Endothelial Injury and Repair in Vasculitis of the Young

University College London

Endothelial injury and repair in vasculitis of the young

Lindsey Anne Clarke

Submitted for the degree of Doctor of Philosophy

Department of Infectious Diseases/Department of Rheumatology

Institute of Child Health, University College London

I, Lindsey Anne Clarke confirm that the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis.

Acknowledgements

I would like to express my thanks to the many people too numerous to mention individually who contributed in some way to this study by their guidance, support, understanding and willingness to donate blood. Without the help of the patients, their families and the clinical staff at Great Ormond Street Hospital this project could not have been carried out – so a big thankyou to you all.

Special thanks must go to my supervisors at ICH Paul Brogan and Nigel Klein, who have kept me focused on the task in hand throughout this project. To Francesca

Arrigoni for her excellent guidance in the lab and all round support throughout the project – a true mentor. To Despina Eleftheriou and Ying Hong who have had the reins of the project passed to them – best of luck for the continuation of the work! To

Markella Ponticos for her help proofreading this manuscript.

The Oliver Bird Rheumatism Programme, which has generously funded me throughout and the head of the programme at UCL, David Isenberg have been a continual source of support and I thank them greatly for the opportunities they have provided. In particular

I thank them for introducing me to my fellow birdies, who I hope will remain good friends for a long time to come.

Finally, I would like to thank my family for understanding my need to be a perpetual student and most importantly Andrew, who knew I would get to the end of this project one day.

Lindsey Anne Clarke, August 2009

Thesis Abstract

The vasculitides are a wide spectrum of disorders which are characterised by vascular inflammation. Endothelial injury can occur as a consequence of inappropriate inflammation and is central to the pathogenesis of these varied diseases. This thesis documents the development of assays for detection of novel biomarkers of endothelial injury and/or activation and subsequent reparative responses in children with primary systemic vasculitis. It focuses in particular on circulating endothelial cells, cellular microparticles, growth factors involved in angiogenesis/vasculogenesis and endothelial progenitor cells.

Circulating endothelial cells are mature endothelial cells which have become detached from the vessel wall and represent a highly damaged vasculature and were found to be significantly higher in children with active primary systemic vasculitis compared to healthy child controls and patients in remission. Microparticles are released from activated cells, including the endothelium and leukocytes. In this study endothelial and monocyte derived microparticles were found to be elevated during active vasculitis.

Growth factors released in response to endothelial injury regulate reparative responses, of which endothelial progenitor cells may play a key role. In this study, patients at disease onset prior to treatment were found to have significantly higher levels of growth factors and endothelial progenitor cells, which decreased with remission inducing therapy. Overall this thesis has investigated the changes in these interlinked biomarkers of injury and repair during active disease, remission and disease flare.

Table of Contents

ACKNOWLEDGEMENTS ...... 3

THESIS ABSTRACT ...... 4

TABLE OF CONTENTS ...... 5

FIGURES ...... 11

TABLES ...... 13

ABBREVIATIONS ...... 14

1 INTRODUCTION ...... 16

1.1 THE ENDOTHELIUM AND ITS FUNCTIONS ...... 17

1.1.1 Control of vessel tone ...... 17

1.1.2 Fluid and solute exchange ...... 18

1.1.3 Haemostasis ...... 19

1.1.4 Angiogenesis and Vasculogenesis ...... 20

1.1.5 Inflammatory responses and the endothelium ...... 22

1.1.6 Conclusions ...... 24

1.2 VASCULITIS – A CLINICAL PERSPECTIVE ...... 24

1.2.1 Primary systemic vasculitis ...... 24

1.2.2 Polyarteritis Nodosa and associated diseases ...... 26

1.2.3 Wegener’s Granulomatosis ...... 27

1.2.4 Takayasu disease ...... 28

1.2.5 Kawasaki disease ...... 28

1.2.6 Henoch-Schönlein purpura ...... 29

1.2.7 Churg-Strauss syndrome ...... 29

1.2.8 Other vasculitic disorders ...... 30

1.2.9 Treating vasculitis ...... 34

1.2.10 Prognosis ...... 35

1.2.11 Conclusions relating to clinical perspectives on vasculitis ...... 36

1.3 THE PATHOGENESIS OF VASCULITIS ...... 36

1.3.1 Environmental triggers ...... 37

1.3.2 Genetics of vasculitis ...... 38

1.3.3 Immunological pathways in vasculitis ...... 40

1.3.4 Conclusions regarding the pathogenesis of vasculitis ...... 52

1.4 ASSESSMENT OF DISEASE ACTIVITY IN VASCULITIS – THE NEED FOR RELIABLE BIOMARKERS ...... 52

1.4.1 Routine clinical tests for diagnosis and assessment of disease activity ...... 53

1.4.2 Novel Biomarkers in vasculitis ...... 57

1.5 AIMS OF THIS THESIS ...... 65

2 METHODS AND MATERIALS ...... 66

2.1 INTRODUCTION ...... 67

2.2 PATIENTS, VASCULITIS CLASSIFICATION AND ASSESSMENT OF DISEASE ACTIVITY ...... 67

2.2.1 Classification of the vasculitic syndromes ...... 67

2.2.2 Assessment of disease activity ...... 68

2.3 MATERIALS...... 69

2.3.1 Reagents for blood collection ...... 69

2.3.2 Fluorochrome conjugated antibodies for flow cytometry...... 69

2.3.3 Tissue culture media ...... 71

2.3.4 Cytokine and protein reagents ...... 72

2.3.5 Bead based extraction reagents ...... 72

2.3.6 ELISA kits ...... 72

2.3.7 Other reagents ...... 72

2.4 PREPARATION OF BLOOD AND TISSUE SAMPLES ...... 73

2.4.1 HUVEC ...... 73

2.4.2 Isolation of PBMCS ...... 77

2.4.3 Isolation of neutrophils ...... 77

2.4.4 Separation of platelet poor plasma ...... 78

2.4.5 Separation of serum ...... 78

2.5 ASSESSMENT OF ENDOTHELIAL ACTIVATION AND INJURY ...... 79

2.5.1 Detection of circulating endothelial cells using immunomagnetic beads ...... 79

2.5.2 Detection of circulating endothelial cells using flow cytometry ...... 80

2.5.3 Detection of cellular microparticles ...... 83

2.5.4 Detection of activated parent cells ...... 84

2.6 ASSESSMENT OF POTENTIAL ENDOTHELIAL REPAIR RESPONSES ...... 85

2.6.1 Measurement of endothelial growth factors ...... 85

2.6.2 Flow cytometry for detection of endothelial progenitor cells ...... 86

2.6.3 Endothelial progenitor cell colony culture ...... 89

2.6.4 Endothelial progenitor cell outgrowth culture ...... 90

2.6.5 Matrigel - assessment of endothelial progenitor cell function ...... 91

2.7 STATISTICS...... 93

3 CIRCULATING ENDOTHELIAL CELLS ...... 95

3.1 SUMMARY ...... 96

3.2 INTRODUCTION ...... 96

3.2.1 Detecting CECs ...... 96

3.2.2 Where do CECs come from and does this influence phenotype? ...... 100

3.3 DETECTION OF CECS USING IMMUNOMAGNETIC BEAD EXTRACTION ...... 104

3.3.1 Reproducibility of the technique ...... 104

3.3.2 Interobserver variability ...... 104

3.3.3 The effect of traumatic venepuncture ...... 106

3.3.4 The effect of storage and time to preparation on CEC enumeration ...... 108

3.4 DETECTING CECS USING FLOW CYTOMETRY ...... 110

3.4.1 The effect of lysis on staining of CD34/CD146/CD45 ...... 110

3.4.2 Using endothelial cells as a positive control for CECs ...... 112

3.4.3 Using counting beads to determine absolute cell number ...... 114

3.4.4 Reproducibility of flow cytometry for endothelial cell enumeration...... 115

3.5 CLINICAL STUDY COMPARING TWO DIFFERENT METHODS OF CEC QUANTIFICATION ...... 117

3.5.1 Patients and methods ...... 117

3.5.2 Results ...... 119

3.6 CONCLUSIONS OF METHODOLOGICAL EXPERIMENTS ...... 131

3.7 EVALUATION OF CECS IN VASCULITIS ...... 133

3.7.1 Patients and methods ...... 133

3.7.2 Results ...... 134

3.7.3 Correlation between CECs and disease activity ...... 138

3.7.4 CECs in individual vasculitic syndromes ...... 145

3.8 DISCUSSION ...... 147

4 CELLULAR MICROPARTICLES ...... 152

4.1 SUMMARY ...... 153

4.2 INTRODUCTION ...... 153

4.2.1 Formation of MPs ...... 153

4.2.2 Detection of MPs...... 154

4.2.3 Function of MPs ...... 155

4.2.4 Endothelial MPs ...... 155

4.2.5 Platelet MPs ...... 158

4.2.6 Leukocyte MPs – Neutrophils, monocytes and lymphocytes ...... 160

4.2.7 Microparticles in vasculitis ...... 162

4.3 DETECTION OF MPS – METHODOLOGICAL OPTIMISATION ...... 163

4.3.1 Preparation of samples ...... 163

4.3.2 Preparation of positive controls for setup experiments ...... 163

4.3.3 Staining of MPs for flow cytometry ...... 165

4.3.4 Flow cytometry of MPS ...... 165

4.3.5 The effect of plasma freeze-thaw on MP quantification ...... 176

4.3.6 Reproducibility of the technique from different plasma volumes ...... 177

4.4 CONCLUSIONS OF METHODOLOGICAL EXPERIMENTS ...... 180

4.5 EVALUATION OF CELLULAR MPS IN VASCULITIS ...... 181

4.5.1 Patients and methods ...... 181

4.5.2 Results ...... 182

4.5.3 Correlation between MPs and disease activity ...... 207

4.5.4 MPs in different vasculitic syndromes...... 209

4.6 DISCUSSION ...... 214

5 GROWTH FACTORS ...... 220

5.1 SUMMARY ...... 221

5.2 INTRODUCTION ...... 221

5.2.1 The VEGF family – the major angiogenic signalling pathway ...... 222

5.2.2 NOTCH-delta signaling – giving endothelial cells direction ...... 226

5.2.3 PDGF – stabilizing new vessels ...... 227

5.2.4 Angiopoietins ...... 228

5.2.5 Thymosin β4 ...... 230

5.3 EVALUATION OF GROWTH FACTORS IN VASCULITIS ...... 232

5.3.1 Patients and methods ...... 232

5.3.2 Results ...... 234

5.3.3 Correlation between growth factors and disease activity ...... 242

5.3.4 Influence of treatment on growth factor responses ...... 242

5.4 DISCUSSION ...... 248

6 ENDOTHELIAL PROGENITOR CELLS ...... 252

6.1 SUMMARY ...... 253

6.2 INTRODUCTION ...... 253

6.2.1 EPCs influence in vivo endothelial repair ...... 254

6.2.2 Detecting EPCs ...... 258

6.2.3 EPCs in health and disease ...... 264

6.2.4 EPCs in vasculitis ...... 264

6.3 OPTIMISATION OF FLOW CYTOMETRY FOR DETECTION OF EPCS ...... 265 9

6.3.1 Optimisation of antibody staining ...... 265

6.3.2 Sample storage - fresh/frozen PBMCS vs whole blood ...... 266

6.3.3 Flow cytometry – instrument settings and gates ...... 267

6.4 EPCS IN VASCULITIS ...... 275

6.4.1 Patients and Methods ...... 275

6.4.2 Results ...... 276

6.4.3 Correlation of EPCs with disease activity ...... 302

6.5 DETECTION OF EPCS USING CELL CULTURE ...... 306

6.5.1 Colony counting ...... 307

6.5.2 Matrigel ...... 309

6.6 DISCUSSION ...... 313

7 DISCUSSION AND CONCLUDING REMARKS ...... 319

DISCUSSION ...... 320

CONCLUDING REMARKS ...... 329

APPENDIX 1 - CLASSIFICATION AND DEFINITIONS OF VASCULITIS ...... 331

APPENDIX 2 - THE BIRMINGHAM VASCULITIS ACTIVITY SCORE ...... 336

APPENDIX 3 – FLOW CYTOMETRY PLANS ...... 338

APPENDIX 4 – MP GENERATION ...... 340

APPENDIX 5 – STANDARD GRAPHS FOR ELISAS ...... 345

APPENDIX 6 – CHANGES IN CECS AND MPS WITH DISEASE ACTIVITY IN INDIVIDUAL PATIENTS ...... 347

REFERENCES ...... 360

PUBLICATIONS FROM THIS THESIS ...... 404

Figures

Figure 2-1: Cultured HUVEC phenotype ...... 75 Figure 2-2: Methodology of Sandwich ELISA ...... 87 Figure 2-3: Methodology of Thymosin-β 4 competition ELISA...... 88 Figure 2-4: EPC colony culture ...... 92 Figure 2-5: Matrigel, incorporation of PBMCS into HUVEC tubules ...... 94 Figure 3-1: Flow cytometric (FC) and immunomagnetic bead (IBE) definitions of CECs ...... 101 Figure 3-2: Phenotype of CECs in vasculitis ...... 103 Figure 3-3: Reproducibility of IBE for detection of CECs ...... 105 Figure 3-4: Bland-Altman plots for testing interobserver variability of IBE for detection of CECs ...... 107 Figure 3-5: Factors influencing CEC counting – traumatic venepuncture...... 109 Figure 3-6: Factors influencing CEC counting – sample storage time and storage temperature ...... 111 Figure 3-7: The effect of red cell lysis on flow cytometric protocols for HUVEC detection ...... 113 Figure 3-8: Absolute enumeration of HUVEC spiked into whole blood ...... 116 Figure 3-9: Bland-Altman statistics for testing agreement between IBE and PBMC-FC for CEC enumeration...... 120 Figure 3-10: Comparison of cell size CECs vs ECs ...... 122 Figure 3-11: FSC/SSC characteristics of ECs and optimisation of flow cytometric gating to detect ECs .. 124 Figure 3-12: Bland-Altman analysis of agreement between EC-optimised FC and PBMC-FC and IBE for CEC enumeration ...... 126 Figure 3-13: The effect of EC number on recovery - sensitivity ...... 129 Figure 3-14: Resolution – sample dilution improves recovery ...... 130 Figure 3-15: The effect of lysis on CEC recovery ...... 132 Figure 3-16: CECs in vasculitis ...... 136 Figure 3-17: CECs in vasculitis – response to treatment ...... 137 Figure 3-18: Correlation with disease activity ...... 139 Figure 3-19: Patient with WG - CECs and disease activity – predictive of disease flare? ...... 141 Figure 3-20: Patient with WG – CECs and disease activity in orbital granuloma ...... 142 Figure 3-21: Patient with PAN – CECs and disease activity ...... 143 Figure 3-22: CECs during autologous stem cell transplantation ...... 144 Figure 3-23: Subdivision into individual diseases - CECs in different vasculitic conditions ...... 146 Figure 4-1: The removal of platelets for generation of platelet poor plasma ...... 164 Figure 4-2: Optimisation of MP gating using 0.3µm, 0.8µm and 3µm latex beads ...... 167 Figure 4-3: MP gating strategy, based on Annexin V positivity ...... 169 Figure 4-4: Recovery of 3µm counting beads...... 171 Figure 4-5: Detecting Flouresbrite beads – recovery of beads alone...... 173 Figure 4-6:Fluoresbrite beads in relation to MPs ...... 174 Figure 4-7:Using an FL1 threshold for MP enumeration ...... 175 Figure 4-8: The effect of freeze thaw on MP enumeration ...... 178 Figure 4-9: Reproducibility of MP enumeration ...... 179 Figure 4-10: Total AnV MPs ...... 184 Figure 4-11: MPs of endothelial origin – CD144 ...... 186 Figure 4-12: MPs of endothelial origin CD105 ...... 187 Figure 4-13: MPs of endothelial origin – E-selectin...... 189 Figure 4-14: MPs of endothelial origin – ICAM-1 ...... 191 Figure 4-15: MPs of platelet origin – CD42a ...... 193 Figure 4-16 MPs of platelet origin – P-selectin ...... 195 Figure 4-17: MPs of neutrophil origin – CD15 ...... 196 Figure 4-18: MPs of neutrophil origin – CD11b ...... 198 Figure 4-19: MPs of neutrophil origin – L-selectin ...... 200 Figure 4-20:MPs of neutrophil origin – CD66b ...... 202 Figure 4-21: MPs of monocyte origin – TF ...... 203 Figure 4-22: MPs of monocyte origin – CD14 ...... 205 Figure 4-23: MPs of lymphocyte origin – CD3 ...... 206

Figure 4-24: Patient with WG - MPs and disease activity – predictive of flare? ...... 210 Figure 4-25: Patient with WG - MPs and disease activity – localised flare ...... 211 Figure 4-26: Patient with PAN – MPs and disease activity ...... 212 Figure 4-27: EMPs in different vasculitis syndromes ...... 213 Figure 5-1: Serum levels of VEGF in vasculitis ...... 235 Figure 5-2:Plasma Angiopoeitin-1 levels in vasculitis ...... 237 Figure 5-3: Plasma Angiopoeitin-2 levels in vasculitis ...... 239 Figure 5-4: Serum levels of Thymosin Beta 4 in vasculitis ...... 241 Figure 5-5: Influence of treatment on endothelial injury markers and clinical parameters ...... 246 Figure 5-6: Influence of treatment on growth factors ...... 247 Figure 6-1: % of mononuclear cells expressing EPC markers in whole blood, fresh or frozen isolated PBMCs ...... 268 Figure 6-2: FSC SSC characteristics of CD34+, CD133+ and VEGFR+ cells ...... 270 Figure 6-3: EPC gating strategy ...... 271 Figure 6-4:CD34+ cells in vasculitis ...... 278 Figure 6-5: CD34+ cells in vasculitis – responses to treatment ...... 279 Figure 6-6: CD34+CD133+ cells in vasculitis ...... 282 Figure 6-7: CD34+CD133+ cells in vasculitis – responses to treatment ...... 283 Figure 6-8: CD34+VEGFR2+CD133+ cells in vasculitis ...... 285 Figure 6-9: CD34+VEGFR2+CD133+ cells in vasculitis ...... 287 Figure 6-10: CD34+VEGFR2+ cells in vasculitis ...... 289 Figure 6-11: CD34+VEGFR2+cells in vasculitis – responses to treatment ...... 291 Figure 6-12: CD34+CD144+ cells in vasculitis ...... 293 Figure 6-13: CD34+CD144+ cell in vascultis – response to treatment ...... 294 Figure 6-14 :CD34+CD144+CD133+ cells in vasculitis ...... 296 Figure 6-15: CD34+CD144+CD133+ cells in vasculitis – response to treatment ...... 298 Figure 6-16:Further CD34+ staining ...... 300 Figure 6-17: CD34+CD14+CD146+ cells in vasculitis ...... 301 Figure 6-18: EPC colony culture ...... 308 Figure 6-19: Matrigel EPC culture ...... 311 Figure 6-20: Matrigel assay results ...... 312

Tables

Table 1-1: Genetic studies in vasculitis ...... 39 Table 1-2: ANCA distribution in the ANCA associated vasculitides ...... 40 Table 2-1: Fluorochrome conjugated antibodies for flow cytometry ...... 70 Table 3-1: CEC detection in human disease ...... 99 Table 3-2: Spearman Rank correlation coefficiants between CECs and conventional markers of disease activity ...... 138 Table 4-1: Spearman Rank correlation coefficiant analysis between different MP phenotypes and other markers of disease activity ...... 208 Table 4-2: Holm-Bonferroni correction for multiple comparisons ...... 209 Table 5-1: Spearman Rank correlation coefficiants of growth factors and markers of disease activity in vasculitis...... 243 Table 5-2: Holm-Bonferroni correction for multiple comparisons relating to correlations of growth factors and other markers of disease activity ...... 244 Table 6-1: How best to enumerate and express EPC number ...... 274 Table 6-2: Spearman rank correlation coefficiants of putative EPC populations – expressed as percentage of PBMCs...... 303 Table 6-3: Spearman rank correlation coefficiants of putative EPC populations – expressed as percentage of CD34+ PBMCs...... 304 Table 6-4: Holm-Bonferroni critical values for significant correlations between EPCs and markers of disease activity ...... 305

Abbreviations

5HT 5-hydroxytryptophan AAV ANCA associated vasculitis/vasculitides ACR American College of Rheumatology AECA anti-endothelial cell antibodies ANCA anti-neutrophil-cytoplasmic antibodies APC antigen presenting cell BD Behçets Disease cANCA cytoplasmic ANCA CMV Cytomegalovirus CSF cerebral-spino-fluid CSS Churg Strauss syndrome EMP endothelial microparticle EPC endothelial progenitor cell EULAR European League Against Rheumatism EUVAS European League Against Vasculitis G-CSF granulocyte colony stimulating factor GM-CSF granulocyte-macrophage colony stimulating factor GOSH Great Ormond Street Hospital HMEC human microvascular endothelial cells HSP Henoch-Schönlein purpura HSPs heat shock proteins HTLV-1 human T-cell lymphotropic virus-1 HUVEC human umbilical vein endothelial cell ICAM-1 intercellular adhesion molecule 1 IL-x interleukin – x IVIG intravenous immunoglobulin JAM junctional adhesion molecule KD Kawasaki disease LPS Lipopolysaccharide MCP-1 monocyte chemoattractant protein-1 M-CSF macrophage colony stimulating factor MHC major histocompatability complex MLCK myosin light chain kinase MMP monocyte microparticle MPA microscopic polyangiitis MPO-ANCA myeloperoxidase- ANCA MPs microparticles NK natural killer NMP neutrophil microparticle NO nitric oxide PAI-1 plasminogen activator inhibitor-1 PAN polyarteritis nodosa pANCA perinuclear ANCA

PECAM-1 platelet/endothelial cell adhesion molecule-1 PMP platelet microparticle PR3-ANCA proteinase-3 ANCA PRES Paediatric Rheumatological European Society PSV primary systemic vasculitis RA rheumatoid arthritis RANTES regulated upon activation normal T cell expressed and secreted SLE systemic lupus erythematosus TCR T cell receptor TD Takayasu's disease TF tissue factor TFPI tissue factor pathway inhibitor TNF tumour necrosis factor VCAM-1 vascular cell adhesion molecule 1 VEGF vascular endothelial growth factor VSM vascular smooth muscle VWf von willebrand factor WG Wegener's Granulomatosus

1 Introduction

1.1. The endothelium and its functions

1.2. Vasculitis – a clinical perspective

1.3. The pathogenesis of vasculitis

1.4. Assessment of disease activity in vasculitis – the need for reliable

biomarkers

1.5. Aims of this thesis

1.1 The endothelium and its functions

The term “vasculitis” refers to inflammation affecting blood vessels. The endothelium, which forms the inner layer of all blood vessels, is both central to the pathogenesis of the disease and can be directly damaged by the inflammatory processes, resulting in impairment of function. This thesis therefore begins with a brief introduction to the endothelium, hilighting some of its key functions that are essential for health but may be impacted upon in vasculitis.

In an adult human the endothelial monolayer consists of approximately 1013 cells covering an area of 1-7m2 (Cines et al., 1998). Endothelial cells are supported by a basement membrane that is surrounded by varying thicknesses of smooth muscle and connective tissue depending on vessel type and function. In the microvasculature where smooth muscle is absent, pericytes also sit within the basement membrane and surround the endothelium with long cytoplasmic processes (Allt and Lawrenson, 2001). Prior to the discovery of the vasodilator prostacyclin in 1976 (Bunting et al., 1976) the endothelium was considered to be an inert barrier that separated blood and tissue.

However, it is now apparent that the endothelium plays a role in many vital functions including: regulation of vessel tone and blood flow, fluid and solute exchange, haemostasis and coagulation, vasculogenesis and angiogenesis and inflammatory responses.

1.1.1 Control of vessel tone

In response to stimulation by substances in the blood (e.g. histamine, angiotensin, 5HT) or mechanical stimuli endothelial cells can release a number of vasoactive factors that relax (e.g. prostacyclin, nitric oxide (NO), hydrogen peroxide (H2O2)) or contract (e.g.

thromboxane A2, endothelin, angiotensin II) surrounding smooth muscle (Feletou and

Vanhoutte, 2006). Endothelial cells can also signal directly with smooth muscle via myoendothelial gap junctions, allowing the spread of membrane potential and small molecules and ions such as calcium (Ca2+) which in turn can alter vascular tone (Feletou and Vanhoutte, 2006).

Global rise in intracellular Ca2+ directly influences the contractile state of vascular smooth muscle (VSM) (House et al., 2008). This occurs via its binding of calmodulin and subsequent activation of myosin light chain kinase (MLCK), which results in the phosphorylation of myosin, allowing interaction with actin and the initiation of contraction (Rang et al., 1999). The ability of the endothelium to activate/inhibit signalling pathways that crosstalk with this system, means that although Ca2+ is important for contraction, the relationship between Ca2+ concentration and the extent of contraction is not as clearly defined in VSM as it is in skeletal or cardiac muscle (Rang et al., 1999).

1.1.2 Fluid and solute exchange

The endothelium acts as a semi-permeable barrier and tightly controls the passage of large molecules and fluids between the blood and interstitial space. Small molecules such as urea, glucose and ions easily pass across in a paracellular manner through interendothelial gap junctions (Lum and Malik, 1994). However, larger molecules such as albumin cannot do this and, in the quiescent endothelium, they pass in an intracellular manner via vesicular transport. Calveolae (pits) in the surface of the endothelium, develop into vesicles which pass across the cell, either independently via fluid-phase uptake or in a receptor dependent manner (Lum and Malik, 1994). In

response to certain stimuli (such as histamine or vascular endothelial growth factor

(VEGF)) interendothelial gap junctions become disrupted and larger molecules can pass through. This transvascular flux of liquid and solutes from the circulation can rapidly lead to the oedema formation that is often associated with inflammation (Mehta and

Malik, 2006).

1.1.3 Haemostasis

Damage to the luminal vessel wall can result in blood loss and haemostasis refers to the process by which it is stopped. It is associated with vasoconstriction, adhesion and activation of platelets at the site of injury and a complex enzymatic activation cascade which culminates in the formation of a stable fibrin plug, effectively sealing the vessel

(Hoffman and Monroe, 2007).

Exposure to the blood of tissue factor (TF), the transmembrane receptor for factor VII, initiates this proteolytic clotting cascade, which ultimately results in the downstream activation of thrombin and its subsequent conversion of fibrinogen to the clot stabilizing polymer fibrin. TF is constitutively expressed on a number of cells in the vessel wall

(e.g. pericytes, fibroblasts and smooth muscle) (Camerer et al., 1996) but not on the endothelium, and thus under normal physiological conditions clotting should not be instigated. However, TF can be expressed on tumour derived endothelium (Engelmann et al., 2003) and on endothelial cells in response to inflammation (Randolph et al.,

1998). It is also found on activated monocytes and is associated with platelets and microparticles in the circulation (Muller et al., 2003). On microparticles in healthy individuals it is thought to be in an inactive form (Furie and Furie, 2008).

Platelets are activated on contact with exposed collagen in the vessel wall and along with leucocytes form part of the thrombus, propagating the enzymatic cascade by acting as a surface upon which TF is exposed and where fibrin generating pathways can occur

(Hoffman and Monroe, 2007). Pathological thrombosis can occur in the absence of substantial endothelial damage, and platelet activation independent of collagen exposure may be an initiating event in this.

During haemostasis the endothelium acts to influence both pro and anticoagulant pathways. It synthesizes and secretes key components of the clotting cascade (e.g. von

Willebrand factor (vWF) and plasminogen activator inhibitor-1(PAI-1)) (Becker et al.,

2000). Conversely the endothelium also generates NO and prostacyclin (inhibiting platelet function), and activates a number of anticoagulant pathways that limit the thrombus formed. These anticoagulant pathways are essential for preventing unnecessary pathological haemostasis and include expression of the surface glycoprotein heparan sulfate, which is a cofactor for the serine protease inhibitor antithrombin III (which can inhibit many of the clotting cascade enzymes) and thrombomodulin, a thrombin receptor which activates the anticoagulant protease protein

C (Adams and Huntington, 2006). Tissue factor pathway inhibitor (TFPI) is constitutively released by the endothelium and interferes in the TF-activated factor VII complex (Lwaleed and Bass, 2006) Tissue plasminogen activator, released in response to procoagulant activity catalyses the conversion of plasminogen to plasmin - a key player in dissolution of the clot once injury has healed (Emeis, 1992).

1.1.4 Angiogenesis and Vasculogenesis

Once haemostasis has prevented blood loss from the vessel wall, repair must occur to

prevent long term dysfunction. In health, the integrity of the endothelium is maintained by a continuous renewal of cells. Turnover of the endothelium is low, with replication rates in the region of 0.1% per day in adults (Hunting et al., 2005), although in healthy growing children it could well be higher.

Until relatively recently angiogenesis, whereby mature endothelial cells already present within the vessel proliferate and migrate, was considered to be the mechanism by which growth, remodeling and repair of damage occurred. Vasculogenesis, whereupon bone marrow derived progenitor cells are recruited to form new vasculature, was considered to occur only during early embryonic development. The concept of post-embryonic vasculogenesis gained impetus in 1997 when Asahara et al described a population of

CD34+ cells which circulated within the peripheral blood, had the potential to form endothelial-like cells in vitro and enhanced repair in an ischaemia-reperfusion model

(Asahara et al., 1997). These cells were termed endothelial progenitor cells (EPCs) and have been the subject of much research in recent years.

Whether EPCs are true endothelial progenitors with the ability to form permanent long lasting endothelium is still open to debate a decade later. One alternative hypothesis is that they promote a localized pro-angiogenic state and are only found transiently at sites of injury (Fazel et al., 2006), although this requires further confirmation. Regardless of the exact mechanisms which contribute to repair, the regulators of these pathways

(growth factors, chemokines and cytokines) a number of which act directly on or are released from the endothelium, are frequently the same in angiogenesis and vasculogenesis and thus these processes are likely to be highly interdependent.

Angiogenesis and vasculogenesis are discussed in more detail in subsequent chapters.

1.1.5 Inflammatory responses and the endothelium

The endothelium responds rapidly to pathological conditions being both affected by and contributing to the inflammatory process, secreting inflammatory cytokines and chemokines and upregulating adhesion molecules, facilitating leukocyte recruitment. In health, these responses are reversible and localized, however sustained or inappropriate activation is associated with endothelial dysfunction and contributes to the development of cardiovascular diseases such as atherosclerosis (Loppnow et al., 2008).

A wide number of cytokines and chemokines can be released from the endothelium in response to inflammatory stimuli such as treatment with lipopolysaccharide (LPS) or tumour necrosis factor-α (TNFα) in vitro, and shear stress, infection or hypoxia in vivo

(Krishnaswamy et al., 1999). These include acute phase response cytokines such as IL-

1 and IL-6, which are responsible for a number of the characteristic manifestations of inflammation including fever, generation of acute-phase reactants by hepatocytes and activation of lymphocytes (Marceau et al., 1992,Ala et al., 1992). Chemokines produced include IL-8, which can recruit and activate neutrophils (Sica et al., 1990a),

MCP-1 (monocyte chemoattractant protein-1) which recruits monocytes, activated T cells, dendritic cells, and natural killer (NK) cells (Sica et al., 1990b) and

CCL5/RANTES (regulated upon activation normal T cell expressed and secreted), which recruits eosinophils. Colony stimulating factors (CSF) released from endothelial cells such as GM-CSF (granulocyte-macrophage-CSF), M-CSF (macrophage-CSF) and

G-CSF (granulocyte-CSF) alter the differentiation, activation, proliferation and migration of macrophages and granulocytes and may even promote the in situ longevity of these recruited inflammatory cells (Rajavashisth et al., 1990).

Leukocyte recruitment to the tissue from the circulation requires cells to migrate through the vessel wall. The endothelium plays a key role in this by upregulating adhesion molecules which interact with leukocytes aiding their recruitment. Endothelial

E and P selectins acting on their carbohydrate ligands on leukocytes result in the leukocytes slowing down and rolling along the endothelial surface. Rolling leukocytes rapidly become activated via immobilized chemokines on the endothelium (Johnston and Butcher, 2002) with upregulation of integrins on their surface. Integrins on leukocytes contribute both to the rolling process (Ley et al., 2007) and to firm adhesion by binding members of the immunoglobulin superfamily on endothelial cells such as intercellular adhesion molecule 1 (ICAM-1) and vascular cell adhesion molecule 1

(VCAM-1) (Petri and Bixel, 2006). Subsequently leukocytes engage other adhesion molecules such as platelet/endothelial cell adhesion molecule-1 (PECAM-1), endothelial cell-selective adhesion molecule (ESAM), CD99 and members of the junctional adhesion molecule family (JAM) and cross the endothelial layer (Ley et al.,

2007).

Studies utilizing neutralising antibodies indicate that different adhesion molecules mediate leukocyte transmigration depending on stimulus, leukocyte cell type or location. In rat mesenteric arteries for example IL-1β but not FMLP induced leukocyte transmigration can be inhibited by the use of PECAM blocking antibodies (Thompson et al., 2000). Leukocytes can cross the endothelium in a paracellular or transcellular manner (Feng et al., 1998). It remains to be determined whether both pathways are of equal physiological relevance or if, as has been observed in vitro, paracellular mechanisms are primary route of migration (Vestweber, 2007).

1.1.6 Conclusions

The endothelium is a dynamic and heterogeneic organ system, with significant differences in response between endothelium of different vessels and even between cells in a similar location to certain stimuli. With such a wide ranging and tightly regulated functionality preservation of a healthy endothelium is crucial to the maintenance of health and normal physiological homeostasis. Inappropriate and sustained inflammatory responses (such as those seen in the vasculitides) are associated with endothelial dysfunction. Endothelial dysfunction is a phenomeneon that is considered to be central to the pathogenesis of the vasculitides and hence this group of disorders will now be discussed in more detail.

1.2 Vasculitis – a clinical perspective

The vasculitides are a group of disorders characterised by the presence of inflammatory cellular infiltrates (neutrophilic, eosinophilic, or mononuclear) in the walls of blood vessels, with resultant tissue ischaemia and necrosis. Vasculitis can affect blood vessels of all sizes and can occur as the primary disease manifestation or secondary to other conditions, such as infection, malignancy or autoimmune diseases such as rheumatoid arthritis (RA), scleroderma, Sjogrens syndrome and systemic lupus erythematosus

(SLE) amongst others.

1.2.1 Primary systemic vasculitis

Classification of the vasculitides has proved difficult because of the absence of sensitive and specific diagnostic tests and an incomplete understanding of the aetiology of most vasculitic syndromes. Patients often have constellations of symptoms and findings that

overlap the clinical features of the various individual diseases. Although many attempts have been made to classify primary systemic vasculitis (PSV) according to type of inflammation, causative mechanisms or associated autoantibodies, the most commonly used system for classifying disease is based on the size of the smallest vessel predominantly involved in the disease process - the Chapel Hill Consensus Criteria

(Jennette et al., 1994).

While the Chapel Hill criteria and the American College of Rheumatology (ACR) criteria (Appendix 1) for classification of vasculitis are widely used, they have never been validated in children. In this context the recently described European League

Against Rheumatism (EULAR)/Paediatric Rheumatology European Society (PRES) endorsed consensus criteria for the classification of childhood vasculitides may be more appropriate (Appendix 1) (Ozen et al., 2006), and are currently undergoing formal validation (Brogan, personal communication). This is based on the Chapel Hill consensus criteria with modifications suggested by consensus relevant to paediatric disease. For the purposes of this thesis which is focused on the downstream consequences of vasculitis (i.e. endothelial injury and repair), classification is perhaps somewhat arbitrary as endothelial injury occurs in PSV irrespective of disease aetiopathogenesis (Brogan et al., 2004a). In addition overlap in the size of vessels affected occurs (“polyangiitis overlap”), for example in Wegener’s granulomatosis (WG), which is classified as a small vessel disease vasculitis can be detected in small and medium- sized arteries and occasionally larger vessels, including the renal artery and aorta

(Shitrit et al., 2002). Therefore in this thesis when data was analysed the cohort was first examined as a whole, and then by subgroup analysis of individual diseases where appropriate. Diseases were classified predominantly using the Chapel Hill Consensus

Criteria, however the ACR and EULAR/PRES consensus criteria were used (out of necessity) as additional classification tools by clinicians experienced in the classification of vasculitis at Great Ormond Street Hospital (GOSH).

The individual vasculitic diseases of the young which have been studied within this thesis will now be described in more detail. The predominant diseases observed in this study have been Polyarteritis Nodosa (PAN) and Wegener’s Granulomatosis (WG), followed by Kawasaki disease (KD) and Takayasu’s Disease (TD). An additional group of patients had definite but “unclassified” vasculitis. Epidemiological evidence suggests that KD and Henloch Schönlein-Purpura (HSP) are the most common vasculitic diseases of childhood. In 2002 incidences for KD and HSP were estimated to be 8.2 per

100 000 children under the age of five (Harnden et al., 2002) and 20.8 per 100 000 children (Gardner-Medwin et al., 2002) respectively, with 0.24 per 100 000 children for all other primary systemic vasculitic conditions (Gardner-Medwin et al., 2002). This is not reflective of the GOSH cohort however, as GOSH is a specialist centre for paediatric vasculitis dealing with more complex cases that are referred from secondary and other tertiary care trusts

1.2.2 Polyarteritis Nodosa and associated diseases

In its classical form, PAN, is a necrotizing vasculitis associated with aneurysmal nodules along the walls of medium-sized muscular arteries, and was first described in detail in 1866 by Kussmaul and Maier (Dillon, 1998). Although overlap occurs with smaller vessel disease, it is distinct from microscopic polyangiitis (see below), and in children occurs more commonly than the anti neutrophil cytoplasmic antibody (ANCA) associated vasculitides (Brogan and Dillon, 2000b). The main clinical features of PAN

are malaise, fever, skin rash, abdominal pain and arthropathy, although testicular pain, myalgia, hypertension, neuropathy, renal failure, organic psychosis and myocardial ischaemia can also occur (Besbas et al., 2000,Brogan and Dillon, 2000b).

