PRT and BRSα1. Sequence analysis also presents a significant advantage over spectral techniques in that baroreflex function can be tracked continuously over time, and may allow for description of baroreflex function, for example, during or just preceding presyncope. Future research should explore this possibility. In this study, we observed key differences in the relationship between cBRS and sBRS derived from sequence analysis, and their relationship to the maximal RRI and FVR responses during orthostasis, respectively. Resting supine cBRS correlates robustly with the HUT-supine ΔcBRS, and therefore correlates reasonably with the maximal RRI response to orthostasis. This is consistent with our understanding of “vagal withdrawal” (the reduction in vagal tone commonly associated with maneuvers that increase sympathetic outflow) during orthostatic stress, with participants with higher resting vagal tone able to “with- draw” to a greater degree (ie. greater “cBRS reserve”). However, it should be noted that baroreflex-mediated cardiac responses have a sympathetic component. However, resting supine sBRS correlates poorly with the HUT-supine ΔsBRS, consistent with the notion of “sympathetic activation” during orthostasis. As sympathetic tone at rest is very low, it is difficult to predict how it will behave during baroreceptor unloading by measuring its activity in the resting state. This is an important distinction, as supine cBRS may be used as a an indication of cardiac baroreflex function during orthostasis, while supine sBRS is less robust. We also assessed cardiac and sympathetic baroreflex function using the neck collar technique. The central finding was that sympathetic vascular baroreflex curves are left-shifted

135 to lower pressures compared to cardiac baroreflex curves, indicating that the sympathetic arm of the baroreflex is better able to buffer rapid reductions in MAP. The vascular responses to baroreceptor unloading were larger in younger participants, and comparable in young women and young men. This is contrary to previous reports of reduced vascular transduction in young women[74, 22], and similar[58] or increased[162] MSNA responses to baroreceptor unloading in young men compared to young women. Conversely, we observed reduced maximal FVR responses during orthostasis in young women compared to young men and older individuals. As we performed an orthostatic stress test and the neck collar protocol in the same individuals, we were able to compare responses between the two tests. There is evidence that the relative contribution of the cardiac and sympathetic baroreflex arms is different for hypotensive vs. hypertensive stimuli[92, 40]. We, therefore, chose to compare the cardiac and vascular responses to a hypotensive neck collar stimulus (+40 mmHg) with a hypotensive stimulus from the orthostatic stress test. The cardiac responses to +40mmHg neck pressure correlated positively with the maximal cardiac responses to orthostatic stress (R2 = 0.16, p = 0.01). However, the FVR responses to +40mmHg neck pressure correlated negatively with the maximal FVR response during tilt (R2 = 0.11, p = 0.048). Thus, large acute FVR responses to simulated hypotension using the neck collar correlated with small maximal FVR responses during orthostasis. The physiology underlying this reciprocal relationship is unclear, but the different baroreceptor location stimulated, or the length of the hypotensive stimulus, may be relevant to the magnitude of the responses. To our knowledge, this is the first time this relationship has been directly observed. However, a previous study examining the vascular responses to a hypotensive stimuli using neck pressure found that young women display comparable vascular responses to young men[92]. Similarly, another study examining the vascular responses to a single burst of MSNA also found that the vascular responses were similar between young men and women[190]. In reconciling these findings with the mounting evidence of reduced vascular transduction in young women compared to young men[74, 22], both authors speculate that the acute vascular responses over a short time period may have influenced their results. If true, this reciprocal relationship between acute vascular responses to baroreflex unloading and maximal responses during orthostasis presents a significant problem for the utility of novel static sympathetic baroreflex tests, at least in the context of healthy individuals. It may underly the poor agreement between the maximal FVR response during orthostasis and static sympathetic baroreflex observed in Chapter 3. Future research should explore the relationship between acute and sustained vascular responses to baroreflex unloading.

136 7.3 The Methodology of Carotid Sinus Massage

In Chapter 5, we show that the current recommendations for the location at which to apply CSM on the neck underestimates the location of the carotid sinus, and likely impacts the magnitude of the responses to CSM. However, we were unable to elicit a hypersensitive response to CSM despite recruiting a large number of older adults and patients, in whom large responses are reported to be prevalent[96]. While we did see evidence that the location of the massage influences the magnitude of the cardiovascular response, we were unable to demonstrate improved diagnostic yield with ultrasound-guided CSM. Future research should examine our protocol in a cohort of participants with demonstrated or suspected CSH. The anatomical data indicating that the area of maximal pulsatility approximates the carotid bifurcation is clear, and alone is sufficient to recommend that CSM be applied to the area of maximal pulsatility when ultrasound is unavailable. Certainly, the current recommendations that CSM be applied at the level of the cricoid cartilage underestimates the carotid bifurcation substantially. Our study further demonstrates that in patients, the cardiovascular responses to ultrasound-guided CSM were larger than CSM applied at the cricoid cartilage. Thus, CSM applied diagnostically (during investigation of CSH) and interventionally (e.g. to interrupt supraventricular tachycardia[32]) should be applied at ultrasound-guided carotid bifurcation, or the area of maximal pulsatility when ultrasound- guidance is unavailable. Few of the healthy participants in our study displayed a cardiovascular response to CSM that was significantly different from sham. It is well-established that the responses to CSM are smaller in younger individuals[96], but to our knowledge this is the first time that CSM has been compared to sham massage in healthy participants. The poor agreement we observed between CSM and other baroreflex tests is therefore not surprising as agreement with a non-response will by definition be poor. The absence of measurable responses in otherwise healthy participants likely contributes to the poor efficacy (5%-15% success rate) of interventional CSM to terminate supraventricular tachycardia[32]. The observed high specificity of diagnostic CSM (70-90%)[100, 91] is in keeping with our observations that otherwise healthy individuals do not display measurable responses to CSM. However, the low sensitivity of the test (10-41%)[91, 100] questions the diagnostic utility of CSM, and may point to CSH as an otherwise normal clinical sign of aging. While this is still debated in the literature[181], the criteria for a “hypersensitive” response to CSM(>3000ms asystole and/or reduced SAP > 50 mmHg)[130] has recently been criticized as too lenient[200]. It remains to be seen what effect a more stringent diagnostic criteria would have on the sensitivity of CSM. It is unclear why digital stimulation of the carotid sinus does not produce meaningful cardiovascular responses, but carotid stimulation with the neck collar does. Carotid stimulation with the neck collar is likely more physiological than massage, as the distending

137 pressure of the carotid arteries is altered in a similar fashion as during changes in arterial pressure. The disparity between the responses to carotid baroreceptor stimulation via CSM vs. neck suction in healthy participants with intact baroreﬂex function raises the question as to why a large proportion of older participants display large responses to CSM.

7.4 The Pathophysiology of Carotid Sinus Hypersensitivity

In Chapter 6, I examined the proposed potential role of chronic sternocleidomastoid denervation in the pathophysiology of CSH[17, 185]. I demonstrated that sensory block of the sternocleidomastoid muscles does not interrupt central integration of sensory and baroreflex afferent information, and the cardiovascular responses to CSM are unchanged[112]. This suggests that the relationship between sternocleidomastoid denervation and CSH is coincidental, not causal[17, 185]. Future investigations should focus on alternative hypotheses regarding the physiology of CSH. Atherosclerotic buildup on the carotid artery vessel walls has been posited to increase vessel stiffness, reduce baroreceptor activity, and upregulate central baroreflex gain with consequent overshoot responses during CSM[134]. Increased arterial stiffening has been observed in patients with CSH[117], and studies examining cBRS in patients with CSH have demonstrated increased BRS using the modified Oxford Method[129], sequence analysis[181], and the neck collar technique[42]. However, the magnitude of increased gain in these studies does not account for the magnitude of the cardiovascular responses to CSM. Further investigations into this proposed phenomenon should focus on the relationship between carotid artery distension and the consequent baroreflex response. Efforts to investigate the mechanism of increased central BRS gain secondary to age- related atherosclerotic changes have been unsuccessful[88, 141]. However, the investigation of increased central gain involves several points of connection involving several neurotrans- mitters (e.g. glutamine, gamma-aminobutyric acid)[41, 207]. Identification of the specific neurological structures responsible for increased central gain may, therefore, be difficult. The second main hypothesis argues that widespread autonomic dysregulation involving altered function of central baroreflex pathways underlies CSH. Evidence in support of this hypothesis include the observation that CSH is common in patients with dementia with Lewy bodies[89], and increased tau accumulation in baroreflex-associated nuclei in patients with CSH[126]. However, it is unclear why these neurological changes differen- tially affect the cardiovascular responses to CSM, with asystole greater 3 seconds and/or reduced SAP greater than 50 mmHg, compared to the modestly increased cBRS with other measures[129, 42, 181]. Furthermore, we have established that a large number of healthy participants do not exhibit measurable responses to CSM. Therefore, any hypothesis attempting to explain the pathophysiology of CSH must explain not an increase in sensitivity

138 of an established baroreﬂex response, but the emergence of a large baroreﬂex response that does not measurably exist in healthy controls.