PAN can occur in a predominantly cutaneous form, associated with painful skin nodules, livedo reticularis, and often following upper respiratory tract infection (Ozen et al., 2004). While cutaneous PAN is generally a milder condition that responds well to non-steroidal anti inflammatory drugs (although in some cases corticosteroids are necessary) and/or antibiotics if infection is suspected, the cutaneous disease can develop into the more serious systemic condition (Till and Amos, 1997). Relapses are common

(up to 25%) and prophylactic penicillin is sometimes used, as infection may be associated with relapse. Those with relapsing disease may have a higher risk of developing systemic vasculitis (Till and Amos, 1997).

Microscopic polyangiitis (MPA, formerly microscopic polyarteritis) differs from classic

PAN by having extensive glomerular involvement and/or pulmonary haemorrhage. It can be defined as a small vessel vasculitis with focal segmental pauci-immune glomerulonephritis, sometimes with alveolar haemorrhage but without granulomatous disease of the respiratory tract (Puechal, 2007). MPA is one of the ANCA-associated vasculitides (AAV), associated with perinuclear ANCA (pANCA) positivity, typically directed against myeloperoxidase (MPO-ANCA) (Bosch et al., 2006).

1.2.3 Wegener’s Granulomatosis

WG is a necrotizing granulomatous vasculitis of the upper and lower respiratory tract,

associated with glomerulonephritis and variable small vessel vasculitis. Upper airway disease can include granulomatous involvement of the trachea leading to stenosis, and of the nasal septum, leading to collapse of the bridge of the nose and “saddle nose” deformity. Lower airway disease can often masquearade as infection (Stegmayr et al.,

2000), but it due to necrotising vasculitis of the pulmonary capillaries. WG is another

ANCA associated disease, although it is predominantly cytoplasmic ANCA staining

(cANCA) that is observed (Bosch et al., 2006), typically against proteinase-3 (PR3-

ANCA), although MPO-ANCA are sometimes observed (Brogan, 2007).

1.2.4 Takayasu disease

World-wide, TD is the third commonest vasculitis of childhood, and some have suggested that it may be related to tuberculosis infection (Dillon, 1998). TD is a giant cell arteritis, a term that relates to the infiltration of giant cells (multinucleated cells formed by the fusion of many cells) seen in biopsies of diseased tissue. TD causes stenosis and aneurysmal dilatation of large arteries, such as the aorta and its major branches. Its clinical features include fever, anorexia, weight loss, arthritis, and later the development of hypertension, heart failure and loss of pulse in some limbs (Kerr et al.,

1994). The disease is often delayed in its diagnosis, especially in the young, and they can be confused with other causes of occlusive vasculopathy such as fibromuscular dysplasia (Tullus et al., 2008).

1.2.5 Kawasaki disease

KD was first described in 1967 by Tomisaku Kawasaki who described 50 Japanese children with an illness characterised by fever, rash, conjunctival injection, erythema and swelling of hands and feet, and cervical lymphadenopathy (Kawasaki, 1967). KD

is more common in children of oriental background, with peak incidence at around 9-11 months of age (Newburger et al., 2003). KD is associated with development of systemic vasculitis complicated by coronary and peripheral arterial aneurysms, and myocardial infarction in some patients. KD is the commonest cause of acquired heart disease in children in the UK and USA (Shulman et al., 1995)

1.2.6 Henoch-Schönlein purpura

HSP (anaphylactoid purpura) is the commonest vasculitis seen in childhood with peak incidence occurring at 4-5 years of age, more commonly in males. It is a multi-system small vessel systemic vasculitis with a prominent cutaneous component, that can also affect joints, gastrointestinal tract and the kidney (Yang et al., 2008). Other organs less frequently involved include the central nervous system, gonads and the lungs. HSP often occurs following an upper respiratory tract infection and the onset of the disorder may be accompanied by systemic symptoms including malaise and mild pyrexia

(Saulsbury, 1999). Of the childhood vasculitides HSP has the most favourable outcome, with the large majority of children requiring no treatment. Recurrence of symptoms occurs in around one third of cases, generally within four months of resolution of the original symptoms and more frequently in those with renal involvement (Saulsbury, 1999).

1.2.7 Churg-Strauss syndrome

Churg-Strauss syndrome (CSS) or allergic granulomatosis is extremely rare in childhood (Boyer et al., 2006). It is described as AAV, although there is no predominant cANCA or pANCA association (Bosch et al., 2006) and in the small number of published studies in children, frequency of ANCA positivity is lower than in

adults (Boyer et al., 2006). The disease can affect the lungs, GI tract and nervous system and often presents initially as severe asthma. Other clinical features include eosinophilia, infiltrates on chest X-ray and extra-vascular granulomata on biopsy (Boyer et al., 2006).

1.2.8 Other vasculitic disorders

According to the EULAR/PRES consensus criteria relating to the classification of paediatric vasculitis the following conditions fall into the category of disease described as “other” (Ozen et al., 2006). No specific definitions are described for these entities in the proporsed EULAR/PRES criteria, thus Chapel Hill consensus or ACR criteria were applied where appropriate.

1.2.8.1 Behçet’s disease

Behçet’s disease (BD) occurs rarely in children. In studies of adults incidence is much higher in the Mediterranean, Middle East and Far East compared to Northern Europe

(Reynolds, 2008). Incidence in adults in Turkey is estimated at 110-370 per 100 000

(Idil et al., 2002,Yurdakul et al., 1988), whereas in the UK it is 0.64 per 100 000

(Chamberlain, 1977). BD is associated with recurrent oral and genital ulcers, relapsing uveitis, skin lesions, arthritis and CNS and GI manifestations. It is recognised that a vasculitic component is an important feature of the disease (Reynolds, 2008,Kutlay et al., 2008).

1.2.8.2 Cogan’s syndrome

Cogan’s syndrome is a rare condition that predominantly affects young adults, although has been reported in children (Olfat and Al-Mayouf, 2001). The syndrome mainly

affects the eyes with inflammation of the cornea (keratitis) leading to eye redness, blurred vision, pain and photophobia and the vestibuloauditory system causing tinnitus, dizziness, hearing loss and nausea and vomiting. Systemic vasculitis can also occur, in predominantly large and medium vessels, typically associated with aortic involvement

(St Clair and McCallum, 1999).

1.2.8.3 Vasculitis secondary to connective tissue diseases

Secondary systemic vasculitis associated with autoimmune connective tissue diseases is a well recognised phenomenon. The extent to which it occurs at a subclinical level may be much more widespread, than previously appreciated, and in children and adults is likely to have long term implications relating to cardiovascular outcomes such as premature atherosclerosis (Bacon et al., 2002).

Autopsy studies in adult patients suffering from RA indicate that a significant proportion have evidence of systemic vasculitis (ranging from 15-31%) (Bely et al.,

1992,Salvarani et al., 1992,Suzuki et al., 1994) however in living patients detection is at much lower levels (<1%) (Watts et al., 2004,Wattiaux et al., 1987,Genta et al., 2006).

The relationship between rheumatic disease and risk of adverse cardiovascular events independent of traditional risk factors is now well recognized. RA doubles the risk of cardiovascular disease and is associated with early mortality (Van et al., 2002,bou-Raya and bou-Raya, 2006) while SLE has a 5-6 times higher risk of a significant coronary event, rising to up to 50 times higher than the general population in young women

(Manzi et al., 1997). The presence of subclinical vasculitis in these patients is an attractive hypothesis to explain this phenomenon, however as discussed in more detail later in this chapter monitoring disease activity in primary systemic vasculitis is fraught

with difficulties in routine clinical practice. Thus assessing subclinical vasculitis in patients with inflammatory disease remains a major challenge, but is likely to be of great importance in view of the poor cardiovascular outcomes associated with autoimmune diseases (Bacon et al., 2002).

1.2.8.4 Vasculitis secondary to infection

Infectious agents such as viruses, bacteria and fungi have been repeatedly proposed as triggering agents for vasculitis, and are more frequently associated with conditions that are ANCA negative (Pagnoux et al., 2006). Some infections are directly responsible for causing vasculitis by their action on the endothelium, either by directly infecting endothelial cells or binding to them, resulting in bystander injury to the surrounding vessel wall. Other infections misdirect the host immune response towards self: if the infecting pathogen expresses structures similar to that seen in the endothelium, molecular mimicry may occur whereby host responses are directed towards the endothelium as well as the pathogen. Complementary peptides (which contain sequences that match the antisense DNA strand of a host peptide) can also induce a host autoimmune response (discussed in more detail in the context of AAV) (Pendergraft, III et al., 2004). Superantigens have also been implicated in vasculitis, particularly KD

(Brogan et al., 2008). These pathogen derived peptides have the ability to activate large proportions of the T cell population (5-30%) by crosslinking MHC class II on antigen presenting cells (APCs) and the T cell receptor (TCR) independent of conventional antigen processing pathways.

Viral infections that have been implicated in vasculitis include hepatitis B, hepatitis C, human immunodeficiency virus, erythrovirus B19, cytomegalovirus (CMV), varicella-

zoster virus (VZV) and human T-cell lymphotropic virus-1 (HTLV-1) (Pagnoux et al.,

2006). Bacterial infections that have been proposed as potential causative agents include strains of Pseudomonas, Legionella, Neisseria, Rickettsiae, Staphylococcus and

Mycobacterium tuberculosis (Pagnoux et al., 2006). Fungal infections of the lung can mimic WG or Churg Strauss syndrome by causing a granulomata to form in vessel walls (El-Zammar and Katzenstein, 2007). In a number of these infection related vasculitides, if the vasculitis is a direct effect of the infective agent (e.g. the necrotizing vasculitis of the lung sometimes seen in Pseudomonas aeruginosa infections) rather than as a downstream consequence after the immune system has been misdirected towards self, then effective treatment of the underlying infection with antimicrobial agents is necessary, and may be sufficient without additional immunosuppression

(Pagnoux et al., 2006).

1.2.8.5 Vasculitis secondary to malignancy

Vasculitis associated with malignancy is extremely rare in children (Pelajo et al., 2007), however it can occur and there a number of ways in which this happens. Vasculitis can occur as the result of an infectious complication in an already immunocompromised patient. Vasculitis can also be triggered directly by the presence of malignancy or metastases invading blood vessels (Racanelli et al., 2008). Alternatively, it can occur as a paraneoplastic phenomenon as a consequence of abeherent immune responses, whereupon treatment of the inciting malignancy can resolve the vasculitis. Often paraneoplastic vasculitis is a cutaneous form that is detected prior to clinical detection of the malignancy. Lastly, it remains possible that both malignancy and vasculitis are triggered by the same environmental stimuli in some cases (Racanelli et al., 2008).

1.2.9 Treating vasculitis

Treatment for systemic vasculitis comprises aggressive immunosuppression and consists of combined corticosteroid and cytotoxic agents such as cyclophosphamide

(oral or pulsed intravenous), azathioprine or methotrexate as well as antiplatelet agents

(usually aspirin), and in some forms of vasculitis pooled human intravenous immunoglobulin (IVIG, particularly for KD (Newburger et al., 1986)) in an attempt to bring the disease into remission (Watts et al., 2005). Plasma exchange may also prove a useful therapy, particularly for severe fulminant forms of small vessel vasculitis (Jayne et al., 2007). The exact regime used depends on the disease and the individual patient.

In some of the milder conditions such as HSP or cutaneous PAN, treatment may only consist of symptomatic relief with analgesics or non steroidal anti inflammatory drugs

(Palit and Inamadar, 2006).

Newer biological agents may be necessary if conventional treatment is ineffective.

These include (amongst others) anti TNFα (Burns et al., 2005a,Booth et al.,

2004a,Brogan et al., 2009), anti IL-1 (Eleftheriou et al., 2009), B cell depletion

(Eleftheriou et al., 2009) or lymphocyte depletion using anti-CD52 (Campath)

(Lockwood et al., 2003). These novel approaches have only been used in relatively small numbers of patients and thus their usage is based on largely anecdotal evidence.

That said, biologic therapy holds promise for recalcitrant disease. If remission is still not achieved, autologous stem cell transplantation may be an option and indeed was performed successfully in one child included in this thesis.

With the more agressive forms of therapy there is significant morbidity associated with long term use of these agents which includes diabetes, infection, bladder cancer,

haemorrhagic cystitis, bone marrow toxicity and infertility. Thus treatment should be tailored to the minimum required to control disease whilst at the same time limiting these unfavourable side effects (Brogan and Dillon, 2000a). Following successful induction of remission, maintenance of remission is achieved using therapy including anti-platelet doses of aspirin and low dose corticosteroids combined with immunosuppressants such as azathioprine, mycophenolate mofetil, or methotrexate sometimes for months or years (Brogan and Dillon, 2000a).

Despite these aggressive therapeutic approaches, there is still a significant mortality and morbidity associated with PSV, including treatment related complications such as opportunistic infections. However it is worth noting that without treatment mortality is extremely high. Historically for patients with WG there was 80% mortality within the first year (Walton E.W, 1958). Thus treatment outcomes for children and adults with vasculitis are vastly improved but require further refinement, a programme of work currently being performed by the EUVAS group (European vasculitis study group).

Lastly it is important to emphasise that many of the treatment protocols for PSV in children, with the exception of KD are based on evidence from clinical trials in adults.

Importantly, however for the first time children with ANCA associated vasculitis have recently been included in EUVAS trials (Brogan, personal communication).

1.2.10 Prognosis

Overall outlook is good for patients who urvive the initial vasculitic insult and any subsequent disease flares, however the long term consequences of vasculitis in childhood are still not clear. Recent studies indicate that there is increased subclinical

atherosclerosis in adult patients with small vessel vasculitis (Chironi et al., 2007) and whether children with these diseases have similar consequences remains to be determined. Coronary artery lesions (including aneurysms and stenoses) in KD predispose to premature coronary atherosclerosis, and peripheral arteries remain stiffer than in healthy controls for years after disease remission in PAN, possibly being indicative of fibrotic changes (Cheung et al., 2002,Suzuki et al., 1986) and/ or ongoing subclinical vascular inflammation (Mitani et al., 2005). So although now most children survive vasculitis, it appears that there could be implications for their cardiovascular health in the longer term (Nakamura et al., 2002).

1.2.11 Conclusions relating to clinical perspectives on vasculitis

There is a wide spectrum of vasculitic disease affecting children. The distinction from other conditions, especially those associated with infection is often difficult, but it is possible to categorise the disorders into clinico-pathological entities even though there is a substantial degree of overlap. Treatment for systemic disease is aggressive immunosuppression, aimed at inducing remission, rather than targeting upstream casual pathology (described in Section 1.3). Outlook for patients on the whole is good, however in the long term, cardiovascular complications may lead to reduced life expectancy, however much work remains to be done in order to identify those patients at greatest risk.

1.3 The pathogenesis of vasculitis

In most patients the definitive upstream causes of vasculitis are unknown, however there are a number of well described inflammatory pathways that culminate in injury to

the endothelium that are beginning to be understood. The predominant theory as to how vasculitis is initiated is the combination of an environmental trigger occurring in an individual with a background genetic susceptibility (Brogan, 2007).

1.3.1 Environmental triggers

A number of epidemiological studies support the hypothesis that an environmental trigger is involved in the development of vasculitis. In the two commonest vasculitic diseases of childhood, HSP and KD, epidemiological evidence points to an infectious cause. Both conditions have seasonal variations in incidence, with KD peaking in the winter and spring months (Bronstein et al., 2000) and HSP in autumn and winter (Yang et al., 2005) often following a respiratory tract infection. WG occurs more commonly in Northern Europe than Southern Europe. In one study by Watts et al although there were significant differences in the incidence of WG between Spain and the UK, there was little variation in HLA-DRB1 region genetics (the human leucocyte antigen region of the genome which is associated with a number of autoimmune diseases) leading the authors to suggest that environmental factors may be more important than genetic risk factors for triggering WG (Watts et al., 2001).

While infections and malignancies (discussed above) may be triggers of vasculitis, there a number of other potential non-infectious environmental factors which have been implicated in disease onset. In 2003 Lane et al demonstrated increased risk of developing PSV (all forms) in Norfolk with exposure to silica, solvents and interestingly farming, which undoubtedly exposes the individual to many chemicals

(pesticides, fertilizers etc) and possible infectious agents from livestock (Lane et al.,

2003). Other studies have also implicated silica and solvents (Tervaert et al., 1998,Pai

et al., 1998) as well as heavy metal exposure (Albert et al., 2004) , a number of different vaccinations (including influenza and hepatitis B) (Somer and Finegold, 1995,Schattner,

2005), many different classes of drugs (Wiik, 2008), corticosteroid withdrawal

(Kinoshita et al., 1999) and allergy desensitisation procedures (Cabrera et al., 1993). A background of allergy or asthma is also associated with increased risk of PSV (all forms) (Lane et al., 2003).

1.3.2 Genetics of vasculitis

In recent years there have been a number of studies examining the genetic background of the different vasculitic diseases. Table 1.1 (adapted from Brogan, 2007) highlights a number of these studies. What is immediately apparent is that the genetic factors which predispose individuals to developing vasculitis are complex and there are a wide variety of candidate genes that may be involved. These genes include those involved in inflammation such as cytokines, chemokines, and their receptors as well as those relating to regulation of the vasculature such as growth factors and NO synthase.

Caveats to these genetic studies include the large numbers of patients with these rare diseases which are required for adequate powering, and the difficulty of robust classification. This has resulted in a number of potentially underpowered studies in KD,

HSP, BD and AAV. However, considerable advances in genome wide association studies hold the promise of major advances in the understanding of complex diseases such as the vasculitides if adequate patient numbers can be studied (numbers in the

1,000s are necessary (McCarthy et al., 2008)). This will ultimately require international cooperation.

Table 1-1: Genetic studies in vasculitis

Gene of Disease Effect interest studied KD Higher expression of ACE insertion allele associated with increased disease susceptibility (Shim et al., 2006)

Angiotensin HSP No association found with ACE, but angiotensinogen (angiotensin precursor) associated with converting disease susceptibility (Ozkaya et al., 2006) enzyme ACE DD genotype predicts persistent proteinurea (Yoshioka et al., 1998)

BD Higher expression of ACE deletion allele than in controls (Turgut et al., 2005) KD Polymorphisms associated with development of disease (Breunis et al., 2006), a more recent study demonstrated no associations (Huang et al., 2008) VEGF

HSP Polymorphisms associated with development of renal complications (Rueda et al., 2006) HSP Polymorphism in iNOS promotor associated with renal involvement (Martin et al., 2005) Nitric oxide although no associations with eNOS polymorphisms (Amoli et al., 2004b) synthase (NOS) KD No association of eNOS or iNOS polymorphisms with susceptibility or coronary artery lesions (Khajoee et al., 2003) KD Polymorphisms in Il-1ra but not IL-1β associated with susceptibility (Wu et al., 2005)

HSP Polymorphisms in both Il-1ra and IL-1β associated with renal involvement (Amoli et al., IL-1/IL-1Ra 2002,Amoli et al., 2004a)

WG No associations found (Zhou et al., 2004) KD Polymorphisms associated with susceptibility (Cheung et al., 2008) and IVIG resistance (Yang et al., 2003) TNFα WG Polymorphisms associated with susceptibility (Spriewald et al., 2005) although another study showed no association (Huang et al., 2000) KD No associations found (Sohn et al., 2001) IL-6 HSP No associations found (Amoli et al., 2007)

BD Polymorphisms associated with susceptibility (Chang et al., 2005) KD Association of promoter polymorphism with increased risk of coronary artery aneurysm (Yang et al., 2003)

IL-10 AAV Polymorphisms associated with CSS but not WG (Wieczorek et al., 2008)

BD Polymorphism associated with susceptibility in UK patients (Wallace et al., 2007) Inositol 1,4,5- trisphosphate-3 Polymorphism in ITPKC ( a negative regulator of T cell signalling) associated with both KD kinase C susceptibility to KD and coronary artery lesions (Onouchi et al., 2008) (ITPKC) Polymorphisms in CCR5 and its ligand CCL3LI associated with susceptibility (Burns et al., CCR5 KD 2005b) Human AAV HLA-DRB4 associated with CSS development (Vaglio et al., 2007) leucocyte antigens KD HLA-B and C polymorphisms associated with susceptibility in Korean cohort (Oh et al., (HLA) 2008). No association with HLA DRB1 (Huang et al., 2007) KD In Children <1 year old polymorphisms in MBL2 (resulting in low MBL levels) were associated with increased risk of coronary artery aneurysms, however conversely at >1 year

of age wild type (normal MBL levels) were at increased risk (Biezeveld et al., 2006) Mannose

binding lectin AAV No associations found (Kamesh et al., 2007) (MBL)

BD MBL2 HYPA haplotype associated with severity and susceptibility (Park et al., 2005b) MMP3 promotor (6A/6A) genotype associated with higher frequency of coronary artery Matrix aneurysms, no associations found with MMP9 polymorphisms (Park et al., 2005a). In a metallo- KD second study no association was found with a different MMP3 promoter polymorphism (- proteinases 439C/G) (Hong et al., 2008) Polymorphism in promoter of PR3 associated with WG susceptibility (Gencik et al., 2000) as Proteinase 3 WG is a polymorphism in the inhibitor of PR3 α-1-antitrypsin which causes deficiency (Elzouki et (PR3) al., 1994) although other studies suggest its influence is minor (Lhotta et al., 1994)

1.3.3 Immunological pathways in vasculitis

1.3.3.1 Autoantibodies

Numerous autoantibodies are associated with vasculitis, however the pathological consequences of antineutrophil cytoplasmic antibodies (ANCA) and anti-endothelial cell antibodies (AECA) have been studied extensively and require mention in detail.

1.3.3.2 Antineutrophil cytoplasmic antibodies

ANCA are detected in 85-95% of adult patients with MPA, WG, CSS, and the localized forms of these diseases (Bosch et al., 2006). The frequency of ANCA positivity may be less for children with these vasculitides however (Akikusa et al., 2007,Belostotsky et al., 2002).

These predominantly IgG autoantibodies target antigens found within neutrophil granules and also within the lysosymes of monocytes (Braun et al., 1991). Although a number of other antigenic targets have been identified the most clinically relevant are those targeting PR3 and MPO. Typically, PR3-ANCA is seen as diffuse staining of the neutrophil cytoplasm (cANCA) whereas MPO-ANCA gives a perinuclear staining

(pANCA) pattern on immunofluorescent microscopy. Table 1.2 summarises the type of

ANCA found in the difference ANCA associated vasculitides in adults.

Table 1-2: ANCA distribution in the ANCA associated vasculitides

% of patients with % of patients with Disease MPO-ANCA PR3-ANCA

Churg Strauss syndrome (CSS) 30-70 <10

Wegener’s Granulomatosis (WG) 10-30 >70

Microscopic Polyangiitis (MPA) 30-70 10-30

Data from (Van Paassen P. et al., 2007)

In comparison to patients positive for MPO-ANCA patients with PR3-ANCA have more organs involved, a faster decline in renal function, more frequent relapses of the disease, and granulomatous necrotizing inflammation in the airways (Franssen et al.,

2000), suggesting different pathological pathways may be involved.

Development of ANCA

The upstream events which lead to development of ANCA remain unknown, however there are a number of hypothesese as to how ANCA develop. As AAV can be initiated by infection, molecular mimicry is one possibility (Benoist and Mathis, 1998), another is autoantigen complementarity a phenomenon described by Pendergraft et al, whereby peptides complementary to PR3 (encoded by the corresponding antisense strand of

DNA) initiate production of antibodies against PR3 (Pendergraft, III et al., 2004). The generation of PR3-ANCA occurs via a 2-stage immune response, the initial response is directed against the complementary peptide, the secondary response again the antibodies generated to the complementary peptide (these will also act on PR3).

Interestingly peptides complementary to PR3 are found in a number of infectious agents, including Staphylococcus Aureus, human herpes virus 4 and rickettsia prowazekii (Pendergraft, III et al., 2004).

A non-infection related hypothesis for ANCA generation relates to neutrophil apoptosis.

When neutrophils undergo apoptosis, their granule contents translocate to the cell surface (Gilligan et al., 1996), incomplete or inappropriate phagocytosis may allow exposure of these self antigens to antigen presenting cells (APCs). However, while injection of apoptotic neutrophils into mice (Rauova et al., 2002) or rats (Patry et al.,

2001) induced ANCA expression, neither species developed vasculitis. Evidence from

in-vitro experiments is conflicting. Although immature dendritic cells can take up apoptotic and necrotic neutrophils, with corresponding increases in maturation markers such as CD83 and MHC class II, the costimulatory molecules (CD86, CD80 and CD40) necessary for T cell activation are downregulated. Thus the overall effect of apoptotic/necrotic neutrophils in this study is tolerogenic, although this was partially reversed by TNFα treatment (Clayton et al., 2003). Thus the exact mechanisms by which defective neutrophil apoptosis could lead to ANCA generation remain poorly defined.

Are ANCA pathogenic?

Regardless of how ANCA production is initiated, there is much evidence to implicate this autoantibody in the pathogenesis of vasculitis from observations made in humans and animal models. Recently animal models have provided some of the most compelling evidence that ANCA are indeed directly pathogenic manner, while in vitro experiments have provided more details at a cellular level on mechanisms by how this can occur.

Proof of concept – do ANCA cause vasculitis in vivo?

As yet only MPO-ANCA have been shown to induce significant vasculitis in animal models. Studies have been performed using PR3-ANCA however it is apparent that the disease pathogenesis in PR3-ANCA associated conditions may be more complicated than just an autoantibody mediated vasculitis (Kallenberg, 2008), corresponding with the differences observed in disease severity between MPO-ANCA and PR3-ANCA in humans (Franssen et al., 2000).

Rats injected with human MPO, generate anti-MPO antibodies and developed a small vessel vasculitis (Little et al., 2005). The IgG isolated from these rats was shown (like human ANCA) to enhance leukocyte-endothelial interactions in vitro. In other studies, injecting MPO knockout mice with mouse MPO allowed generation of anti-mouse

MPO IgG (removing the potential confounding factor of using a human peptide) (Xiao et al., 2002). When these antibodies were injected into wildtype or recombinase activating gene-2 (RAG2-/-) knockout mice (which have no mature B or T cells) they developed a disease similar to AAV with focal necrotizing and crescentic glomerulonephritis, and in some mice a systemic small vessel vasculitis (Xiao et al.,

2002). The presence of disease in RAG-/- mice indicated that T cells were not required to induce the acute lesions. Neutrophils however were necessary as prior injection of a neutrophil depleting antibody prevented disease (Xiao et al., 2005). Correspondingly, priming neutrophils with LPS prior to the IgG dose exacerbated the disease (Huugen et al., 2005). Interestingly, the non classical complement activation pathway is also implicated in this model as C4 (classical) but not C5 (common) or factor B (non- classical) knockout mice develop disease after transfer of anti-mouse-MPO IgG (Xiao et al., 2007a).

In contrast, evidence that PR3-ANCA are directly pathogenic is so far conflicting.

When mice were injected with human PR3-ANCA, they developed antibodies to PR3 in a two stage immune response similar to that described for complimentary peptides.

Mice initially develop antibodies to human PR3-ANCA, followed by antibodies against these (anti-anti-human-PR3-ANCA) which have the ability to bind PR3 (Tomer et al.,

1995). These mice developed lung disease but not glomerulonephritis. However, when

PR3-ANCA was generated in PR3 knockout mice (using mouse PR3), the injection of

these antibodies into wild type mice did not result in spontaneous vasculitis (Pfister et al., 2004). The only difference observed was an exacerbated response to localised

TNFα injection (Pfister et al., 2004). Rats injected with human/mouse chimeric PR3, also generated PR3-ANCA, however, despite this being able to bind rat PR3, vasculitis did not develop (van der Geld et al., 2007).

Perhaps the most robust evidence yet that ANCA can cause vasculitis in humans comes from two reports that have documented trans-placental transfer of IgG MPO-ANCA from mothers with active MPA. The two neonates described developed pulmonary renal syndrome soon after birth (Bansal and Tobin, 2004,Schlieben et al., 2005).

How do ANCA cause vascular injury?

The predominant mechanism by which ANCA induce endothelial injury is via neutrophil activation and adhesion to the endothelium, followed by neutrophil degranulation and subseqeunt dyseregulated apoptosis (Bosch et al., 2006). ANCA can also activate other cell types involved in the inflammatory processes, although these pathways are less well defined (Ralston et al., 1997).

Although MPO and PR3 are not found on the surface of resting neutrophils (Gilligan et al., 1996), priming of neutrophils with substances such as TNFα or LPS (at a concentration that does not induce full activation) or even isolated purified ANCA results in their translocation to the cell surface (Csernok et al., 1994,Hattar et al., 2001).

In patients with active AAV circulating neutrophils are primed; they have enhanced expression of surface PR3 (Muller Kobold et al., 1998a), the activation markers CD66b,

CD63, and CD64 (Muller Kobold et al., 1998b) and increased basal superoxide production (Harper et al., 2001).

When ANCA bind their target antigens on the neutrophil surface, they also crosslink with Fc receptors causing cell activation (Porges et al., 1994). Neutrophil-endothelial adhesion changes from rolling to firm in an integrin dependent manner, followed by respiratory burst and degranulation. Degranulation releases potentially damaging reactive oxygen species, chemokines, cytokines and proteolytic enzymes in close vicinity to the endothelium (Harper et al., 2004). NO generated independent of classical

NO synthase (NOS) pathways is also released (Tse et al., 2001), although the consequences of the increased availability of NO is unclear. While NO is a potent vasodilator and has been shown to reduce neutrophil endothelial interactions, it is also cytotoxic, increases vascular permeability, forms free radicals and can modulate cytokine production in leucocytes, thus could conceivably be contributing to endothelial injury in this context (Tripathi et al., 2007).

Neutrophil activation by ANCA is followed by apoptosis, however in vitro experiments indicate this is accelerated and dysregulated; while neutrophils showed typical features of apoptosis such as shrunken nuclei and internucleasomal cleavage of DNA phosphatidylserine (PS) does not become externalized (Harper et al., 2000). The externalisation of PS is essential for recognition of apoptotic cells by phagocytes, which engulf and safely remove them preventing release of damaging cell contents. This poor clearance explains the leucocytoclasia (accumulation of unscavenged apoptotic or necrotic neutrophils or their fragmented nuclei in the tissues around the vessels) often seen in vasculitic lesions.

As well as neutrophil activation ANCA have the potential to initiate endothelial injury by other mechanisms. Both PR3-ANCA and MPO-ANCA can act directly on their target enzymes altering their function. PR3 degrades extracellular matrix proteins such as elastin and collagen type IV. Its natural inhibitor is α1-antitrypsin which forms complexes with PR3 causing it extensive structural reorganization and irreversible loss of function. PR3-ANCA inhibits both the proteolytic activity of PR3 and its complex formation with α1-antitrypsin (van der Geld et al., 2002,Dolman et al., 1993) which may impair both regulation and clearance of PR3 (Dolman et al., 1993). MPO is involved in the reaction between hydrogen peroxide and chloride ions which generates hypochlorous acid (HOCl), one of the reactive oxygen species that contributes to the neutrophil oxidative burst. In vitro, MPO-ANCA can activate MPO and may directly initiate neutrophil oxidative burst, resulting in the release of a number of EC damaging

ROS (Guilpain et al., 2007).

PR3-ANCA can act directly on the endothelium, despite endothelial cells not expressing it (or MPO either) and the mechanism by which this occurs remains elusive

(Pendergraft et al., 2000). Consequences include activation and upregulation of adhesion molecules such as ICAM-1 (de et al., 1997a) and VCAM-1 (Mayet et al.,

1996), expression of TF and release of IL-1 (de et al., 1997b). Both PR3-ANCA and

MPO-ANCA can activate monocytes, resulting in production of IL-8 (Ralston et al.,

1997) and increasing expression of CD14 (LPS receptor) and CD18 (a 2 integrin receptor)(Nowack et al., 2000). Interestingly glomerular epithelial cells are also reported to express PR3 mRNA, raising the possibility of direct glomerular epithelial injury by ANCA (Schwarting et al., 2000)

1.3.3.3 Anti-endothelial cell antibodies

A number of autoantibodies directed against the endothelium have been detected in vasculitis. The antigens targeted by AECA encompass a broad spectrum and remain poorly identified; AECA positive sera displays a broad reactivity against EC from many different locations (Praprotnik et al., 2001) and different species (reacting with bovine and murine endothelial cells) (Shoenfeld, 2002) as well as a number of other targets, including extracellular matrix (Belizna and Tervaert, 1997). IgG, IgM, and IgA isotypes of those antibodies have all been reported (Praprotnik et al., 2001).

AECA from the sera of patients with WG are able to activate endothelial cells directly, up-regulating expression of adhesion molecules such as E-selectin and inducing production of proinflammatory cytokines and chemokines including IL-1, IL-6, IL-8, and MCP-1 (Del et al., 1996) via NFκB activation (Blank et al., 1999). AECA have also been shown to induce apoptosis of endothelial cells (Bordron et al., 1998) and upregulate TF (Tannenbaum et al., 1986).

Mice immunized with IgG from the sera of patients with WG after any ANCA present has been depleted, develop a vasculitic disorder characterized by lymphocytic infiltration in vessel walls, particularly in the lungs and kidney (Damianovich et al.,

1996). As in the previously described ANCA models injection of human IgG results in an idiopathic response, whereby antibodies are initially generated against the human

IgG, then a second response generates antibodies against these. The second stage antibodies have the ability to bind to the target endothelial antigens.

1.3.3.4 T cells

T cells are activated when APCs present specific antigen to them via the MHC molecules on their surface, in combination with co-stimulatory signals. T cells can broadly fall under the categories of T helper cells (CD4+), whereby they release cytokines promoting the immune response, cytotoxic T cells (CD8+) whereby their function is to remove altered-self cells, or regulatory T cells whereupon they suppress immune responses (Goldsby RA et al., 2003a). Memory cells of these subtypes are also found, as are natural killer cells, a T cell subtype which recognizes and eliminates host cells missing MHC class I molecules (Krzewski and Strominger, 2008). CD4+ T helper cells can secrete different profiles of cytokines. Th1 type cells produce TNFα and IFNγ which are responsible for promoting the cellular immune system, activating macrophages and cytotoxic cells, whilst Th2 type cells release IL-4, IL-6, IL-10 which activates the humoral immune system, inducing B cell proliferation and antibody production (Goldsby RA et al., 2003a). A recently described population of T helper cells, Th17s produce Il-17, IL-21 and IL-22. IL-17 in particular is a potent chemoattractant of neutrophils and can induce production of other inflammatory cytokines and chemokines (Kolls and Linden, 2004). There has been much interest in this novel T cell population, with evidence indicating that IL-17 is involved in the pathogenesis of many autoimmune conditions (Bettelli et al., 2007).

In AAV T cells can be activated by MPO and PR3 (King et al., 1998) and in patients there is evidence of activated T cells irrespective of phase of disease or therapy as demonstrated by reduced CD28 (co-stimulatory molecule) and increased CD69 (an early T-cell activation marker) on CD3 T cells, and CD38 on CD8+ cytotoxic T-cells

(present on mitogen stimulated T cells) (Christensson et al., 2000). In WG activated

CD4+ cells have a predominantly Th1 cytokine profile (Ludviksson et al., 1998), although in disease remission CD4+ cells are skewed towards a Th2, Th17 response

(Abdulahad et al., 2008). In WG there is also evidence of a PR3 specific Th17 population (Abdulahad et al., 2008). Both CD4+ and CD8+ T cells are found in lesions in AAV, as are the adhesion molecules and cytokines that are required for their recruitment. In pauci-immune glomerulonephritis (so called because of lack of immune complex deposition) cellular infiltrates at the site of injury include activated T cells, indicative of a casual role in this particular disease manifestation (Cunningham et al., 1999).

In KD activated T cells have been found in skin biopsies and coronary vascular lesions

(Terai et al., 1990), however there are conflicting reports regarding whether peripheral blood T cells are activated in acute KD. While some indicate increased activation

(Barron et al., 1988), others have demonstrated no overall difference from controls, although subfamilies of T cells bearing particular Vβ segments may be activated preferentially (Brogan et al., 2008).

T cells also play a pivotal role in giant cell arteritis (not a disease of childhood, however the mechanism bears mention) and are believed to direct a response to arterial antigens -

T cells in the arterial wall have identical sequences at the third complementary determining region (CDR3) of their T-cell receptors suggesting a common antigen

(Weyand et al., 1994).

Regulatory T cells are a topic of great interest in the context of autoimmune disease.

They are characterised by expression of CD4, CD25, and the transcription factor

FOXP3, are anergic to proliferative responses in vitro and do not express key cytokines, such as IL-2 or IFN-γ in response to stimulation. Functionally, they suppress proliferation of effector T cells, which in vitro requires activation and cell contact but is not via IL-4, IL-10, or transforming growth factor-β (TGF-β) production (Torgerson,

2006). Studies in KD indicate that regulatory T cells are reduced in acute disease

(Furuno et al., 2004), while in BD (Hamzaoui et al., 2006) and WG they are elevated, however their functional capacity to suppress proliferation was found to be decreased in

WG (Abdulahad et al., 2007).

1.3.3.5 Immune complexes, complement and cryoglobulins

Immune complexes occur when antibodies bind to soluble antigen – the mechanism of type III hypersensitivity reactions. Once formed immune complexes can be deposited in various tissues, instigating an inflammatory response. Immune complexes are found in a number of small vessel vasculitides including HSP, PAN and cryoglobulinaemia and are a major cause of associated glomerulonephritis (Guillevin and Dorner, 2007).