7.5 Conclusion

The agreement of non-invasive cardiac and sympathetic baroreflex tests is variable and should be considered when planning studies involving baroreflex measurement. Our novel metrics of dynamic sympathetic baroreflex sensitivity provide direct measurement of vascular sympathetic function, with results consistent with previous literature establishing sex- and age-related differences in vascular baroreflex function. However, these methods have yet to be validated with microneurography, which should be a focus for future researchers interested in developing vascular baroreflex tests. Following positive validation, these techniques should also be applied to clinical populations such as patients with POTS, vasovagal syncope, or orthostatic intolerance. The physiology underlying the cardiovascular responses to CSM is perplexing. Our observations that healthy individuals with intact baroreflex function do not display measurable responses to CSM further complicates the problem, as CSH appears to therefore involve the emergence of a cardiovascular response, rather than sensitization of an existing one. Future studies should seek to unravel this mystery, particularly as it appears to be a relatively common product of aging[96].

7.6 Key Concepts

• The agreement and repeatability of common non-invasive baroreflex tests is variable, and care should be taken when selecting and interpreting baroreflex tests. In general, cardiac baroreflex tests display better agreement and repeatability than sympathetic baroreflex tests. Dynamic baroreflex measures display high agreement and repeatability, and enable characterization of cardiac and vascular baroreflex function concurrently.

• Simultaneous measurement of cardiac and sympathetic vascular baroreflex function provides valuable insight into the baroreflex control of blood pressure. Young women display attenuated FVR and sBRS responses to orthostasis, which likely impacts their ability to maintain blood pressure during orthostasis. Sympathetic baroreflex sigmoid curves are left-shifted compared to cardiac baroreflex sigmoid curves, suggesting that sympathetic vascular responses may be more important for the baroreflex regulation of blood pressure during hypotensive stimuli such as orthostatic stress. The baroreflex- mediated vascular responses to orthostasis are negatively correlated with those from simulated hypotension with the neck collar, while the cardiac responses are positively correlated, indicating that the two methods have differential effects on the cardiac and

139 sympathetic arms of the baroreflex. We have identified different baroreflex strategies for the maintenance of blood pressure based on vascular responses during orthostasis.

• The current recommendations for CSM to be performed at the level of the cricoid cartilage underestimates the location of the carotid sinus, and likely impacts the magnitude of the responses to CSM. Performing CSM at the location of maximal pulsatility approximates the carotid sinus well, and may be used by physicians who are unable to perform ultrasound-guided CSM. Ultrasound-guided CSM at the level of the carotid bifurcation produces larger responses in patients with syncope. A large proportion of healthy controls do not display responses to CSM that are signiﬁcantly diﬀerent from sham massage.

• Neural block of the sternocleidomastoid muscles does not increase cardiovascular responses to CSM in healthy young controls. This suggests that the previously reported association between sternocleidomastoid denervation and CSH in older adults[17] may be coincidental, rather than causal, and highlight the need for further research to identify the pathophysiological mechanisms underlying CSH.

140 Bibliography

[1] Howraa Abed, Ball Patrick A, and Wang Le-Xin. Diagnosis and management of postural orthostatic tachycardia syndrome: A brief review. Journal of Geriatric Cardiology, 9(1):61–67, 2012.

[2] Wayne O. Adkisson and David G. Benditt. Pathophysiology of reﬂex syncope: A review. Journal of Cardiovascular Electrophysiology, 28:1088–1097, 2017.

[3] Philip N Ainslie and James Duﬃn. Integration of cerebrovascular CO2 reactivity and chemoreﬂex control of breathing : mechanisms of regulation , measurement , and interpretation. Am J Physiol Regul Integr Comp Physiol, 296:R1473–R1495, 2009.

[4] S Akselrod. Spectral analysis of ﬂuctuations in cardiovascular parameters: a quantitative tool for the investigation of autonomic control. Trends in pharmacological sciences, 9(1):6–9, jan 1988.

[5] S Akselrod, D Gordon, J B Madwed, N C Snidman, D C Shannon, and R J Cohen. Hemodynamic regulation: investigation by spectral analysis. The American journal of physiology, 249(4 Pt 2):H867–75, oct 1985.

[6] S Akselrod, D Gordon, F A Ubel, D C Shannon, A C Berger, and R J Cohen. Power spectrum analysis of heart rate ﬂuctuation: a quantitative probe of beat-to- beat cardiovascular control. Science (New York, N.Y.), 213(4504):220–2, jul 1981.

[7] Vaibhav Amin and Behzad B Pavri. Carotid Sinus Syndrome. Cardiology in Review, 23(3):130–134, 2015.

[8] David R Anderson, Kenneth P Burnham, and William L Thompson. Null Hypothesis Testing: Problems, Prevalence, and an Alternative David. The Journal of Wildlife Management, 64(4):912–923, 2000.

[9] Melanie Barantke, Timothy Krauss, Jasmin Ortak, Wolfgang Lieb, Michael Rep- pel, Christof Burgdorf, Peter P. Pramstaller, Heribert Schunkert, and Hendrik Bon- nemeier. Eﬀects of gender and aging on diﬀerential autonomic responses to orthostatic maneuvers. Journal of Cardiovascular Electrophysiology, 19:1296–1303, 2008.

[10] Hal Barron, Adam Fitzpatrick, and Nora Goldschlager. Head-up tilt testing: Do we need to give an added push? American Journal of Medicine, 99:689–690, 1995.

[11] David G Benditt, J Gert van Dijk, Richard Sutton, Wouter Wieling, Joseph C Lin, Scott Sakaguchi, and Fei Lu. Syncope. Current problems in cardiology, 29(4):152–229, apr 2004.

141 [12] G G Berntson, J T Bigger, D L Eckberg, P Grossman, P G Kaufmann, M Malik, H N Nagaraja, S W Porges, J P Saul, P H Stone, and M W van der Molen. Heart rate variability: origins, methods, and interpretive caveats. Psychophysiology, 34(6):623– 48, nov 1997.

[13] G Bertinieri, M Di Rienzo, A Cavallazzi, A U Ferrari, A Pedotti, and G Mancia. Evaluation of baroreceptor reﬂex by blood pressure monitoring in unanesthetized cats. The American Journal of Physiology, 254(2 Pt 2):H377–83, feb 1988.

[14] J T Bigger, J L Fleiss, R C Steinman, L M Rolnitzky, R E Kleiger, and J N Rottman. Frequency domain measures of heart period variability and mortality after myocardial infarction. Circulation, 85(1):164–71, jan 1992.

[15] A P Blaber, Y Yamamoto, and R L Hughson. Methodology of spontaneous baroreﬂex relationship assessed by surrogate data analysis. The American journal of physiology, 268(18):H1682–H1687, 1995.

[16] J. J. Blanc, C. L’Her, A. Touiza, B. Garo, E. L’Her, and J. Mansourati. Prospec- tive evaluation and outcome of patients admitted for syncope over a 1 year period. European Heart Journal, 23(10):815–820, 2002.

[17] J J Blanc, G L’Heveder, J Mansourati, S H Tea, P Guillo, and D Mabin. Assessment of a newly recognized association. Carotid sinus hypersensitivity and denervation of sternocleidomastoid muscles. Circulation, 95(11):2548–51, jun 1997.

[18] J M Bland and D G Altman. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet, 1(8476):307–10, feb 1986.

[19] J Martin Bland and Douglas G Altman. Measuring Agreement in Method Comparison Studies. Statistical Methods in Medical Research, 8:135–160, 1999.

[20] Istvan Bonyhay, Marcelo Risk, and Roy Freeman. High-pass ﬁlter characteristics of the baroreﬂex–a comparison of frequency domain and pharmacological methods. PloS one, 8(11):e79513, jan 2013.

[21] Walter F. Boron and Emile L. Boulpaep. Medical Physiology. 2e edition, 2008.

[22] L. J.B. Briant, A. E. Burchell, L. E.K. Ratcliﬀe, N. Charkoudian, A. K. Nightingale, J. F.R. Paton, Michael J. Joyner, and E. C. Hart. Quantifying sympathetic neuro- haemodynamic transduction at rest in humans: insights into sex, ageing and blood pressure control. Journal of Physiology, 594(17):4753–4768, 2016.

[23] M Brignole, M Disertori, C Menozzi, A Raviele, P Alboni, MV Pitzalis, P Delise, E Puggioni, M Del Greco, V Malavasi, M Lunati, M Pepe, and D Fabrizi. Management of syncope referred urgently to general hospitals with and without syncope units. Europace, 5:263–266, 2003.

[24] M. Brignole, C. Menozzi, A. Del Rosso, S. Costa, G. Gaggioli, N. Bottoni, P. Bartoli, and R. Sutton. New classiﬁcation of haemodynamics of vasovagal syncope: Beyond the VASIS classiﬁcation: Analysis of the pre-syncopal phase of the tilt test without and with nitroglycerin challenge. Europace, 2(1):66–76, 2000.

142 [25] Michele Brignole, Paolo Alboni, David G Benditt, Lennart Bergfeldt, Jean-Jacques Blanc, Paul Erik Bloch Thomsen, J Gert van Dijk, Adam Fitzpatrick, Stefan Hohn- loser, Jean Janousek, Wishwa Kapoor, Rose Anne Kenny, Piotr Kulakowski, Giulio Masotti, Angel Moya, Antonio Raviele, Richard Sutton, George Theodorakis, Andrea Ungar, and Wouter Wieling. Guidelines on management (diagnosis and treatment) of syncope–update 2004. Europace : European pacing, arrhythmias, and cardiac electrophysiology : journal of the working groups on cardiac pacing, arrhythmias, and cardiac cellular electrophysiology of the European Society of Cardiology, 6(6):467–537, nov 2004. [26] C. M. Brown and R. Hainsworth. Forearm vascular responses during orthostatic stress in control subjects and patients with posturally related syncope. Clinical Autonomic Research, 10(2):57–61, 2000. [27] CM Brown and Roger Hainsworth. Shifts During Orthostatic Stress Intolerance. Clinical Autonomic Research, 9:69–73, 1999. [28] A Cevese, G Gulli, E Polati, L Gottin, and R Grasso. Baroreflex and oscillation of heart period at 0.1 Hz studied by alpha-blockade and cross-spectral analysis in healthy humans. The Journal of physiology, 531(Pt 1):235–44, feb 2001. [29] David J. Clark and Roger A. Fielding. Neuromuscular contributions to age-related weakness. Journals of Gerontology - Series A Biological Sciences and Medical Sci- ences, 67 A(1):41–47, 2012. [30] Victoria E Claydon and Roger Hainsworth. Increased postural sway in control subjects with poor orthostatic tolerance. Journal of the American College of Cardiology, 46(7):1309–13, oct 2005. [31] R.H. Clayton, A.J. Bowman, G.A. Ford, and A. Murray. Measurement of baroreflex gain from heart rate and blood pressure spectra: a comparison of spectral estimation techniques. Physiological measurement, 16:131–139, 1995. [32] Nikolas A. Collins and George L. Higgins. Reconsidering the effectiveness and safety of carotid sinus massage as a therapeutic intervention in patients with supraventricular tachycardia. American Journal of Emergency Medicine, 33:807–809, 2015. [33] Victor Convertino. Neurohumoral mechanisms associated with orthostasis: reaffirma- tion of the significant contribution of the heart rate response. Frontiers in physiology, 5(June):236, jan 2014. [34] Victor A Convertino. Gender differences in autonomic functions associated with blood pressure regulation. Am J Physiol, 275(6 pt 2):1909–1920, 1998. [35] Victoria Cooper, Mark W Elliott, Stan B Pearson, Claire M Taylor, and Roger Hainsworth. Daytime variability in carotid baroreflex function in healthy human subjects. Clinical autonomic research : official journal of the Clinical Autonomic Research Society, 17(1):26–32, feb 2007. [36] Victoria Cooper and Roger Hainsworth. Carotid baroreflex testing using the neck collar device. Clinical autonomic research official journal of the Clinical Autonomic Research Society, 19(2):102–112, may 2009.

143 [37] VL Cooper and R Hainsworth. Carotid baroreceptor reflexes in humans during orthostatic stress. Experimental Physiology, 86(5):677–681, 2001. [38] VL Cooper and R Hainsworth. Effects of head-up tilting on baroreceptor control in subjects with different tolerances to orthostatic stress. Clinical science London England, 103(3):221–226, 2002. [39] P. Cortelli, C. Lombardi, P. Montagna, and G. Parati. Baroreflex modulation during sleep and in obstructive sleep apnea syndrome. Autonomic Neuroscience: Basic and Clinical, 169(1):7–11, 2012. [40] Daniel P Credeur, Seth W Holwerda, Leryn J Boyle, Lauro C Vianna, Areum K Jensen, and Paul J Fadel. Effect of aging on carotid baroreflex control of blood pressure and leg vascular conductance in women. American journal of physiology. Heart and circulatory physiology, 306(10):H1417–25, may 2014. [41] Roger A.L. Dampney. Resetting of the baroreflex control of sympathetic vasomotor activity during Natural Behaviors: description and conceptual model of central mechanisms. Frontiers in Neuroscience, 11(461), 2017. [42] T C Dehn, C A Morley, and R Sutton. A scientific evaluation of the carotid sinus syndrome. Cardiovascular research, 18(12):746–51, dec 1984. [43] C P Doe, M J Drinkhill, D S Myers, D A Self, and R Hainsworth. Reflex vascular responses to abdominal venous distension in the anesthetized dog. The American journal of physiology, 271(3 Pt 2):H1049–56, sep 1996. [44] T J Ebert and A W Cowley. Baroreflex modulation of sympathetic outflow during physiological increases of vasopressin in humans. The American journal of physiology, 262(5 Pt 2):H1372–8, may 1992. [45] T J Ebert, B J Morgan, J a Barney, T Denahan, and J J Smith. Effects of aging on baroreflex regulation of sympathetic activity in humans. The American journal of physiology, 263(3 Pt 2):H798–H803, 1992. [46] D L Eckberg. Temporal response patterns of the human sinus node to brief carotid baroreceptor stimuli. The Journal of physiology, 258(3):769–82, jul 1976. [47] D L Eckberg, M S Cavanaugh, A L Mark, and F M Abboud. A simplified neck suction device for activation of carotid baroreceptors. The Journal of laboratory and clinical medicine, 85(1):167–73, jan 1975. [48] K. M. El-Bedawi, M. M. A. E. Wahbha, and R. Hainsworth. Cardiac pacing does not improve orthostatic tolerance in patients with vasovagal syncope. Clinical Autonomic Research, 4(5):233–237, oct 1994. [49] Ciaran Finucane, Mary Paula Colgan, Clodagh O’Dwyer, Collette Fahy, Orla Collins, Gerry Boyle, and Rose Anne Kenny. The accuracy of anatomical landmarks for locating the carotid sinus. Age and Ageing, 45(6):904–907, 2016. [50] James P Fisher, Areum Kim, Colin N Young, Shigehiko Ogoh, Peter B Raven, Niels H Secher, and Paul J Fadel. Influence of ageing on carotid baroreflex peak response latency in humans. The Journal of physiology, 587(Pt 22):5427–39, nov 2009.