Endothelial injury mediated by immune complexes involves recruitment of inflammatory cells and also activation of classical complement pathways (Claudy,

1998). C1q, the first component of complement, binds to the Fc portion of IgG or IgM antibodies in the deposited immune complexes resulting in activation of other components of the enzymatic cascade which culminates in the formation of a membrane attack complex and subsequent cell lysis (Goldsby RA et al., 2003b). Some components

(such as C5a) have chemotactic properties and can recruit neutrophils, monocytes and lymphocytes (Guo and Ward, 2005). C5a can also activated neutrophils, inducing oxidative burst and degranulation (Guo and Ward, 2005). The deposition of immune

complexes in the vessel wall can therefore target these activated inflammatory cells to the endothelium.

Cryoglubulins are immunoglobulins that persist in the serum, precipitate with cold temperature, and resolubilize when rewarmed (Cacoub et al., 2002). They are associated with some forms of vasculitis and also induce immune complex deposition with resultant vascular injury (Cacoub et al., 2002).

1.3.3.6 Endothelial immune interactions – amplifying inflammation?

As previously mentioned, endothelial cells play a pivotal role in inflammatory processes and it is thus not surprising that they can play a role in amplifying the vasculitic process.

Activation of the endothelium induces upregulation of adhesion molecules to facilitate leukocyte recruitment and adhesion (Rao et al., 2007) and release of pro-inflammatory cytokines which act in both an autocrine and paracrine manner; increasing endothelial activation and activating surrounding cells such as neutrophils to induce further injury

(Krishnaswamy et al., 1999).

In PAN it has been shown that while early on in lesional formation there is increased expression of E-selectin and VCAM-1 at the site of lesions, as disease progresses adhesion molecules are downregulated in the lesion but upregulated on surrounding new formed microvasculature (Coll-Vinent et al., 1998). Interestingly animal models indicate that upregulation of adhesion molecules occurs normally in newly formed vessels (Vallien et al., 2000). Thus reparative responses such as angiogenesis could exacerbate disease if inflammation is not controlled by facilitating in recruitment of more inflammatory cells to the site of injury (Coll-Vinent et al., 1998).

1.3.4 Conclusions regarding the pathogenesis of vasculitis

Complex pathways involving immune cells, immune complexes and pro-inflammatory pathways as well as the endothelium itself are involved in the pathogenesis of vasculitis, although the upstream mechanisms which initiate and maintain endothelial injury remain to be fully elucidated. Whatever the exact mechanisms involved in different vasculitis syndromes, the common end result is injury to the endothelium. This concept will be examined in further detail in the subsequent chapters of this thesis.

1.4 Assessment of disease activity in vasculitis – the need for reliable biomarkers

After initial disease is brought into remission, it is not uncommon in a number of the vasculitides for repeated recurrent flares of the disease to occur. This is a particular problem for PAN and the AAV (Gayraud et al., 2001). In this context it is thus important clinically to be able to distinguish between vasculitic disease activity in which escalation of immunosuppression is required or treatment related manifestations such as opportunistic infection, where further immunosuppresion is unneccessary or could be detrimental.

In the late 1990s, an expert working group at the NIH (National Institutes of Health,

USA) was established to define the concept of biomarkers and their role in defining clinical outcomes and endpoints for trials. They defined a biomarker as: "a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” (Biomarkers definitions working group, 2001). In the context of vasculitis 52

there is great interest in finding novel biomarkers that could be used to routinely assess disease activity in a minimally invasive manner, a number of which are discussed in detail below.

It would have been virtually impossible to assess all the potential biomarkers described herein within the vasculitis cohort at GOSH. Therefore this thesis focuses on biomarkers associated with endothelial injury, a phenomenon common to all of the vasculitides, irrespective of causal pathology as discussed in Section 1.3. Biomarkers that reflect endothelial activation and injury therefore have the potential to monitor disease activity over a heterogeneous cohort (Brogan et al., 2004a). While it is difficult to access the tissue directly, any factors released by the endothelium enter the blood, thus providing a window to the affected organ. Before consideration is given to these novel biomarkers, some description of routine clinical assessment of vasculitis disease activity is required.

1.4.1 Routine clinical tests for diagnosis and assessment of disease activity

1.4.1.1 Initial diagnosis

The symptoms of vasculitis are shared with a wide number of other conditions and therefore numerous investigations are performed in order to distinguish it from other diseases. In the first instance these include haematological tests such as blood counts, erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), assessment of renal, liver and thyroid function, screens for infectious diseases, screening for prothrombotic clotting factors and certain genetic mutations as well as a multitude of immunological tests, including autoantibodies (antinuclear antibodies (ANA), anti double stranded DNA antibodies (DS-DNA), rheumatoid factor (RF), antibodies to

extractable nuclear antigens such as Ro and La (ENAs), anticardiolipin antibodies, anti- glomerular basement membrane antibodies (GBM)), immunoglobulins, cryoglobulins, factors involved in the complement pathways (MBL, C3/C4/C5a) and cytokines

(Damianovich et al., 1996,Rees et al., 2007). Radiological investigations, biopsy and further testing to exclude rare conditions (such as malignancy) and to further confirm the presence or absence of vasculitis would then be dependent on the individual patient symptoms and organ involvement (Rees et al., 2007).

Radiological investigations are most useful in the medium to large vessel vasculitides, in smaller vessel diseases the techniques may not be sufficiently sensitive to detect abnormalities. At GOSH digital subtraction arteriography, magnetic resonance imaging and angiography (MRI/MRA) and positron emission tomography (PET) may be used for initial diagnostic assessment. Of these, arteriography is the most invasive, whereupon a contrast agent is injected into the arteries at the site of interest (selective visceral and/or cerebral arteriography), via a catheter and guide wires usually inserted through the femoral artery (Brogan et al., 2002). This allows X-rays to be taken which will show changes within the vessel lumen, demonstrating aneurysms occlusions and stenosis (Brogan et al., 2002).

MRI and MRA are relatively non-invasive techniques which can provide additional information about the vessel wall (MRI) and lumen (MRA). MRI can give indication of thickening of the vessel walls including oedema which may be indicative of inflammation (Pipitone et al., 2008). MRA is used to assess aneurysms, occlusions and stenosis, however it can overestimate the degree of stenosis and thus have higher false positives than conventional areriography (Gotway et al., 2005). Conversely MRA lacks

sufficient sensitivity for the detection of changes affecting smaller muscular arteries frequently observed in PAN (Brogan et al., 2002).

PET scanning utilizes a radiolabelled glucose analogue (18fluorodeoxyglucose (18FDG)) to highlight cells which have a higher level of glucose uptake due to increased metabolic activity. At sites of active vasculitis uptake is high (Belhocine et al., 2003), however although it is unlikely to be as large a confounding factor in paediatric cohorts as it is in adults, atherosclerotic lesions are also sites of high metabolic activity (Yun et al., 2001) and thus great care must be taken when interpreting the scans. Co-registration of PET with computed tomography (PET-CT) vastly improves anatomical resolution and is useful for TD, although radiation exposure is very high (Henes et al., 2008).

Evidence of vasculitic disease activity on a tissue biopsy is frequently referred to as the

“gold standard” for confirming active disease (Mukhtyar et al., 2006). However due to the localised nature of the disease, a negative result on biopsy does not indicate that disease is absent. For example in patients with well-documented active WG, only around 50% of nasal biopsies demonstrated evidence of active vasculitis (Del Buono and Flint, 1991). Ultimately diagnosis is based on the ability to assimilate the evidence pointing to vasculitis and to exclude other conditions that may mimic it through a broad array of clinical investigations.

1.4.1.2 Routine assessment of disease activity following diagnosis

As treatment for vasculitis is often toxic, a well defined diagnosis at outset is essential.

In addition, the ability to monitor disease activity is equally important due to the relapsing nature of a number of the vasculitides and the need for rapid intervention if or

when disease returns. Careful clinical evaluation of individual patients, with assessment of changes in symptoms and biological and radiological markers is the only way in which disease activity and response to treatment can be assessed. This is the principle behind the Birmingham Vasculitis Activity Score (BVAS) (Appendix 2, discussed in

Section 2.2.2) which is now routinely used to quantify disease activity in clinical trials and increasingly in routine clinical practice as well (Luqmani et al., 1994).

Commonly used biological markers of inflammation such as ESR and CRP are non specific and cannot easily distinguish between infection or disease activity. During active disease, levels of these inflammatory markers may also remain normal, particularly in episodes of localised disease and granulomatous flares (Salvarani et al.,

2003,Sproson et al., 2007), so caution must be taken when interpreting such information.

In the ANCA-associated vasculitides routine monitoring of ANCA levels is commonplace, both PR3 and MPO-ANCA can be detected using ELISA or indirect immunofluorescence (Hagen et al., 1996). However, the usefulness of ANCA as a biomarker remains controversial; while transient rises in ANCA often occur during relapse many patients who have a rise in ANCA remain well (Mukhtyar et al., 2008).

Although histopathological evidence on biopsy is a reliable confirmation of active disease in vasculitis, it is not a practical measure for routinely assessing disease activity due to its invasive nature. The same applies for radiological investigations, while changes in radiological images are indicative of ongoing disease activity, the techniques can be extremely invasive and are also expensive, time consuming and in some

instances carry a significant radiation level. Moreover, contrast agents used in

MRI/MRA protocols have recently been associated with nephrogenic systemic fibrosis in patients with renal impairment (Agarwal et al., 2008,Penfield, 2008). Thus it is with caution that MRI/MRA should be performed in patients with renal vasculitis.

1.4.2 Novel Biomarkers in vasculitis

Due to the complexity of diagnosing and assessing vasculitis there has been much interest in finding novel biomarkers which may prove useful in this context. Possible biomarkers for vasculitis that have been investigated range from autoantibodies, to soluble adhesion molecules, cytokines, growth factors and cellular fragments.

1.4.2.1 Autoantibodies

Anti-endothelial cell antibodies

As described in Section 1.3 AECA may have a pathological role in vasculitis. AECA are commonly detected using a cell based ELISA with HUVEC as the substrate and their presence has been observed in a number of the vasculitides, including WG (Savage et al., 1991), PAN and BD (Navarro et al., 1997), TD (Park et al., 2006) and KD

(Fujieda et al., 1997). However some studies completely contradict these findings with no difference in the proportion of samples positive for AECA in patients or controls

(Dinc et al., 2003,Varagunam et al., 1993). A small number of studies have shown

AECA levels fluctuate correspondingly with disease activity in vasculitis (Park et al.,

2006), however other groups have shown no such relationship (Sebastian et al., 2007).

The discrepancies in findings have been in part attributed to the non standardised methodologies used to detect AECA; while the majority of groups use variations on cellular ELISA, flow cytometry and a number of other methods have also been

employed (Belizna and Tervaert, 1997). While experimental evidence points to AECA as a potential mechanism of endothelial activation both in vivo and in vitro, the contrasting observations in patients do not currently support its role as a biomarker of disease activity.

Anti-heat shock protein antibodies

Heat shock proteins (HSPs) are highly evolutionary conserved proteins that are upregulated in response to a number of environmental stresses (including, hypoxia, heat-shock and inflammation) (Tsan and Gao, 2004). Many family members also function constitutively, acting as molecular chaperones – stabilizing in folding and assisting in the intracellular trafficking of peptides, including complex formation.

Under stress, their increased expression is protective assisting in cellular repair and inhibiting a number of apoptotic signalling pathways (Lanneau et al., 2008).

Because of the highly conserved structure of HSPs throughout species, the presence of antibodies to bacterial and human homologue HSPs in vasculitis has been proposed as evidence of molecular mimicry causing disease, particularly in KD (Sireci et al., 2000) and BD (Tanaka et al., 1999). As HSPs can bind with cell membrane peptides, it has been suggested that HSPs may be a target of AECA (Jamin et al., 2005). As with other autoantibodies in vasculitis the evidence regarding anti-HSPs is conflicting, while some studies show elevated levels of antibodies to specific HSPs in patients with vasculitis such as alpha-enolase in KD (Chun et al., 2008) others show no differences in expression from controls (Sherer et al., 2008). As yet no study has examined the changes in anti-HSPs with treatment and thus the lack of reproducible data and the

possibility that it may only be relevant in a small proportion of patients indicates this may not prove a suitable biomarker in a varied cohort.

Other autoantibodies

Other autoantibodies that have been shown to be elevated in vasculitis include antiphospholipid antibodies and anti glomerular basement membrane antibodies.

Neither of these are specific for vasculitis, nor have been studied in the context of disease progression, however both may influence the type of disease complications which occur. Antiphospholipid antibodies are a heterogenous groups of autoantibodies specific for anionic phospolipids (including anti-cardiolipin, lupus anticoagulant and anti-β2 glycoprotein), and have been documented in WG (Hergesell et al., 1993) and

PAN (de la Fuente and Grana, 1994), with a small number of studies suggesting an association with adverse thrombotic events (Morelli et al., 1998). Antibodies targeting the glomerular basement membrane are rare, but their presence is associated with renal and pulmonary complications (Goodpasture’s disease). While these may not provide a direct assessment of disease activity, their presence or absence could provide useful insight for treatment in certain vasculitides, albeit the vast minority of cases (Rutgers et al., 2005).

1.4.2.2 Soluble adhesion molecules

Adhesion molecules mediate the attachment of cells to each other, playing a pivotal role in the recruitment of inflammatory cells to the site of injury in vasculitis. Proteolytic cleavage of the extracellular portion of cell membrane expressed adhesion molecules results in a soluble form being released into the circulation and explains in part the transient expression of adhesion molecules on activated cells (Gearing and Newman,

1993). There is also evidence that some adhesion molecules (such as P-selectin) may have alternatively spliced forms which lack a trans-membrane domain and thus are produced in a soluble form (Ishiwata et al., 1994).

L-selectin is constitutively expressed on leukocytes, but is rapidly shed from the surface of activated cells. In-vitro the soluble form can inhibit lymphocyte-endothelial interactions (Schleiffenbaum et al., 1992). In patients with active AAV (Ara et al.,

2001), PAN (Coll-Vinent et al., 1997) and KD (Takeshita et al., 1997) L-selectin levels were lower than in controls, however in KD levels rose during the convalescent stages of the disease (Takeshita et al., 1997).

P-selectin is a preformed molecule which is stored in the granules of platelets and the

Weibel-Palade bodies of the endothelium, and is involved in the interaction between platelets, the endothelium and leukocytes, thus playing a key role in vascular inflammation. The soluble form has been shown to be upregulated in active BD

(Turkoz et al., 2005), and in the subacute phase of KD, correlating with platelet count

(Takeshita et al., 1997). In AAV and PAN P-selectin levels did not differ from controls

(Ara et al., 2001,Coll-Vinent et al., 1997).

E-selectin is highly selective, only being found on activated endothelial cells in a transient manner. Its soluble form is detected in the supernatants of TNFα and IL-1 stimulated ECs (Pigott et al., 1992). Correspondingly it is upregulated in active vasculitis, in AAV (Ara et al., 2001), PAN (Coll-Vinent et al., 1997) and KD (Takeshita et al., 1997) and declines with remission inducing treatment. In KD it showed

correlation with disease activity and was highest in patients with coronary artery lesions

(Furui et al., 2002).

Both ICAM-1 and VCAM-1 are released in a soluble form from activated ECs in vitro

(Pigott et al., 1992). Unlike E-selectin their expression is not specific to the endothelium, however they are both upregulated in vasculitis (Wang et al.,

2006,Stegeman et al., 1994,Verity et al., 1998) and in some studies have shown correlation with disease activity (Stegeman et al., 1994), although not in all (Wang et al., 2006,Ara et al., 2001).

While soluble adhesion molecules have been widely studied in the vasculitides the data is ambiguous, with good correlation with disease activity reported in some studies(Furui et al., 2002,Stegeman et al., 1994), but not in others (Ara et al., 2001). The predominant methodology for detection of these molecules is ELISA, which may not distinguish between biologically active and inactive forms of the adhesion molecules and may also be detecting adhesion molecules associated with membrane fragments (microparticles) rather than true soluble molecules.

1.4.2.3 Other peptides – cytokines, chemokines, clotting and growth factors

Unsurprisingly a number of the vasculitides are associated with elevations in inflammatory cytokines such as IL-1, TNFα, IFNγ Il-17, Il-23 and Il-6 (Ohlsson et al.,

2004,Chi et al., 2008,Grau et al., 1989) and chemokines such as RANTES, fractalkine, eotaxin, MCP-1 and IL-8 (Torheim et al., 2005,Bjerkeli et al., 2007,Dhawan et al.,

2006). Both are produced by a wide variety of cell types, including the endothelium and recruited leukocytes. The actual cytokine profile in individual patients is extremely

varied and as in the case of elevated ESR and CRP, cytokines and chemokines are not specific to vasculitis and can be upregulated in inflammation without the presence of active vasculitic disease. Thus their use as a biomarker for disease activity would require careful interpretation. However with the increasing use of biological therapies to Il-1 and TNFα in patients refractory to other treatments, the presence of high levels of these cytokines may provide valuable clinical insight regarding indication for treatment with these agents. Interestingly in AAV at least, although anti-TNFα therapy

(infliximab) induced remission in 10 patients with active disease, levels of serum TNFα were not significantly different to controls, nor was there a significant change observed with treatment in individual patients (Booth et al., 2004b). Further studies will be necessary to clarify any diagnostic potential of these cytokines.

Von Willebrand Factor (VWf) is released in large amounts from Weibel-Palade bodies by activated and damaged endothelium. It is important in coagulation, binding exposed collagen in the vessel wall, platelet receptors to form complexes even under high shear stress, and clotting factor VII preventing its inactivation (Ruggeri, 2007). In vasculitis a number of studies have assessed its potential as a biomarker of disease activity with promising results. VWf correlates with disease activity in WG (D'Cruz et al., 1999) and

HSP (Yilmaz et al., 2005), although in acute KD despite being elevated compared to healthy controls there were no differences to febrile controls (Lin et al., 2005) suggesting a lack of specificity for vasculitis. Another peptide involved in haemostasis is thrombomodulin which acts as a receptor for thrombin and functions as a natural anticoagulant. It is found on endothelial cells and is cleaved on cellular activation. In vasculitis it has been to shown to correlate with disease activity in WG and MPA

(Hergesell et al., 1996) (Boehme et al., 1996) although like VWf is unlikely to be specific for vasculitis and is upregulated in infection (Van de Wouwer M. et al., 2004).

In response to vascular injury, the endothelium secretes numerous growth factors to assist in repair. This important topic is discussed in more detail in Chapter 5. In vasculitis a number of studies have examined VEGF levels in patient plasma or sera and found it to be upregulated in active disease in HSP (Topaloglu et al., 2001), BD (Shaker et al., 2007) and in KD (Terai et al., 1999), which decreased on recovery. Data also suggests in KD that high VEGF may be a risk factor for development of coronary artery lesions (Ohno et al., 2000). Interestingly in KD the antiangiogenic growth factor endostatin was shown to be lower in active disease (Takeshita et al., 2005).

1.4.2.4 Cells and cellular fragments

When cells become activated, membrane blebs or microparticles (MPs) are shed and can be detected in the circulation (Piccin et al., 2007). Thus cellular MPs could provide valuable information about activated cell populations and hypothetically may be able to predict disease flare prior to significant vascular injury by early detection of activated or primed cells before irreversible vascular injury occurs. MPs are discussed in detail in

Chapter 4 where they are assessed as a potential biomarker of disease activity in vasculitis.

If endothelial cells become sufficiently activated, they may become apoptotic or necrotic and whole cells may become detached from the vessel wall. Circulating endothelial cells (CECs) have been detected in a number of conditions associated with

vascular injury (Blann et al., 2005) and are discussed in detail in Chapter 3, where they have been assessed as a potential biomarker of vasculitic disease activity.

Conversely endothelial injury must be followed by endothelial repair in order to maintain health, and markers of putative reparative response such as endothelial progenitor cells (EPCs) and the growth factor signalling pathways involved in their recruitment may also provide insight into disease activity. This important area is discussed in more detail in Chapters 5 and 6.

While these markers of endothelial activation and injury are of particular relevance in the vasculitis cohort, they also have significance in a number of other conditions affecting the vasculature, including atherosclerosis. Although this latter condition is not so relevant in paediatric cohorts (where atherosclerotic changes are rare), it may be of greater significance in adult cohorts. That said, as will be discussed later in this thesis, both endothelial MPs and CECs have been studied in adults with vasculitis with promising results (Erdbruegger et al., 2008).

1.5 Aims of this thesis

The aim of this thesis is to evaluate novel putative biomarkers of disease activity in primary systemic vasculitis. As the endothelium is the target organ of the disease, this objective has been addressed by focusing on surrogate markers of endothelial activation and injury and subsequent reparative responses. The biomarkers studied in subsequent chapters of this thesis include circulating endothelial cells and endothelial microparticles, growth factors associated with angiogenesis and vasculogenesis and endothelial progenitor cells. A major factor of work involving children is the small volumes of blood available on which to perform such tests and thus much care was taken to minimise blood volumes necessary for study.

2 Methods and Materials

2.1. Introduction

2.2. Patients, vasculitis classification and assessment of disease activity

2.3. Materials

2.4. Preparation of blood and tissue samples

2.5. Assessment of endothelial activation and injury

2.6. Assessment of potential endothelial repair responses

2.7. Statistics

2.1 Introduction

This section details materials and methods which are used in more than one chapter.

Where methodologies have been adapted or developed specifically for this study, they are discussed in greater detail in the relevant chapter.

2.2 Patients, vasculitis classification and assessment of disease activity

All patients enrolled in the study were assessed by rheumatologists at GOSH. Full informed consent was obtained for all patients and controls (parental) and children’s assent was sought where appropriate on enrolment in the study.

2.2.1 Classification of the vasculitic syndromes

As discussed in Section 1.2 vasculitis was classified predominantly according to the

Chapel Hill Consensus Criteria (Jennette et al., 1994), however the American College of

Rheumatology (ACR) criteria (Fries et al., 1990) and the EULAR/PRES endorsed consensus criteria (Ozen et al., 2006) for vasculitis of the young were applied as additional classification tools. The Chapel Hill criteria classifies vasculitis on the size of the vessel predominantly affected and also forms the basis of the EULAR/PRES criteria, which contains further modifications relating to the paediatric population.

These include; the exclusion of diseases not relevant in childhood and inclusion of diseases that are not addressed by the adult system, an extra category of disease entitled

“other diseases” and the use of vessel “predominantly affected” to categorise patients whose disease overlaps vessel sizes (Ozen et al., 2006). Appendix 1 lists all criteria.

2.2.2 Assessment of disease activity

Disease activity was assessed using a modified paediatric version of the Birmingham

Vasculitis Activity Score (BVAS) (Luqmani et al., 1994), a scoring system that is used to give an overall view of disease activity in vasculitis patients (Appendix 2). It has never been formally validated in children. However, in a previous report from our group a modified paediatric version of BVAS was found to significantly correlate with the conventional acute phase marker CRP (as in adults) (Brogan et al., 2004a,Luqmani et al., 1994).

The aim of the BVAS scoring system is to give a uniform assessment of current disease activity in groups of patients with different forms of vasculitis and is not specifically designed for any single vasculitic syndrome, although a specific form of BVAS for

Wegener’s Granulomatosis has been described for use in adult patients (Jayne,

2001,Stone et al., 2001). At time of scoring, disease activity is only considered for 1 month previously. The score is weighted more heavily towards biochemical and objective evidence of active vasculitis (such as organ involvement e.g. gut or kidneys) rather than symptoms which have a subjective element (e.g. myalgia, arthralgia).

Despite the inherent potential limitations and scorer subjectivity of the BVAS this system does provide an overall clinical score of vasculitic disease activity and has proved a successful tool for use in clinical trials (Jayne, 2001).

In this study, active vasculitis was defined as a BVAS score greater than 0, where the scored criteria were newly appearing or worse in the preceding 4 weeks and attributable to vasculitis, having excluded other causes such as infection. The maximum score

attainable was 63. Disease remission was defined as a score of 0/63. The modified

BVAS used in this thesis is shown in Appendix 2.

2.3 Materials

2.3.1 Reagents for blood collection

After blood was obtained it was aliquoted into one or more of the following collection bottles depending on use; 1.4ml 3.2% trisodium citrate (Sarstedt), 5ml EDTA

(International scientific supplies), 5ml polypropylene containing no anticoagulant or sterile 20ml Universal bottles containing 40µl of preservative-free heparin (Monoparin,

CP Pharmaceuticals Ltd, 1000 U/ml).

2.3.2 Fluorochrome conjugated antibodies for flow cytometry

The fluorochrome conjugated antibodies used are listed in Table 2.1. Antibodies were diluted using 0.01M PBS + 0.1% azide for staining of peripheral blood mononuclear cells (PBMCS) and Human umbilical vein endothelial cells (HUVEC) and sterile filtered RPMI 1640 with 5% FCS for staining of microparticles (MPs). Where antibodies were used in whole blood they were added neat to give the final dilution given in Table 2.1. All antibodies were titrated alongside their manufacturers recommended isotype control to determine their optimal concentration and minimise non-specific binding. FC receptor blocking reagent (Miltenyi biotech) was used to further reduce non specific antibody binding.

Table 2-1: Fluorochrome conjugated antibodies for flow cytometry

Specificity Isotype Conjugate Clone Dilution Source Usage

CD34 IgG1 PERCP 8G12 1:10 BD E,EC,C

CD133 IgG1 PE AC133 1:10 Miltenyi E

CD14 IgG1 FITC Mem-18 1:20 Serotec E

CD14 IgG1 PE Mem-18 1:20/1:15 Serotec E,M

CD144 Poly FITC - 1:15 Serotec E

CD144 IgGIgG1 PE 16B1 1:20/1:15 eBiosciences E,M CD45 IgG1κ FITC HI30 1:20 eBiosciences C

CD146 IgG1 κ PE P1H12 1:20 BD E,EC,C

CD3 IgG1 CY UCHT1 1:20/1:15 BD E,M

KDR IgG1 BIOTIN KDR-2 1:10 Sigma E

CD105 IgG1 PE SN6 1:15 Serotec M

CD62e IgG1κ PE 68-5H11 1:15/1:10 BD M,EC

CD62p IgG1κ PE AK4 1:15 BD M

CD62L IgG1κ PE-CY5 DREG-56 1:15/1:20 BD M, N

CD11b IgG1κ PE ICRF44 1:15/1:20 BD M,N

CD15 IgG1κ PE HLDA8 1:15/1:20 Biolegend M,N

CD54 IgG1κ PE-CY5 HA58 1:15 BD M

CD66b IgG1 PE Birma17C 1:40/1:25 IBGRL M,N

CD42a IgG1κ Percp Beb1 1:15 BD M

TF IgG1κ PE HTF-1 1:15 BD M

Use in this thesis E = endothelial progenitor cells, C = Circulating endothelial cells, M = microparticles, EC = in vitro endothelial activation, N = in vitro neutrophil experiments

2.3.3 Tissue culture media

The following tissue culture media was used: RPMI-1640 culture medium containing

2mM L-glutamine (Invitrogen), MCDB-131 (Invitrogen) and the EGM-2 (endothelial growth media) bullet kit (Cambrex). The EGM-2 bullet kit consisted of EBM-2

(endothelial basal media) with the following supplementation (concentrations were not given) human recombinant Epidermal Growth Factor (hEGF), human Fibroblast

Growth Factor - Basic (hFGF-B), Vascular Endothelial Growth Factor (VEGF),

Ascorbic Acid (Vitamin C), Hydrocortisone, Long human Recombinant Insulin-like

Growth Factor (R3-IGF-1), Heparin, Fetal Bovine Serum (FBS), Gentamicin and

Amphotericin.

Further supplementation to the culture media depended on the culture conditions as described in the individual methodologies, however the following were also used; penicillin and streptomycin (10,000U and 10,000µg, diluted 1:100, Invitrogen), antibiotic-antimycotic liquid 100x (10,000U penicillin, 10,000 µg streptomycin, 25 µg amphotericin B, diluted 1:100, Invitrogen), Vascular Endothelial Growth Factor

(VEGF, R&D), basic human fibroblast growth factor (bFGF, R&D), heparin and fetal calf serum (FCS, Invitrogen). In most experiments FCS was heat inactivated at 56o C for 60 minutes to destroy complement. Variations in batch quality of FCS can dramatically influence HUVEC growth, therefore batches of FCS used for HUVEC culture were tested prior to purchase and bought in large quantities. To enhance cell attachment or to act as a growth scaffold the following were used: Attachment factor

(PAA systems), fibronectin (100ng/ml, Calbiochem), MatrigelTM, phenol red free (BD).

2.3.4 Cytokine and protein reagents

The following cytokines were used: recombinant TNF- (100ng/ml, Autogen Bioclear), recombinant Il-1β (100ng/ml, Autogen Bioclear). The following peptides were used: N- formyl-methyl-leucyl-phenylalanine (FMLP) (1uM, Sigma), LPS (100ng/ml, Sigma).

The following fluorochrome-conjugated proteins were used: annexin-V (FITC; 1:100 dilution, BD); Ulex europaeus I Agglutinin (UEA-I FITC, 10mg/ml, Sigma),

Streptavidin (FITC, 1:150 dilution, Serotec). Type II collagenase (Sigma) was used for isolation of HUVEC from umbilical cords. Trypsin-EDTA (10x, 0.5%, Invitrogen) was used for detachment of HUVEC from culture.

2.3.5 Bead based extraction reagents

DYNAL bead technology (Invitrogen) was used for isolation of CECs: Dynabeads Pan

Mouse IgG were labeled with an unconjugated IgG1 antibody to CD146, clone s-endo 1

(BioCytex). The Dynal MPC-L magnet which holds 1-8ml tubes was also used.

2.3.6 ELISA kits

The following ELISA kits were used to determine levels of growth factors in patients plasma and serum: Angiopoeitin-1 (R&D), Angiopoeitin-2 (R&D), Vascular endothelial growth factor VEGF) (R&D), Thymosin β4 (immundiagnostik). All test kits were used as per manufacturer recommendation.

2.3.7 Other reagents

For isolation of PBMCs from blood samples, the density centrifugation separation reagent Lymphoprep (Axis-Shield) was used. To isolate neutrophils a similar product

Polymorphprep was used (Axis-Shield). The following latex beads were used to 72

investigate forward and side scatter (FSC/SSC) flow cytometric characteristics of small particles: 0.3, 0.8, 3µm latex beads, (Sigma). Countbright beadsTM (Invitrogen) were used as in an internal standard for absolute enumeration of circulating endothelial cells and endothelial progenitor cells in whole blood samples.

2.4 Preparation of blood and tissue samples

All human blood and tissue samples were collected with full ethical approval and informed consent.

2.4.1 HUVEC

2.4.1.1 Isolation of HUVEC

Umbilical cords were collected from the maternity unit at the West Middlesex hospital,

London with ethical approval and informed consent. HUVEC were isolated from freshly collected cords in the following manner: Using tweezers cords were removed from the collection bottle (containing RPMI 1640 + 100U/ml Penicillin + 100 µg /ml

Streptomycin + 0.25 µg/ml Amphotericin) and sprayed liberally with IMS. Any blood inside the cord was squeezed out and discarded. One end was clamped approximately

2cm in and the end cut off. At the other end, a further 1-2cm was removed and a Kwill

(Universal Hospital Supplies) was inserted into the umbilical vein and clamped into position. A syringe filled with RPMI 1640 (with antibiotics and antimycotic as above) was squeezed into the cord via the Kwill and used to fill it. The clamped end of the cord was then released and the vein contents washed out into a clean waste beaker. The vein was then filled with warm (37oC) collagenase type II solution (0.1%, dissolved in

RPMI-1640 and passed through a 0.2µm tissue culture filter, aliquoted and stored at - 73

20°C). Both ends were clamped and the cord left for 10 minutes in the incubator. The contents of the vein were then emptied into a sterile collection bottle and a further wash through with RPMI (containing antibiotics only + 5% heat inactivated FCS) was performed. The collected vein contents containing HUVEC were then centrifuged for 8 minutes at 450g creating a cell pellet.

2.4.1.2 HUVEC culture

Pelleted HUVEC were resuspended in MCDB131 (containing 20% heat inactivated

FCS + antibiotics) and placed in a 25 cm2 flask, precoated (30 minutes prior to use) with

o attachment factor for optimal cell attachment. Cells were cultured at 37 C with 5% CO2 and inspected each day for growth. The media was changed every 3 days until cells were required, passaging as necessary at about 80% confluence. Initally HUVEC were easily identifiable as small clusters of oval adherent cells when inspected under phase- contrast microscopy (Figure 2.1a). As they reached confluence their appearance was characteristically “cobblestone” (Figure 2.1b). In some cultures there was significant overgrowth of smooth muscle-like or fibroblast-like cells (which are characteristically spindle-shaped), cultures containing these and those that had not grown well were discarded.

2.4.1.3 HUVEC culture passage

HUVEC that were near-confluent were washed three times in warmed PBS to remove non-adherent cells and any FCS. Cells were then washed once in 0.5ml 0.05% Trypsin-

EDTA solution (per 25 cm2 flask) which was then removed and a further 1ml Trypsin-

EDTA solution added. Cells were inspected under phase-contrast microscopy, and when rounding and becoming dislodged (typically after 30 seconds), the flask was tapped 74

Figure 2-1: Cultured HUVEC phenotype

Figure 2.1a: Soon after they were placed in culture HUVEC were easily identifiable as small clusters of oval adherent cells when inspected under phase-contrast microscopy. Figure 2.1b: As

HUVEC reached confluence their appearance was characteristically “cobblestone” . Scale bars represent 50µm

against the bench top to aid removal. If cells did not detach, flasks were placed in the incubator for a further 30 seconds. The detached cells were quickly resuspended in warmed HUVEC culture medium and transferred to a 50-ml sterile conical tube. These were then seeded into 6 or 12 well flat-bottom tissue culture plates, precoated in attachment factor. Tissue culture plates were then incubated as before.

When cord supply was plentiful HUVEC at passage 1 and 2 were frozen down, to provide a continous supply of endothelial cells if fresh cords were unavailable. After passage, viable HUVEC were counted with trypan blue (to exclude dead cells) in a haemocytometer. A minimum of 5 x 106 cells were resuspended in heat inactivated

FCS containing 10% DMSO and frozen in a freezing container placed in a -80oC freezer. For use, cells were defrosted rapidly at 37oC and decanted straight into a T25 flask precoated with attachment factor and containing EGM-2 media with all additives.

Details of specific conditions used to grow HUVEC for various experimental procedures are described in the relevant sections in subsequent chapters. HUVEC morphology and phenotype, especially in terms of induction of cell adhesion molecules in response to pro-inflammatory stimuli may alter after serial passages. The cells used throughout this study were therefore always first to third sub-culture passage.

2.4.1.4 Spiking of whole blood samples and PBS with HUVEC

In these experiments HUVECs at passage 1 or 2 were detached as described above using 0.05% trypsin-EDTA and resuspended in PBS + 5% heat inactivated FCS + 0.1% azide. Cells were centrifuged to a pellet, resuspended in a small volume of PBS

(additives as above) and counted as described above. The HUVEC were serially diluted 76

to give stock concentrations from which 100l of cells could be added to 1ml blood or

1ml PBS. Whole blood or PBS was spiked with 0, 100, 1000, 10000 and 100000

HUVECs/ml. Cells were carefully mixed in the blood or PBS sample by gentle pipetting. Equal volumes of spiked whole blood or PBS were used for flow cytometry or immunomagnetic bead extraction in the comparative experiments described in

Chapter 3.

2.4.2 Isolation of PBMCS

PBMCs were isolated using density gradient media (Lymphoprep). 10ml of blood collected in preservative free heparin was mixed with 10ml RPMI 1640 containing L- glutamine with added antibiotics (penicillin 100U/ml and streptomycin 100g/ml) and layered over 10ml of lymphoprep in a 50ml conical tube. The sample was then centrifuged at room temperature for 25 minutes at 700g with no brake. A wide tipped

Pasteur pipette was used to aspirate PBMCs from the buffy coat layer formed into a fresh 50ml conical tube, which was then washed with RPMI 1640 containing L- glutamine (with 10% heat inactivated FCS and antibiotics as before). Cells were centrifuged at 500g for 10 minutes and the pellet re-suspended in RPMI 1640 (with FCS and antibiotics). Cells were counted as described previously. To allow batch analysis aliquots of cells (concentration 5-10x106 cells/ml) were resuspended in heat inactivated

FCS with 10% DMSO and frozen in a freezing container placed in a -80oC freezer for use in later experiments.

2.4.3 Isolation of neutrophils

Neutrophils were also isolated using density gradient media (Polymorphprep). 6ml of blood was layered directly over 6ml of polymorphprep in a 15ml falcon tube. The 77

sample was then centrifuged at 500g for 30minutes. The different gradient preparation resulted in a separation of 2 bands of cells, the lower of which constituted neutrophils with the upper band corresponding to PBMCs. A wide tipped Pasteur pipette was used to remove the upper band first, allowing better access to the lower band of interest, which was then removed in the same way. To reduce neutrophil activation, cells were resuspended in PBS and kept on ice until use.

2.4.4 Separation of platelet poor plasma

Blood collected in 3.2% trisodium citrate was spun at 5000g for 5 minutes in a microcentrifuge in 1.5ml polypropylene tubes. The supernatant was decanted and then respunded in order to obtain platelet poor plasma (PPP). The PPP was then aliquoted and stored at -80oC until use. PPP was used for detection of microparticles as described in 2.5.3 and in ELISAs for detection of Angiopoetin-1 and Angiopoetin-2.

2.4.5 Separation of serum

Uncoagulated blood was allowed to stand for 2-4 hours. The sample was then centrifuged at 1500g for 10mins, the supernatant aliquoted and samples stored at -80oC until future use. Serum was used in ELISAs for detection of VEGF and Thymosin β4.