144 [51] D I Fisiologia. Some observations on the eﬀects of stimulating the stretch receptors in the carotid artery of man. Journal of Physiology, 137(Supplement):45P, 1957.

[52] A F Folino. Cerebral autoregulation in neurally mediated syncope: victim or execu- tioner? Heart (British Cardiac Society), 92(November 2005):724–726, 2006.

[53] Robert R Freedman, Subhash C Sabharwal, and Nagaraj Desai. Sex Diﬀerences in Peripheral Vascular Adrenergic Receptors. Circulation Research, 61(4):581–585, 1987.

[54] Roy Freeman. Assessment of cardiovascular autonomic function. Clinical Neurophys- iology, 117(4):716–730, 2006.

[55] M A Frey and G W Hoﬄer. Association of sex and age with responses to lower-body negative pressure. Journal of applied physiology (Bethesda, Md. : 1985), 65(4):1752–6, 1988.

[56] M A Frey, C M Tomaselli, and W G Hoﬄer. Cardiovascular responses to postural changes: diﬀerences with age for women and men. Journal of clinical pharmacology, 34(5):394–402, 1994.

[57] Qi Fu, Sarah Witkowski, and Benjamin D. Levine. Vasoconstrictor reserve and sympathetic neural control of orthostasis. Circulation, 110(18):2931–2937, 2004.

[58] Qi Fu, Sarah Witkowski, Kazunobu Okazaki, and Benjamine Levine. Eﬀects of gender and hypovolemia on sympathetic neural responses to orthostatic stress. Am J Physiol Regul Integr Comp Physiol, 289:R109–R116, 2005.

[59] Emily Garmon and Scott Dulebohn. Topical, Local, and Regional Anesthesia and Anesthetics. StatPearls, 2017.

[60] David S. Goldstein and William P. Cheshire. Beat-to-beat blood pressure and heart rate responses to the Valsalva maneuver. Clinical Autonomic Research, 27(6):361–367, 2017.

[61] DS Goldstein, D Horwitz, and HR Keiser. Comparison of Techniques for Measuring Baroreﬂex Sensitivity in Man. Circulation, 66(2):432–439, 1982.

[62] R W Gotshall, P F Tsai, and M A Frey. Gender-based diﬀerences in the cardiovascular response to standing. Aviation, space, and environmental medicine, 62(9 Pt 1):855–9, sep 1991.

[63] S. Gouveia, M. G. Scotto, G. D. Pinna, R. Maestri, M. T. La Rovere, and P. J. S. G. Ferreira. Spontaneous baroreceptor reflex sensitivity for risk stratification of heart failure patients: optimal cut-off and age effects. Clinical Science, 129(12):1163–1172, 2015.

[64] R Grasso, F Schena, G Gulli, and a Cevese. Does low-frequency variability of heart period reﬂect a speciﬁc parasympathetic mechanism? Journal of the autonomic nervous system, 63(1-2):30–8, mar 1997.

[65] Jan T. Groothuis, Dick H J Thijssen, Miriam Kooijman, Rebecca Paulus, and Maria T E Hopman. Attenuated peripheral vasoconstriction during an orthostatic challenge in older men. Age and Ageing, 37:680–684, 2008.

145 [66] Ilja Guelen, Berend E Westerhof, Gertrude L van der Sar, Gert A van Montfrans, Fer- dinand Kiemeneij, Karel H Wesseling, and Willem Jan W Bos. Validation of brachial artery pressure reconstruction from ﬁnger arterial pressure. Journal of Hypertension, 26(7):1321–1327, 2008.

[67] Giosub Gulli, Victoria L Wight, Roger Hainsworth, and Antonio Cevese. Spectral and cross-spectral autoregressive analysis of cardiovascular variables in subjects with diﬀerent degrees of orthostatic tolerance. Clinical Autonomic Research, 11:19–27, 2001.

[68] Giosuè Gulli, Victoria Louise Cooper, Victoria Elizabeth Claydon, and Roger Hainsworth. Prolonged latency in the baroreﬂex mediated vascular resistance response in subjects with postural related syncope. Clinical autonomic research oﬃcial journal of the Clinical Autonomic Research Society, 15(3):207–212, jul 2005.

[69] Seung Min Ha, Young Suk Cho, Gyu Chong Cho, Choong Hyun Jo, and Ji Young Ryu. Modiﬁed carotid sinus massage using an ultrasonography for maximizing vagal tone: A crossover simulation study. American Journal of Emergency Medicine, 33(7):963– 965, 2015.

[70] R Hainsworth and Y M Al-Shamma. Cardiovascular responses to stimulation of carotid baroreceptors in healthy subjects. Clinical science (London, England : 1979), 75(2):159–65, aug 1988.

[71] R. Hainsworth and K. M. El-Bedawi. Orthostatic tolerance in patients with unexplained syncope. Clinical Autonomic Research, 4:239–244, 1994.

[72] Roger Hainsworth. Exercise training in orthostatic intolerance. Q J Med, 91:715–717, 1998.

[73] Roger Hainsworth. Pathophysiology of syncope. Clinical autonomic research : oﬃcial journal of the Clinical Autonomic Research Society, 14 Suppl 1:18–24, oct 2004.

[74] Emma C Hart, Nisha Charkoudian, B Gunnar Wallin, Timothy B Curry, John Eise- nach, and Michael J Joyner. Sex and ageing diﬀerences in resting arterial pressure regulation: the role of the β-adrenergic receptors. The Journal of physiology, 589(Pt 21):5285–97, dec 2011.

[75] Emma C. Hart, Nisha Charkoudian, B. Gunnar Wallin, Timothy B. Curry, John H. Eisenach, and Michael J. Joyner. Sex diﬀerences in sympathetic neural-hemodynamic balance implications for human blood pressure regulation. Hypertension, 53(3):571– 576, 2009.

[76] G D S Hirst and F R Edwards. Sympathetic Neuroeﬀector Transmission in Arteries and Arterioles. Physiological Reviews, 69(2):546–604, 1989.

[77] Chih Cheng Huang, Paola Sandroni, David M. Sletten, Stephen D. Weigand, and Phillip A. Low. Eﬀect of age on adrenergic and vagal baroreﬂex sensitivity in normal subjects. Muscle and Nerve, 36(5):637–642, 2007.

146 [78] a M Humm and C J Mathias. Abnormal cardiovascular responses to carotid sinus massage also occur in vasovagal syncope - implications for diagnosis and treatment. European journal of neurology : the oﬃcial journal of the European Federation of Neurological Societies, 17(8):1061–7, aug 2010.

[79] B W Hyndman, R I Kitney, and B M Sayers. Spontaneous rhythms in physiological control systems. Nature, 233(5318):339–41, oct 1971.

[80] Masashi Ichinose, Takeshi Nishiyasu, Elisabeth Lambert, and Bakeridi Heart. Arterial baroreﬂex control of muscle sympathetic nerve activity under orthostatic stress in humans. Frontiers in Physiology, 3(314):1–10, 2012.

[81] D L Janosik, H Genovely, C Fredman, and P Bjerregaard. Discrepancy between head- up tilt test results utilizing diﬀerent protocols in the same patient. American heart journal, 123(2):538–41, feb 1992.

[82] J R C Jansen, J J Schreuder, J P Mulier, N T Smith, J J Settels, and K H Wesseling. A comparison of cardiac output derived from the arterial pressure wave against ther- modilution in cardiac surgery patients. British Journal of Anaesthesia, 87(2):212–222, 2001.

[83] Soﬁe Jansen, Jaspreet Bhangu, Sophia de Rooij, Joost Daams, Rose Anne Kenny, and Nathalie van der Velde. The Association of Cardiovascular Disorders and Falls: A Sys- tematic Review. Journal of the American Medical Directors Association, 17(3):193– 199, 2016.

[84] D L Jardine. Vasovagal syncope: new physiologic insights. Cardiology clinics, 31(1):75–87, feb 2013.

[85] Michael J Joyner, Jill N Barnes, Emma C Hart, B Gunnar Wallin, and Nisha Chark- oudian. Neural Control of the Circulation : How Sex and Age Diﬀerences Interact in Humans. Comprehensive Physiology, 5(January):193–215, 2015.

[86] W N Kapoor and N B Brant. Evaluation of Syncope by Upright Tilt Testing with Isoproterenol: A Nonspeciﬁc Test. Annals of Internal Medicine, 116:358–363, 1992.

[87] R A Kenny, A Ingram, J Bayliss, and R Sutton. Head-up tilt: a useful test for investigating unexplained syncope. Lancet (London, England), 1(8494):1352–5, jun 1986.

[88] R A Kenny, C C Lyon, A M Ingram, J Bayliss, S L Lightman, and R Sutton. Enhanced vagal activity and normal arginine vasopressin response in carotid sinus syndrome: implications for a central abnormality in carotid sinus hypersensitivity. Cardiovascular research, 21(7):545–50, jul 1987.

[89] R a Kenny, F E Shaw, J T O’Brien, P H Scheltens, R Kalaria, and C Ballard. Carotid sinus syndrome is common in dementia with Lewy bodies and correlates with deep white matter lesions. Journal of neurology, neurosurgery, and psychiatry, 75(7):966– 71, 2004.

147 [90] R A Kenny, D O Shea, S W Parry, Victoria Wing, Victoria Infirmary, and Queen Vic- toria Road. The Newcastle protocols for head-up tilt table testing in the diagnosis of vasovagal syncope , carotid sinus hypersensitivity , and related disorders. Heart, 83:564–569, 2000. [91] Simon R J Kerr, Mark S Pearce, Carol Brayne, Richard J Davis, and Rose Anne Kenny. Carotid sinus hypersensitivity in asymptomatic older persons: implications for diagnosis of syncope and falls. Archives of internal medicine, 166(5):515–20, mar 2006. [92] Areum Kim, Shekhar H Deo, Lauro C Vianna, George M Balanos, Doreen Hartwich, James P Fisher, and Paul J Fadel. Sex differences in carotid baroreflex control of arterial blood pressure in humans : relative contribution of cardiac output and total vascular conductance. Am J Physiol Heart Circ Physiol, 301:2454–2465, 2011. [93] Barry J. Kneale, Philip J. Chowienczyk, Sally E. Brett, D. John Coltart, and James M. Ritter. Gender differences in sensitivity to adrenergic agonists of forearm resistance vasculature. Journal of the American College of Cardiology, 36(4):1233–1238, 2000. [94] Panagiotis Kougias, Sarah M Weakley, Qizhi Yao, Peter H Lin, and Changyi Chen. Arterial baroreceptors in the management of systemic hypertension. Med Sci Monit, 16(1):1–15, 2010. [95] C T Paul Krediet, David L Jardine, and Wouter Wieling. Dissection of carotid sinus hypersensitivity: the timing of vagal and vasodepressor effects and the effect of body position. Clinical science (London, England : 1979), 121:389–396, 2011. [96] Paul Krediet, Steve W Parry, David L Jardine, David G Benditt, Michele Brignole, and Wouter Wieling. The history of diagnosing carotid sinus hypersensitivity: why are the current criteria too sensitive? Europace : European pacing, arrhythmias, and cardiac electrophysiology : journal of the working groups on cardiac pacing, arrhythmias, and cardiac cellular electrophysiology of the European Society of Cardiology, 13(1):14–22, jan 2011. [97] Tom A Kuusela, Timo J Kaila, and Mika Kähönen. Fine structure of the low-frequency spectra of heart rate and blood pressure. BMC physiology, 3:11, oct 2003. [98] M T La Rovere, J T Bigger, F I Marcus, A Mortara, and P J Schwartz. Baroreflex sensitivity and heart-rate variability in prediction of total cardiac mortality after myocardial infarction. ATRAMI (Autonomic Tone and Reflexes After Myocardial Infarction) Investigators. Lancet (London, England), 351:478–84, 1998. [99] Maria Teresa La Rovere, Gian Domenico Pinna, and Grzegorz Raczak. Baroreflex sensitivity: measurement and clinical implications. Annals of noninvasive electrocardiology : the official journal of the International Society for Holter and Noninvasive Electrocardiology, Inc, 13(2):191–207, apr 2008. [100] Gustavo de Castro Lacerda, Roberto Coury Pedrosa, Renato Côrtes de Lacerda, Marcela Cedenilla dos Santos, Maurício de Andrade Perez, Alfredo Brasil Teixeira, and Aristarco Gonçalves de Siqueira-Filho. Cardioinhibitory carotid sinus hypersensitivity: prevalence and predictors in 502 outpatients. Arquivos brasileiros de cardi- ologia, 90(3):148–55, mar 2008.

148 [101] Tomi Laitinen, Juha Hartikainen, Esko Vanninen, Leo Niskanen, Ghislaine Geelen, and Esko Länsimies. Age and gender dependency of baroreflex sensitivity in healthy subjects. Journal of Applied Physiology, 84(2):576–583, 1998. [102] Tomi Laitinen, Leo Niskanen, Ghislaine Geelen, Esko Länsimies, and Juha Har- tikainen. Age dependency of cardiovascular autonomic responses to head-up tilt in healthy subjects. Journal of applied physiology, 96:2333–2340, 2004. [103] L E Lamb, H C Green, J J Combs, S A Cheeseman, and J Hammond. Incidence of loss of consciousness in 1,980 Air Force personnel. Aerospace medicine, 31:973–88, dec 1960. [104] N A Lassen. Cerebral blood flow and oxygen consumption in man. Physiological reviews, 39(2):183–238, apr 1959. [105] Dominique Laude, Jean-Luc Elghozi, Arlette Girard, Elisa-Beth Bellard, Malika Bouhaddi, Paolo Castiglioni, Catherine Cerutti, Andrei Cividjian, Marco Di Rienzo, Jacques-Olivier Fortrat, Ben Janssen, John M Karemaker, Georges Lefthériotis, Gi- anfranco Parati, Pontus B Persson, Alberto Porta, Luc Quintin, Jacques Regnard, Heinz Rüdiger, and Harald M Stauss. Comparison of various techniques used to estimate spontaneous baroreflex sensitivity (the EuroBaVar study). American journal of physiology. Regulatory, integrative and comparative physiology, 286(1):R226–R231, 2004. [106] Alexandre Laurin, Matthew G. Lloyd, Tesshin Hachiya, Mitsuru Saito, Victoria E. Claydon, and Andrew Blaber. New indices from microneurography to investigate the arterial baroreflex. Physiological Reports, 5(12):1–11, 2017. [107] P Lelorier, G J Klein, A Krahn, R Yee, A Skanes, and J K Shoemaker. Combined Head-Up Tilt and Lower Body Negative Pressure as an Experimental Model of Or- thostatic Syncope. J Cardiovasc.Electrophysiol., 14:920–924, 2003. [108] A B Levin. A simple test of cardiac function based upon the heart rate changes induced by the Valsalva maneuver. The American journal of cardiology, 18(1):90–9, jul 1966. [109] Johannes J Van Lieshout. Exercise training and orthostatic intolerance : a paradox ? Journal of Physiology, 551.2:401, 2003. [110] Roberta Lima, Marion Wofford, and Jane F Reckelhoff. Hypertension in Post- menopausal Women. Curr Hypertens Rep, 14(3):254–260, 2012. [111] Ruth D. Lipman, Julie K. Salisbury, and J. Andrew Taylor. Spontaneous indices are inconsistent with arterial baroreflex gain. Hypertension, 42(4 I):481–487, 2003. [112] MG Lloyd, JM Wakeling, MS Koehle, RJ Drapala, and VE Claydon. Carotid sinus hypersensitivity: block of the sternocleidomastoid muscle does not affect responses to carotid sinus massage in healthy young adults. Physiological reports, 5(19):1–9, 2017. [113] S W Lord, R H Clayton, M C Hall, J C Gray, A Murray, J M McComb, and Rose Anne Kenny. Reproducibility of three different methods of measuring baroreflex sensitivity in normal subjects. Clinical science (London, England : 1979), 95(5):575–81, nov 1998.