2.5 Assessment of endothelial activation and injury

2.5.1 Detection of circulating endothelial cells using immunomagnetic beads

2.5.1.1 Preparation of immunomagnetic beads

Dynabeads (pan mouse IgG) were resuspended and 350µl was added to a 5ml polypropylene tube. The tube was then placed into the magnet (MPC-L) for 2 minutes to allow the beads to attach to the side of the tube. The supernatant was discarded using a Pasteur pipette and the beads resuspended in 1ml PBS (+0.1% BSA). The tube was placed in the magnet again and the supernatant discarded as before. The beads were resuspended in 950ul of PBS (+ 0.1% BSA) and 400ul of s-endo-1 antibody were added

(25µg/ml concentration). The tube was sealed and the sample mixed on a roller mixer for 2 hrs at 4oC. Beads were washed as before three times and resuspended in a final volume of 1ml, ready for use.

2.5.1.2 CEC extraction from peripheral blood

CECs were detected using the consensus protocol as described by Woywodt et al

(Woywodt et al., 2006a). Briefly 1ml of venous blood collected into a blood bottle containing EDTA was mixed with 1ml buffer (phosphate buffered saline (PBS) containing 0.1% bovine serum albumin (BSA) and 0.6% sodium citrate) and 20μl Fc receptor (FcR) blocking reagent and incubated for 5 minutes at room temperature. 50μl of anti-CD146 coated immunomagnetic beads was then added and the sample incubated at 4oC whilst rotating for 30 minutes. Bead-bound cells were separated using the magnet as described above and washed three times with buffer. Cells were then re-suspended in

100μl of buffer containing 10μl of 2mg/ml FITC labelled Ulex Europeus lectin and

incubated in the dark at room temperature for one hour. Cells were then washed again three times and re-suspended in a final volume of 200μl buffer.

2.5.1.3 Counting CECS

CECs in the sample were counted using a Nageotte chamber on a fluorescent microscope and were defined as the following; Ulex bright cells, greater than 10μm in size with more than 5 magnetic beads attached, a definition put forward in the 2006 consensus paper (Woywodt et al., 2006a). A Nageotte counting chamber is designed for counting of rare events and holds 50µl of sample over a grid consisting of 40 lines.

CECs were counted over 20 lines of the grid (corresponding to 25µl or 1/8 of the sample) if there were high numbers, which made it difficult to count or over all 40 lines

(corresponding to 50µl or 1/4 of the sample) if there were low numbers. Counts were then multiplied by the appropriate factor to get a total value per ml of whole blood.

2.5.2 Detection of circulating endothelial cells using flow cytometry

Enumeration of CECs has been carried out using both the immunomagnetic bead extraction as described above and flow cytometry (FC), however only one publication has demonstrated whether they give comparable results and this was not in patients with vasculitis (Goon et al., 2006). Thus both methods were used in this thesis for comparison. The flow cytometric definition of CECs definition of putative CECs as suggested by Goon et al in their comparative paper and other groups was thus: cells falling within a mononuclear cell gate that express both CD146 and CD34, but which are negative for CD45 (Goon et al., 2006,Bertolini et al., 2006,Del et al., 2004,Khan et al., 2005).

2.5.2.1 Preparation of blood samples for flow cytometry of CECs

1ml of blood collected in EDTA was lysed using 10ml of FACS lysis solution (1:10 dilution of 10x FACS lyse, BD). After centrifugation to a pellet and decantation of the supernatant, non-specific antibody binding was blocked using 20l FcR blocking reagent (Miltenyi) and 200l of mouse serum (Sigma-Aldrich) for 20 minutes at room temperature. Cells were then incubated with the following fluorochrome labelled anti- human antibodies: CD45 FITC, CD146 PE and CD34 Percp for 30 minutes. Cells were washed once using PBS + 5% heat inactivated FCS + 0.1% azide, centrifuged to pellet the cells and then re-suspended in 200l of fixative with 10% formaldehyde and 1% azide (BD, cellFIXtm) and made up to 1ml with buffer immediately prior to FC.

2.5.2.2 Flow cytometry instrument settings

Samples were run on a Becton Dickinson FACSCalibur flow cytometer. FC was performed using FSC and SSC instrument settings optimised for detection of PBMCs as previously described (Goon et al., 2006), using a linear FSC/SSC scale with FSC voltage set at E00. After the optimisation steps described in Chapter 3 FC was also performed as for larger cells such as HUVEC, using a logarithmic FSC scale and a linear SSC, with a FSC voltage set at E00 as previously described (Brogan et al.,

2004b). CECs were gated firstly on their FSC and SSC characteristics and then on cells negative for CD45 staining. CECs were then considered to be all events bright for

CD146 and CD34 in this region. Non-stained samples and fluorochrome matched isotype controls (IgG1 FITC (eBiosciences), IgG1 PE (BD) and IgG1 Percp (BD)) were used to define positivity. The gating strategy used is shown in Chapter 3 in

Figures 3.1 and Figure 3.11.

2.5.2.3 Absolute enumeration of CECs using flow cytometry

In order to obtain absolute CEC counts, samples were either spiked with CountbrightTM absolute counting beads as per manufacturer’s instructions, run to completion or had the flow rate calculated prior to the sample being run using 3µm latex beads (see below).

To enumerate using countbrightTM beads 50µl was spiked into the sample. 50µl contained approximately 50,000 beads (depending on batch). Therefore absolute cell count could be determined by calculating the volume of sample analysed by calculating the proportion of beads run through the system using the following equation:

Beads added (50,000) CECs counted x = CECs/ml Beads counted

In order to estimate the flow rate of the machine 3µm latex beads were diluted in sterile filtered distilled water to give a final concentration of 20,000 beads/µl. 10 µl of this stock was then added to 240µl of distilled water. The number of beads acquired in a set time at a set flow speed (usually medium flow) was used to estimate the flow rate (in

µl/second) using the following calculation:

Beads counted x Volume (250µl) ÷ Time (s) = Flow rate (µl/s) Beads added

(200,000)

Once the flow rate was estimated the number of CECs in a sample could be calculated by working out the volume of sample that would be run in a set time at that flow rate and then calculating how many CECs would be in the whole sample.

2.5.3 Detection of cellular microparticles

Samples of platelet poor plasma (PPP) were thawed rapidly at 37oC and a known volume was spun at 15000g for 60 minutes to pellet the microparticles (MPs) at room temperature in a temperature controlled centrifuge. The supernatant was decanted and the MPs resuspended in annexin V buffer (BD). 35µl was then added to wells on a 96 well plate (polypropylene) containing 5µl of annexin V FITC diluted 1:10 (BD) and

10µl of a PE conjugated, PerCP conjugated or PE-CY5 conjugated antibody diluted to give a final concentration in 50µl as in Table 2.1 (1 in 3 for most antibodies, 1 n 5 for

CD144 and 1 in 8 for CD66b). MPs were stained, rocking gently, in the dark at room temperature for 10 minutes. 200µl of Annexin V buffer was then added to each well to end the incubation and samples were then transferred to FACS tubes. The FC plans used for MP detection are shown in Appendix 3. A 22 tube plan was used for samples with a volume greater than 300µl, while smaller samples were run using 12 tubes.

2.5.3.1 Flow cytometry instrument settings for microparticle detection

Samples were analysed on a BD Facscalibur, using a logarithmic scale with FSC voltage set at E01. Instrument settings were initially optimised using 0.3, 0.8 and 3µm latex beads, this process is described in detail in Chapter 4. As described previously

(Combes et al., 1997) MPs were gated on their Annexin V (AnV) positivity as demonstrated in Figure 4.3. As AnV is a protein and not an antibody, it therefore does not have an isotype control so unlabeled MPs were used to establish positivity cutoff.

In order to determine whether observed noise in the system leading to high event rate in some samples had a significant effect on MP detection, MPs were analysed with and without a threshold based on AnV binding (discussed in Chapter 4). Histograms based on the gated AnV positive population were obtained for the fluorochrome conjugated

antibody binding and positivity was defined by the isotype control using 2.5 % positivity as a cutoff.

2.5.3.2 Absolute enumeration of microparticles

MP samples were run at a medium flow rate for 45 seconds, with a known number

(200,000) of 3µm latex beads run for 45 seconds at the same flow rate before and after each patient’s samples, allowing absolute MP enumeration. By using the equation below, which incorporates the beads run in 45 seconds and the exact volume of plasma used the absolute number of annexin V binding microparticles per ml of plasma was determined:

1000 Beads added (200,000) Annexin V +ve no of FACs tubes MPs/ml of x events x sample divided into x Vol of plasma = Average beads counted plasma spun down (µl)

2.5.4 Detection of activated parent cells

When MPs were generated from isolated cell populations during setup experiments activation of the parent cell was confirmed in the following manner. After the cells were pelleted and the supernatant was decanted for MP enumeration, the parent cells were resuspended in PBS (containing 0.1% azide and 5% heat inactivated FCS) to a concentration of 2 x 106 cells/ml. 100µl of cell suspension was added to wells on a 96 well plate. The plate was then spun at 500g for 5 minutes at 4oC. The supernatant was flicked off and the plate blotted. 10µl of diluted antibody or control (final concentration as in table 2.1) was added to the pelleted cells. Cells were stained for 30 minutes at

4oC, then washed 3 times with PBS using the spin down, flick and blot technique

described above. After the final wash cells were resuspended in 200µl of Cellfix. The phenotype of the MPs generated from different cell populations and their activated parent cells is shown in Appendix 4.

2.6 Assessment of potential endothelial repair responses

2.6.1 Measurement of endothelial growth factors

Growth factors associated with endothelial reparative responses were measured using then following ELISA kits: VEGF (R&D), Angiopoetin 1 (R&D), and 2 (R&D), and thymosin beta 4 (immunodiagnostik). The three R&D kits were based on the sandwich

ELISA principle while the thymosin beat 4 kit was based on a competitive ELISA principle.

In a sandwich ELISA a monoclonal antibody specific to the peptide of interest is coated onto each well of a 96 well plate. Standards and samples are then pipetted into the wells in duplicate and any peptide of interest becomes immobilised. Unbound sample is then washed away and an enzyme linked secondary monoclonal specific for the peptide of interest is added. This then binds the immobilised peptide and after a wash step, the introduction of a substrate for the enzyme allows a colour to develop, the reaction is stopped and the plate then read at a specific wavelength to determine the colour intensity (Figure 2.2). The intensity of colour (measured as optical density (OD)) of the standards allows a standard graph to be plotted so the test samples can be compared to it and have their concentration calculated. Examples of the standard graphs obtained

using these kits can be seen in Appendix 5. VEGF was measured in serum (100µl per well, undiluted), Angiopoetin 1 was measured in platelet poor plasma (50µl per well, diluted 1 in 15) as was Angiopoeitin-2 (50µl per well diluted 1 in 5).

In a competition ELISA (Figure 2.3) the peptide of interest is coated onto the wells of the 96 well plate and competes with the peptide in the standards and samples to bind an enzyme linked substrate. In the case of thymosin β4, the substrate was monomeric actin. Thus in samples which have a high OD the substrate had bound the plate and thus there was a low concentration of thymosin β4 in the sample, conversely samples with low OD have high thymosin β4. The standards with a known concentration for thymosin beta 4 were used to generate a standard curve with which to compare the test samples. Unfortunately although the kit purchased claimed to be fully optimised for measurement of TB4 in human serum, children had in some cases 10 fold high thymosin β 4 level than adults and were off the standard curve. Further dilution of the sample was necessary and demonstrated that all samples required a 1 in 5 dilution before running on the plate. Examples of the standard curves generated are in Appendix

2.6.2 Flow cytometry for detection of endothelial progenitor cells

Frozen PBMCs were thawed rapidly at 37oC, then transferred to warmed FCS, to which warm RPMI 1640 (with antibiotics) was added to give a final FCS concentration of 10-

20%. PBMCs were then counted and diluted to 2 x 106 cells per ml (in some samples a pelleting step was necessary to concentrate cells), 50µl of FcR blocking agent was added per ml and the cells incubated for 10 minutes at room temperature. 150µl of the cell suspension was added to wells on a 96 well plate, subsequently the plates were spun at 500g for 5 mins at 4oC and the supernatant flicked off and the plate blotted.

Figure 2-2: Methodology of Sandwich ELISA

Figure 2.2: Sandwich ELISAs for VEGF and Angiopoietins-1 and 2 were performed using the above protocol, adapted from R&D instruction manuals.

Figure 2-3: Methodology of Thymosin-β 4 competition ELISA

Figure 2.3: Thymosin beta-4 competition ELISA methodology was performed as described in the protocol above, adapted from Immunodiagnostik instruction manual. Although the kit was optimised for adult samples, it became apparent that children have much higher levels of TB4 and thus serum samples had to be diluted 1 in 5 before they could be analysed using this kit

Cells were then stained with 10µl of diluted antibody (diluted using PBS + 0.1% azide; table 1a) and incubated for 30 minutes at 4oC. Cells were washed using 200µl of FACs wash (PBS + 0.1% azide + 5% heat inactivated FCS) and spun down and the supernatant flicked off as before. The secondary antibody for KDR, Streptavidin FITC was then added to the required wells and incubated for 10mins at 4oC. Cells were then washed twice as before and finally resuspended in 150µl of fixative with 10% formaldehyde with 1% azide (BD, cellFIXtm).

Using flow cytometry the endothelial progenitor cell (EPC) markers KDR, CD133 and

CD34 were examined alongside the endothelial markers CD144, CD146 and the lymphocyte marker CD3, the monocyte marker CD14 and the pan leukocyte marker

CD45. Non stained samples and fluorochrome matched isotype controls were used to define positivity. EPCs were expressed as a percentage of the mononuclear cell gate, as a percentage of total CD34+ cells and in the vasculitis patients as an estimate of absolute EPC number based on the full blood count obtained from the clinical labs as described in Chapter 6. Briefly the leukocyte count is multiplied by the % of viable cells expressing CD34. It is important to note that this was not used as an absolute count of EPCs but as a crude tool to estimate absolute EPC numbers in order to determine whether trends observed in individual patients were reflected when the different ways of expressing EPC number were compared.

2.6.3 Endothelial progenitor cell colony culture

Hill et al described a methodology for counting colonies believed to be derived from

EPCs. In a large study these colonies correlated with flow mediated dilatation (a measure of endothelial function) (Hill et al., 2003). This technique was thought to be

dependent on the batch of FCS used, which was developed into the commercially available colony counting assay – EndocultTM (Stemcell technologies). This technique has now become widely used in clinical studies and is thought to predict cardiovascular mortality (Hill et al., 2003). Work done previously within our group by Dr Francesca

Arrigoni had resulted in a growth factor supplemented media which emulated the growth characteristics of endocult, independent of FCS batch. Thus this media was used for assessment of EPC colonies as follows: PBMCs were isolated as described in

2.4.4 and 5 x 106 cells were cultured in RPMI 1640 (with 20% FCS (not heat inactivated) and antibiotics as before) for 2 days per well on a fibronectin coated 6 well plate. The non-adherent cells (hypothesized to contain the non differentiated stem cell population) were then removed and replated onto a fibronectin coated 24 well plate at a concentration of 1 x 106 cells per well. The replated cells and the adherent population remaining on the 6 well plate were cultured in MCDB131 containing the following additives L-glutamine (50ng/mL), endothelial cell growth serum (50µg/ml,Sigma),

VEGF (20ng/mL), bFGF (5ng/mL), 20% FCS (not heat inactivated), heparin (5U/mL) and penicillin (100 U/mL) and streptomycin (100µg/ml). Media was changed every 3 days until day 7 when colonies were counted. Colonies consisted of a central core of rounded cells surrounded by spindle like outgrowth cells (Figure 2.4a).

2.6.4 Endothelial progenitor cell outgrowth culture

Although a number of groups have used colony counting as a measure of EPC function, other groups have derived endothelial-like monolayers from adherent cell populations, in much longer cultures (20+ days) and these have been termed late outgrowth EPCs

(Lin et al., 2000). In line with most studies which have grown these late outgrowths, the commercially available endothelial media EGM-2 was used. PBMCs were isolated as

described in Section 2.4.4 and in order to grow late outgrowths 10 x 106 cells were added to fibronectin coated wells on a 6 well plate. Cells were cultured in EGM-2 with supplements as described in Section 2.3.3. After the first 4 days, media was changed every 3 days. As with the Hill protocol described above colonies formed between 4-7 days in culture (Figure 2.4b). In some experiments cells were replated on fresh fibronectin using trypsin as described for passage of HUVEC to determine optimal conditions for growth of monolayers (Figure 2.4c and d).

2.6.5 Matrigel - assessment of endothelial progenitor cell function

A number of studies have shown that EPCs can incorporate into tubule networks formed by endothelial cells on extracellular matrix formulations (MatrigelTM) (Bompais et al.,

2004,Hur et al., 2004). The following protocol used for assessing incorporation of

PBMCs into tubule networks is based on work previously established by Dr Francesca

Arrigoni and Dr Neil Halliday within the group.

Isolated PBMCs were cultured for 9 days in EGM-2 on a fibronectin coated 24 well plate with 5 x 106 cells added per well as described above. EGM-2 cultures were thus the same age as the non-adherent colony cultures when lifted. At day 9 cells were stained with Di-I-acetylated low density lipoprotein (DiI-AC-LDL) 1.6µl/well for 4 hrs at 37oC. An hour before the cells were ready, 100µl of defrosted matrigel (defrosted for

3+ hours at 4oC) was added using cold (left overnight at -20oC) pipette tips into each well of a cold (-20oC overnight) 24 well plate on ice, taking care to spread evenly across the well. The matrigel was left to set at room temperature for 1 hour. After 1 hour, 1ml of EGM-2 (without hydrocortisone) was added to each well. EGM-2 cells and HUVEC were then lifted using trypsin as previously described and 4 x 104 EGM-2 cells and 4 x

Figure 2-4: EPC colony culture

Figure 2.4a: After 7 days in culture after the preplating step the EPCS had formed colonies which consisted of a central cluster of rounded cells surrounded by spindle-like outgrowths. Figure 2.4b: cells cultured for 7 days in EGM-2 had a similar phenotype. Figure 2.4cd: When these EGM-2 cells were left for longer, they went on to form “late outgrowth” EPCS which resembled a HUVEC-like monolayer at around 18-20 days in culture. Scale bars represent 50µm.

104 HUVEC or 4 x 104 HUVEC alone were added to each well. Tubule growth and

PBMC incorporation was assessed after 18 hours in culture, using the HUVEC alone as a control (Figure 2.5ab).

2.7 Statistics

A number of parametric and non-parametric statistical analyses were used in this thesis.

Description of individual tests and their applications is given in the relevant chapters.

While parametric tests were used whenever possible, there were a number of instances where data were not normally distributed (and could not be easily transformed to normality) and thus non-parametric tests were used. Parametric tests used were student’s t-test and paired t-tests. Non-parametric tests used were the Kruskall-Wallis test, the Wilcoxon signed-ranks test, and Spearman rank correlation coefficient. Bland-

Altman statistics were performed where comparison of methodology was required, including assessment of interuser variability. In several chapters of this thesis multiple comparisons were performed simultaneously, so in order to reduce the risk of falsely finding tests positive a post-hoc Holm-Bonferroni correction was performed. Graphs were plotted using Microsoft Excel 2007 and Graphpad Prism version 4. All statistics were calculated using Microsoft Excel 2007, and SPSS versions 15.

As the cohort described herein is highly heterogenous, data was colour coded dependent on disease classification, in all the graphs within this thesis the following coding applies:

Figure 2-5: Matrigel, incorporation of PBMCS into HUVEC tubules

Figure 2.5a: When HUVEC alone are cultured on matrigel, they form characteristic lattices. Scale bar represents 100µm. Figure 2.5b: When cells cultured for 9 days in EGM-2 are added alongside

HUVEC they adhere/incorporate in the tubules formed the extenct of incorporation is assesssed using DiI acetylated LDL to visualise these cells. Scale bar represents 50µm.

3 Circulating endothelial cells

3.1. Summary

3.2. Introduction

3.3. Detection of CECs using IBE

3.4. Detection of CECs using FC

3.5. Comparison of FC and IBE for Detection of CECs

3.6. Conclusions of methodological experiments

3.7. Evaluation of CECs in vasculitis

3.8. Discussion

3.1 Summary

As discussed in the introduction to this thesis, assessment of disease activity in vasculitis can be challenging. The gold standard technique for detecting active inflammatory disease (tissue biopsy) is invasive and impractical for repeated assessment, particularly due to the patchy nature of the disease where false negative results can occur. This chapter describes the methodology used to detect circulating endothelial cells (a novel biomarker of endothelial injury) and demonstrates which of the two commonly used techniques is more appropriate for patients with vasculitis. The latter part of this chapter then examines the relationship between circulating endothelial cells and disease activity in children with primary systemic vasculitis. Some of the data presented herein are now published (Clarke et al., 2008,Shantsila and Lip, 2008).

3.2 Introduction

Circulating endothelial cells (CECs) are mature endothelial cells which have become detached from the vessel wall. In health, these cells are found in low levels in the circulation and are believed to represent physiological endothelial turnover

(Erdbruegger et al., 2006), however in conditions associated with endothelial injury they are elevated (Woywodt et al., 2006a).

3.2.1 Detecting CECs

The concept that mature endothelial cells can be detected within peripheral blood is not new. CECs were first described in the 1950s in blood smears (Li et al.,

1958,Shanberge, 1955) and were considered likely to be the result of traumatic venepuncture and not of pathological consequence. The CECs described by Li et al 96

occurred both singly and in sheets of cells and their detection relied on observations of morphology and size under light microscopy. In the early 1970s the link between these

CECs in blood smears and vascular injury was proposed and these cells were subsequently identified in various conditions including myocardial infarction (Hladovec et al., 1978) and hypertension (Hladovec and Prerovsky, 1989). All these early studies were reliant on morphology alone and were technically difficult and subjective.

Further refinement of the blood smear methodology included the use of fluorescent antibodies to endothelial markers such as Von Willebrand factor (VWf) or cell adhesion molecules (Percivalle et al., 1993). However there remained a lack of specificity of cell surface antigens and the use of more specific intracellular antigens required cell permeabilisation, which increased the potential for cell loss from samples. By the early

1990s antibodies to more specific surface markers for ECs had been developed, including the monoclonal S-endo-1 against CD146, a highly expressed constitutive EC cell surface marker associated with the cytoskeleton. CD146 is more specific for ECs

(although it is found on some activated T cells and melanoma cells) than other EC surface markers such as CD31 (PECAM-1, also found on platelets, monocytes, neutrophils and T cells), CD54 (ICAM-1, also found on monocytes and lymphocytes) or

CD105 (endoglin, also found on activated monocytes, macrophages and erythrocyte precursers) and soon became the antibody of choice for CEC detection. Its high expression on most ECs importantly allowed the use of immunomagnetic beads for enrichment of CECs from blood samples (George et al., 1992).

In 2006 Woywodt et al described a consensus protocol for detection of CECs using immunomagnetic bead extraction (IBE) and it is described in detail in Section 2.5.1

(Woywodt et al., 2006a). After CECs are extracted from peripheral blood using CD146 bound to immunomagnetic beads they are stained with a fluorescently labelled lectin

Ulex Europeus (Ulex). Ulex selectively binds alpha-L-fucose containing glycoproteins, that can be found on the surface of human ECs and is more sensitive than vWF as it can bind ECs shown to be vWF negative (Holthofer et al., 1982). As with CD146, Ulex is not endothelial specific and has been shown to bind many cell types including gut epithelial cells (Narita and Numao, 1992) and megakarocytes (Liu and Liu, 1990) as well as the blood type H(O) antigen on red blood cells (Ito et al., 1990). Therefore

CECs must be identified by the expression of both of these two endothelial markers to discriminate from other cell populations. Microscopy also allows discrimination of

CECs based on size and morphology from smaller cells such T cells (which can express

CD146) and endothelial progenitor cells (EPCs) (CD146 and ULEX can both stain

EPCs) (Delorme et al., 2005,Porat et al., 2006).

Using the IBE methodology high levels of CECs have been detected in a number of conditions associated with vascular injury including vasculitis in adults, as demonstrated in Table 3.1. In AAV, CECs correlated with disease activity (BVAS score) (Woywodt et al., 2003b) and in studies in patients with coronary artery disease it was noted that the highest levels of CECs occurred in those patient with the most severe and acute disease (Mutin et al., 1999). Low levels have been documented in healthy individuals of all ages in all studies, suggesting that high levels of CECs may be a marker of extreme EC injury.

A commonly used alternative methodology to IBE for enumeration of CECs uses flow cytometry (FC), although as yet the consistent reproducibility seen with IBE has not

Table 3-1: CEC detection in human disease

Condition Method Case Control Study

Sickle cell anaemia IBE 13.2-22.8 2.6 (Solovey et al., 1997)

Septic Shock IBE 30 2 (Mutunga et al., 2001)

Rickettsial infection IBE 5-1600 <3 (George et al., 1993)

Renal transplantation IBE 24-72 6 (Woywodt et al., 2003a)

Bone marrow transplantation IBE 16-44 8 (Woywodt et al., 2004)

Acute MI IBE 7.5 0 (Mutin et al., 1999)

Diabetes IBE 35-126 10 (McClung et al., 2005)

Stroke IBE 10.8-20.7 2.7 (Nadar et al., 2005),

Hypertension IBE 10 7 (Boos et al., 2007)

AAV IBE 20-5700 5 (Woywodt et al., 2003b)

KD IBE 15 6 (Nakatani et al., 2003)

Breast cancer 6800- 1200- FC (Mancuso et al., 2001) & lymphoma 39100 7900

Systemic sclerosis FC 243-375 69-110 (Del et al., 2004)

SLE FC 89 10 (Rajagopalan et al., 2004)

Acute coronary syndrome FC 8-20 6-10 (Goon et al., 2006)

Breast cancer FC 6-16 6-10 (Goon et al., 2006)

Median CEC value (range used where given)

been demonstrated with this technique. Widely differing reference ranges have been reported in healthy individuals across different laboratories (Table 3.1) and moreover no consensus has been reached for FC methodology, including the choice of cell surface markers examined to evaluate CECs. FC quantification, however, offers a number of potential advantages over IBE including: multiple CEC surface markers can be

examined simultaneously; exclusion of leukocytes using the pan leukocyte marker

CD45; and further characterisation of sub-populations of CECs such as those in either an activated, or apoptotic state. Consequently attempts to optimise FC methodology for

CEC detection in different health and disease states are clearly of considerable importance. In this context, a recent FC protocol for CEC enumeration has been described that demonstrated agreement between CEC counts measured using FC methodology and IBE in healthy adults and patients with primary breast cancer and acute coronary syndromes (Goon et al., 2006).

In order to determine which methodology would be most appropriate for the paediatric vasculitis cohort, both the IBE consensus protocol (Woywodt et al., 2006a) and an FC protocol based on the aforementioned paper (Goon et al., 2006) was established. As described in the methods chapter, the FC definition of a CEC that was used has been described by a number of studies: cells that express both CD146 and CD34, but are negative for CD45 (Del et al., 2004,Goon et al., 2006,Bertolini et al., 2006,Khan et al.,

2005) (Figure 3.1a-c). The definition of a CEC using IBE was as follows; Ulex bright cells, greater than 10m in size with more than 5 CD146 conjugated magnetic beads attached, (Figure 3.1d-f) (Woywodt et al., 2006a).

3.2.2 Where do CECs come from and does this influence phenotype?

CECs can be detected in both the arterial and venous circulation without significant differences in values obtained (Woywodt et al., 2006a). This observation raises interesting questions as to the origins of CECs and whether they can pass through the capillary beds (unlikely, in the opinion of this author, when considering the size of the capillary lumen and the size of CECs), or whether CECs represent widespread (arterial

100

Figure 3-1: Flow cytometric (FC) and immunomagnetic bead (IBE) definitions of CECs

Figure 3.1a-c: CECs as defined by FC are mononuclear cells (figure 3.1a, R1) which are negative for CD45, (figure 3.1b, R2) and positive for CD34 and CD146 (figure 3.1c, R3). FC data shown is from a patient with active Wegener’s Granulomatosis. Figure 3.1d-f CECs as defined by IBE are

ULEX bright, have greater than 5 beads attached and are greater than 10µm in size. These CECs are from 2 patients with active disease, the first (figure 3.1d) with Wegener’s granulomatosis, the second (figure 3.1e) with PAN. Figure 3.1f Demonstrates how several CECs may clump together.

Although it is apparent that a number of cells make up this sheet, it is still only counted as 1 CEC.

This CEC is from the same patient with PAN in (figure 3.1e).

101

and venous) or whether CECs represent widespread (arterial and venous) endothelial dysfunction or even elevated systemic turnover originating from a localised disease.

Interestingly the vessel site of origin of CECs may vary between different conditions. A number of studies have used the proposed microvascular specific antigen CD36

(Solovey et al., 1997,Swerlick et al., 1992) to assess the extent to which CECs have come from microvasculature. In healthy adult controls half the CECs had a microvascular phenotype, whereas in sickle cell anaemia this was higher at 78%

(Solovey et al., 1997). Interestingly in acute coronary syndrome no CD36 staining was observed, suggesting a macrovascular origin (Mutin et al., 1999). Neither of these findings in disease is unexpected when the nature of the two diseases is considered.

CECs have also been shown to have widely differing phenotypes. Relating to size and morphology CECs consist of both sheets of cells or individual cells, and in some samples there are numerous brightly staining fragments, which are too small to consider whole cells but are of endothelial origin (Woywodt et al., 2006a). As Figure 3.2 demonstrates such findings have all been demonstrated in the GOSH vasculitis cohort described in this thesis. The majority of CECs detected using IBE in adults with vasculitis are of a necrotic or apoptotic nature (Woywodt et al., 2003b). In other conditions this may not be the case. Both Lin et al and Mutunga et al were able to culture CD146+ CECs isolated from peripheral blood from healthy controls and

Mutunga et al demonstrated that CECs from a proportion of patients with septic shock could also be cultured (Lin et al., 2000,Mutunga et al., 2001).

102

Figure 3-2: Phenotype of CECs in vasculitis

Figure 3.2a: CECs from patients with PSV can include sheets of many endothelial cells. Figure

3.2b: CECS were also seen as single cells Figure 3.2cd: Frequently samples from PSV patients contain many CD146 bead bound and Ulex bright events that were too small to be counted as CECs

- shown at low magnification (figure 3.2c: arrows indicate bright events large enough to be counted as CECs) and at higher magnification (figure 3.2d). These may be fragments of whole cells or endothelial microparticles.

103

3.3 Detection of CECs using immunomagnetic bead extraction

CECs were isolated and stained as described in Chapter 2.5.1. Each sample where practical, was counted by two individuals blinded to the patients clinical state. CEC enumeration is reliant on user observation and therefore has an element of subjectivity.

Whilst strict counting guidelines as described above will minimize this, duplicate counting by two independent observers is advised, particularly for validation of the technique.

3.3.1 Reproducibility of the technique

In order to evaluate the reproducibility of the IBE technique, CECs from 1 healthy adult control and 1 patient with active WG were prepared from 3 individual 1ml samples. As a positive control, blood was spiked with 1000 HUVEC/ml and again 3 samples were prepared from this. Figure 3.3 demonstrates that there was greater variation in count for samples with high CECs (i.e. from the HUVEC spiked into blood and from the patient) and little variation in absolute CEC count in healthy adult control samples. Despite this variation at high counts, results were all consistantly high and thus a single sample was considered a sufficiently robust representation of clinical CEC counts from patients in subsequent studies.

3.3.2 Interobserver variability

As mentioned previously, due to subjectivity in CEC counting bias relating to interobserver variability could occur. In samples counted by more than one observer,

Bland-Altman analysis was performed to determine the degree of interobserver

104

Figure 3-3: Reproducibility of IBE for detection of CECs

Figure 3.3: Blood samples from 1 healthy adult control and 1 patient with active Wegener’s granulomatosus were divided into 3 and CECs enumerated in each sample. As a further positive control a sample of the healthy control blood was spiked with 1000 HUVEC/ml. The mean CEC count in the healthy adult control was 1 CECs/ml SD 2; in the patient with active Wegener’s granulomatosis the mean was 768 CECs/ml SD 128; and in the HUVEC spiked sample the mean was 757 CECs/ml SD 72.

105

variability between myself and four other individuals who had been trained to count

CECs. Figure 3.4a demonstrates that there was greater interobserver variability between samples which had high CEC counts (overall mean difference -26.7, 95% CI

+/- 331.5), however this is undoubtedly influenced by the multiplication factor used as part of the nageotte chamber counting protocol.

As described in Chapter 2 (section 2.5.1.3) when samples have high CECs less lines are counted on the nageotte chamber (since counting the full 40 lines would be prohibitively time consuming and impractical for routine clinical used) and therefore the multiplication factor is greater. Thus in figure 3.4b when the absolute nageotte count was considered, the interobserver variability was low, the mean difference was 0.2

CECs counted (95% CI, +/- 36.2). While there was higher interobserver variability at the high range of the CEC counts, the ability to enumerate exactly in this disease context is somewhat less relevant than the ability to distinguish between grossly elevated and low CEC counts, where agreement was much better. Particularly when monitoring individuals for disease activity in clinical studies.

3.3.3 The effect of traumatic venepuncture

As has previously been documented (Woywodt et al., 2006a) if venepuncture is traumatic then CECs may appear falsely elevated. This is of particular relevance in paediatric cohorts as it is often technically more difficult to obtain peripheral blood from small children than from adults. In order to reduce the effect of venepuncture, many groups discard the first 5ml of blood taken, however this may not be appropriate or possible for paediatric samples, which are typically of smaller volume. The influence of traumatic venepuncture was investigated in 1 healthy adult control (a sample size of 3

106

Figure 3-4: Bland-Altman plots for testing interobserver variability of IBE for detection of CECs

Figure 3.4a Primary observer (LC = myself) CEC counts compared to 4 different individuals trained to count CECs. Each different observer is indicated by a different coloured cross. CECs were counted in 18 patients with vasculitis and 13 healthy child controls. Overall, the mean difference between the CEC counts I obtained and those obtained by others was -26.7, 95% CI +/-

331.5. Notably levels of agreement was better at low CEC counts and the wide confidence interval influenced by 3 outliers. This analysis does not take into account differences in the multiplication factor used to calculate the final count, therefore any differences are amplified in samples with high

CEC counts. Figure 3.4b thus compares absolute nageotte count which gives a mean difference of

0.2, 95% CI +/- 36.3.

107

would have been desirable, however volunteers were not forthcoming) with ethical approval and informed consent. Blood was collected first using 3 x 5ml syringes from the control arm. Traumatic venepuncture was then achieved in the other arm by scraping the needle backwards and forwards in the vessel, once it had been inserted. As in the contralateral control arm, blood was then collected in 3 x 5ml syringes.

Although traumatic venepuncture elevated CECs from the 16 cells/ml seen in the contralateral arm to a maximum of 128 cells/ml (Figure 3.5), CECs did not increase to the levels observed in adult active vasculitis patients (up 5,000 CECs/ml) (Woywodt et al., 2003b). It was therefore concluded that as long as child controls were age matched the effect of the more traumatic venepuncture expected in children would not detract from the elevations seen in active disease, although could account for more

“background noise” in both paediatric controls and vasculitis patients. Also it can be seen in this individual that discarding the first 5mls of blood after traumatic venepuncture did not make a difference to the CEC count as both the second and third syringes of blood gave a similar CEC count to the first 5mls sampled (Figure 3.5).

However, this may have been a consequence of the excessive trauma occurring in this individual case. Thus in line with other studies (Woywodt et al., 2006a) for all future blood samples the first 5mls of blood were not used for CEC quantification (where possible) but rather for other assays described in this thesis, minimizing waste of precious paediatric blood samples.

3.3.4 The effect of storage and time to preparation on CEC enumeration

In the same individual who had kindly agreed to consent to traumatic venepuncture, after the 1ml of blood required for IBE was removed the remainder of the first 5ml

108

Figure 3-5: Factors influencing CEC counting – traumatic venepuncture.

150 traumatic control (contralateral arm) 125

100

75 CECs/ml

0 first second third Syringe

Figure 3.5 demonstrates the effect of traumatic venepuncture on CEC enumeration in 1 individual.

15 mls of blood was taken in 3 x 5ml syringes from both arms. In the second arm, venupuncture was traumatic. CEC count was much higher under traumatic conditions than in the control arm, however notably, discarding the first 5 mls of blood would not have made a difference to the CEC count under highly traumatic conditions as the second and third samples from each arm were of similar levels to the first.

109

sample was divided into 2 and left at RT or 4oC. CECs were then counted in 0.5ml of blood at 2hrs, 4hrs and 24hrs to demonstrate any alterations in CEC count caused by sample storage time (Figure 3.5a and b, represented by diamonds).

The same time course experiment was also performed on 3 vasculitis patient samples

(Figure 3.5 a and b, represented by crosses). As can be seen in Figures 3.6 a and b, as has previously been described CEC counts had declined by 24hrs in all individuals studied (Woywodt et al., 2006a). The temperature at which the sample was stored did not appear to influence this, although room temperature samples were more difficult to read, with higher background noise possibly due to greater fragmentation/degradation of cells than samples stored at 4oC. Thus, where possible all bloods were processed immediately, usually within a 2 hour window and kept cold at 4oC if they could not be processed straight away.

3.4 Detecting CECS using flow cytometry

For the purposes of this thesis the methodology as described by Goon et al was closely followed, as described in Section 2.5.2. (Goon et al., 2006)

3.4.1 The effect of lysis on staining of CD34/CD146/CD45

In the protocol described by Goon et al, 1ml of whole blood underwent red cell lysis prior to addition of antibodies (to reduce the volume of antibody required for each sample). This is a recognized but uncommon methodology for whole blood FC and thus in order to make sure the cells were stained optimally experiments were performed

110

Figure 3-6: Factors influencing CEC counting – sample storage time and storage temperature

Figure 3.6a and b: 3 samples of blood from active vasculitis patients and one from the healthy adult control who had undergone traumatic venupuncture were left at room temperature (Figure 3.6a) or

4oC (Figure 3.6b) for 0, 2, 4 and 24 hrs before they were prepared for CEC detection. By 24 hrs

CEC counts were decreased in all individuals. Thus blood was prepared as close to time of sampling as possible for all future clinical studies.