149 [114] Phillip A. Low and Victoria A. Tomalia. Orthostatic hypotension: Mechanisms, causes, management. Journal of Clinical Neurology (Korea), 11(3):220–226, 2015.

[115] Daniela Lucini, Carlo Palombo, Mara Malacarne, and Massimo Pagani. Relation- ship between carotid artery mechanics and the spontaneous baroreﬂex. Journal of Hypertension, 30:1809–1816, 2012.

[116] D A Ludwig, J Vernikos, C E Wade, and V A Convertino. Blood pressure changes during orthostatic stress: evidence of gender diﬀerences in neuroeﬀector distribution. Aviation, space, and environmental medicine, 72(10):892–8, oct 2001.

[117] Kenneth M Madden, Chris Lockhart, and Karim Khan. Arterial stiﬀness and the response to carotid sinus massage in older adults. Aging clinical and experimental research, 22(1):36–41, 2010.

[118] Roberto Maestri, Gian Domenico Pinna, Andrea Mortara, Maria Teresa La Rovere, and Luigi Tavazzi. Assessing Baroreﬂex Sensitivity in Post-Myocardial Infarction Patients: Comparison of Spectral and Phenylephrine Techniques. Journal of the American College of Cardiology, 31(2):344–351, feb 1998.

[119] Alberto Malliani. Heart rate variability: from bench to bedside. European journal of internal medicine, 16(1):12–20, feb 2005.

[120] Alberto Malliani, Massimo Pagani, Federico Lombardi, and Sergio Cerutti. Cardiovas- cular Neural Regulation Explored in the Frequency Domain. Circulation, 84(2):482– 493, 1991.

[121] Giuseppe Mancia. Inﬂuence of Carotid Baroreceptors on Vascular Responses to Carotid Chemoreceptor Stimulation in the Dog. Circulation Research, 36:270–277, 1975.

[122] G J Martin, S L Adams, H G Martin, J Mathews, D Zull, and P J Scanlon. Prospective evaluation of syncope. Annals of emergency medicine, 13(7):499–504, jul 1984.

[123] S J McIntosh, J Lawson, and R A Kenny. Intravenous cannulation alters the speciﬁcity of head-up tilt testing for vasovagal syncope in elderly patients. Age and ageing, 23(4):317–9, jul 1994.

[124] Marte R Mellingsaeter, Mellingsæter V Wyller, Torgeir B Wyller, and Anette H Ran- hoﬀ. Gender diﬀerences in orthostatic tolerance in the elderly. Aging clinical and experimental research, 25:659–665, 2013.

[125] Milos Milic, Ping Sun, Fujun Liu, Claudia Fainman, Joel Dimsdale, Paul J Mills, and Michael G Ziegler. A comparison of pharmacologic and spontaneous baroreﬂex methods in aging and hypertension. Journal of hypertension, 27(6):1243–51, jun 2009.

[126] V. M. Miller, R. a. Kenny, J. Y. Slade, a. E. Oakley, and R. N. Kalaria. Medullary autonomic pathology in carotid sinus hypersensitivity. Neuropathology and Applied Neurobiology, 34(4):403–411, 2008.

150 [127] K D Monahan, F A Dinenno, D R Seals, C M Clevenger, C A Desouza, and H Tanaka. Age-associated changes in cardiovagal baroreflex sensitivity are related to central arterial compliance. American journal of physiology. Heart and circulatory physiology, 281:H284–9, 2001. [128] Kevin D Monahan. Effect of aging on baroreflex function in humans. American Journal Of Physiology, 293:3–12, 2007. [129] C A Morley, T C Dehn, E J Perrins, S L Chan, and R Sutton. Baroreflex sensitivity measured by the phenylephrine pressor test in patients with carotid sinus and sick sinus syndromes. Cardiovascular research, 18(12):752–61, dec 1984. [130] Angel Moya, Richard Sutton, Fabrizio Ammirati, Jean-Jacques Blanc, Michele Brig- nole, Johannes B Dahm, Jean-Claude Deharo, Jacek Gajek, Knut Gjesdal, Andrew Krahn, Martial Massin, Mauro Pepi, Thomas Pezawas, Ricardo Ruiz Granell, Fran- cois Sarasin, Andrea Ungar, J Gert van Dijk, Edmond P Walma, and Wouter Wieling. Guidelines for the diagnosis and management of syncope (version 2009). European heart journal, 30(21):2631–71, nov 2009. [131] Krzysztof Narkiewicz, Bradley G. Phillips, Masahiko Kato, Dagmara Hering, Leszek Bieniaszewski, and Virend K. Somers. Gender-selective interaction between aging, blood pressure, and sympathetic nerve activity. Hypertension, 45(4):522–525, 2005. [132] Jean-Pierre Ndayisaba, Alessandra Fanciulli, Roberta Granata, Susanne Duerr, Flo- rian Hintringer, Georg Goebel, Florian Krismer, and Gregor K. Wenning. Sex and age effects on cardiovascular autonomic function in healthy adults. Clinical autonomic research : official journal of the Clinical Autonomic Research Society, 25(5):317–326, 2015. [133] Philip C Njemanze. Critical limits of pressure-flow relation in the human brain PC Njem. Stroke, 23:1743–1747, 1992. [134] D O’Mahony. Pathophysiology of carotid sinus hypersensitivity in elderly patients. Lancet, 346(8980):950–2, oct 1995. [135] S Omboni, G Parati, M Di Rienzo, W Wieling, and G Mancia. Blood pressure and heart rate variability in autonomic disorders: a critical review. Clinical autonomic research : official journal of the Clinical Autonomic Research Society, 6(3):171–82, jun 1996. [136] M Pagani, F Lombardi, S Guzzetti, O Rimoldi, R Furlan, P Pizzinelli, G Sandrone, G Malfatto, S Dell’Orto, and E Piccaluga. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circulation research, 59(2):178–93, aug 1986. [137] M. Pagani, V. Somers, R. Furlan, S. Dell’Orto, J. Conway, G. Baselli, S. Cerutti, P. Sleight, and Alberto Malliani. Changes in autonomic regulation induced by physical training in mild hypertension. Hypertension, 12(6):600–610, dec 1988. [138] H A Palmero, T F Caeiro, D J Iosa, and J Bas. Baroreceptor Reflex Sensitivity Index Derived from Phase 4 of the Valsalva Maneuver. Hypertension, 3(6 Pt 2):134–137, 1981.

151 [139] G Parati, J P Saul, M Di Rienzo, and G Mancia. Spectral analysis of blood pressure and heart rate variability in evaluating cardiovascular regulation. A critical appraisal. Hypertension, 25(6):1276–86, jun 1995.

[140] J Parlow, J P Viale, G Annat, R Hughson, and L Quintin. Spontaneous cardiac barore- ﬂex in humans. Comparison with drug-induced responses. Hypertension, 25(5):1058– 68, may 1995.

[141] S W Parry, M Baptist, J J Gilroy, N Steen, and R a Kenny. Central alpha2 adreno- ceptors and the pathogenesis of carotid sinus hypersensitivity. Heart (British Cardiac Society), 90(8):935–936, 2004.

[142] S W Parry, D Richardson, D O’Shea, B Sen, and R A Kenny. Diagnosis of carotid sinus hypersensitivity in older adults: carotid sinus massage in the upright position is essential. Heart (British Cardiac Society), 83(1):22–3, jan 2000.