111

to determine whether the lysis step was influencing staining of these cell surface markers. Figures 3.7a-c demonstrate that there were no significant differences in the percentage expression of CD45, CD146 or CD34 within the FC mononuclear cell gate in 3 patients if cells were stained before or after lysis confirming the validity of this methodology.

3.4.2 Using endothelial cells as a positive control for CECs

HUVEC had provided a useful positive control while establishing the IBE assay. It was therefore ascertained whether HUVEC could be used as a positive control in FC protocols for CEC detection. Unfortunately HUVEC detached from cultured monolayers and resuspended in PBS as described in section 2.4.3 could not be detected using the previously published flow cytometric instrument settings for CECs. This was because HUVEC had greater forward and side scatter (FSC and SSC) than peripheral blood mononuclear cells. However when gated optimally HUVEC were detected by FC and were negative for the pan leukocyte marker CD45 (Figure 3.7d) and expressed high levels of CD146 (Figure 3.7e) and at first passage CD34 (although this diminished with later passage) (Figure 3.7 f and g). Thus HUVEC could be used as a positive control for assessment of sensitivity and reproducibility of FC as long as the FSC SSC setting were optimized to incorporate them. This important observation has significant implications for the detection of CECs using FC (Clarke et al., 2008) and will be discussed in more detail later in this chapter.

112

Figure 3-7: The effect of red cell lysis on flow cytometric protocols for HUVEC detection

Figure 3.7 a-c demonstrates that the pre-staining red cell lysis protocol as described by Goon et al, does not significantly alter the % of monunuclear cells expressing the 3 CEC markers: CD45,

CD146 and CD34. Figure 3.7 d-g demonstrates that HUVEC can be used as a positive control for the CECs described by Goon et al, as they are negative for CD45 (Figure 3.7d) and express high

CD146 (Figure 3.7e). HUVEC only express CD34 when examined at early passage (Figure 3.7f), by P3 HUVEC are negative for CD34 (Figure 3.7g).

113

3.4.3 Using counting beads to determine absolute cell number

In the paper from Goon et al, (Goon et al., 2006) the flow rate of the flow cytometer was determined prior to each sample being run using a sample spiked with a known number of latex counting beads, a standard technique for cell counting using FC. This allowed a calculation of the quantity of sample analysed in a set time and thus could be used to calculate an absolute CEC number. However, samples have to be run for a long period of time as CECs are rare events in peripheral blood and thus variations in flow rate may occur, which could significantly affect the result.

To determine whether this would alter results significantly an experiment was performed that enumerated HUVEC spiked into whole blood using different methods for calculating absolute HUVEC counts. Whole blood was spiked with 100,000

HUVEC/ml and comparative experiments (n=3) were performed. In these experiments three different counting strategies were investigated: 1. Samples were run to completion. 2. The flow rate was calculated prior to samples being run using latex beads. 3. An internal standard counting bead was used (Countbrighttm beads,

Invitrogen). These methods are described in Chapters 2.5.2 and 2.5.3.

Figure 3.8 demonstrates that enumerating samples using Countbrighttm beads resulted in a higher (but non-statistically significant) percentage recovery than calculating the flow rate prior to running the samples (NS, p=0.06). Running samples to completion also gave better recovery than calculating flow rate from latex beads, however this technique would not be practical for routine clinical studies - the data files produced were large and in order to run a 1ml sample to completion dry it took over an hour, even at a high flow rate. This experiment confirmed that the use of an internal standard such as

114

CountbrightTM beads was more accurate over the long period of time required to collect sufficient sample for rare event analysis and therefore would be the methodology of choice for future studies relating to FC for CEC analysis in this thesis.

3.4.4 Reproducibility of flow cytometry for endothelial cell enumeration

As described previously whole blood was spiked with 100,000 HUVEC/ml. The sample was then divided into 3 and the number of HUVEC in each sample was again enumerated using Countbrighttm beads. The mean recovery of HUVEC was 53% (range

51-56%) demonstrating the reproducibility of the technique. However this recovery of

53% of spiked HUVEC was low. The reasons for this are explained in more detail later in this chapter, but are due to poor optimisation of the FC technique for this particular experimental setup. As the aim of this series of preliminary experiments was to evaluate the previously published FC methodology in vasculitis patients we deliberately did not change the methodology from that described by Goon et al before evaluating it in patients (other than to utilize countbrightTM for enumeration as opposed to a calculated flow rate) (Goon et al., 2006).

115

Figure 3-8: Absolute enumeration of HUVEC spiked into whole blood

Figure 3.8 demonstrates the % recovery of 100,000 HUVEC spiked into samples of healthy adult controls blood. Samples were enumerated by 1. Running to completion. 2.By calculating the flow rate prior to running the sample using latex beads 3. By using an internal standard added to the sample to calculate the proportion run (Countbright TM beads). Countbright TM beads as an internal standard gave the best % recovery compared to the other two techniques (NS, vs completion p=0.11 and vs calculated flow p=0.06). While not statistically significant, this difference could be clinically relevant and thus Countbright TM beads were used for CEC enumeration in future FC studies.

116

3.5 Clinical study comparing two different methods of

CEC quantification

3.5.1 Patients and methods

3.5.1.1 Patients

25 children with primary systemic vasculitis were studied for this part of the study.

These included 12 males and 13 females of median age 13.1 years (range 2.3 -16.4 years). The vasculitis type (using the Chapel Hill Consensus Criteria) (Jennette et al.,

1994) was PAN n=9, WG n=4, MPA n= 1 and unclassified vasculitis n=11. Disease activity was assessed using a modified paediatric Birmingham Vasculitis Activity Score

(BVAS), with a value of 0/63 defining clinical remission (no clinical disease activity) as previously described (Brogan et al., 2004a,Luqmani et al., 1994) (Appendix 2). To establish if the assays compared performed similarly in health and disease, in addition to the 25 vasculitis patients 20 healthy adult controls were also recruited (2 males, 18 females, age range 24-40 years). All blood used for analysis was drawn using a 21 gauge needle with minimal trauma and with the first 5 mls discarded if possible as previously described (Woywodt et al., 2006a). Blood was processed immediately after collection as described below.

3.5.1.2 Detection of CECs using IBE and FC

This was performed as describe in Chapter 2 (sections 2.5.1 and 2.5.2). The IBE methodology was identical to the previously published consensus paper (Woywodt et al., 2006a). FC was performed using forward and side scatter (FSC/SSC) instrument settings optimised for detection of peripheral blood mononuclear cells (PBMCs; Figure

3.1) as previously described (Goon et al., 2006) (PBMC-FC). A second FC protocol was designed to optimally capture larger cells such as human umbilical vein endothelial 117

cells (HUVEC) as previously described (Brogan et al., 2004b). This is hereafter referred to as the EC-optimised-FC gating protocol.

As previously described, CECs were gated firstly on their FSC/SSC characteristics and then on cells negative for CD45 (Figure 3.1). Non-stained samples and fluorochrome matched isotype controls (IgG1 FITC, eBiosciences; IgG1 PE, BD; and IgG1 Percp,

BD) were used to define positivity. All FC instrument settings were optimised using detached HUVEC or pulmonary artery ECs (PAECs) in PBS, or whole blood spiked with 100000/ml HUVEC. The PAECs (first passage) were obtained from the main pulmonary artery of the explanted lungs of a child with pulmonary hypertension and were a kind gift from Dr Sue Hall (Institute of Child Health, UCL).

3.5.1.3 Statistics

The Mann-Whitney U test (non-paired samples) or Wilcoxon signed rank test (paired samples) were used to compare CEC counts between experiments. Levels of agreement for CEC counts between different methodologies were summarised using Bland-Altman plots (Bland and Altman, 1986). All blood spiking experiments were repeated in triplicate, unless otherwise stated. All statistical analyses were performed in SPSS version 15.0. Statistical significance was set at p<0.05.

118

3.5.2 Results

3.5.2.1 Agreement between IBE and PBMC-FC

The overall median IBE CEC count in vasculitis patients (n=25) and controls (n=20) combined was 32 CECs/ml, range 0-800. In contrast the PBMC-FC CEC count in the same individuals (n=45) was significantly lower at 26 CECs/ml, range 0-326 (p=0.014); paired analyses in 45 individuals. The Bland-Altman plot in Figure 3.9 summarises the levels of agreement between IBE and PBMC-FC for vasculitis patients (n=25, denoted as crosses) and controls (n=20, denoted as circles). Overall there was poor agreement between IBE and PBMC-FC based on these 45 paired observations in the 45 individuals: the mean difference between the two methods (IBE - PBMC-FC) was 78

CECs/ml, 95% confidence interval of the difference +/-365 CECs/ml.

Subgroup analysis of these 45 paired analyses revealed that most of this poor agreement arose in the vasculitis patients. Measurements from paired samples in these patients

(n=25; 10 with active vasculitis [BVAS>0]; 15 with vasculitis in remission [BVAS=0]) demonstrated significantly higher numbers of CECs with the IBE method than with

PBMC-FC: with IBE the median CEC count in the vasculitis patients was 80 CECs/ml, range 0-800, and the corresponding median CEC count using PBMC-FC was 31

CECs/ml, range 4-326 (p=0.004). Subgroup analysis of healthy adult controls (n=20) showed that agreement between IBE and PBMC-FC was better, but remained weak: the mean difference in CEC count between the two methods (IBE – PBMC-FC) was 17

CECs/ml, 95% CI +/-127 CECs/ml.

119

Figure 3-9: Bland-Altman statistics for testing agreement between IBE and PBMC-FC for CEC enumeration

Figure 3.9 is a Bland Altman plot to assess the agreement between the two methodologies described for detection of CECs. 25 vasculitis patients (denoted as crosses) and 20 healthy adult controls

(denoted as circles) were studied. Overall there was poor agreement between IBE and PBMC-FC based on these 45 paired observations in the 45 individuals: the mean difference between the two methods (IBE - PBMC-FC) was 78 CECs/ml, 95% confidence interval of the difference +/-365

CECs/ml.

120

3.5.2.2 Optimisation of flow cytometric gating for detection of endothelial cells

Having demonstrated consistently higher CEC counts with IBE compared to FC in patients with vasculitis, it was decided to investigate whether the mononuclear gate used in the FC protocol was optimal for detection of CECs. As had previously been observed HUVEC have different FSC SSC characteristics to mononuclear cells, thus in order to determine whether CECS might also have different FSC SSC characteristics their size was compared to both HUVEC and a second cultured primary endothelial cell

- pulmonary artery ECS (PAEC) which had been explanted from the pulmonary artery of the lungs of a child with pulmonary hypertension.

Cultured ECs are of similar size to CECs obtained from patients with vasculitis

Phase contrast and immunofluorescent microscopy of HUVEC and PAECs isolated using IBE from spiked whole blood or PBS was performed. This demonstrated that these cultured ECs overlapped in size and exhibited similar morphology to CECs extracted from the peripheral blood of patients with active vasculitis (Figure 3.10a-e).

The diameter of CECs from patients was determined by measurement of all the CECs in

5 random fields from 5 patients with active vasculitis (WG=3, PAN=2). In contrast,

CD146+ cells extracted from isolated PBMCs using immunomagnetic beads were significantly smaller than CECs (median diameter of CD146+ PBMCs and CECs were

6.8m versus 21.5m respectively (p<0.0001; Figure 1a). These observations emphasised the significant difference in size and morphology between CECs and

CD146+ PBMCs as defined by light and fluorescent microscopy, which would be likely to influence the flow cytometric FSC/SSC characteristics of these cell populations.

121

Figure 3-10: Comparison of cell size CECs vs ECs

Figure 3.10a-e: CECs were compared to pulmonary artery endothelial cells (PAECs), venous endothelial cells (HUVEC) and peripheral blood mononuclear cells (PBMCs) isolated from spiked blood samples using immunomagnetic beads coated with CD146. Figure 3.10a: Demonstrates the comparative cell diameter of CD146+ PBMCs, PAECs, CECs from vasculitis patients (5 children, 5 random fields, all CECs counted in field) and HUVEC. Of note was the wider range in diameter of

CECs compared with detached cultured ECs, overlapping both HUVEC and PAECs. Figure 3.10b:

Fluorescent microscopy of a single CD146+ PBMC (arrowed) bound to CD146-coated immunomagnetic beads; Figure 3.10c: PAEC with rosette of CD146-coated immunomagnetic beads; Figure 3.10d: CEC from a child with active PSV. CECs are characterised by rosetting of immunomagnetic beads and uptake of FITC labelled lectin from Ulex europaeus and are significantly larger than PBMCs at approximately 20-40m in size. Figure 3.10e: HUVEC demonstrating similar size to CECs (scale bar, 20m).

122

EC forward and side scatter characteristics

The FSC/SSC characteristics of detached cultured PAECs and HUVEC were examined using the previously described PBMC-FC instrument settings (Goon et al., 2006).

(Figure 3.1) Using these settings the majority of these ECs (>80% PAEC, >95%

HUVEC) were detected outside of the previously proposed CEC gate (Figure 3.11).

Therefore, having demonstrated that CECs from patients were of similar size and morphology as both PAECs and HUVEC (Figure 3.10), the FC instrument settings were optimised to capture both EC populations and mononuclear events, in order to recover all possible CEC events. The EC-optimised-FC gating strategy is illustrated in Figure

3.11.

3.5.2.3 Comparison of EC-optimised-FC with PBMC-FC

When the EC-optimised-FC gating strategy was compared with PBMC-FC to detect

CECs in children with vasculitis (paired observations in n=25 patients) and healthy controls (paired observations in n=20 subjects) greater numbers of CECs were detected using the EC-optimised protocol: the median EC-optimised-FC count was 39 CECs/ml, range 5-331; versus PBMC-FC CEC count 26 CECs/ml, range 0-326 (p=0.000). There was correspondingly poor agreement between the two FC protocols as indicated in the

Bland-Altman plot in Figure 3.12a. Mean difference between the two techniques (EC- optimised-FC – PBMC- FC) was 18 CECs/ml, 95%CI +/-63 CECs/ml.

3.5.2.4 Comparison of EC-optimised-FC with IBE for enumeration of CECs

The median CEC value obtained with the EC-optimised-FC protocol from all 45 subjects (25 vasculitis patients; 20 healthy adult controls) was comparable to IBE: the the two methods on Bland- Altman analysis: the mean difference (IBE – EC-optimised-

123

Figure 3-11: FSC/SSC characteristics of ECs and optimisation of flow cytometric gating to detect ECs

Figure 3.11a demonstrates the FSC SSC profile of whole blood and the mononuclear cell gate.

When pulmonary artery ECs (PAECs) (Figure 3.11b) and HUVEC (Figure 3.11c) were examined using the same instrument settings the majority of all ECs (>80% PAECs, >95% HUVEC) fall outside the previously proposed CEC gate. Figure 3.11d: EC-optimised-FC gate using logarithmic

FSC/SSC settings. When 100000 HUVEC were spiked into whole blood they formed a discrete population distinguishable from blood cells. The FC instrument settings were thus adjusted to design a FSC/SSC gate that was optimised to capture endothelial cells for both HUVEC (arrowed,

H=HUVEC); and PAECs(data not shown). N=neutrophils, M=monocytes, L=lymphocytes.

124

IBE median count was 32 CECs/ml, range 0-800; versus EC-optimised-FC CEC count of 39 CECs/ml range 5-331(p=0.25). Despite this, overall agreement was poor between

FC) was 60 CECs/ml, 95% CI +/-374 CECs/ml, figure 3.12b. Again much of this poor agreement arose from the children with active vasculitis and high CEC counts

(CECs>100/ml) as shown in the Bland-Altman plot (Figure 3.12b).

When the optimised FC gating strategy was used to detect CECs in children with vasculitis (n=24 patients) greater numbers of CECs were detected (median count 39

CECs/ml, range 9-330) (p=0.023). However these values remained considerably lower than those obtained concurrently using IBE (median count 84 CECs/ml, range 0-800

CECs/ml) (p=0.018). Again, lower levels of agreement were observed in children with more active vasculitis and high CEC counts (CECs > 100/ml) with better agreement between the two methods noted at lower CEC counts as shown in the Bland-Altman plot (Figure 3.12b). In healthy adult controls consistently higher numbers of CECs were recovered using the optimised FC gating strategy (median count 36 CECs/ml, range 5-166) (p = 0.001). Again there was no significant difference between this and the results obtained using IBE (p=0.179).

3.5.2.5 Factors influencing the sensitivity of IBE and FC for recovery of ECs

A series of in vitro experiments were undertaken to investigate further why there was poor agreement between IBE and FC (PBMC-FC and EC-optimised-FC) CEC enumeration in the clinical samples, particularly in those patients with active vasculitis.

PBS or whole blood was spiked serially with 100-100000 HUVEC/ml and the sensitivity of EC recovery using IBE or FC was evaluated under different experimental conditions. Using HUVEC spiked into WB was a better model of CEC enumeration, as

125

Figure 3-12: Bland-Altman analysis of agreement between EC-optimised FC and PBMC-FC and IBE for CEC enumeration

Figure 3.12a: shows the agreement between EC-optimised-FC (with FSC/SSC optimised to gate on

ECs) compared to the previously published PBMC-FC protocol. The mean difference between EC- optimised-FC and PBMC-FC was 18 CECs/ml, 95% CI +/-63 CECs/ml. Figure 3.12b: IBE compared to EC-optimised-FC. Again levels of agreement between IBE and FC were poor, particularly in patients with high CEC counts. Mean difference between IBE and EC-optimised-FC was 60 CECs/ml, 95% CI +/-374 CECs/ml.

126

it would mean that confounding cell populations and non specific binding relevant in both FC and IBE would also be present. However there was concern that the HUVEC might suffer complement mediated degradation and thus the PBS control was necessary.

As expression of CD34 on HUVEC is low (Muller et al., 2002) and also dependent on passage (Figure 3.7f and g) all CD45-/CD146+ events were considered to be EC in these experiments.

Effect of EC numbers on quantification accuracy of IBE and FC

The PBMC-FC gating protocol (Goon et al., 2006) failed to detect any ECs when 100

HUVEC/ml were added to whole blood (Figure 3.13a). EC recovery using this FC protocol was also low when between 1000-100 000 EC were added to whole blood: mean EC recovery (n=3 experiments) was 0.2-0.9% (Figure 3.13a). Detection of

HUVEC was similar in both PBS and whole blood using these instrument settings

(Figure 3.13b).

Using the EC-optimised-FC protocol to recover cultured ECs from PBS or whole blood resulted in greater recovery of CD45-/CD146+ cells (Figure 3.13a and b). Larger numbers of HUVEC (100000 HUVEC/ml) added to whole blood were easily identified as a distinct population from both the lymphocyte and monocyte populations (Figure

3.11d) although recovery was low with a mean of 38% (range 8-56%, n=5 experiments). Recovery using EC-optimised-FC was greatest at 1000 HUVEC/ml with mean recovery of 72% (range 21-154%, n=5 experiments; Figure 4). However, when fewer HUVEC (100 HUVEC/ml) were added considerable variation in recovery (36-

708%) was observed indicating sub-optimal sensitivity and specificity at this level of detection. Recovery of spiked HUVEC from PBS using the EC-optimised-FC protocol

127

was better than from whole blood with mean recovery of 63% for 1000 HUVEC/ml,

90% for 10000 HUVEC/ml and 70% for 100000 HUVEC/ml (n=3 experiments).

In contrast, the sensitivity of EC recovery using IBE in these experiments was between

25%-112% of the cells added to both whole blood and PBS, with greatest sensitivity at

1,000 cells per ml in WB (80% recovery from whole blood, range 69-94%, n=3 experiments) and at 100 cells/ml in PBS (101%, range 88-112%). IBE demonstrated lower recovery at higher EC counts: when 10000 HUVEC/ml were added, recovery was

44% in WB, (range 27-56%) and 43% in PBS (range 41-47%); and was even lower for

100000 HUVEC/ml at 28% in WB, range 14-37% and 45% in PBS (range 30-55).

Flow cytometric resolution: counting of rare events and effect of sample dilution

In order to determine whether the flow cytometer was “missing” HUVEC events in whole blood because of a phenomenon known as coincidence the final volume of sample for FC analysis was diluted to obtain a lower event acquisition rate.

Coincidence occurs if the sample is too concentrated and the total cell events acquired per second are too great to resolve all individual rare events (Jaroszeski and Radcliff,

1999). In case the presence of enumeration beads was influencing this, samples were run to completion to evaluate its effect. As Figure 3.11d demonstrated HUVEC are easily distinguished from whole blood on their CD146 expression alone, so in these experiments HUVEC were defined as CD146-high cells. Diluting the final flow volume to a concentration of 1/2 or 1/5, then running tubes to completion at medium flow rate increased the recovery of CD146-high HUVEC from spiked blood (100,000 cells/ml;

Figure 3.14). However, by increasing the final volume of sample the time taken to acquire the data was considerably increased, up to 90 minutes.

128

Figure 3-13: The effect of EC number on recovery - sensitivity

Figure 3.13ab: Whole blood (figure 3.13a) and PBS (figure 3.13b) were spiked with HUVEC and the cells recovered using IBE (grey bars, n=3 experiments) and the two different FC protocols (EC- optimised-FC, black bars, n=5 experiments; or PBMC-FC, white bars, n=3 experiments). Using FC

HUVEC were defined as CD45-/CD146+. On the whole HUVEC recovery from whole blood or PBS using IBE was comparable to EC-optimised-FC across the range of cells added. EC-optimised-FC

HUVEC recovery was sub-optimal at 100 HUVEC/ml, however, since the percentage recovery was

>100% in some experiments. Of note, previously proposed PBMC-FC strategies failed to detect any

CD45-/CD146+ cells in these experiments. Error bars demonstrate the standard error of the mean.

129

Figure 3-14: Resolution – sample dilution improves recovery

Figure 3.14: Diluting the final flow volume of blood spiked with HUVEC (100,000 cells/ml of blood) resulted in increased HUVEC recovery. This is due to the phenomenon known as coincidence, events are missed because the flow cytometer is unable to resolve them as flow rate is too fast.

Dilution of the sample improved this. Error bars demonstrate standard error of the mean. Mean recovery of neat sample was 43% which increased to 63% when dilute 1in2 and 66% when diluted 1 in 5.

130

Lysis of ECs in flow cytometric protocols limits CEC recovery from vasculitis patients.

When samples from patients with active vasculitis (n=3) were treated with FC red cell lysis buffer prior to enumeration using IBE there was a significant reduction in the number of CECs recovered. This reduction was observed using both BD FACS lyse, and a non-fixative ammonium chloride lysis buffer in 3 individual patients with active vasculitis. The median IBE CEC count from these patients for the non-lysed sample was 376 CECs/ml (range 216-552) which fell to 152 CECs/ml (range 144-168) for samples pre-treated with BD FACS lyse; and 120 CECs/ml (range 64-176) for samples pre-treated with ammonium chloride lysis buffer (Figure 3.15ab). Repeating these experiments using detached HUVEC added to PBS revealed that pre-treatment with lysis buffer did not reduce EC recovery Figure 3.15c.

3.6 Conclusions of methodological experiments

As this series of experiments has demonstrated, considerable optimization of FC protocols is still required in order to detect CECs with accuracy. Technical factors inherent to FC which influence CEC quantification include FSC/SSC gating strategy which necessarily require optimisation for EC capture; limited FC reliability of rare event measurement; variations in FC flow rate during long sample acquisition and loss of activated and necrotic CECs during the lysis and washing steps in preparation for FC.

For the purposes of this thesis it was concluded that detection of CECs using IBE methodology as described by Woywodt et al in their 2006 consensus paper (Woywodt et al., 2006a) would the most appropriate methodology for enumeration of CECs in the

GOSH vasculitis cohort, described in Section 2.5.

131

Figure 3-15: The effect of lysis on CEC recovery

Figure 3.15ab: When blood samples taken from children with active vasculitis (n=3) were lysed using BD FACs lyse (Figure 3.15a) or an ammonium chloride based lysis (AC lysis) (Figure 3.15b) prior to IBE there was a reduction in the number of cells recovered that was not seen when viable cells such as HUVEC were lysed prior to IBE at all concentrations of spiked HUVEC (Figure

3.15c).

132

3.7 Evaluation of CECs in vasculitis

As described previously, CECs obtained from patients with active vasculitis were affected by the lysis stages necessary for flow cytometry. Thus in this thesis CECs as a biomarker of endothelial injury in children with vasculitis were evaluated using IBE.

3.7.1 Patients and methods

3.7.1.1 Patients

42 children with primary systemic vasculitis were studied. These included 18 males and

24 females of median age 10.8 years (range 0.3 -16.6 years). The vasculitis type (using the Chapel Hill Consensus Criteria) (Jennette et al., 1994) was polyarteritis nodosa

(PAN) n=18, Wegener’s granulomatosis (WG) n=8, microscopic polyangiitis (MPA) n=

1, Kawasaki’s disease (KD) n = 6, Takayasu’s disease (TD) n = 3, Beçhets disease (BD) n = 1 and unclassified vasculitis n=5. Disease activity was assessed using a modified paediatric BVAS (Appendix 2). 25 healthy child controls (6 males, 19 females, median age 8.9 years, range 1.5-15.8), 13 febrile controls (7 males, 6 females, median age 7.3 years range 1.5- 14.0) and 20 healthy adult controls were also recruited (2 males, 18 females, age range 24-40 years). The febrile control group consisted of patients with a variety of autoimmune or other inflammatory disorders (including infection) without the presence of vasculitis.

3.7.1.2 Methods

CECS were isolated from EDTA coagulated peripheral venous blood as described in section 2.5.1. All blood used for analysis was drawn using a 21 gauge needle with minimal trauma and where possible the first 5 mls of blood were discarded (and used in

other assays). Blood was processed immediately after collection.

133

3.7.1.3 Statistics

Median CEC counts were compared between groups using a Kruskall Wallis test to check for overall differences. Individual Mann Whitney U tests for non-parametric data were used for specific inter-group comparisons. The Wilcoxon signed rank test was performed to compare groups with paired samples. Spearman Rank correlation coefficiants were examined to determine the levels of correlation between CECs and other markers of clinical disease activity. In all data graphs patient classification was denoted using colour coding as indicated below:

3.7.2 Results

3.7.2.1 CECs are elevated during active disease and decline with treatment

Kruskall Wallis testing indicated there was an overall significant differences between test groups (p<0.0001). CECs were significantly higher in 32 children with active vasculitis (as defined by a BVAS score >0, median value 208 CECs/ml) compared to 13 febrile disease controls (non-syndromic, inflammatory disorders, median value 36

CECs/ml, p=0.0009), 25 healthy child controls (median value 32 CECs/ml, p<0.0001) and 20 healthy adult controls (median value 20 CECs/ml, p<0.0001). There were no significant differences between 28 patients with inactive vasculitis (median value 32

CECs/ml) and these control groups.

134

Notably, there were a number of individual patients whose disease was considered clinically inactive, yet had elevated CECs, possibly indicative of subclinical endothelial injury (Figure 3.16a). The data shown in Figure 3.16a includes paired samples between individual patients with active and inactive disease and thus these groups could not be compared without their exclusion. Only the first observation made in each individual patient was included in this analysis. CECs were significantly higher in 32 patients with active vasculitis compared to 11 patients in remission (p<0.0001, median CEC count was 208 CECs/ml vs 16 CECs/ml).

In 19 individual patients with paired analyses (active vasculitis vs inactive vasculitis),

CECs were significantly higher during active disease compared to when they were considered to be clinically in remission (BVAS score = 0, median time to remission 3.5 months, p=0.01, Figure 3.16b). Of these 19 patients, 16 had frequent and detailed follow up over time for up to 18 months from disease onset and pre-treatment (prior to immunomodulatory therapy) or from a flare of their disease while on treatment (10 patients were followed up from disease onset - solid line, 6 from a disease flare - dotted line). Ultimately CEC numbers declined in all individuals although in some patients this took longer than in others (Figure 3.17). At median followup of 3 and 12 months from pretreatment at disease onset (10 individuals) CECs were significantly lower than at onset (p=0.005 and p=0.014). At 12 months 2 patients had flared, with corresponding elevation in CEC number (Figure 3.17).

135

Figure 3-16: CECs in vasculitis

Figure 3.16a: Circulating endothelial cells were significantly higher in 32 children with active vasculitis compared to 13 febrile disease controls (non syndromic, inflammatory disorders)

(p=0.0009), 25 healthy child controls (p<0.0001) and 20 healthy adult controls (p<0.0001). There were no significant differences between 28 patients with inactive disease and these control groups, however it is notable that there are a number of individual patients with inactive disease and elevated CECs, possibly indicative of subclinical endothelial injury. Figure 3.16b: In 19 individual patients with paired analysis, CECs were significantly higher when patients had active disease compared to when they were considered inactive (median time to remission 3.5 months, p=0.01).

136

Figure 3-17: CECs in vasculitis – response to treatment

Figure 3.17:16 individual patients had their circulating endothelial cells measured from initiation of treatment after disease onset or flare (dotted lines indicate patients followed from a flare while on treatment, solid lines indicate patients who were observed from pre-treatment). At median followup of 3 and 12 months from pre-treatment CECs were significantly lower than at onset (p=0.005 and p=0.014). At 12 months 2 patients had flared, with corresponding elevation in CEC number.

137

3.7.3 Correlation between CECs and disease activity

Spearman Rank correlation coefficiant analysis was performed between CEC count and the traditional markers of active disease; BVAS, ESR, CRP, platelet count, albumin and white cell count. Correlations were performed on all values obtained from the cohort of

44 patients. Table 3.2 summarises the results. CECs in vasculitis correlated significantly with disease activity as determined using a paediatric modified BVAS scoring system, although not significantly with other markers. Figure 3.18 demonstrates the relationship between CECs, BVAS , ESR and CRP visually. As can be seen in Figure 3.18a-c there were a number of patients with low ESR or CRP and low disease activity as defined by BVAS that had high CEC levels, again possibly indicative ongoing subclinical endothelial injury (Figure 3.18).

Table 3-2: Spearman Rank correlation coefficiants between CECs and conventional markers of disease activity

Marker of disease activity vs CECs Marker of disease activity vs CECs

BVAS r = 0.48 Platelet count r = 0.059 p<0.0001 p = 0.556 n = 110 n = 102

ESR r = 0.12 Albumin r = -0.160 p = 0.23 p = 0.11 n = 101 n = 99

CRP r = 0.134 White cell count r = 0.117 p = 0.17 p = 0.233 n = 107 n = 105

Significant correlations highlighted in blue

138

Figure 3-18: Correlation with disease activity

Figure 3.18a: Circulating endothelial cells correlated with disease activity in vasculitis as determined using a paediatric modified BVAS scoring system (p<0.0001, R=0.47) based on 105 measurements in 42 patients. Figure 3.18b and c CECs did not correlate significantly with ESR

(p=0.09, based on 98 measurements in 41 patients) or CRP (p=0.2, based on 102 measurements in

42 patients), as the graphs show there are a number of patients with low inflammation as determined by ESR or CRP or disease activity as defined by BVAS that have high CEC levels, indicative of subclinical endothelial injury.

139

Of the 16 patients who had detailed repeated measurements over time, a number had flares of active disease during the study period. Figures 3.19-3.22 hilight 4 interesting cases, the first three are patients studied from disease onset, the fourth a patient who underwent autologous stem cell transplantation. Figure 3.19 a-d shows the changes in

CEC count alongside ESR, CRP and BVAS over time from treatment in a 12 year old female with WG. As can be seen she had a disease flare (arrowed) in which CECs mirrored the more traditional methods used for assessing disease activity. In Figure

3.20 a-d an 11 year old female with WG was also followed up over time, unfortunately her CECs were not observed at the onset of her disease, however she entered an episode of severe, localized and (presumed) granulomatus flare (orbital mass) and as can be seen her CECs rose, but did not rise to the high levels seen in severe active systemic disease

(Figure 3.20). In Figure 3.21 a-d a 10 year old female with PAN was followed closely at the initiation of disease. After initially entering remission her ESR and CRP became elevated, however her CECs remained normal. This was subsequently proven not to be a flare of her disease but an associated urinary tract infection. Figure 3.22 a-d shows the changes in CEC number that occured in one individual who underwent autologous stem cell transplantation for recalcitrant vasculitic disease occurring secondary to a familial fever. CEC numbers declined from 3,136 to 20 after a week of pre-conditioning treatment (Campath®). CEC numbers remained low after the transplant, until the patient contracted a cytomegalovirus (CMV) infection, which corresponded with a rise in CEC count. CECs infected with CMV have been documented by a number of groups in patients with CMV infections (Pooley, Jr. et al., 1999,Percivalle et al., 1993). Other examples of the changes in CECs in individual patients with other markers of disease activity over time from treatment are found in Appendix 6.

140

Figure 3-19: Patient with WG - CECs and disease activity – predictive of disease flare?

Figure 3.19a indicates the change in CEC number over time from treatment in a 12 year old female with WG. Figures 3.19 b-d indicate the changes in disease activity as determined using BVAS scoring and the changes in the inflammatory markers ESR and CRP. CEC levels mirrored the changes seen in the more traditional clinical methods of assessing disease activity in this individual.

Rx = treatment, BVAS = Birmingham vasculitis activity score, ESR = erythrocyte sedimentation rate, CRP = c reactive protein.

141

Figure 3-20: Patient with WG – CECs and disease activity in orbital granuloma

Figure 3.20a indicates the change in CEC number during a granulomatus flare in an 11 year old female with WG. Figures 3.20 b-d indicates the changes in disease activity as determined using

BVAS scoring and the changes in the inflammatory markers ESR and CRP. Although CECs increased slightly, they were within the healthy child range.

142

Figure 3-21: Patient with PAN – CECs and disease activity

Figure 3.21a-d indicates the changes in CEC number and traditional markers of disease activity after treatment in a 10 year old female with PAN. At 6 months into her disease she experienced a spike in ESR and CRP (Figures c and d) subsequently proven to be due to a urinary tract infection

(UTI), not active disease – BVAS and CECS remained low.

143

Figure 3-22: CECs during autologous stem cell transplantation

Figure 3.22a-d indicates the changes in CEC number and traditional markers of disease activity in a patient who underwent autologous stem cell transplantation for recalcitrant vasculitic disease occurring secondary to a familial fever. CECs were high, prior to Campath® pre-conditioning and remained low during the transplant, until the patient contracted a CMV infection. Unfortunately

ESR and CRP were not measured as frequently in this patient as the research bloods as they are not part of the routine bloods performed in transplantation patients.

144

3.7.4 CECs in individual vasculitic syndromes

When the combined cohort of vasculitis patients was subdivided into individual diseases, the results, although based on small numbers were worthy of some comment.

During active disease patients with KD and TD (diseases predominantly affecting large and medium vessels) had lower CECs than patients with PAN or WG (disease predominantly affecting small to medium vessels) (Figure 3.22a). This interesting result which warrants further investigation, could indicate that this biomarker could depend on how widespread the overall surface area of the endothelium affected by the disease is. Bearing in mind that by far the largest surface area of endothelium is made up of smaller vessel endothelium (Ganong WF, 2001) this explanation may not be an oversimplification. During clinically defined remission (BVAS =0, Figure 3.22b) the majority of patients had low CECs. There were two notable exceptions: in the WG group, three patients had high CECs despite clinical remission; the single patient with

KD had coronary artery aneurysms and although systemic inflammation had subsided, was likely to be undergoing extensive vascular remodeling.

145

Figure 3-23: Subdivision into individual diseases - CECs in different vasculitic conditions

Active

Disease

Clinical

Remission

Figure 3.23a: When the patient cohort was divided into the different disease classifications, although the numbers were low, patients with Takayasu’s disease (TD) and Kawasaki disease (KD) had lower levels of CECs in active disease than those patient with Polyarteritis nodosa (PAN) or

Wegener’s granulomatosis (WG). Figure 3.23b: Interestingly in those patients in clinical remission, the group with the highest proportion of individuals who had elevated CECs was the WG group, a disease affecting predominantly small vessels and notorious for its relapsing nature and poor clinical outcome.

146

3.8 Discussion

Circulating endothelial cells are elevated at times of active disease in PSV and correlate with disease activity although inconsistently with routine clinical markers of inflammation. In individual patients, CECs were elevated during active disease and declined over time with remission inducing treatment (Figures 3.16 and 3.17). CECs tended to be higher in patients with diseases affecting predominantly smaller vessels, although this observation was preliminary (Figure 3.18). In the small number of patients who entered a flare of systemic vasculitic disease while on treatment, CECs correspondingly increased (Figure 3.19, Appendix 6 patients 3,4,7,10 and 13), although in localised flare this was not to the high levels seen in systemic disease (Figure 3.20,

Appendix 6 patients 6 and 8), a finding that has also been observed in adults with WG

(Woywodt et al., 2006b).

The significant differences in CEC number between active disease and healthy age matched and febrile controls, suggest that CECs may be a reliable marker of active vasculitis and endothelial injury. Vasculitis patients are a highly heterogeneous group and it is thus not surprising that some patients with active disease do not have high elevation of CECs, perhaps related to the time of sampling in relation to initiation of treatment and the location and extent (systemic versus localised) of the vasculitic injury.

As discussed in the introduction to this chapter, it remains unclear where CECs originate and how they enter the venous circulation. Of some concern in this patient population was the small number of patients who had elevated levels of CECs but no clinical signs of disease activity. It may be that this is indicative of an increased endothelial turnover and vascular remodeling rather than acute injury, an hypothesis that makes sense considering the time it takes CECs to return to normal levels in some

147

individual patients, despite bringing the inflammation under control. A less optimistic hypothesis, however is that in these patients endothelial injury is ongoing, despite the clinical impression of disease remission.

The poor correlation with clinical markers of inflammation is consistant with the conflicting results observed in previous studies in adult, whereby strong correlations have been observed between CECs and BVAS in ANCA associated disease (Woywodt et al., 2003b,Woywodt et al., 2006b) but only a weak correlation was been observed with CRP (Woywodt et al., 2006b). As discussed in the introduction to this thesis, levels of ESR and CRP may remain normal during active vasculitis, particularly in episodes of localised disease and granulomatous flares (Salvarani et al., 2003,Sproson et al., 2007) and thus this result is unsurprising, indicating as mentioned above that endothelial injury can occur in the absence of inflammation.

Unfortunately the technique used in this thesis for detection of CECs is not yet suitable for large scale automated routine clinical testing, because of its time consuming nature and the training required to reliably read samples. Development of a flow cytometry based assay optimised for detection of necrotic cell populations was not a direction in which this project went due to time constraints and the availability of a well established and reproducible assay for reliable detection of CECS. However flow cytometry protocols can be relatively simply automated and future work based on this would be of considerable interest. A multiparametric approach for detection of CECs would allow a much more defined population, that could provide more understanding of the exact nature of CECs.

148

As red cell lysis used in FC was one of the major causes of loss of the fragile CEC population, the use of similar techniques to those used for examination of apoptotic and necrotic cells using FC such as no wash/no lyse protocols (Darzynkiewicz et al., 1997) could overcome this limitation. However without red cell lysis there is much background noise due to non removal of red cells, and in order to distinguish cells of interest from erythrocytes, a threshold based on fluorescence must be used. As yet these techniques are only described for rare event analysis based on very bright staining populations with metabolic dyes rather than antibody staining (Rehse et al., 1995) and thus may not be suitable for detection of CECs. Another potential method for overcoming these problems (in order to benefit from the multiparametric potential of

FC) would be to enrich the CEC population prior to preparation for FC, thus removing the majority of superfluous cells and much of the background noise. This could be done using FC compatible magnetic beads as described recently by Widemann et al.

Interestingly, although this protocol gave comparable values to IBE, statistically the results were consistently higher when healthy individuals, patients who had undergone coronary angioplasty or renal transplantation were studied (Widemann et al., 2008).

The authors attributed this to the loss of fragile cells populations during the washing steps required for IBE and thus this finding is extremely promising for future studies in vasculitis, whereby conventional FC methodology resulted in considerable loss on handling of CECs (Clarke et al., 2008).

An alternative multiparameter identification technique which has been developed into a commercially available CEC counting system involves automated microscopy

(Celltracks endothelial cell kit, Immunicon). CECs are isolated from peripheral blood using CD146 immunomagnetic beads then stained with DAPI, anti-CD105-PE and anti-

149

CD45-APC. The stained cells are transferred to a counting chamber and are scanned with an automated microscope. Events positive for DAPI and CD105 and negative for

CD45 have composite images taken which an operator can then distinguish on morphology. Not surprisingly the setup costs for this kind of system are prohibitively expensive. However, with the right software and using a microscope with a motorised stage and filter set, it may be possible to emulate the technology at a fraction of the cost.

Unfortunately time restraints did not permit the development of either of these avenues of work and thus for the purposes of this thesis and to examine CECs within the vasculitis patient cohort, IBE as described by the 2006 consensus paper was therefore the methodology of choice (Woywodt et al., 2006a).

While IBE has proved to be more reproducible and robust than FC for assessment of

CECs in vasculitis, excluding a single study in KD (Nakatani et al., 2003) this is the first study to examine a paediatric cohort with necrotising vasculitis in detail. Thus potential paediatric related sources of error needed to be considered. In particular the effect of traumatic venepuncture and the inability to guarantee that all samples would have the first 5 mls of blood discarded due to the small samples typically obtained for research in children. Although the data is preliminary, due to lack of willing volunteers, traumatic venepuncture did considerably elevate the CECS in a single healthy adult volunteer above the median values normally obtained in healthy adult controls (20

CECs/ml in this study, comparable with previous studies of around 10-20 CECs/ml)

(Blann et al., 2005). However the range seen in active AAV (20-5700 CECs/ml)

(Woywodt et al., 2003b) was much higher than this and the experimental venepuncture performed in this study was far more traumatic than would be hoped under normal conditions. Interestingly discarding the first 5 mls of blood in this individual, would

150

not have influenced the CECs enumerated. Although it is difficult to draw firm conclusions from these preliminary findings, as long as the same conditions are applied to the control group as to the patient group it is possible that these child related confounding issues will not affect our the results, because CECs are so grossly elevated in children and adults with active vaculitis.

In conclusion it appears that in vasculitis in children CECs reflect the extreme endothelial damage associated with this condition and are a useful tool for assessing disease activity. Further development of the methodology towards a routine clinical test would be an obvious future direction of this work.

151

4 Cellular Microparticles

4.1. Summary

4.2. Introduction

4.3. Detection of MPs-methodological optimisation

4.4. Conclusions of methodological experiments

4.5. Evaluation of cellular MPs in vasculitis

4.6. Discussion

152

4.1 Summary

This chapter describes the methodology used to detect cellular derived microparticles.

Like CECs this technique can provide insight into the activation state of the endothelium, and may be an appropriate biomarker of disease activity in vaculitis.

However, due to the multiparametric nature of the technique used, insight can be gained at the same time relating to the activation state of a number of different cells (eg neutrophils, monocytes) which are also pivotal to disease activity in vasculitis. The latter part of this chapter examines in detail the relationship between microparticles and disease activity in children with primary systemic vasculitis and further expands on previously published data from the group (Brogan et al., 2004a).

4.2 Introduction

Microparticles (MPs) are membrane derived vesicles 0.1-2µm in size that are released during physiological or pathological cell or platelet activation (Piccin et al., 2007). MPs can express surface markers of their cell of origin and in contrast to quiescent cells can express phosphatidylserine (PS) on their outer leaflet (Hugel et al., 2005), a consequence of the membrane remodelling or in more pathological situations the loss of integrity necessary for their formation.

4.2.1 Formation of MPs

Cellular and platelet plasma membranes consist of a phospholipid bilayer, which in unstimulated cells has an asymmetrical distribution of its components. Uncharged phospholipids such as phophatidylcholine (PC) and sphyingomyelin (SM) are predominantly expressed on the outer leaflet and negatively charged phospholipids such 153

as PS and phosphatidylethanolamine (PE) found on the inner leaflet. Membrane remodelling and subsequent PS expression on the outer leaflet of cells promotes the activation of clotting factors (Bevers et al., 1982) and recognition of the cell by phagocytes and is considered to be a key stage in apoptosis (Fadok et al., 1992).

Membrane composition is tightly regulated by enzymes, known as flippases, floppases and scramblases (Piccin et al., 2007). Flippases inwardly direct the negatively charged phospholipids in an ATP dependent manner, whereas floppases act in an outward manner (Daleke, 2003). Scramblases degrade the bilayer phospholipid gradients, acting in both directions. A Ca2+ activated scramblase has been described which plays an important role in membrane reorganization in response to cell stimulation (Daleke,

2003), the resultant membrane disorganisation results in PS expression on the surface of the cell. As transient rises in Ca2+ can alter the membrane asymmetry and activate enzymes involved in cytoskeleton remodeling, such as calpain, it is believed to play a key role in MP formation (Wiedmer et al., 1990). Thus Ca2+ ionophores can be used to generate MPs.

In vitro it is relatively easy to stimulate isolated cell populations to generate MPs by the use of cytokines of ionophores. In vivo the stimuli which generate MPs and their functions remains unclear, however it is apparent that they are a consequence of cell activation, apoptosis and necrosis and are the product of remodeling of the cytoskeleton and loss of membrane asymmetry.

4.2.2 Detection of MPs

The majority of studies which have detected MPs have used flow cytometry (FC) based

154

protocols. This allows for a multi-parameter based assessment of the MP population.

However there is much variation in methodology and no consensus protocol has been described (Enjeti et al., 2007). An alternative technique utilizes sandwich ELISA, whereupon MPs are immobilized on the ELISA plate by PS binding. However this does not provide as much information relating to phenotype as FC allows, and also does not allow analysis of PS negative MPs. Thus in this thesis FC was employed to study MPs.

4.2.3 Function of MPs

The phenotype and function of MPs from different cellular origins remains to be fully elucidated, however it is apparent that different cellular MPs and the circumstances under which they are generated can exert different functional effects. MPs are generated in response to numerous stimuli and while some groups hypothesise that their role is in inter-cellular signaling, others suggest it is a consequence of cell survival. The expression of high levels of PS on the outer surface of all phenotypes of MPs indicates widespread procoagulant potential, as PS promotes the coagulation cascade (Hugel et al., 2005,Toti et al., 1996). Elevated levels of MPs from endothelium, platelets, mononuclear cells and neutrophils have been documented in a number of diseases associated with cardiovascular injury, including vasculitis in both adults (Daniel et al.,

2006,Erdbruegger et al., 2008) and children (Brogan et al., 2004a) and will now be discussed in more detail.

4.2.4 Endothelial MPs

Endothelial MPs (EMPs) were first described by Hamilton et al in 1990, being produced in response to complement mediated attack on HUVEC (Hamilton et al.,

1990). The resultant MPs generated were less than 1µm in size and bound activated

155

factor V a key component of the coagulation pathway (Hamilton et al., 1990). Cultured

ECs have subsequently been shown to produce MPs in response to various stimuli including inflammatory cytokines such as TNFα (Combes et al., 1999) and IL-1α (bid

Hussein et al., 2003), CRP (Wang et al., 2007), thrombin (Sapet et al., 2006) and serum from patients with antiphospholipid syndrome (APS) (Combes et al., 1999), thrombocytopenic purpura (Jimenez et al., 2001) and multiple sclerosis (Minagar et al., 2001).

Elevated levels of EMPS have been detected in many diseases associated with vascular injury including vasculitis (Brogan et al., 2004a), acute coronary syndrome (Bernal-

Mizrachi et al., 2003), multiple sclerosis (Minagar et al., 2001) and type 2 diabetes where they are predictive of associated coronary artery disease (Koga et al., 2005).

Thus are not specific to the vasculitides, but reflect activation and injury.

4.2.4.1 Phenotype of EMPs

There is no agreed consensus as to what defines an EMP other than the presence of endothelial cell membrane antigens on the surface of a MP. More often than not MPs are defined by the presence of PS (shown by binding of AnnexinV (AnV)) on the surface of particles with low FSC, however some groups consider AnV negative populations as well. EC markers that have been examined on MPs include CD144,

CD146 (Faure et al., 2006), CD31, CD105, CD51, ICAM-1, E-selectin and CD106

(Jimenez et al., 2003).

Jimenez et al demonstrated that when ECs were deprived of growth factors to induce apoptosis the MPs generated expressed CD31 and CD105 (Jimenez et al., 2003), however after cellular activation with TNFα the MPs produced had much higher

156

expression of the inducible activation markers E-selectin, ICAM-1 and CD106 (Jimenez et al., 2003). Thus examining a combination of markers may give insight into the circumstances of MP generation. For the purposes of this thesis, the constitutive endothelial markers CD144 and CD105 and the inducible endothelial markers ICAM-1 and E-selectin were examined.

CD144 is specific for ECs in adults, however can be found on haematopoeitic stem cells in the foetus (Kim et al., 2005) and endothelial progenitor cells post-natally (Rehman et al., 2004). It is involved in maintaining the tight cell-cell contact between ECs, thus significant EC disruption would necessitate it being exposed. CD105 or endoglin is part of the TGFβ receptor complex on endothelial cells and like CD144 it is not frequently found on cells circulating in the blood, other than leukaemic cells (Duff et al., 2003). It is upregulated in hypoxia (Sanchez-Elsner et al., 2002) and on proliferating ECs (Duff et al., 2003). Although CD31 and CD146 are alternative EC constitutive markers, they are found on haematopoietic cell types which are more prevalent and may generate MPs in inflammatory conditions; CD146 is found on activated T cells and CD31 is found on platelets, neutrophils, T cells and NK cells. E-selectin is highly specific for activated

ECs and has only been observed on other cell types in vitro; TNFα stimulated foetal astrocytes (Hurwitz et al., 1992) or on CD4+ T cells after antibody stimulation of membrane-bound TNF (Harashima et al., 2001). In vitro its expression is transient, peaking at 4hrs after stimulation (Pober et al., 1986). ICAM-1 is constitutively expressed on EC, but is also upregulated on activation, peaking at 4hrs after IL-1 stimulation (Dustin et al., 1986). It is less EC specific though and is also expressed on activated leucocytes.

157

4.2.4.2 Functional significance

Microparticles generated in vitro from cultured endothelial cells such as HUVEC or microvascular endothelial cells (MVEC) have been shown to activate monocytes (Jy et al., 2004b) and enhance monocyte-endothelial adhesion (Huber et al., 2002), suppress

HUVEC growth in a matrigel angiogenesis assay (Mezentsev et al., 2005) and promote coagulation in vitro and in vivo (bid Hussein et al., 2008). In isolated perfused human microvessels, the addition of EMPs reduced acetylcholine induced endothelium dependent vasodilation and in the same study, when EMPs were injected into both rats and mice acute lung injury occurred (Densmore et al., 2006).

4.2.5 Platelet MPs

Platelet MPs (PMPs) are the population which have been most extensively studied to date and make up the predominant MP population in healthy individuals (Diamant et al.,

2004). Stimulation of platelets by a wide variety of agents (such as thrombin, collagen or ADP) induces the aggregation necessary for haemostasis and can also result in MP release (Diamant et al., 2004). PMPs are also released in vitro by shear stress

(Miyazaki et al., 1996).

Elevated levels of PMPs have been found in patients suffering thrombotic complications or vascular injury including acute myocardial infarction (Matsumoto et al., 2002), peripheral arterial disease (Zeiger et al., 2000), vasculitis (Brogan et al., 2004a), hypertension (Preston et al., 2003) and in malignancy (Kim et al., 2003). Conversely patients with bleeding disorders, such as Castamann’s disease have reduced PMPs

(Castaman et al., 1997) and in the rare genetic disease Scott’s syndrome where PS is

158

unable to be exposed on the surface of cells, PMPs cannot be generated at all (Toti et al., 1996).

4.2.5.1 Phenotype of PMPs

As with EMPS, PMPs have been assessed based on their expression of PS in combination with a platelet surface marker such as; CD42a, CD42b, CD41, CD61,

CD63 or P-selectin (Piccin et al., 2007). For the purposes of this thesis CD42a, which is constitutive and selectively expressed on platelets and the adhesion molecule P- selectin which is found on platelets although less selectively (it is also expressed on endothelial cells (Merten and Thiagarajan, 2004)) were chosen. CD42a or glycoprotein

IX is part of the vWF receptor complex (consisting of 4 subunits – gp1bα and β, gpIX and gpV) which is involved in mediating the tethering of platelets to exposed vWF on the subendothelium. The complex also interacts with thrombin, P-selectin and coagulation factors (Du, 2007). P-selectin mediates rolling of platelets and leukocytes on activated endothelium and intereactions between platelets and leukocytes (Merten and Thiagarajan, 2004).

4.2.5.2 Function of PMPs

In vitro PMPs have been shown to initiate and propagate coagulation (Warkentin et al.,

1994), activate neutrophils (Jy et al., 1995), monocytes (Barry et al., 1998) and endothelial cells (Barry et al., 1997) and mediate the cell-cell interactions between them

(Barry et al., 1998). PMPs have been shown to transfer platelet antigens (Majka et al.,

2007) and other receptors between cell types such as CXCR4, which can make cells susceptible to the HIV virus (Rozmyslowicz et al., 2003). In lung cancer PMPs can induce metastasis (Janowska-Wieczorek et al., 2005). Conversely under some

159

conditions PMPs have demonstrated anticoagulant properties, via acceleration of inactivation of activated factor V by protein C (Tans et al., 1991) and by their expression of proteins which inhibit coagulation, such as TF pathway inhibitor

(Steppich et al., 2005), suggesting a complex role in the balance of haemostasis.

4.2.6 Leukocyte MPs – Neutrophils, monocytes and lymphocytes

As neutrophils play a key role in the pathogenesis of AAV, it is perhaps not surprising that elevated levels of neutrophil derived MPs (NMPs) have been described in vasculitis

(Daniel et al., 2006). Elevated levels of NMPS have been also been documented in sepsis (Joop et al., 2001). Both studies examined the immunoglobulin surface marker

CD66b. For the purposes of this thesis CD15, CD11b, and L-selectin were examined in addition to CD66b. CD15 is a constutively expressed adhesion molecule found in high levels on neutrophils and is involved in phagocytosis and chemotaxis (Skubitz et al.,

1989). CD11b is an integrin involved in neutrophil and monocyte adhesion with the endothelium and is upregulated on stimulation of neutrophils (Finn et al., 1994). L- selectin is involved in leucocyte-endothelial interactions and is downregulated on neutrophil stimulation (Finn et al., 1994).

Microparticles derived from monocytes (MMPs) have previously been detected by their high expression of CD14 in combination with AnV (Kanazawa et al., 2003). CD14 is part of the LPS receptor complex and is found on a small proportion of neutrophils.

Stimulated monocytes can synthesis high levels of tissue factor (TF). While both endothelial cells and platelets can release TF from intracellular granules on stimulation, it is hypothesized that monocytes may be the predominant source of TF positive MPs, which have the potential to form complexes with platelets and PMPs, propagating

160

coagulation (Osterud, 2001). Elevated levels of CD14+ MPs have been documented in lung cancer (Kanazawa et al., 2003) and type 2 diabetes (Omoto et al., 2002).

Elevations in lymphocyte microparticles (LMPs) have been documented in preeclampsia (VanWijk et al., 2002) and HIV(Aupeix et al., 1997) and has been postulated to be associated with high levels of lymphocyte apoptosis in these conditions. CD4 and

CD8 MPs were examined in these studies. For the purposes of this thesis CD3 positive

MPs were assessed. CD3 is part of the T cell receptor complex and is constitutively expressed on mature T cells (Goldsby RA et al., 2003a).

4.2.6.1 Function of leukocyte MPs

Leukocyte derived MPS have been less well studied than those derived from endothelium or platelets. It is apparent that they are associated with inflammation and infection and there is evidence accumulating that they may also have functional significance. MMPs express high levels of TF, which can be transferred to other cell populations (Osterud, 2001) indicating a role in coagulation. LMPs have been shown to activate monocytes (Scanu et al., 2008) propagating inflammation. The data relating to

NMPs is more complex; while NMPs can stimulate endothelial cells to release IL-6

(Mesri and Altieri, 1998), perhaps potentiating inflammatory reponses, conversely they can act in an anti-inflammatory manner. In vitro NMPs generated from N-formyl- methyl-leucyl-phenylalanine (FMLP) stimulated adherent neutrophils (cocultured with

HUVEC),can inhibit neutrophil-endothelial interactions (Dalli et al., 2008). In vivo, mouse NMPs generated in the same manner (using mouse endothelium) inhibited IL-1β induced inflammatory infiltrate in a mouse air pouch model (Dalli et al., 2008).

161

4.2.7 Microparticles in vasculitis

To date, there have been 3 studies and one case study examining MPs in vasculitis, the first by Brogan et al demonstrated an elevation in E-selectin and CD105 expressing

EMPs and CD42a (although not P-selectin) expressing PMPs in children with active

PSV compared to healthy controls (Brogan et al., 2004a). Daniel et al demonstrated an elevation in NMPs expressing CD66b and PMPs expressing CD41a in adult patients with AAV during active disease compared to controls. During remission patients also had higher NMPs than controls (Daniel et al., 2006). Recently Erdbruegger et al examined EMPs in adults with AAV and demonstrated an elevation E-selectin and

CD105 EMPs during active disease compared to patients in remission and healthy controls (Erdbruegger et al., 2008). The case study described by Kumpers et al, demonstrated eleveated EMPs, PMPs and leucocyte MPs (CD45+) at onset of Churg-

Strauss syndrome that declined with treatment (Kumpers et al., 2008).

While Erdbruegger et al followed changes in EMPs for 6 months from treatment and demonstrated a rapid decline, the two other studies only examined a small number of patients before and after treatment. Thus a comprehensive study relating to changes in

MPs in relation to treatment and disease flare has yet to be performed. The study described herein will begin to examine MPs in this context.

162

4.3 Detection of MPs – methodological optimisation

4.3.1 Preparation of samples

The methodology described is based on work previously optimized and published by the group (Brogan et al., 2004a) and thus few changes to the basic protocol were required. Platelet poor plasma was generated as described in Section 2.4.7. Figure 4.1 demonstrates the extent to which platelets were removed from the sample after each spin. 50µl samples of blood prior to spinning, plasma after 1 x 5 mins spin at 5,000g and after 2x 5 mins spins at 5,000g were diluted in 200µl of sterile filtered (0.22µm)

FACS buffer and run for 45 seconds on a BD Facscalibur, using logarithmic instrument settings, with FSC amplitude of E01. As can be seen in Figure 4.1a-c while 1 spin removes a large proportion of platelets (Figure 4.1b), 2 spins removes almost all of them (Figure 4.1). The removal of platelets from plasma is an important step in the protocol as to (some extent) they have overlapping FSC/SSC profiles to MPs making

FC analysis more difficult in the presence of platelet contamination. Moreover, the freeze thawing of platelets may disrupt their membrane and could result in false positive

MP events (see section of effect of freeze-thawing below).

4.3.2 Preparation of positive controls for setup experiments

In order to detect the different MP populations in vasculitis patients MPs were generated from activated parent cells of different types. For EMPs HUVEC were isolated and cultured as described in section 2.4.1. After first passage they were grown on 6 well plates until 80% confluence, whereupon they were serum starved overnight and then stimulated with 100ng/ml TNF for 24hrs. To generate NMPs neutrophils were isolated using polymorphprep as described in section 2.4.5 and stimulated for 30 mins-

4hrs using FMLP (10ng/ml), TNFα (100ng/ml) or LPS (10ng/ml) or combinations of 163

Figure 4-1: The removal of platelets for generation of platelet poor plasma

Figure 4.1a-c: Using instrument settings optimised for MP detection (Voltage E01, logarithmic scale) the platelet population can be easily seen within whole blood. (Figure 4.1a, circled). After 1 and 2 x 5000g spins, 50µl samples of the plasma obtained were diluted with 200µl of sterile filtered

PBS + 0.1% azide and run for 45seconds at low flow. Figure 4.1b: After 1 x 500g spin some platelets are removed. Figure 4.1c: After 2 spins most platelets are removed. Absolute platelet count dropped from 250 to 27 after the second spin in this representative experiment.

164

FMLP with another reagent. To generate MMPS and LMPs, PBMCs were isolated as described in Section 2.4.4 and stimulated using 100ng/ml TNF for 30 mins-4hrs. To detect a mixed population of MPs, including platelet, monocyte, lymphocyte and neutrophil whole blood was stimulated with LPS 100ng/ml. All stimulation experiments of blood and cells in suspension were carried out in polypropylene tubes to prevent MPs sticking to the plasticware. Activation of parent cells was confirmed using FC as 2.5.4.

The phenotype of the MPs generated and their activated parent cells is shown in

Appendix 4.

4.3.3 Staining of MPs for flow cytometry

A known and exact volume of sample containing MPs – either platelet poor plasma or culture media from experiments (which had been centrifuged at 500g for 5 mins to remove cells) was centrifuged at 13000g for 60 mins to pellet the MPs. This was necessary to allow samples to be resuspended in AnV buffer (containing 2.5mM CaCl2) as AnV binding is Ca2+ dependent and thus there must be optimal Ca2+ present to allow binding to occur. Once the supernatant had been carefully decanted the MPs were resuspended in sufficient Annexin V buffer to allow 35µl to be added to wells on a 96 well plate containing 5 µl of Annexin V FITC and 10 µl of a PE/PE-CY5 or Percp conjugated antibody as described in Chapter 2.5.3. After 10mins of staining at room temperature 200µl of Annexin V buffer was added to stop staining and allow flow cytometry of sample. The full MP staining plan is shown in Appendix 3.

4.3.4 Flow cytometry of MPS

MPs were analysed on a BD Facscalibur using logarithmic instrument settings with a

FSC voltage of E01.

165

4.3.4.1 Optimisation of instrument settings using latex beads

To establish the limits of detection and appropriate instrument settings 0.3, 0.8 and 3µm latex counting beads were used. It is important to note that while these beads give an approximate estimate of size they are not directly comparable with MP size as they are of different composition and therefore have different optical properties (the refraction and reflection of the laser beam in the flow cytometer which contribute to FSC and SSC will be different when considering latex particles to cell membranes). Thus MPs should be FSC/SSC gated based on their discrete populations and not based entirely on where beads fall on the FC (Furie and Furie, 2006). As can be seen in Figure 4.2a, 3µm latex beads lie within a similar FSC location to platelets when whole blood was examined and thus could be used as an approximate upper boundary of the FC region of interest.

In order to investigate the lower limits of detection, 0.8µm, 0.3µm beads and sterile filtered water alone were also examined. Sterile filtered water was used to examine the absolute limits of detection. The noise detected using 0.2µm filtered water was presumed to be electrical noise and therefore a FSC threshold was established to exclude the peak of this (Figure 4.2b). Once this threshold was established the 0.8 µm and 0.3µm beads were examined. As can be seen from the histograms, while the 0.8µm beads are easily distinguishable from noise (Figure 4.2c) the 0.3µm beads are not

(Figure 4.2d). Thus, as MPs are close to the limits of detection of the flow cytometer, it is apparent that they cannot be entirely distinguished from this noise based on their FSC and SSC characteristics alone, emphasizing the importance of using AnV as a constitutive MP marker. Figure 4.3 shows the gating strategy used for detection of

MPs. As can be seen a number of events positive for AnV fall within this overlapping noise region.

166

Figure 4-2: Optimisation of MP gating using 0.3µm, 0.8µm and 3µm latex beads

Figure 4.2a demonstrates the FSC profile of 3µm latex beads and shows that they sit at the upper end of the platelet region that was described in figure 3.16a-c. This was used as the upper limit of the region of interest. Figure 4.2 b indicates the lower limits of detection by using sterile filtered distilled water only and highlights the extent of background electrical noise within the system. The blue line indicates the FSC threshold used to remove the peak of the background noise Figure 4.2c shows that 0.8µm beads can be easily distinguished from both the noise and the 3µm bead population. 0.3µm beads however cannot be distinguished from the background noise (figure 4.2d)

167

4.3.4.2 Absolute enumeration of MPs

As described previously in Chapter 3 there are 3 commonly used methods for absolute enumeration of flow cytometric events: 1. Allowing samples to run to completion;

2.Using an internal standard such as counting beads; or 3.calculating the flow rate for a set collection time prior to running the samples. While using an internal standard had proved more accurate for enumerating CECs, the presence of beads within the MP sample might influence the level of back ground noise making it more difficult to reliably count MPs accurately. Thus this issue needed further clarification.

Optimisation of detection of 3µm latex beads

In previous studies, 3µm latex beads had been used as an internal standard (Brogan et al.,

2004a). In the first instance it was important to establish that these counting beads were being detected with accuracy. 3µm latex beads were diluted in sterile filtered distilled water to give a final concentration of 20,000 beads/µl. 10 µl of this stock (200,000 beads) was then added to 240µl of sterile filtered water. Samples (n=3) were then run to completion at a medium flow rate using the instrument settings as established in

Figures 4.2 and 4.3. Using these instrument settings beads recovery was poor, with mean recovery of only 8.9% (range 6.7-11.4%).

The manufacturer’s recommendations for the use of latex beads suggested that the use of a detergent may prevent beads clumping together, a phenomenon which could explain this poor recovery. Thus an experiment was performed to determine if this could be the cause of poor recovery. Bead clumping can occur over time, so all bead samples used in these experiments were prepared immediately prior to use in FC experiments. Beads were diluted in sterile filtered distilled water or 0.05 % TWEEN to

168

Figure 4-3: MP gating strategy, based on Annexin V positivity

Figure 4.3a shows the FSC vs Annexin V positivity of MPs from a patient with active vasculitis. The

MPs form a discrete population at the lower end of the FSC axis that can be gated on, notably a number of events fall within the electrical noise region described in Figure 4.2. When a histogram is drawn based on this gating, (figure 4.3b) this gated MP populations sample can be seen to have high numbers of CD144 positive MPs. The black line represents the negative isotype control antibody, and the red line the CD144 antibody staining. The marker was set at 2.5% positivity of the isotype control alone, the non-specific 2.5% being subtracted from the CD144% staining to obtain the final value of 38% in this example.

169

give a stock solution of 20,000 beads/µl. 10µl of each stock solution (200,000) was then added to 240µl of sterile filtered distilled water, 0.05% TWEEN or AnV buffer.

As Figure 4.4b demonstrates the recovery of beads was poor under all conditions, which indicated that bead clumping may not be the reason for poor recovery. Alternative explanations thus included suboptimal instrument settings.

The aforementioned phenomenon of flow cytometric “coincidence”, whereby too many individual events are occurring simultaneously to detect with accuracy, could conceivably influence recovery. As Figure 4.4a demonstrates, the flow cytometric instrument settings are at the limits of detection and there is much background noise

(circled) which contributes to the events occuring. In order to determine if this phenomenon was influencing the ability to recover beads a FSC threshold was set

(Figure 4.4c). As Figure 4.4d shows, this increased the average bead recovery from

8.9% to 111% (range 89-133%). Thus in order to enumerate counting beads accurately the instrument settings must be optimised to the counting beads. This was problematic as using the FSC threshold show in figure 4.4c would mean that the majority of the MPs would go undetected: as can be seen in figure 4.3a the MP population has a low FSC.

Thus it was apparent that the use of these 3µm beads as an internal standard was not appropriate in this experimental setting for reliable MP quantification.

As the background noise was having such a detrimental effect on bead enumeration

(possibly via coincidence or another undefined mechanism) this observation raised questions regarding the effect of “noise” on MP counts. MPs are detected using AnV positivity to define them, so using a threshold based on the MP AnV phenotype could reduce background noise and may allow a FITC labeled counting bead to be used as an

170

Figure 4-4: Recovery of 3µm counting beads

Figure 4.4: When 200,000 latex beads were spiked into 240µl of sterile distilled water and acquired using MP flow cytometric instrument setting (Figure 4.4a) recovery was poor with mean recovery

8.9% (range 6.7-11.4%). To assess if this was due to bead clumping the stock sample of beads was diluted in the presence of detergent as recommended by the manufacturer (0.05% TWEEN) - white bars (Figure 4.4b) and spiked into distilled water, Annexin V buffer or more detergent. This had no affect on bead recovery. There was much background noise in the sample (circled in Figure 4.4a) so a FSC threshold to cut out the noise was set (Figure 4.4c). This resulted in good recovery of beads (Figure 4.4d, white bar) mean recovery 111% (range 89-133%).

Using fluorescence thresholds to reduce background noise

171

internal standard. As had been established for the non-labelled latex counting beads it was important to determine that the FITC labelled beads (3µm flouresbrite yellow green microparticles, Polysciences) were being recovered with accuracy. As before the beads were diluted in sterile distilled water to give a final concentration of 20,000 beads/µl, of which 10µl (200,000 beads) was added to 240µl of sterile filtered distilled water.

Figure 4.5a-c demonstrates that when instrument settings were optimised to beads

(figure 4.5a) and noise (circled) was excluded using an FL1 threshold (figure 4.5b) bead recovery was good (mean recovery 106%, range 96-111%) (figure 4.5c). In order to determine if this could be used as an internal standard 10µl of the stock solution was added to 240µl of a sample containing MPs stained with AnV only. As Figures 4.6a and b indicate, the bright fluorescence of the beads required the FL1 voltage to be decreased, making the MP population appear compressed (Figure 4.6b). Thus it was decided that the most appropriate method for MP enumeration was to use the 3µm latex counting beads with FSC threshold applied (as in Figure 4.4c) to determine the flow rate of the FACS machine before and after each patient sample was run and thus calculate a mean value.

While fluoresbright beads could not be used as an internal standard due to their high signal when compared to MPs, in order to determine whether the use of an FL1 threshold improved the recovery of AnV positive MPs, all samples were run using the previously published instrument settings (Brogan et al., 2004a) and then again using an

FL1 threshold. Figure 4.7 demonstrates the total Annexin V MP results obtained in 20 healthy child controls.

172

Figure 4-5: Detecting Flouresbrite beads – recovery of beads alone

Figure 4.5a-c demonstrates that when FC instrument settings were optimised to include fluoresbrite beads (figure 4.5a) and the noise (circled) was excluded using an FL1 threshold (figure 4.5b) bead recovery was good (mean value 106%, range 96-111%) (figure 4.5c).

173

Figure 4-6:Fluoresbrite beads in relation to MPs

Figure 4.6: When beads were used as an internal standard in the presence of Annexin V positive

MPs, if instrument settings were optimised to the MPs the beads were off the FL1 scale (Figure

4.6a, arrowed). In order to recover beads at the same time as MPs FL1 voltage was reduced

(Figure 4.6b, beads circled). However this shifted the MP population considerably.

174

Figure 4-7:Using an FL1 threshold for MP enumeration

Figure 4.7: When an FL1 threshold was used to remove background noise from the system, (figures

4.7a and b) the calculated number of Annexin V positive MPs per ml of plasma was increased

(figure 4.7c). Figure 4.7c shows the recovery of AnnexinV MPs in 20 healthy child controls. When an FL1 threshold was used, the total number of MPs recovered was significantly higher

175

It is apparent that the use of an FL1 threshold increases the sensitivity of the technique by removing the majority of the noise within the system. It does however prevent analysis of AnV negative events. While phosphatidylserine (PS) is widely thought to be found on all MPs it has been suggested that there are large populations within the circulation which do not express PS (Horstman et al., 2004). Therefore for the purposes of this thesis it was decided to enumerate MPs using a modification of the settings described by Brogan et al, with the addition of an FL1 threshold as described above.

However data was also collected using the previously published instrument settings to allow future study of the AnV negative population (Brogan et al., 2004a). Use of fluorescence thresholds have been previously described by others for detection of MPs

(Jimenez et al., 2001).

It is important to note that these instrument settings are entirely dependent on the particular flow cytometer (in this case a FACscalibur, BD) used for assaying these MPs.

It is possible that the use of a different machine would not need all of the rigorous steps taken above to reduce background noise. However, when MP samples were run on a second flow cytometer (FACSscan, BD) to investigate this, it was found that the use of an FL1 threshold still increased the number of AnV positive events recovered and thus hilighted that the steps taken to optimize this assay are of important consideration when performing rare event analysis under any circumstances.

4.3.5 The effect of plasma freeze-thaw on MP quantification

As mentioned previously the removal of platelets from plasma is an important step in the protocol as all samples were frozen. While desirable (and possible with this

176

technique) to get same day results, the staining process is time consuming, and storage of frozen plasma for analysis in batches is more efficient.

In order to determine if freeze-thawing altered the MPs obtained from patients a comparison between fresh and frozen platelet poor plasma was performed. As figure

4.8 demonstrates there was no difference between fresh and frozen total MP number based on experiments in 5 samples. MPs were stained and enumerated as described above and in chapter 2.5.3, but only with AnV. This finding is in keeping with other groups who found no difference due to freeze thaw (Enjeti et al., 2007). There is one study that indicates the freeze thaw process may be detrimental to EMPs (Jy et al.,

2004a), however in order to establish this finding with certainty a much greater sample size, including different MP generation conditions, different MP populations and the concentrations frozen must be considered. Time restraints meant that this investigation could not be performed to such a detailed level.

4.3.6 Reproducibility of the technique from different plasma volumes

Platelet poor plasma from 1 healthy adult control was frozen in 3 separate aliquots, each with a different volume (range 150-500µl). As Figure 4.9 demonstrates each aliquot gave similar mean results for total MP enumeration (based on 5 sample tubes analysed using flow cytometry as described above), verifying acceptable reproducibility of the technique of different volumes of frozen plasma.

177

Figure 4-8: The effect of freeze thaw on MP enumeration

Figure 4.8: MPs were enumerated from 5 individual patients, on freshly isolated platelet poor plasma and platelet poor plasma that had been frozen for 24hrs at -80oC. Freeze thawing of platelet poor plasma gave no significant differences in absolute Annexin V positive MP number

(n=5).

178

Figure 4-9: Reproducibility of MP enumeration

Figure 4.9: Platelet poor plasma was obtained from one healthy individual. 3 plasma aliquots of different sizes (range 150-500µl) were frozen down and the MPs enumerated, based on 5 individual flow cytometric sample readings. There was no significant difference between the samples for total

AnnexinV positive MP events.

179

4.4 Conclusions of methodological experiments

What is apparent from this series of experiments is that detection of MPs is fraught with problems relating to the sensitivity of FC, as it involves using the machine at the limits of its detection. However, MP populations can be detected and phenotyped with careful and systematic optimisation of FC instrument settings.

To determine whether the MP populations of interest could be detected by FC, MP controls from different parent populations as positive controls were generated. For all the cell surface markers used in this study MPs expressing them were able to be generated in vitro (staining profiles are in Appendix 4), thus if they were present in patients, then their expression should be detectable. However, the MPs thus generated may not be representative of pathophysiological conditions and thus may not be entirely representative results of the MPs found within the peripheral blood in health and disease. While this is a point for consideration it does not detract from the validity of the FC work as FC is the only technique with which a multiparametric approach can be taken to enumerate subtypes of MPs and has been used widely in different states of health and disease.

For the purposes of this thesis MP samples were run twice, once using an FL1 threshold to reduce background noise, (the data which is presented in the next section of chapter) and once using previously published instrument settings (Brogan et al., 2004a) so that they could be compared with previously published data from within our group (this data is not included within this thesis).

180

4.5 Evaluation of cellular MPs in vasculitis

4.5.1 Patients and methods

4.5.1.1 Patients

40 children with primary systemic vasculitis were studied. These patients were the same individuals described in Chapter 3, allowing MPs to be compared to CECs. These included 17 males and 23 females of median age 11.1 years (range 1 -16.6 years). The vasculitis type (using the Chapel Hill Consensus Criteria) (Jennette et al., 1994) was polyarteritis nodosa (PAN) n=18, Wegener’s granulomatosis (WG) n=8, microscopic polyangiitis n= 1, Kawasaki’s disease = 5, Takayasu’s disease (TD) = 2, Beçhets disease = 1 and unclassified vasculitis n=5. Disease activity was again assessed using a modified paediatric BVAS (Appendix 2). 20 healthy child controls (8 males, 12 females, median age 8.5 years, range 1.5-15.8), 10 febrile controls (6 males, 4 females, median age 6.8 years, range 1.5-13.5) and 20 healthy adult controls were also recruited

(2 males, 18 females, age range 24-40 years). As with the vasculitis patients, the control individuals were the same as those studied in Chapter 3.