[143] Eric C. Peterson, Zhengfeng Wang, and Gavin Britz. Regulation of cerebral blood ﬂow. International Journal of Vascular Medicine, 2011, 2011.

[144] Gian Domenico Pinna, Roberto Maestri, Grzegorz Raczak, and Maria Teresa La Ro- vere. Measuring baroreﬂex sensitivity from the gain function between arterial pressure and heart period. Clinical science, 103(1):81–8, jul 2002.

[145] B Pomeranz, R J Macaulay, M A Caudill, I Kutz, D Adam, D Gordon, K M Kilborn, A C Barger, D C Shannon, and R J Cohen. Assessment of autonomic function in humans by heart rate spectral analysis. The American journal of physiology, 248(1 Pt 2):H151–3, jan 1985.

[146] J T Potts and P B Raven. Eﬀect of dynamic exercise on human carotid-cardiac baroreﬂex latency. American Journal of Physiology - Heart and Circulatory Physiol- ogy, 37(3):H1208–H1214, 1995.

[147] Clare L Protheroe, Henrike Rianne J C Ravensbergen, Jessica A Inskip, and Victoria Claydon. Tilt testing with combined lower body negative pressure: a "gold standard" for measuring orthostatic tolerance. Journal of visualized experiments : JoVE, (73):e4315, jan 2013.

[148] R G Querry, S A Smith, M Strømstad, K Ide, N H Secher, and P B Raven. Anatom- ical and functional characteristics of carotid sinus stimulation in humans. American journal of physiology. Heart and circulatory physiology, 280(5):H2390–8, may 2001.

[149] Satish R. Raj. Postural tachycardia syndrome (POTS). Circulation, 127(23):2336– 2342, 2013.

[150] M Ranucci, A Porta, V Bari, V Pistuddi, and M T La Rovere. Baroreﬂex sensitivity and outcomes following coronary surgery. PLoS ONE, 12(4):1–11, 2017.

[151] Chester A. Ray and Kevin D. Monahan. The vestibulosympathetic reﬂex in humans: Neural interactions between cardiovascular reﬂexes. Clinical and Experimental Phar- macology and Physiology, 29(1-2):98–102, 2002.

152 [152] H. W. Robbe, L. J. Mulder, H. Ruddel, W. Langewitz, J. B. Veldman, and G. Mulder. Assessment of baroreceptor reﬂex sensitivity by means of spectral analysis. Hyperten- sion, 10(5):538–543, nov 1987.

[153] Carlo Rostagno, Massimo Felici, Sabina Caciolli, Giuseppe Olivo, Marco Comeglio, Giorgio Galanti, and Gian Gastone Neri Serneri. Decreased baroreﬂex sensitivity assessed from phase IV of valsalva maneuver in mild congestive heart failure. Angiology, 50(8):655–664, 1999.

[154] L Rudas, A A Crossman, C A Morillo, J R Halliwill, K U Tahvanainen, T A Kuusela, and D L Eckberg. Human sympathetic and vagal baroreﬂex responses to sequential nitroprusside and phenylephrine. The American journal of physiology, 276(5 Pt 2):H1691–8, may 1999.

[155] Martin Huth Ruwald, Morten Lock Hansen, Morten Lamberts, Carolina Malta Hansen, Michael Vinther Højgaard, Lars Køber, Christian Torp-Pedersen, Jim Hansen, and Gunnar Hilmar Gislason. The relation between age, sex, comorbidity, and pharmacotherapy and the risk of syncope: a Danish nationwide study. Europace : European pacing, arrhythmias, and cardiac electrophysiology : journal of the working groups on cardiac pacing, arrhythmias, and cardiac cellular electrophysiology of the European Society of Cardiology, 14(10):1506–14, oct 2012.

[156] Dirk P Saal and J Gert van Dijk. Classifying syncope. Autonomic neuroscience : basic & clinical, 184:3–9, sep 2014.

[157] P Sándor. Nervous control of the cerebrovascular system: doubts and facts. Neuro- chemistry international, 35(3):237–59, sep 1999.

[158] R. Schondorf and Phillip A. Low. Gender related diﬀerences in the cardiovascular responses to upright tilt in normal subjects. Clinical Autonomic Research, 2:183–187, 1992.

[159] Christine Schrezenmaier, Wolfgang Singer, Nicolette Muenter Swift, David Sletten, James Tanabe, and Phillip A. Low. Adrenergic and vagal baroreﬂex sensitivity in autonomic failure. Archives of Neurology, 64(3):381–386, 2007.

[160] Christopher E Schwartz and Julian M Stewart. The arterial baroreﬂex resets with orthostasis. Frontiers in physiology, 3(December):461, jan 2012.

[161] Win-Kuang Shen, Robert S. Sheldon, David G. Benditt, Mitchell I. Cohen, Daniel E. Forman, Zachary D. Goldberger, Blair P. Grubb, Mohamed H. Hamdan, Andrew D. Krahn, Mark S. Link, Brian Olshansky, Satish R. Raj, Roopinder Kaur Sandhu, Dan Sorajja, Benjamin C. Sun, and Clyde W. Yancy. 2017 ACC/AHA/HRS Guideline for the Evaluation and Management of Patients With Syncope: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society. Circulation, 136:e25–e59, 2017.

[162] J. K. Shoemaker, C. S. Hogeman, M. Khan, D. S. Kimmerly, and L. I. Sinoway. Gender aﬀects sympathetic and hemodynamic response to postural stress. American journal of physiology. Heart and circulatory physiology, 281(5):H2028–H2035, 2001.

153 [163] P E Shrout and J L Fleiss. Intraclass correlations: uses in assessing rater reliability. Psychological bulletin, 86(2):420–8, mar 1979.

[164] J.O. Skarphedinsson, M. Elam, L. Jungersten, and B.G. Wallin. Sympathetic nerve traﬃc correlates with the release of nitric oxide in humans: Implications for blood pressure control. Journal of Physiology, 501(3):671–675, 1997.

[165] Ioanna Skrapari, Nicholas Tentolouris, and Nicholas Katsilambros. Baroreﬂex function: determinants in healthy subjects and disturbances in diabetes, obesity and metabolic syndrome. Current diabetes reviews, 2(3):329–38, aug 2006.

[166] H. S. Smyth, P. Sleight, and G. W. Pickering. Reﬂex Regulation of Arterial Pressure during Sleep in Man: A Quantitative Method of Assessing Baroreﬂex Sensitivity. Circulation Research, 24(1):109–121, jan 1969.

[167] Monica Solbiati, Robert Sheldon, and Colette Seifer. Managing syncope in the elderly: The not so simple faint in aging patients. Canadian Journal of Cardiology, 32:1124– 1131, 2016.

[168] Statistics Canada. Population Projections for Canada, Provinces and Territories. 2009-2036. Number 91. 2010.

[169] Statistics Canada. Annual Demographic Estimates : Canada, Provinces and Territo- ries. (91):1–179, 2015.

[170] P M Stevens. Cardiovascular dynamics during orthostasis and the inﬂuence of in- travascular instrumentation. The American journal of cardiology, 17(2):211–8, feb 1966.

[171] P M Stevens and L E Lamb. Eﬀects of lower body negative pressure on the cardiovascular system. The American journal of cardiology, 16(4):506–15, oct 1965.

[172] Sarah L Stevens, Sally Wood, Constantinos Koshiaris, Kathryn Law, Paul Glasziou, Richard J Stevens, and Richard J McManus. Blood pressure variability and cardiovascular disease: systematic review and meta-analysis. British Medical Journal, 354(i4098):1–8, 2016.

[173] Julian M. Stewart, Marvin S. Medow, Richard Sutton, Paul Visintainer, David L. Jar- dine, and Wouter Wieling. Mechanisms of vasovagal syncope in the young: Reduced systemic vascular resistance versus reduced cardiac output. Journal of the American Heart Association, 6(1):1–12, 2017.

[174] Y Sugiyama, T Matsukawa, A S Shamsuzzaman, H Okada, T Watanabe, and T Mano. Delayed and diminished pressor response to muscle sympathetic nerve activity in the elderly. J Appl Physiol, 80(3):869–875, 1996.

[175] Benjamin C. Sun. Quality-of-Life, Health Service Use, and Costs Associated With Syncope. Progress in Cardiovascular Diseases, 55:370–375, 2013.

[176] Benjamin C. Sun, Jennifer A. Emond, and Carlos A. Camargo. Direct medical costs of syncope-related hospitalizations in the United States. American Journal of Cardi- ology, 95:668–671, 2005.

154 [177] G Sundlöf and B G Wallin. Human Muscle Nerve Sympathetic Activity at Rest. Relationship to Blood Pressure and Age. Journal of Physiology, 274:621–637, 1978.

[178] Richard Sutton. Carotid sinus syndrome : Progress in understanding and management. Global Cardiology Science and Practice, 18:1–8, 2014.

[179] C Swenne. Baroreﬂex sensitivity: mechanisms and measurement. Netherlands heart journal : monthly journal of the Netherlands Society of Cardiology and the Nether- lands Heart Foundation, 21(2):58–60, feb 2013.

[180] Marek Sykora, Jennifer Diedler, Peter Turcani, Werner Hacke, and Thorsten Steiner. Baroreﬂex: A new therapeutic target in human stroke? Stroke, 40:678–683, 2009.

[181] Maw Pin Tan, Rose Anne M Kenny, Tom J Chadwick, Simon R J Kerr, and Steve W Parry. Carotid sinus hypersensitivity: disease state or clinical sign of ageing? Insights from a controlled study of autonomic function in symptomatic and asymptomatic subjects. Europace : European pacing, arrhythmias, and cardiac electrophysiology : journal of the working groups on cardiac pacing, arrhythmias, and cardiac cellular electrophysiology of the European Society of Cardiology, 12(11):1630–6, dec 2010.

[182] Indu Taneja, Christopher Moran, Marvin S Medow, June L Glover, Leslie D. Mont- gomery, and Julian M Stewart. Diﬀerential eﬀects of lower body negative pressure and upright tilt on splanchnic blood volume. Am J Physiol Heart Circ Physiol, 292:1420– 1426, 2007.

[183] Task Force. Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Task Force of The European Society of Cardiology and The North American Society of Pacing and Electrophysiology. European heart journal, 17:354–381, 1996.

[184] J. A. Taylor, D. L. Carr, C. W. Myers, and D. L. Eckberg. Mechanisms Underlying Very-Low-Frequency RR-Interval Oscillations in Humans. Circulation, 98(6):547–555, aug 1998.

[185] S H Tea, J Mansourati, G L’Heveder, D Mabin, and J J Blanc. New insights into the pathophysiology of carotid sinus syndrome. Circulation, 93(7):1411–6, apr 1996.

[186] Dick H J Thijssen, Mark A Black, Kyra E Pyke, Jaume Padilla, Greg Atkinson, Ryan A Harris, Beth Parker, Michael E Widlansky, Michael E Tschakovsky, and Daniel J Green. Assessment of ﬂow-mediated dilation in humans : a methodological and physiological guideline. Am J Physiol Heart Circ Physiol, 300:2–12, 2011.

[187] Andrea Ungar, Martina Rafanelli, Iacopo Iacomelli, Maria Angela Brunetti, Alice Ceccoﬁglio, Francesca Tesi, and Niccolò Marchionni. Fall prevention in the elderly. Clinical Cases in Mineral and Bone Metabolism, 10(6):91–95, 2013.

[188] United States Census Bureau. United States Census Bureau World Population Sum- mary, 2017.

[189] M.E.A.M. van Kleef, M.C. Bates, and W. Spiering. Endovascular Baroreﬂex Ampli- ﬁcation for Resistant Hypertension. Current Hypertension Reports, 20(5):1–9, 2018.

155 [190] L. C. Vianna, E. C. Hart, S. T. Fairfax, N. Charkoudian, M. J. Joyner, and P. J. Fadel. Inﬂuence of age and sex on the pressor response following a spontaneous burst of muscle sympathetic nerve activity. AJP: Heart and Circulatory Physiology, 302(11):H2419–H2427, 2012.

[191] E. R. Vogel, P. Sandroni, and P. A. Low. Blood pressure recovery from Valsalva maneuver in patients with autonomic failure. Neurology, 65(10):1533–1537, 2005.

[192] Naoki Wada, Wolfgang Singer, Tonette L. Gehrking, David M. Sletten, James D. Schmeizer, and Phillip A. Low. Comparison of Baroreﬂex Sensitivity to Fall and Rise in Blood Pressure Induced by the Valsalva Maneuver. Clinical science (London), 127(5):307–313, 2014.

[193] M M Wahbha, C A Morley, Y M Al-Shamma, and R Hainsworth. Cardiovascular reﬂex responses in patients with unexplained syncope. Clinical science (London, England : 1979), 77(5):547–53, dec 1989.

[194] L L Watkins, P Grossman, and A Sherwood. Noninvasive assessment of baroreﬂex control in borderline hypertension. Comparison with the phenylephrine method. Hy- pertension, 28(2):238–43, aug 1996.

[195] Lana L. Watkins, Claudia Fainman, Joel Dimsdale, and Michael G. Ziegler. Assess- ment of baroreﬂex control from beat-to-beat blood pressure and heart rate changes: A validation study. Psychophysiology, 32(4):411–414, jul 1995.

[196] Joseph P Weir. Quantifying test-retest reliability using the intraclass correlation coeﬃcient and the SEM. Journal of strength and conditioning research / National Strength & Conditioning Association, 19(1):231–40, feb 2005.

[197] Rachel Wells, Andrew J. Spurrier, Dominik Linz, Celine Gallagher, Rajiv Mahajan, Prashanthan Sanders, Amanda Page, and Dennis H. Lau. Postural tachycardia syndrome: Current perspectives. Vascular Health and Risk Management, 14:1–11, 2018.

[198] K H Wesseling, J R C Jansen, J J Settels, and JJ Schreuder. Computation of aortic ﬂow from pressure in humans using a nonlinear, three-element model. J Appl Physiol, 74(5):2566–2573, 1993.

[199] D D White and L D Montgomery. Pelvic blood pooling of men and women during lower body negative pressure. Aviation, space, and environmental medicine, 67(6):555–9, jun 1996.

[200] W Wieling, C T P Krediet, D Solari, F J de Lange, N van Dijk, R D Thijs, J G van Dijk, M Brignole, and D L Jardine. At the heart of the arterial baroreﬂex: a physiological basis for a new classiﬁcation of carotid sinus hypersensitivity. Journal of internal medicine, 273(4):345–58, apr 2013.

[201] Wouter Wieling, Frederik J de Lange, and David L Jardine. The heart cannot pump blood that it does not receive. Frontiers in physiology, 5:360, jan 2014.

[202] R. Willeput, C. Rondeux, and A. De Troyer. Breathing aﬀects venous return from legs in humans. Journal of Applied Physiology, 57(4):971–976, oct 1984.

156 [203] World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA : the journal of the American Medical Association, 310(2191-2194), 2013.

[204] Fumio Yamazaki and Ryoko Sone. Whole-body heating slows carotid baroreﬂex response in human subjects. European Journal of Applied Physiology, 94(5-6):690–696, 2005.

[205] H. Yang, W. H. Cooke, K. S. Reed, and J. R. Carter. Sex diﬀerences in hemodynamic and sympathetic neural ﬁring patterns during orthostatic challenge in humans. Journal of Applied Physiology, 112:1744–1751, 2012.

[206] B. J. Yates, Philip S. Bolton, and Vaughan G. Maceﬁeld. NIH Public Access. Compr Physiol, 4(2):851–887, 2014.

[207] B Silvano Zanutto, Max E Valentinuzzi, and Enrique T Segura. Neural set point for the control of arterial pressure: role of the nucleus tractus solitarius. Biomedical engineering online, 9:4, jan 2010.

157