4.5.1.2 Methods

MPs were analysed as described above and in Chapter 2.5.3. Depending on the volume of platelet poor plasma obtained samples were either analysed using the 22 tube FC plan or the 12 tube FC plan shown in Appendix 3. As platelet and endothelial microparticles have particular relevance in the vasculitides, these were prioritised in the 12 tube plan.

Other markers are less well defined in MP populations and thus these were analysed in a smaller cohort, where plasma sample size was not a limiting factor. Adult healthy controls were only analysed on the 12 tube plan due to time constraints.

181

4.5.1.3 Statistics

Median MP counts were compared between groups using the Kruskall Wallis test to check for overall differences. Individual Mann Whitney U tests for non parametric data were performed to compare between specific groups. The Wilcoxon signed rank test was performed to compare medians of paired samples. Spearman Rank correlation coefficiants were performed to assess correlation between MPs and other markers of disease activity, including CECs. In all data graphs patient classification was denoted using colour coding as indicated below:

4.5.2 Results

4.5.2.1 Total Annexin V positive MPs

These data are presented in Figure 4.10a-c. Kruskall Wallis testing indicated there was an overall significant difference between groups (p=0.005). Total AnnexinV positive

MPs were significantly higher in 30 patients with active vasculitis (median value 0.98 x

106 MPs/ml) compared to 20 healthy child controls (median value 0.65 x 106 MPs/ml, p=0.03) and 20 healthy adult controls (median value 0.30 x 106 MPs/ml, p<0.001), although there was no significant difference compared 10 febrile controls (median value

0.70 x 106 MPs/ml p=0.09). There were no significant differences between 27 patients with inactive vasculitis (median value 0.65 x 106 MPs/ml) and febrile controls or healthy child controls. Healthy adults also had significantly lower MPs than children

182

with inactive vasculitis (p=0.007) and child controls (p=0.04) (Figure 4.10a). When paired data between active and inactive vasculitis was excluded (the first observation made in individual patients was used in this analysis) there was no significant difference between 30 patients with active disease and 11 patients in remission (median value 0.86 x 106 MPs/ml, data not shown). In 17 individual patients there was no significant difference between total MP number during active disease and remission (median time to remission 4.5 months) (Figure 4.10b). Of these 17 patients, 13 had frequent and detailed follow up for up to 18 months from pre-treatment at disease onset (10 patients, solid line) or from flares of their disease (3 patients, dotted line). At median followup of

3 and 12 months AnV+ MPs were not significantly different from pretreatment levels

(Figure 4.10c).

4.5.2.2 MPs of endothelial origin

These data are presented in Figures 4.11-4.14. Kruskall Wallis testing indicated that there was an overall significant difference between groups for MPs expressing CD144

(p=0.002). No other endothelial marker showed overall significance, CD105 (p=0.1) E- selectin (p=0.2), ICAM-1 (p=0.06) (Figure 4.11-4.14a). Direct comparisons between individual patient and control groups will be presented below.

CD144+ EMPs

MPs expressing the constitutive marker CD144, were significantly higher in 30 patients with active vasculitis (median value 0.35 x 106 MPs/ml) compared to 20 healthy child controls (median value 0.09 x 106 MPs/ml, p=0.02) and 20 healthy adult controls

(median value 0.04 x 106 MPs/ml, p=0.007), but were not significantly different when compared with 10 febrile controls (median value 0.15 x 106 MPs/ml).

183

Figure 4-10: Total AnV MPs

Figure 4.10a: Kruskall Wallis testing indicated there was an overall significant difference between groups (p=0.005). Total AnnexinV positive MPs were significantly higher in 30 patients with active vasculitis compared to healthy child controls (n=20, p=0.03) and healthy adult controls (n=20, p<0.001), there were no significant differences compared 10 febrile controls (p=0.09). There were no significant differences between 27 patients with inactive vasculitis disease and febrile controls or healthy child controls. Healthy adults also had significantly lower MPs than patients with inactive vasculitis

(p=0.007) and child controls (p=0.04). Figure 4.10b:In 17 individual patients paired analysis revealed there was no significant difference between total AnV+ MPs during active disease and remission

(median time to remission 4.5 months). Figure 4.10c: 13 individual patients were followed from disease onset (n=10, solid line) or flare (n=3, dotted line). At median followup of 3 and 12 months from disease onset AnV+ MPs were not significantly different from pretreatment levels. Rx = treatment.

184

There were no significant differences in CD144+ MPs between 27 patients with inactive disease (median value 0.03 x 106 MPs/ml) and these control groups (Figure 4.11a).

When paired data between active and inactive vasculitis were excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 30 patients with active disease and 11 patients in remission (median value 0.23 x 106 MPs/ml, data not shown). In 17 individual patients however, paired analysis revealed CD144 MPs were significantly higher during active disease compared to remission (p=0.001, median time to remission 4.5 months, Figure

4.11b). Of these 17 patients, 13 were followed up for up to 18 months from initiation of treatment at disease onset (10 patients, solid line) or from flare (3 patients, dotted line).

At median followup of 3 and 12 months from disease onset there was no significant difference from pretreatment levels of CD144+ MPs (Figure 4.11c).

CD105+ EMPs

MPs expressing CD105 (endoglin), were not significantly different in the same 30 patients with active disease (median value 0.11 x 106MPs/ml) compared to 20 healthy child controls (median value 0.11 x 106MPs/ml), 10 febrile child controls (median value

0.13 x 106MPs/ml or 20 adult controls (median value 0.03 x106MPs/ml). However, MPs expressing CD105 were significantly higher in 27 patients with inactive disease (median value 0.22 x106MPs/ml) compared to healthy child (p=0.038) and adult (p=0.028) controls (Figure 4.12a). When paired data between active and inactive vasculitis were excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 30 patients with active disease and 11 patients in remission (median value 0.20 x 106 MPs/ml, p=0.09, data not shown).

185

Figure 4-11: MPs of endothelial origin – CD144

Figure 4.11a: Kruskall Wallis testing indicated there was an overall significant difference between groups (p=0.002). CD144+ MPs were significantly higher in 30 patients with active disease compared to healthy child controls (n=20, p=0.02) and healthy adult controls (n=20, p=0.007), but not febrile controls (n=10). Figure 4.11b: In 17 individual patients, paired analysis revealed CD144 MPs were significantly higher during active disease compared to when patients were in remission (p=0.001, median time to remission 4.5months). Figure 4.11c: 13 individual patients were followed from disease onset

(n=10, solid lines) or flare(n=3, dotted lines). At median followup of 3 and 12 months from disease onset

(n=10) CD144+ MPs were not significantly different from pretreatment levels. Rx = treatment.

186

Figure 4-12: MPs of endothelial origin CD105

Figure 4.12a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.1). However when individual analysis was performed although CD105+ MPs were not significantly different between 30 patients with active vasculitis and control groups, they were significantly higher in 27 patients with inactive vasculitis compared to healthy child (n=20, p=0.04) and adult controls (n=20, p=0.03) although not febrile controls (n=10). Figure 4.12b: In 17 individual patients, paired analysis revealed CD105+ MPs were not significantly different during active disease compared to when patients were in remission (p=0.3, median time to remission 4.5months). Figure

4.12c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) CD105+ MPs were not significantly different from pretreatment levels. Rx = treatment.

187

In 17 individual patients paired analysis revealed no significant difference between active and inactive disease (p=0.3, median time to remission 4.5 months, Figure 4.12b).

13 of these patients were followed up from initiation of treatment at disease onset

(n=10, solid line) or from flare (n=3, dotted line). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of CD105+ MPs (Figure 4.12c).

E-selectin+ EMPs

MPs expressing the inducible activation marker E-selectin were significantly higher in

30 patients with active disease (median value 0.075 x 106MPs/ml) compared to 20 healthy adult controls (p=0.015, median value 0.004 x 106MPs/ml), although not child controls, febrile (median value 0.07 x 106MPs/ml) or healthy (median value 0.04 x

106MPs/ml). There were no significant differences between patients with inactive vasculitis and these control groups (Figure 4.13a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 30 patients with active disease and 11 patients in remission (median value 0.049 x 106 MPs/ml, data not shown). There was also no significant differences between active and inactive disease in 17 individual patients where paired samples were available for analysis (Figure

4.13b). 13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines).

At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of E-selectin+ MPs (Figure 4.13c).

188

Figure 4-13: MPs of endothelial origin – E-selectin

Figure 4.13a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.1). However when individual analysis was performed E-selectin + MPs were significantly higher in 30 patients with active disease compared to 20 healthy adult controls (p=0.015), although not child controls, febrile or healthy. Figure 4.13b: In 17 individual patients paired analysis revealed there was no significant differences between active and inactive disease (p=0.17, median time to remission 6 months). Figure 4.13c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) E-selectin+

MPs were not significantly different from pretreatment levels. Rx = treatment.

189

ICAM-1+ EMPS

There were no significant differences in MPs expressing the inducible activation marker

ICAM-1 between 26 patients with active vasculitis (median value 0.05 x 106 MPs/ml) compared to 20 healthy child controls (median value 0.02 x 106 MPs/ml) or 3 febrile controls(median value 0.16 x 106 MPs/ml) . There were no significant differences between 18 patients with inactive vasculitis (median value 0.06 x 106 MPs/ml) and these controls groups either (Figure 4.14a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 26 patients with active disease and 4 patients in remission (median value 0.06 x 106 MPs/ml). Paired analysis in 14 individual patients showed no significant differences between and inactive disease (Figure 4.14b). 13 individuals were followed from pre-treatment at disease onset (n=10, solid lines) or diseaseflare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of ICAM-1+ MPs (Figure 4.14c).

4.5.2.3 MPs of platelet origin

These data are presented in figures 4.15-16. Kruskall Wallis testing indicated that there were no overall significant differences between groups for MPs expressing CD42a

(p=0.7) or P-selectin (p=0.4). Direct comparisons between individual patient and control groups will be presented below.

CD42a+ PMPs

When individual analysis was performed between groups there were no significant differences in the number of CD42+ MPs between 30 patients with active vasculitis

190

Figure 4-14: MPs of endothelial origin – ICAM-1

Figure 4.14a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.06). When individual analysis was performed ICAM-1+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were no significant differences between inactive vasculits and these control groups.Figure 4.14b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease (p=0.3, median time to remission 6 months). Figure 4.14c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and

12 months from disease onset (n=10) ICAM-1+ MPs were not significantly different from pretreatment levels. Rx = treatment.

191

(median value 0.12 x 106 MPs/ml) compared to 10 febrile controls (median value 0.04 x

106 MPs/ml), 20 healthy child controls (median value 0.16 x 106 MPs/ml) or 20 healthy adult controls (median value 0.15 x 106 MPs/ml). There were no significant differences between patients with inactive vasculitis (median value 0.18 x 106 MPs/ml) and these controls groups (Figure 4.15a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 30 patients with active disease and

11 patients in remission (median value 0.26 x 106 MPs/ml, data not shown). Paired analysis in 17 individual patients showed no significant differences between and inactive disease (Figure 4.15b). 13 of these 17 individuals were followed up for up to

18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset

(n=10) there was no significant difference from pretreatment levels of CD42a+ MPs

(Figure 4.15c).

P-selectin+ PMPs

There was also no significant differences for P-selectin+ MPs between 26 patients with active vasculitis (median value 0 x 106 MPs/ml) compared to child controls (3 febrile, median value 0 x 106 MPs/ml and 20 healthy median value 0 x 106 MPs/ml). There were no significant differences between 18 patients with inactive disease (median value

0.1 x 106 MPs/ml) and these control groups (Figure 4.16a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between

26 patients with active disease and 4 patients in remission (median value 0.01 x 106

MPs/ml, data not shown). Paired analysis in 14 individual patients indicated P-selectin+

192

Figure 4-15: MPs of platelet origin – CD42a

Figure 4.15a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.06). When individual analysis was performed ICAM-1+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were no significant differences between inactive vasculitis and these control groups.Figure 4.15b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease (p=0.3, median time to remission 6 months). Figure 4.15c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and

12 months from disease onset (n=10) CD42a+ MPs were not significantly different from pretreatment levels. Rx = treatment.

193

PMPs were signficiantly higher during remission than during active disease (p=0.01,

Figure 4.16b). 13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of P-selectin+ MPs (Figure 4.15c).

4.5.2.4 MPs of leukocyte origins

These data are presented in Figures 4.17-4.23. Kruskall Wallis testing indicated that there was no overall significant differences between groups for MPs expressing the neutrophil markers CD15 (p=0.3), CD11b (p=0.2), L-selectin (p=0.4), or CD66b

(p=0.8), the monocyte markers TF (p=0.2) and CD14 (p=0.2) or the T cell marker CD3

(p=0.1). Comparison between individual patient and control groups are presented below.

CD15+ NMPs

When individual analysis between specific groups was performed CD15+ MPs were not significantly different between 26 patients with active vasculitis (median value 0.025 x

106 MPs/ml) and child controls (3 febrile, median value 0.03 x 106 MPs/ml and 20 healthy, median value 0.04 x 106 MPs/ml). There was no significant difference between

18 patients with inactive vasculitis (median value 0.02 x 106 MPs/ml) and these control groups (Figure 4.17a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 26 patients with active disease and 4 patients in remission (median value 0.045 x 106 MPs/ml, data not shown). Paired analysis in 14 individuals showed no significant difference between active and inactive

194

Figure 4-16 MPs of platelet origin – P-selectin

Figure 4.16a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.06). When individual analysis was performed P-selectin+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls.

There were no significant differences between inactive vasculits and these control groups.Figure 4.16b:

In 14 individual patients paired analysis revealed P-selectin+ MPs were significantly higher during remission than active disease (p=0.01, median time to remission 6 months). Figure 4.16c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) P-selectin+ MPs were not significantly different from pretreatment levels. Rx = treatment.

195

Figure 4-17: MPs of neutrophil origin – CD15

Figure 4.17a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.3). When individual analysis was performed CD15+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculits and these control groups (AAV patients – red dots). Figure 4.17b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease, subgroup analysis of 7 patients with AAV (red lines) also showed no significant difference Figure 4.17c:: 13 individual patients were followed from disease onset

(n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) CD15+ MPs were not significantly different from pretreatment levels, analysis of 4 patient with AAV(red lines) also demonstrated no significant difference at these timepoints. Rx = treatment.

196

disease (Figure 4.17b). Subgroup paired analysis of 7 individuals with AAV also disease (Figure 4.17b). Subgroup paired analysis of 7 individuals with AAV also indicated no significant difference between active and inactive disease. 13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of

3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of CD15+ MPs (Figure 4.15c), there was also no differences observed in the AAV subgroup (Figure 4.17c, AAV – red lines, n=4).

CD11b+ NMPs

CD11b+ MPs were not significantly different between 26 patients with active vasculitis

(median value 0.27 x 106 MPs/ml) and child controls (3 febrile, median value 0.37 x 106

MPs/ml and 20 healthy, median value 0.18 x 106 MPs/ml). There was no significant difference between 18 patients with inactive vasculitis (median value 0.15 x 106

MPs/ml) and these control groups (Figure 4.18a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 26 patients with active disease and 4 patients in remission (median value 0.26 x 106 MPs/ml, data not shown). Paired analysis in 14 individuals showed no significant difference between active and inactive disease (Figure 4.18b). Subgroup paired analysis of 7 individuals with AAV also indicated no significant difference between active and inactive disease.

13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of CD11b+ MPs (Figure 4.15c), there was also no differences observed in the AAV subgroup (Figure 4.18c, AAV – red lines, n=4). 197

Figure 4-18: MPs of neutrophil origin – CD11b

Figure 4.18a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.2). When individual analysis was performed CD11b+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculitis and these control groups (AAV patients – red dots).. Figure 4.18b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease, subgroup analysis of 7 patients with AAV (red lines) also showed no significant difference Figure 4.18c: : 13 individual patients were followed from disease onset

(n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) CD11b+ MPs were not significantly different from pretreatment levels subgroup analysis of

4 patients with AAV (red lines) also demonstrated no significant difference at these timepoints.. Rx = treatment.

198

L-selectin+ NMPs

L-selectin+ MPs were not significantly different between 26 patients with active vasculitis (median value 0.05 x 106 MPs/ml) and child controls (3 febrile, median value

0.08 x 106 MPs/ml and 20 healthy, median value 0.02 x 106 MPs/ml). There was no significant difference between 18 patients with inactive vasculitis (median value 0.025 x

106 MPs/ml) and these control groups (Figure 4.19a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) there was no significant difference between 26 patients with active disease and 4 patients in remission (median value 0.035 x 106 MPs/ml, data not shown). Paired analysis in 14 individuals showed no significant difference between active and inactive disease (Figure 4.19b). Subgroup paired analysis of 7 individuals with AAV also indicated no significant difference between active and inactive disease.

13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of L-selectin+ MPs (Figure 4.19c), there was also no differences observed in the AAV subgroup (Figure 4.19c, AAV – red lines, n=4).

CD66b+ NMPs

CD66b+ MPs were not significantly different between 26 patients with active vasculitis

(median value 0.005 x 106 MPs/ml) and child controls (3 febrile, median value 0 x 106

MPs/ml and 20 healthy, median value 0 x 106 MPs/ml). There was no significant difference between 18 patients with inactive vasculitis (median value 0 x 106 MPs/ml) and these control groups (Figure 4.20a). When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used

199

Figure 4-19: MPs of neutrophil origin – L-selectin

Figure 4.19a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.4). When individual analysis was performed L-selectin+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculits and these control groups (AAV patients – red dots).. Figure 4.19b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease, subgroup analysis of 7 patients with AAV also showed no significant difference (red lines). Figure 4.19c: 13 individual patients were followed from disease onset

(n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) L-selectin+ MPs were not significantly different from pretreatment levels, subgroup analysis of 4 patient with AAV (red lines) also demonstrated no significant difference at these timepoints..

Rx = treatment.

200

in this analysis) there was no significant difference between 26 patients with active disease and 4 patients in remission (median value 0.005 x 106 MPs/ml, data not shown).

Paired analysis in 14 individuals showed no significant difference between active and inactive disease (Figure 4.20b). Subgroup paired analysis of 7 individuals with AAV also indicated no significant difference between active and inactive disease (although

6/7 patients did have higher CD66b+ MPs during active disease).

13 of these 17 individuals were followed up for up to 18 months from disease onset and initiation of treatment (n=10, solid lines) or disease flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) there was no significant difference from pretreatment levels of L-selectin+ MPs (Figure 4.19c), there was also no differences observed in the AAV subgroup (Figure 4.19c, AAV – red lines, n=4).

When 10 individuals were followed from disease onset, there were no significant differences from pre-treatment to median followup of 3 and 12 months, there was also no differences observed in the AAV subgroup (Figure 4.20c, AAV – red lines, n=4).

TF+ MMPS

TF+ MPs were only observed in a small number of individuals (2 patients with active vasculitis, 5 in remission, 1 febrile control and 3 healthy controls). Hence no significant differences were observed between 26 patients with active vasculitis, or 18 patients with inactive disease and child controls (3 febrile, 20 healthy, median value for all groups 0 x

106 MPs/ml, Figure 4.21a). Paired analysis in 14 individuals showed no significant differences between active and inactive disease (Figure 4.21b) and when 13 individuals were followed from disease onset prior to treatment (n=10, solid lines) or disease flare

(n=3, dotted lines), only 4 individuals had any detectable TF+ MPs (Figure 4.21c).

201

Figure 4-20:MPs of neutrophil origin – CD66b

Figure 4.20a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.8). When individual analysis was performed CD66b+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculits and these control groups (AAV- red dots).

Figure 4.20b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease, subgroup analysis of 7 patients AAV (red lines) also showed no significant difference (although 6/7 patients had a increased CD66b+ MPs during active disease) Figure

4.20c: : 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and 12 months from disease onset (n=10) CD66b+ MPs were not significantly different from pretreatment levels, subgroup analysis of 4 patients with AAV also revealed no differences at these timepoints. Rx = treatment.

202

Figure 4-21: MPs of monocyte origin – TF

Figure 4.21a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.2). When individual analysis was performed TF+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculits and these control groups. The majority fo individuals had no detectable TF+ MPs. Figure 4.21b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease Figure 4.21c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). TF+

MPs were only detected in 4 individuals. Rx = treatment.

CD14+ MMPs 203

MPs expressing CD14, were significantly higher in 30 patients with active vasculitis

(median value 0.51 x 106 MPs/ml) compared to 20 healthy child controls (median value

0.24 x 106 MPs/ml, p=0.037) or 20 healthy adult controls (median value 0.17 x 106

MPs/ml, p=0.02), although not 10 febrile controls (median value 0.25 x 106 MPs/ml).

There were no significant differences in CD14+ MPs between 27 patients with inactive disease (median value 0.23 x 106 MPs/ml) and these control groups (Figure 4.22a).

In 17 individual patients paired analysis also revealed no significant difference between active vasculitis and remission (Figure 4.22b) Of these 17 patients, 13 were followed up for up to 18 months, from disease onset (n=10, solid lines) or disease flare (n=3, dotted lines) at median followup of 3 and 12 months from disease onset, there was no signficant difference from pretreatment levels of CD14+ MPs (Figure 4.22c).

CD3+ LMPs

CD3+ MPs were only observed in a small number of individuals (3 patients in remission and 1 healthy child controls), hence no significant differences were observed between 26 patients with active vasculitis, or 18 patients with inactive disease and child controls (3 febrile, 20 healthy, median value for all groups 0 x 106 MPs/ml, Figure

4.23a). Paired analysis, demonstrated only 2/14 individuals had any change in their

CD3+ MPs (Figure 4.23b) and when 13 individuals were followed from disease onset

204

Figure 4-22: MPs of monocyte origin – CD14

Figure 4.22a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.2). However when individual analysis was performed CD14+ MPs were significantly higher in 30 patients with active disease compared to 20 healthy child controls (p=0.038) or 20 adult controls

(p=0.02), although not 3 febrile controls. There no significant differences between inactive vasculits and these control groups. Figure 4.22b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease. Figure 4.22c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines). At median followup of 3 and

12 months from disease onset (n=10) CD14a+ MPs were not significantly different from pretreatment levels. Rx = treatment.

205

Figure 4-23: MPs of lymphocyte origin – CD3

Figure 4.23a: Kruskall Wallis testing indicated that there was no overall significant difference between groups (p=0.1). When individual analysis was performed CD3+ MPs were not significantly different between 26 patients with active disease and 20 healthy child controls or 3 febrile controls. There were also no significant differences between inactive vasculits and these control groups. The majority of individuals had no detectable CD3+ MPs. Figure 4.23b: In 14 individual patients paired analysis revealed there was no significant differences between active and inactive disease Figure 4.21c: 13 individual patients were followed from disease onset (n=10, solid lines) or flare (n=3, dotted lines).

CD3+ MPs were only detected in 4 individuals. Rx = treatment.

206

(n=10, solid lines) or flare (n=3, dotted lines), only 4 individuals had any detectable

CD3+ MPs (Figure 4.23c).

4.5.3 Correlation between MPs and disease activity

Spearman Rank correlation coefficiant analysis was performed between all the phenotypes of MPs assessed and BVAS, CRP, ESR, CEC, white cell count, neutrophil count, platelet count and serum albumin levels, Table 4.1 summarises the results – significant correlations are highlighted in blue. Table 4.2 shows the corrected P-values obtained from a post-hoc Holm-Bonferroni correction. The significant correlations observed on individual tests did not withstand the rigours of multiple comparison correction and thus no significant correlations were observed between different MP phenotypes and other markers of disease activity.

When 16 patients were followed up in detail over time from first initiation of treatment a number had periods of disease flare. While different populations of MPs may not be significantly different from healthy controls they did change over time in individual patients. Figures 4.24 - 4.26 highlights some of the changes seen in MP populations over time from disease onset and initiation of treatment in the same 3 patients that were described earlier in this thesis in the context of CECs. In the two patients who flared

(Figure 4.24 – systemic flare and figure 4.25, localized granulomatous flare) E-selectin+

MPs rose before the flare and prior to CD144 becoming elevated. In the same patients,

CD14 MPs also rose prior to flare and were followed by CD15, suggesting that different

MPs may reflect different stages of disease activity and underlying pathology. In the patient who did not have a flare, but a UTI, there was no rise in EMPs or NMPs and

MMPS. These data are very preliminary but warrants further study in a large cohort of

207

patients. Further examples of changes in MPs and CECs in individual patients are given in Appendix 6.

Table 4-1: Spearman Rank correlation coefficiant analysis between different MP phenotypes and other markers of disease activity

MP Neutrophil Platelet CEC BVAS ESR CRP WCC Albumin population count count count

Total AnV r = -0.015 r = -0.072 r = 0.005 r = 0.065 r = 0.065 r = 0.145 r = -0.003 r = 0.081 MP p = 0.89 p = 0.52 p = 0.97 p = 0.56 p = 0.56 p = 0.19 p = 0.98 p = 0.43 n = 88 n = 83 n = 88 n = 85 n = 85 n = 82 n = 82 n = 81

CD144 r = 0.038 r = -0.001 r = 0.040 r = 0.084 r = -0.006 r = 0.155 r = -0.011 r = 0.232 p = 0.73 p = 0.99 p = 0.71 p = 0.44 p = 0.96 p = 0.17 p = 0.92 p = 0.037 n = 88 n = 83 n = 86 n = 85 n = 85 n = 82 n = 81 n = 81 CD105 r = -0.011 r = -0.21 r =- 0.132 r = -0.111 r = -0.092 r =- 0.026 r = 0.001 r = 0.006 p = 0.3 p = 0.055 p = 0.23 p = 0.31 p = 0.40 p = 0.82 p = 0.99 p = 0.96 n = 88 n = 83 n = 86 n = 85 n = 85 n = 82 n = 81 n = 81 E-selectin r = 0.121 r = 0.070 r = 0.02 r = -0.001 r = -0.009 r = 0.155 r = -0.051 r = 0.092 p = 0.26 p = 0.53 p = 0.84 p = 0.99 p = 0.94 p = 0.17 p = 0.65 p = 0.41 n = 88 n = 83 n = 86 n = 85 n = 85 n = 82 n = 81 n = 81 ICAM-1 r = -0.138 r = 0.044 r = 0.072 r = 0.077 r = 0.002 r = 0.122 r = -0.060 r = -0.141 p = 0.24 p = 0.72 p = 0.54 p = 0.52 p = 0.99 p = 0.31 p = 0.63 p = 0.25 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 CD42a r = -0.235 r = -0.284 r = -0.245 r = -0.087 r = -0.158 r = 0.072 r = 0.108 r = -0.094 p = 0.029 p = 0.010 p = 0.024 p = 0.43 p = 0.15 p = 0.52 p = 0.34 p = 0.41 n = 87 n = 82 n = 85 n = 86 n = 84 n = 81 n = 80 n = 80 P-selectin r = -0.136 r = -0.063 r = -0.082 r = -0.29 r = -0.25 r = -0.030 r = 0.181 r = 0.028 p = 0.25 p = 0.60 p = 0.50 p = 0.013 p = 0.033 p = 0.81 p = 0.63 p = 0.82 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 CD15 r = 0.073 r = 0.061 r = 0.048 r=-0.217 r = -0.110 r = 0.008 r = -0.075 r = 0.010 p = 0.53 p = 0.61 p = 0.69 p = 0.07 p = 0.354 p = 0.94 p = 0.54 p = 0.93 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 CD11b r = 0.071 r = 0.145 r = 0.176 r= 0.097 r = -0.004 r = -0.026 r = -0.150 r = -0.044 p = 0.54 p = 0.22 p = 0.14 p = 0.42 p = 0.98 p = 0.83 p = 0.22 p = 0.72 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 L-selectin r = 0.078 r = -0.040 r = 0.016 r= 0.148 r = 0.019 r = -0.038 r = -0.174 r = 0.058 p = 0.51 p = 0.74 p = 0.89 p = 0.21 p = 0.88 p = 0.76 p = 0.15 p = 0.63 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 CD66b r = 0.157 r = 0.094 r = 0.077 r = 0.233 r = 0.262 r = -0.042 r = -0.212 r = 0.087 p = 0.18 p = 0.43 p = 0.52 p = 0.047 p = 0.025 p = 0.73 p = 0.081 p = 0.48 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 TF r = 0.057 r = -0.023 r = -0.082 r = -0.341 r = -0.247 r = 0.116 r = 0.042 r = 0.031 p = 0.63 p = 0.84 p = 0.49 p = 0.003 p = 0.035 p = 0.34 p = 0.73 p = 0.80 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 CD14 r = -0.003 r = -0.166 r =- 0.096 r = 0.005 r = -0.049 r = 0.161 r = 0.076 r = 0.111 p = 0.98 p = 0.13 p = 0.38 p = 0.96 p = 0.66 p = 0.15 p = 0.50 p = 0.33 n = 88 n = 83 n = 86 n = 85 n = 85 n = 82 n = 81 n = 81 CD3 r = -0.166 r = -0.150 r = -0.049 r = -0.114 r = -0.182 r = -0.061 r = 0.106 r = -0.241 p = 0.16 p = 0.21 p = 0.68 p = 0.34 p = 0.12 p = 0.62 p = 0.39 p = 0.048 n = 75 n = 72 n = 73 n = 73 n = 73 n = 70 n = 69 n = 68 Significant correlations are highlighted in blue 208

Table 4-2: Holm-Bonferroni correction for multiple comparisons

Significant test P value Adjusted critical value (Holm- Bonferroni) TF vs WCC 0.003 0.0004464 CD42a vs ESR 0.01 0.0004505 P-selectin vs WCC 0.013 0.0004545 CD42a vs CRP 0.024 0.0004587 CD66b vs neutrophil count 0.025 0.000463 CD42a vs BVAS 0.029 0.0004673 P-selectin vs neutrophil count 0.033 0.0004717 TF vs neutrophil count 0.035 0.0004762 CD144 vs CECs 0.037 0.0004808 CD66b vs WCC 0.047 0.0004854 CD3 vs WCC 0.048 0.0004902

4.5.4 MPs in different vasculitic syndromes

When the vasculitis cohort was classified into the different, there were no obvious differences in expression of the different MP populations in active or inactive disease between diseases as had been for CECs. However, Figure 4.27 demonstrates the expression of EMPs (CD144+, CD105+, E-selectin+, ICAM-1+) during active disease.

Interestingly, whereas CECs were all low in patients with KD (Figure 3.22), their EMPs were high.

209

Figure 4-24: Patient with WG - MPs and disease activity – predictive of flare?

Figure 4.24a indicates the changes in BVAS anc CEC number over time from intitiation of treatment at disease onset in a 12 year old female with WG (ANCA+). Figure 4.24b indicates the changes in endothelial MPs (CD144+ and E-selectin+). Figure 2.24c indicates changes in monocytes and neutrophil MPS (CD14+ and CD15+ respectively). Notably in this individual, E-selectin+ MPs mirrored changes in CECs, whilst CD144+ MPs remained high despite clinical remission.

210

Figure 4-25: Patient with WG - MPs and disease activity – localised flare

Figure 4.25a indicates the changes in BVAS and CEC number over time from intitiation of treatment at disease onset in an 11 year old female with WG (ANCA+) who suffered a granulomatous orbital flare.

Figure 4.25b indicates the changes in endothelial MPs (CD144+ and E-selectin+). Figure 2.25c indicates changes in monocytes and neutrophil MPS (CD14+ and CD15+ respectively). Notably in this individual, E-selectin+ MPs, CD14+ MPs and CD15+ MPs rose prior to the clinical flare, whilst

CD144MPs and CEC rose at the time of clinical detection of flare.

211

Figure 4-26: Patient with PAN – MPs and disease activity

Figure 4.26a indicates the changes in BVAS and CEC number over time from intitiation of treatment at disease onset in an 12 year old female with PAN. Figure 4.26b indicates the changes in endothelial MPs

(CD144+ and E-selectin+). Figure 2.25c indicates changes in monocytes and neutrophil MPS (CD14+ and CD15+ respectively). Notably in this individual, while CD14+ and CD15+ were elevated prior to treatment and fell with treatment, CD144+ and E-selectin MPs rose after initiation of treatment, after this CD144+ closely tracked changes in CECs. There was no elevation in any MP during the patients

UTI.

212

Figure 4-27: EMPs in different vasculitis syndromes

Figure 4.27a-d: When the patient cohort was divided into the different disease classifications, although very preliminary, patients with KD had higher EMPs than other diseases. This is in striking contrast to their CECs (Chapter 3).

213

4.6 Discussion

Endothelial (CD144+) and monocyte derived microparticles (CD14+) are elevated at times of active vasculitic disease activity as assessed by BVAS score. Although significant differences in a number of other MP populations between health and disease were not observed, preliminary data in individual patients demonstrated that different biomarkers altered with changes in disease activity.

The finding that EMPs are elevated during active vasculitis is comparable to the other studies, performed in vasculitis which examined EMPs (Brogan et al.,

2004a,Erdbruegger et al., 2008). However, these studies did not use CD144 as endothelial marker and the data presented herein did not reproduce these groups findings that CD105+ and E-selectin+ MPs were significantly elevated during active vasculitis. Differences in methodology relating to the use of an FL1 threshold in this study to enhance sensitivity could account for this, as well as differences in patient populations. Erdbruegger et al examined adults with AAV, whereas Brogan et al examined 29 children with active PSV, 15 of which had KD. As can be seen in the preliminary data presented in Figure 4.27, in this study KD patients appeared to have higher EMPs (of all phenotypes) than other vasculitic syndromes which could have skewed this previous result.

Both CD105 (endoglin) and CD144 are considered to be constitutive endothelial markers and it is interesting that they gave completely different results with peaks at different time points in some individual patients (appendix 6). As discussed in the introduction to this chapter, these markers have very different roles, CD144 forms part of the intercellular EC-EC contact complex and thus its expression would require

214

significant cellular disruption to release it. This could explain why this showed correlation with CEC levels (although this did not withstand the more stringent post-hoc testing for multiple comparisons). CD105 is upregulated on proliferating/angiogenic cells and thus may be indicative of a more reparative phase of the disease rather than an activatory phase. The finding that E-selectin+ and ICAM-1+ EMPs were not elevated in active disease is perhaps not surprising when the transient nature of their expression on activated endothelial cells is considered.

Another possibility for the differences in EMPs between this and previous studies relates to the observation that of the patients studied there were a number with active vasculitis prior to treatment that had very low levels of endothelial derived MPs

(examples are shown in Figures 4.25-26, and can also be observed in the longitudinal studies in Figure 4.11-14). It may be that EMPs are sequestered out of the plasma by binding to activated endothelium, platelets and/or cells of the blood. This situation has been observed in other diseases associated with severe endothelial injury including multiple organ dysfunction associated with sepsis (Joop et al., 2001). Thus when patients are on treatment and the underlying inflammation is suppressed, MPs may be observed in greater numbers and perhaps could provide insightful information regarding the risk of relapse prior to overt clinical disease flare. Although numbers are very preliminary, increases in E-selectin+ MPs occurred prior to and during flares in a small number of patients (Figure 4.25, and Appendix 6 patients 4, 8 and 12), whilst CD144+

MPs were elevated during active disease. In some individuals EMPs were higher during disease flare than at onset (Figure 4.26, Appendix 6 patients 2,5,8,10 and 13).

CD14+ and CD15+ MPs were also elevated during disease flare in individual patients

(Figure 4.25-26, Appendix 6 patients 1,5,8,9,12 and 13)

215

The study by Daniel et al examined CD66b+ neutrophil MPs and CD41a+ platelet MPs in adults with AAV (WG, MPA, CSS) (Daniel et al., 2006). Significant differences between healthy children and patients with active vasculitis for any of the NMPs or

PMPs studied herein were not observed in this study, however there are number of differences between this study and the one performed by Daniel et al. Primarily Daniel et al did not discriminate MPs based on phosphatidylserine (PS) expression using AnV and so may have included MPs that were excluded herein. Also the clinical criteria for active vasculitis used by Daniel et al was patients prior to cytotoxic treatment with

BVAS >7.5 - so adults with highly active systemic disease. Unfortunately the rarity of the vasculitides in children makes this exclusion criteria inappropriate for this cohort.

In contrast to the data from Daniel et al, prior to correction for multiple comparisons

PMPs in this study actually correlated negatively with disease activity, which if is a true finding rather than a false error could indicate that PMPs are sequestered from the plasma during active disease.

MPs which were observed in very low levels in all cohorts included CD3, however other studies had not examined this T cell marker in patients before and it may be that this is not expressed on T-cell derived MPs. This finding is supported by an in vitro study (Fritzsching et al., 2002) where low levels of CD4 and CD3 were found in MPs derived from isolated T cells. It may be that the MPs produced from activated T cells lack PS on their surface and are in the AnV negative fraction of MPs, which were not analysed in this study. In this context another marker that was found in low levels was

TF, no other study has examined this in vasculitis, however the presence of TF positive

MPs are associated conditions which are often associated with vascular damage (Morel et al., 2006), so this is somewhat surprising, however the antibody was titrated with care

216

and worked well on activated parent cells so it may be a genuine finding within the parameters of the cohort analysed.

Another factor which needs to be considered is that these parameters were being tested simultanously in the patient populations and therefore there is a risk of identifying false positives. While correcting for multiple comparisons may prevent the investigator falsely finding significant results it is equally important that the investigator does not discredit true findings by over conservative correction of the data. While the Holm

Bonferroni is a less conservative variation of the Bonferroni test, it is not without its limitations and there are varying opinions on the validity of making such corrections.

While some statisticians believe that a correction for multiple comparisons must be followed rigidly, others believe careful interpretation of the data is more crucial to understanding which results are likely to be true positives. Especially when there are only small numbers in each test group. In the case of the elevation of CD144+ EMPs in this patient population and correlation with other markers of disease activity, as other studies performed have indicated elevation of EMPs in patients with vasculitis it is likely that this is a true result and not a false positive.

What is apparent from this study of MPs is that in order to understand the role of different MP phenotypes in the context of vasculitis, their assessment at several timepoints over the disease course in order to examine the subtle changes may provide greater information relating to disease activity rather than large cross-sectional studies comparing groups of patients either in remission or with active disease.

217

As with all methods there are technical pitfalls and scope for further optimization. One major confounder relating to FC for detection of MPs is that it is unlikely that the whole

MP population is observed. As MPs range in size from 0.1-2µm in diameter (Piccin et al., 2007) it is likely that the smaller MPs are beyond the limits of detection of FC.

While ultimately it may not be as important in conditions where large elevations of MPs are observed, it may give an inaccurate representation of their phenotype and a skewed population profile. For example some cells may have a tendency to produce smaller

MPs and therefore where a population is not observed by FC, it doesn’t necessarily mean it was not present (e.g. CD3 MPs).

If it were not for the time constraints of a PhD study, other methodologies could have been employed including pull down ELISA. However only one parameter could be measured using this approach per well on the ELISA plate and thus would not be as simple to generate multi-parametric data. ELISA plates are coated with Annexin V which MPs from platelet poor plasma samples bind to. A secondary antibody to a cell surface marker associated with the MP of interest can then be used to enumerate the

MPs in the sample (Jy et al., 2004a). However, as the data presented herein indicates,

MPs can express a wide varety of cell surface markers and thus choosing a single marker to base the assay around would not be as informative as the multiparamet approach that FC offers. ELISA can however be used in a more functional capacity to investigate the behavior of MPs and in fact there are now commercially available kits to examine the prothrombotic potential of MPs (e.g. ZYMUPHEN MP-activity kit,

Aniara).

218

While the current protocol described had been used previously within the group there is still much controversy regarding the centrifugation protocols used for MP enumeration within the literature and as yet no consensus has been reached (Enjeti et al., 2007).

Some groups centrifuge at a lower speed but for longer to generate platelet poor plasma.

Some do not pellet the MPs at all. Feasibly all the spinning protocols particularly those associated with genreation of MP pellets could result in damage and loss of fragile MPs.

Thus a protocol which involves a minimal amount of centrifugation and handling would be preferable. Currently the widespread use of Annexin V for detection of PS on the surface and its’ reliance on Ca2+ for binding necessitates the high speed pelleting spin to remove anticoagulated plasma with low Ca2+ content. However use of alternative stains for PS, such as monoclonal antibodies directed against it (Mourdjeva et al., 2005) may allow development of a potentially more robust methodology.

219

5 Growth Factors

5.1. Summary

5.2. Introduction

5.3. Evaluation of growth factors in vasculitis

5.5. Discussion

220

5.1 Summary

As the endothelium is such a dynamic organ, it must be protected from the consequences of injury which could conceivably lead to loss of function. As has been demonstrated in Chapters 3 and 4 during active vasculitis injury and activation of the endothelium is significant, thus it is possible that repair mechanisms would be activated in these patients. The concept of endothelial repair in response to injury has rapidly gained credence in a number of human diseases, including coronary artery disease. This chapter examines some of the important signals that could provide evidence of an attempt of endothelial repair in the face of vasculitic injury.

Growth factors released from the endothelium and surrounding cells are pivotal in activating the mechanisms by which repair can be instigated, including recruitment of endothelial progenitor cells from the bone marrow (discussed in more detail in Chapter

6) and activation of angiogenic pathways in the surrounding endothelium, smooth muscle and pericytes. As well as reflecting a potential reparative response they could also reflect the extent of injury to the endothelium and as such could be used as a surrogate biomarker of disease activity. This chapter details the assessment of a number of growth factors associated with angiogenic and vasculogenic responses in patients with vasculitis, and considers their potential for use as a biomarker of disease activity.

5.2 Introduction

Angiogenesis, the remodelling of existing blood vessels to form new ones has been well studied in the development of tumour vasculature and is regulated by an extremely sensitive interplay of growth factors and inhibitors. A number of key signalling 221

pathways have been identified that play a role in the formation of new vessels including vascular endothelial growth factor (VEGF) and it’s receptor family; platelet derived growth factor (PDGF) and its receptor family; Delta-notch signalling and the angiopoietins (Ang-1-Ang-4) and their receptors Tie1 and Tie2 (Ferrara and Kerbel,

2005). Other signalling molecules that can impact on this system include peptides such as the thymosins and various cytokines and chemokines. Vasculogenesis, the de novo formation of blood vessels has been studied predominantly in embryonic development and perhaps unsurprisingly the same regulatory pathways are involved.

5.2.1 The VEGF family – the major angiogenic signalling pathway

The mammalian VEGF family consists of five isoforms of VEGF (A-D and placental growth factor PlGF) and three tyrosine kinase receptors which are expressed predominantly in the vasculature (Olsson et al., 2006). VEGFR1 and VEGFR2 are highly expressed by vascular endothelial cells whereas VEGFR3 is prominently expressed in lymphatic endothelial cells, although it can also be expressed on activated vascular endothelial cells. Other cell types that VEGFRs are found on include haematopoetic stem cells, monocytes, osteoblasts and neurons (Tammela et al., 2005).

The VEGF family is further complicated by the presence of splice variants of each of the different ligand isoforms which may have different functional effects depending on their ability to interact with VEGFR co-receptors, one variant of VEGF-A (VEGF-

A165b) for example is a negative regulator of VEGF activity (Woolard et al., 2004). The

VEGF-A165 isoform is the most abundant and widely studied of the VEGF isoforms and

is principal effector of VEGF action (others include VEGF-A189 and VEGF-A206 which are mostly bound to cell surface or extracellular matrix and VEGF-A121 which like

222

VEGF-A165 is a soluble secreted form) (Tammela et al., 2005). VEGFR1 is bound by

VEGF-A, VEGF-B and PlGF, VEGFR2 is bound by VEGF-A and cleaved forms of

VEGF-C and VEGF-D, while VEGFR3 is bound by VEGF-C and VEGF-D.

Experimental data implicates it is the interactions of VEGF-A with VEGFR1 and

VEGFR2 that are most important in embryonic vascular development and postnatal angiogenesis and vasculogenesis (Karamysheva, 2008), and it is these pathways that most recent research studies have focused on. A recent publication has implicated a novel role for VEGFR3 in the early vessel sprouting stages of vessel development

(Tammela et al., 2008).

Targeted deletions of all of the VEGF receptors (VEGFR1-3) in mice results in embryonic lethality associated with defective vasculogenesis and vessel development

(Olsson et al., 2006). Whilst VEGFR2-/- is associated with deficiency in EC development and a lack of vessels (Shalaby et al., 1995), VEGFR1-/- results in excessive endothelial proliferation and obstruction of vessel lumen (Fong et al., 1995). Knock out

(-/-) or depletion (+/-) of VEGF-A gives a similar phenotype to the VEGFR2-/-. Mice which exclusively express individual splice variants of VEGF-A have wide variations in phenotype, and studies have implicated VEGF-A164 (Human equivalent is VEGF-A165 as mice VEGF variants are one amino acid residue shorter than the corresponding human isoforms) as the most important isoform as these are normal and healthy (Maes et al., 2004). Mice expressing only VEGF-A120 die soon after birth from impaired myocardial angiogenesis with reduced capillary branching (Carmeliet et al.,

1999,Ruhrberg et al., 2002) whilst those expressing VEGF-A188 have stunted growth due to defects in bone formation and vascularisation (Maes et al., 2004). Conversely overexpression of VEGF-A also results in embryonic lethality (Miquerol et al., 2000),

223

highlighting the essential role of tightly regulated VEGF-A interactions in early developmental angiogenesis and vasculogenesis.

Knockouts of other VEGF isoforms are viable, but with the exception of VEGF-D-/- all have significant defects; VEGF-B-/- have reduced heart size with dysfunctional coronary vasculature and an impaired recovery from induced myocardial infarction (Bellomo et al., 2000) although a later study suggested phenotype was less severe and limited to impaired atrial conduction (Aase et al., 2001). PlGF-/- which also interacts with

VEGFR1 has a vascular phenotype with impaired angiogenesis and wound healing

(Carmeliet et al., 2001). VEGF-C-/- mice have defective lymphatic development

(Karkkainen et al., 2004).

On ligand binding VEGFR form homo or heterodimers which allow their signal to be transduced via activation of receptor tyrosine kinases. Activated tyrosine kinases recruit and activate an array of second messengers including the mitogen activated protein kinases (MAPK) phosphoinositide -3 kinases (PI3K) and phospholipase Cγ (PLCγ), which in turn activate numerous signaling cascades and complex crosstalk between them (Olsson et al., 2006). Negative regulation of the signal is important for limiting the target cell responses and phosphotyrosine phosphatases such as src-homology phosphatases (SHP1 and 2) which dephosphorylate and deactivate the tyrosine kinase signaling are involved (Guo et al., 2000), as is receptor internalization and degradation

(Singh et al., 2005).

Neuropilins (NRP1 and NRP2), another family of transmembrane receptors can also interact with VEGF although their intracellular domain is short and they do not signal

224

(Takahashi and Shibuya, 2005). Knockout studies indicate that while NRP1 is important in vascular development (Kawasaki et al., 1999) NRP2 is important in the lymphatic system (Yuan et al., 2002) and this reflects their expression profile

(Karamysheva, 2008). In vitro studies show that NRP1 complexes with VEGFR2, enhancing the proliferation and migration of cells in response to VEGF (Whitaker et al.,

2001).

Pathways that VEGF signaling have been implicated in postnatally include endothelial cell proliferation, survival, motility and migration, vascular permeability and recruitment of bone marrow derived stem cells. VEGFR signaling is dependent on the availability of the VEGF ligands, which can be released from a number of different cell types including epithelial cells, tumour cells and macrophages (Berse et al., 1992), smooth muscle cells (Ferrara et al., 1991), fibroblasts (Pertovaara et al., 1994), neutrophils (Gaudry et al., 1997) and from the endothelium itself (Lee et al., 2007).

Interestingly the autocrine VEGF signal is necessary for endothelial cell survival although not for angiogenesis, demonstrated in endothelial specific VEGF knockout mice where marked endothelial apoptosis and thromboembolic events were observed, yet overall vascular branched structure was no different to wildtype (Lee et al., 2007).

VEGF is upregulated by a variety of stimuli, including hypoxia, growth factors, cytokines, oestrogen, nitric oxide and mutations in its promoter associated with tumours

(Takahashi and Shibuya, 2005). Circulating VEGF is upregulated in pathological conditions associated with vascular injury and dysfunction including myocardial infarction (Seko et al., 1997), diabetes and sickle cell aneamia (Solovey et al., 1999) and

225

has previously been observed in active vasculitis in HSP (Topaloglu et al., 2001),

Behçet’s disease (Shaker et al., 2007) and in KD (Terai et al., 1999).

5.2.2 NOTCH-delta signaling – giving endothelial cells direction

The Notch family consist of 4 transmembrane receptors (Notch 1-4) and their membrane bound ligands Delta-like-1, -3 and -4 and Jagged-1 and -2. While VEGF directly stimulates the endothelium in angiogenesis and promotes the direction of migration, the response cannot be identical in all cells, otherwise new vessel formation would not be able to occur in the directional manner that it does. So called “tip cells” which form the tip of the sprouting blood vessel are necessary and signaling between

Delta-like ligand-4 (Dll4) and Notch receptors 1 and 4 have been shown to be important in their selection, allowing the VEGF gradient to direct new vessel formation (Thurston and Kitajewski, 2008).

Notch signaling occurs in many cell types and has a fundamental role in determining the differentiation patterns of individual cells in close proximity. One mechanism by which this occurs is lateral inhibition, whereby one cell presents a Notch ligand (Delta or

Jagged) to an adjacent Notch receptor expressing cell, resulting in activation of the

Notch pathway in one cell and suppression in the adjacent cell. Notch 1 knockout is embryonically lethal in mice and is associated with defective angiogenic vascular remodeling. Mice develop early vessels (the primary vascular plexus) suggesting it is not influencing vasculogenesis, however these do not develop further into a complex organized network of vessels (Krebs et al., 2000). Mice lacking Notch 4 develop normally, however dual Notch 1/Notch 4 knockouts often have a more severe vascular phenotype (some can lack entire major vessels) than the Notch 1 knockout (Krebs et al.,

226

2000) and overexpression of Notch 4 results in embryonic lethality (Uyttendaele et al.,

2001). Thus appropriate Notch 4 signalling is also necessary. Dll4 knockout is embryonic lethal and the majority of heterozygous knockouts do not survive (Gale et al., 2004).

5.2.3 PDGF – stabilizing new vessels

The PDGF family consists of 4 ligands (PDGFA-D) which form functional hetero or homodimers and 2 receptors (α and β). PDGFs form part of the same receptor/ligand family as VEGFs, it is generally accepted that the PDGFs influence mesenchymal cells, whereas VEGFs influence endothelial cells and both PDGFS and VEGFS are selective for their own receptors. However a recent publication has demonstrated that VEGF-A can bind to and activate PDGF receptors in bone-marrow-derived mesenchymal stem cells and thus this long held view that the growth factors do not cross-signal has become further complicated (Ball et al., 2007).

PDGFs act in a paracrine manner, in angiogenesis PDGF-B is highly secreted by endothelial cells, particularly those in the tip region and influences the recruitment of pericytes and vascular smooth muscle (VSM) cell precursors expressing PDGFR-β

(Hellstrom et al., 1999) which will stabilise the newly formed vessel. Knockouts of both PDGF-B and PDGFR-β result in pericyte and VSM cell deficiency seen from early in embryonic development. While the cells develop, they do not spread along newly developed vessels and the result is embryonic lethality due to vascular defects including endothelial hyperplasia, abnormal junctions, and excessive luminal membrane folds

(Hellstrom et al., 2001). The stabilization of newly formed vessels by pericytes involves their release of TGFβ which inhibits further endothelial proliferation and

227

migration, downregulates VEGFR2 expression and promotes differentiation of pericyte progenitors (Karamysheva, 2008). In vitro this occurs on contact between endothelial cells and pericytes (Sato and Rifkin, 1989).

5.2.4 Angiopoietins

The angiopoietin family (Angiopoietins-1,-2,-3 and -4 (Ang-1-4)), like the VEGF family, bind to receptor tyrosine kinases (RTK) – namely Tie-1 and Tie-2 (Fiedler and

Augustin, 2006). The Tie receptors are predominantly expressed within the vasculature and haematopoeitic stem cells, although Tie-2 expression can be detected on a subset of monocytes (Venneri et al., 2007). The known functional effects of the angiopoietins are predominantly mediated via Tie-2, as yet Tie-1 signalling pathways have been poorly elucidated and it was considered an orphan receptor. Yuan et al have recently demonstrated that Ang-1 can induce Tie1 signalling, although this is dependent on Tie2 expression, its consequence is a down regulation of the Tie2 dependent EC survival

(Yuan et al., 2007).

Deletion of Tie-2 in mice results in embryonic lethality that is associated with reduction in vascular complexity and branching (Sato et al., 1995). A similar phenotype is seen when Ang-1 (Suri et al., 1996) is deleted, or when Ang-2 is overexpressed

(Maisonpierre et al., 1997), indicating that Ang-2 acts as an antagonist of Ang-1 signalling via the Tie-2 receptor (at least in embryonic development). Deletion of Ang-

2 gives a relatively mild phenotype at birth, although mice usually die within 2 weeks due to severe chylous ascites (lymph fluid leaking into the peritoneal cavity) due to defective lymphatic development (Gale et al., 2002). Embryonic vascular development is relatively normal, however postnatal angiogenesis is defective, indiciating a key role

228

for Ang-2 in remodeling of the vasculature (Gale et al., 2002). While Ang-1 is constitutively expressed by many cell types including smooth muscle, fibroblasts and some tumor cells, Ang-2 is expressed predominantly by endothelial cells (as well as some tumour cells and retinal cells) (Fiedler and Augustin, 2006).

Ang-1 acting on Tie-2, induces phosphorylation of the receptor, resulting in activation of a complex cascade of second messengers, including PI3K and Akt (DeBusk et al.,

2004). Akt promotes cell survival and inhibits the action of the forkhead transcription factor FKHR-1, which induces Ang-2 production (Daly et al., 2004). Thus Ang-1 signalling inhibits production of its antagonist Ang-2. It also reduces vascular permeability, inhibits NFkB activation and thus suppresses inflammation (Hughes et al.,

2003) and promotes association between pericytes and the endothelium helping stabilize vessels. The constitutive agonist action of Ang-1 on Tie-2 is considered to be crucial to maintaining the endothelium in a non-activated quiescent state (Fiedler and Augustin,

2006).

Stimulation of the endothelium results in release of Ang-2 from Weibel-palade bodies, resulting in disruption of the resting state of the endothelium and priming it for angiogenesis or inflammation (Fiedler and Augustin, 2006). Ang-2 also enhances

TNFα induced leukocyte endothelial interactions via increased upregulation of adhesion molecules (Fiedler et al., 2006). The action of Ang-2 is altered by the presence of

VEGF, in its absence Ang-2 contributes to vascular regression, but in its presence Ang-

2 stimulates angiogenesis (Maisonpierre et al., 1997). In vitro, Ang-2 treatment of human microvascular endothelial cells (HMECs) disrupts barrier function, promoting actin stress fibre formation, resulting in intercellular gaps and increased permeability

229

(Parikh et al., 2006). While antagonism of Ang-1 Tie2 signalling may explain some of these responses, it is also conceivable that Ang-2 may also signal by alternative as yet unknown pathways.

Elevated Ang-2 is associated with conditions of inflammation and cardiovascular injury including sepsis (Siner et al., 2008), hypertension (Patel et al., 2008), diabetes (Lim et al., 2005) and acute coronary syndromes (Lee et al., 2004). After trauma Ang-2 can be elevated in circumstances when VEGF is not (Ganter et al., 2008). At the time of study no groups had examined Angiopoietin levels in patients with vasculitis. However a recent publication has indicated that Ang-2 levels are a potential biomarker of disease activity in adults with AAV (Kumpers et al., 2009).

5.2.5 Thymosin β4

The Thymosins are a family of peptides which were first isolated from the thymus

(Hannappel, 2007). They were originally thought to be thymic hormones, however it has become apparent that they are ubiquitously expressed. The β-thymosins sequester

G-actin, regulating its phosphorylation and thus cytoskeletal remodeling (Hannappel,

2007). Thymosin β4 (Tβ4) is the most abundant member in most cell types, representing approximately 70–80% of total beta thymosins and has been implicated in angiogenic and vasculogenic pathways (Smart et al., 2007b).

Tβ4 has been shown to influence VEGF levels, act as an endothelial chemoattractant and regulate a number of genes which can influence angiogenesis and cell motility, including matrix metalloproteinases, adhesion molecules and cytoskeletal related peptides (Smart et al., 2007b). In tissues with high Tβ4, VEGF levels are also elevated

230

(Cha et al., 2003), correspondingly in tissues where Tβ4 has been knocked out, VEGF expression is also low (Smart et al., 2007a). Whether this is a direct interaction between

Tβ4 and VEGF or whether Tβ4 influences an upstream inducer of VEGF such as hypoxia inducible factor (HIF) remains unknown. Another member of the thymosin beta family, Tβ10 which has anti-angiogenic effects reduces VEGF levels (Mu et al.,

2006). A small number of studies have examined plasma or serum levels of Tβ4 in humans, Tβ4 increases in rhinovirus infection (Hsia et al., 1989), interestingly is lower in serum extracted from umbilical cords than from adults (Weller et al., 1988) and a recent study has indicated it is elevated in patients with psoriasis (Plavina et al., 2008).

In this chapter the growth factors which have been shown to directly drive angiogenesis

(VEGF, Ang-1, Ang-2 and Tβ4) as opposed to those described which are more important in influencing direction and vessel stabilisation have been measured in patients with vasculitis. Many of these studies (Ang-1, Ang-2 and Tβ4) represent data which have hitherto not been reported in paediatric vasculitis. At the time of writing, of these 3 growth factors only Ang-2 has been reported in adults with AAV (Kumpers et al., 2009).

231

5.3 Evaluation of growth factors in vasculitis

5.3.1 Patients and methods

5.3.1.1 Patients

40 children with primary systemic vasculitis were studied. The majority of these patients (n=36) were the same individuals described in Chapters 3 and 4, allowing GFs to be compared with CECs and MPs. These included 16 males and 24 females of median age 9.8 years (range 0.1 -19.0 years). The vasculitis type (using the Chapel Hill

Consensus Criteria) (Jennette et al., 1994) was polyarteritis nodosa (PAN) n=17,

Wegener’s granulomatosis (WG) n=9, microscopic polyangiitis (MPA) n=1,

Kawasaki’s disease (KD) n=5, Takayasu’s arteritis (TD) n=2, Beçhets disease (BD) n=1 and unclassified n=6. Disease activity was assessed using a modified paediatric

Birmingham Vasculitis Activity Score (BVAS), as described previously. 18 healthy child controls (4 males, 14 females), 14 febrile controls (6 males, 8 females with a variety of autoimmune or other inflammatory disorders without the presence of vasculitis) and 20 healthy adult controls were also recruited (2 males, 18 females, age range 24-40 years). As with the vasculitis patients, the control individuals were the same as those studied in Chapters 3 and 4.

5.3.1.2 Methods

Platelet poor plasma and serum were collected and stored as described in section 2.4.

Platelet poor plasma was used in ELISA for Ang-1 and Ang-2, using ELISA kits from

R&D as per manufacturer’s instructions (as described in section 2.6.1). Serum was used in ELISA for detection of VEGF-A165 using an ELISA kit from R&D as described in section 2.6.1 Serum was also used for detection of Tβ4 using an ELISA kit from immunodiagnostik (as described in section 2.6.1). Samples were defrosted at room

232

temperature while the ELISA kit reagents were being prepared. Ang-1 required a 1 in

15 dilution of sample, Ang-2 required a 1 in 5 dilution of samples (as per manufacturers instructions). VEGF serum was used neat and Tβ4 was diluted 1 in 5 before addition to the test plate. As Ang-1 levels were frequently falling below the detection limit of the kit and the high level of dilution was necessary to prevent Ang-1 complex formation

(R&D, technical support, personal communication) fewer samples had their Ang-1 detected than Ang-2. Examples of calibration curves are given in Appendix 5.

5.3.1.3 Statistics

Median growth factor levels were compared between groups using the Kruskall Wallis test to check for overall differences. Individual Mann Whitney U tests for non parametric data were performed to compare between specific groups. The Wilcoxon signed rank test was performed to compare medians of paired samples. Spearman Rank correlation coefficiants were performed to assess correlation between growth factors and other markers of disease activity, including CECs. In all data graphs patient classification was denoted using colour coding as indicated below:

233

5.3.2 Results

5.3.2.1 VEGF

Kruskall Wallis testing indicated there were significant differences between test groups

(p=0.04). Individual analysis between groups revealed VEGF was significantly higher in 20 patients with active vasculitis (median value 579pg/ml) compared to 22 healthy child controls (median value 194pg/ml, p=0.02) and 9 healthy adult controls (median value 231pg/ml, p=0.03). There were no significant differences between patients with active vasculitis and 9 febrile controls (median value 395 pg/ml) or 17 patients with inactive disease (median value 241pg/ml) and these control groups. The data presented in Figure 5.1a represents the overall vasculitis cohort and when tested a significant difference between patients with active and inactive vasculitis (p=0.04) was observed, however this analysis includes paired data. When paired data between active and inactive vasculitis was excluded (the first observation made in each individual patient was used in this analysis) the difference between 18 patients with active disease

(median value 578 pg/ml) and 6 patients in remission (median value 137pg/ml) was no longer significant (p=0.07). As before the colour of symbols and lines indicates disease classification of individual patients.

In 13 individual patients VEGF levels were significantly higher during active disease than when patients were in remission (p=0.013) (median time to remission 3 months)

(Figure 5.1b). 10 patients were followed over time from initiation of treatment at disease onset, at median follow up of 3 and 12 months VEGF was significantly lower than at onset than at followup (p=0.02 and p=0.047 respectively, Figure 5.1c).

234

Figure 5-1: Serum levels of VEGF in vasculitis

Figure 5.1a:VEGF was significantly higher in 20 patients with active vasculitis compared to 22 healthy child controls (p=0.02), 9 healthy adult controls (p=0.03) and 17 patients with inactive disease (p=0.04).

There were no significant differences between patients with active disease and 9 febrile controls or 17 patients with inactive disease and any of the control groups. Figure 5.1b: In 13 individual patients VEGF levels were significantly higher during active disease compared to remission (p=0.025, median time to remission 3 months). Figure 5.2c: 13 individual patients were followed from initiation of treatment , at median followup of 3 months VEGF was significantly lower than at onset (p=0.015) although not at 12 months (p=0.06). Symbol and line colours reflect individual patients disease classification.

235

5.3.2.2 Angiopoietin-1

Kruskall Wallis testing indicated there were no overall significant differences between test groups (p=0.07). When test groups were considered individually Ang-1 was not significantly different in 13 patients with active vasculitis (median value 2670 pg/ml) compared to 14 healthy child controls (median value 835pg/ml, p=0.1) and 3 healthy adult controls (median value 0pg/ml, not enough data to compare statistically). There were no significant differences between 12 patients with inactive vasculitis (median value 1051pg/ml) and these control groups (Figure 5.2a. Ang-1 was comparable with reference ranges for the kit in adult controls. The data presented in Figure 5.2a represents the overall vasculitis cohort, when tested no significant difference was observed between patients with active and inactive vasculitis (p=0.5). There were not enough samples to exclude paired data between the two test groups. As before the colour of symbols and lines indicates disease classification of individual patients.

In 12 individual patients Ang-1 levels were not significantly different between active disease and remission (p=0.6) (median time to remission 3.5 months, Figure 5.2b). 12 individual patients were followed from initiation of treatment at disease onset or flare

(dotted lines indicate flare). In 11 patients followed over time from initiation of treatment, at median follow up of 3 and 12 months Ang-1 was not significantly different to levels at onset (p=0.3 and p=0.2 respectively, Figure 5.2c).

5.3.2.3 Angiopoeitin-2

Kruskall Wallis testing indicated there were no overall significant differences between the test groups (p=0.2). When test groups were considered individually Ang-2 was not significantly different between 23 patients with active vasculitis (median value 1739 pg/

236

Figure 5-2:Plasma Angiopoeitin-1 levels in vasculitis

Figure 5.2a: Ang-1 levels were not significantly different in 12 patients with active vasculitis compared to 14 healthy child controls (p=0.07), 11 patients with inactive vasculitis (p=0.5) and 3 healthy adult controls (not enough data to compare). There were no significant differences between patients with inactive vasculitis and these control groups. Figure 5.2b: In 13 individual patients Ang-1 levels were not significantly different between active disease and remission (p=0.44) (median time to remission 3.5 months). Figure 5.2c: 12 patients were followed over time from initiation of treatment, at median follow up of 3 and 12 months Ang-1 was not significantly different to levels at onset (p=0.3 and p=0.2 respectively). Symbol and line colours reflect individual patients disease classification.

237

ml) compared to 12 febrile controls (median value 1798pg/ml, p=0.27), 19 healthy child controls (median value 1512 pg/ml, p=0.15) and 20 healthy adult controls (median value 1382pg/ml, p=0.167, comparable with reference ranges for the kit). There were no significant differences between 25 patients with inactive vasculitis and healthy child controls or adult controls however Ang-2 was significantly higher in febrile controls

(p=0.04). The data in figure 5.3a represents the whole vasculitis cohort, on testing no significant differences were observed between active and inactive disease. When paired samples between these groups were excluded there remained no significant difference between 23 patients with active vasculitis and 10 in remission (p=0.2). As before the colour of symbols and lines indicates disease classification of individual patients.

In 17 individual patients Ang-2 was significantly higher during active disease than in remission (p=0.045, median time to remission 5.5 months, Figure 5.3b). 13 patients were followed over time from initiation of treatment at disease onset (solid lines) or during flare (dotted lines), at median follow up of 3 and 12 months from disease onset and initiation of treatment Ang-2 was significantly lower than at onset (p=0.001 and p=0.04 respectively, Figure 5.3c).

5.3.2.4 Thymosin β4

Kruskall Wallis testing indicated no overall significant differences between the test groups (p=0.2). When test groups were considered individually Tβ4 was not significantly different between 17 patients with active vasculitis (median value 5.87

µg/ml) compared to 5 febrile controls (median value 6.7µg/ml p=0.56), 9 healthy child controls (median value 8.61µg/ml p=0.78) or 8 healthy adult controls (median value3.84µg/ml, p=0.07). There were no significant differences between patients with inactive vasculitis and any of these controls groups or with patients with active disease 238

Figure 5-3: Plasma Angiopoeitin-2 levels in vasculitis

Figure 5.3a: Ang-2 was not significantly different between 23 patients with active vasculitis compared to

12 febrile controls, 19 healthy child controls and 20 healthy adult controls. There were no significant differences between patients with inactive disease and healthy child controls or adult controls however

Ang-2 was significantly higher in febrile controls (p=0.04). Figure 5.3b: In 17 individual patients, Ang-2 was significantly higher during active disease than in remission (p=0.045, median time to remission 5.5 months) Figure 5.2c: In 12 patients followed over time from initiation of treatment, at median follow up of 3 and 12 months Ang-2 was significantly lower than at disease onset (p=0.001 and p=0.04 respectively). Symbol and line colours reflect individual patients disease classification.

239

(this analysis included paired data between active and inactive disease as there was not enough data for statistical analysis otherwise). Healthy child controls had significantly higher levels of Tβ4 than healthy adult controls (p=0.022, Figure 5.4a).

In 12 individual patients there were no significant differences in Tβ4 between active and inactive disease (p= 0.7, median time to remission 3 months) (Figure 5.4b). 11 patients were followed over time from initiation of treatment at disease onset (solid line) or flare

(dotted line). In 10 patients at median follow up of 3 and 12 months from disease onset there were no significant differences from initiation of treatment (p=0.8 and p=0.19,

Figure 5.4c).

240

Figure 5-4: Serum levels of Thymosin Beta 4 in vasculitis

Figure 5.4a: Tβ4 was not significantly different between 17 patients with active vasculitis and 5 febrile controls, 9 healthy child controls or 8 healthy adult controls. There were no significant differences between patients with inactive vasculitis and any of these controls groups. Healthy child controls had significantly higher levels of Tβ4 compared to healthy adult controls (p=0.022). Figure 5.4b: In 12 individual patients there were no significant differences in Tβ4 between active and inactive disease

(median time to remission 3 months) Figure 5.4c: 11 patients were followed over time from initiation of treatment at disease onset (solid lines) or flare (dotted lines). At median follow up of 3 and 12 months there were no significant differences from initiation of treatment. Symbol and line colours reflect individual patients disease classification.

241

5.3.3 Correlation between growth factors and disease activity

Spearman Rank correlation coefficiant analysis was performed between the 4 growth factors assessed and the traditional markers of active disease; BVAS, CRP, ESR, platelet count, albumin, white cell count as well as the previously demonstrated novel markers of endothelial injury and cellular activation CEC count, EMPs and PMPs in patients with vasculitis. Table 5.1 summarises the results. Table 5.2 shows the corrected P-values obtained from a post-hoc Holm-Bonferroni correction. In this instance a number of the significant correlations observed on individual tests withstood the rigours of multiple comparison correction and therefore are likely to be true significant results and not false positives.

Serum VEGF and Plasma Ang-2 both correlated significantly with the inflammatory markers ESR and CRP, as well as the BVAS score of disease activity. Interestingly neither growth factor correlated with circulating endothelial cells. Although the following did not withstand correction for multiple comparisons Ang-2 correlated inversely with EMPs (CD105+) Ang-1 correlated with platelet count and TB4 only correlated with VEGF. Importantly VEGF did not correlate with platelet count or

PMPS. Both can act as stores of VEGF and thus platelet degranulation may influence serum levels (Jelkmann, 2001), however this did not appear to affect this result.

5.3.4 Influence of treatment on growth factor responses

The result that growth factors did not correlate with endothelial injury as assessed by

CECs or EMPs was unexpected, especially as both VEGF and Ang-2 correlated with the inflammatory markers ESR and CRP and the clinical score of disease activity. As immunosuppression influences the release of chemokines and cytokines from a number

242

Table 5-1: Spearman Rank correlation coefficiants of growth factors and markers of disease activity in vasculitis.

Marker of VEGF ANG-1 ANG-2 TB4 disease activity BVAS r = 0.329 r = 0.158 r = 0.344 r = 0.22 p = 0.018* p = 0.34 p = 0.002* p = 0.2 n = 51 n = 38 n = 76 n = 33 ESR r = 0.529 r = 0.035 r = 0.432 r = 0.140 p<0.001* p = 0.83 p <0.001* p = 0.45 n = 48 n = 38 n = 74 n = 31 CRP r = 0.460 r = 0.086 r = 0.503 r = 0.27 p = 0.001* p = 0.6 p <0.001* p = 0.14 n = 49 n = 38 n = 75 n = 32 WCC r = 0.508 r = 0.034 r = 0.137 r = 0.337 p <0.001* p = 0.84 p = 0.24 p = 0.055 n = 50 n = 39 n = 75 n = 33 Platelet count r = 0.274 r = 0.44 r = -0.015 r = 0.215 p = 0.06 p = 0.006* p = 0.9 p = 0.245 n = 47 n = 37 n = 72 n = 31 Albumin r = -0.502 r = -0.190 r = -0.370 r = -0.027 p <0.001* p = 0.266 p = 0.001* p = 0.09 n = 46 n = 36 n = 72 n = 31 CEC count r = 0.150 r = 0.182 r = 0.016 r = 0.058 p = 0.35 p = 302 p = 0.89 p = 0.765 n = 41 n = 34 n = 70 n = 29 EMP (CD144+) r = 0.061 r = 0.127 r = -0.148 r = 0.109 p = 0.69 p = 0.428 p = 0.2 p = 0.55 n = 44 n = 41 n = 76 n = 32 EMP (CD105+) r = 0.047 r = -0.62 r = -0.308 r = -0.22 p = 0.76 p = 0.7 p = 0.007* p = 0.23 n = 44 n = 41 n = 76 n = 32 EMP (eselectin+) r = 0.217 r = 0.025 r = -0.103 r = 0.132 p = 0.16 p = 0.88 p = 0.378 p = 0.47 n = 44 n = 41 n = 76 n = 32 PMP (CD42a+) r = -0.235 r = -0.056 r = -0.214 r = 0.137 p = 0.13 p = 0.73 p = 0.066 p = 0.46 n = 43 n = 41 n = 75 n = 32 Total MP r = -0.111 r = -0.129 r = -0.067 r = -0.022 p = 0.473 p = 0.42 p = 0.566 p = 0.9 n = 44 n = 41 n = 76 n = 32 VEGF r = 0.315 r = 0.306 r = 0.369 p = 0.09 p = 0.039* p = 0.023* n = 30 n = 46 n = 38 ANG-1 r = 0.315 r = 0.032 r = 0.178 p = 0.09 p = 0.84 p = 0.38 n = 30 n = 30 n = 26 ANG-2 r = 0.306 r = 0.032 r = 0.269 p = 0.039* p = 0.84 p = 0.13 n = 46 n = 30 n = 33 TB4 r = 0.369 r = 0.178 r = 0.269 p = 0.023* p = 0.38 p = 0.13 n = 38 n = 26 n = 33 Significant correlations are highlighted in blue

243

Table 5-2: Holm-Bonferroni correction for multiple comparisons relating to correlations of growth factors and other markers of disease activity

Significant Test P value Adjusted critical value

(Holm-Bonferroni)

WCC vs VEGF <0.001 0.000925926

Albumin vs VEGF <0.001 0.000943396

ESR vs ANG-2 <0.001 0.000961538

CRP vs ANG-2 <0.001 0.000980392

ESR vs VEGF 0.001 0.001

CRP vs VEGF 0.001 0.001020408

Albumin vs ANG-2 0.001 0.001041667

BVAS vs ANG-2 0.002 0.00106383

Platelet count vs ANG-1 0.006 0.001086957

EMP (CD105+) vs ANG-2 0.007 0.001111111

BVAS vs VEGF 0.018 0.001136364

TB4 vs VEGF 0.023 0.001162791

ANG-2 vs VEGF 0.039 0.001190476

of cell types, we hypothesised that treatment might be influencing the release of growth factors and thus the data was reanalysed to take this into account. 29 individual patients whose growth factors had been measured and had active disease as defined via clinical scores (BVAS) were then divided into two subgroups: those who were treatment naive

(i.e. had received no treatment, n= 13), and those who had flared while on immunosuppressive treatment (n=16). In those individuals who had had repeated measurements taken while disease was still active, it was the first sample that was used in this analysis. 244

There were no significant differences in markers of endothelial injury/activation (CECs and EMPs) between 9 patients who were treatment naive and 12 patients who were on immunosuppressive treatment (Figures 5.5a and b). Patients who were treatment naive had significantly higher BVAS scores (p=0.03) and CRP (p=0.01) although not ES).

(Figures 5.5c-e) Correspondingly the treatment naive patients also had significantly higher levels of VEGF (p=0.007), Ang-1 (p=0.03) and Ang-2 (p=0.0003), than those who were on immunosuppressive therapy. This may indicate that immunosuppressive treatment influences release of these growth factors in response to disease activity.

245

Figure 5-5: Influence of treatment on endothelial injury markers and clinical parameters

Figure 5.5a-e: When patients with active disease had their treatment status considered (treatment naive or on treatment), there was no significant differences in CECs or EMP between patients who were treatment naive (no Rx) or on immunosuppressive therapy (on Rx) when they flared (Figure 5.5a and b).

There were significant differences between the two groups when clinical parameters were considered:

BVAS (p=0.01, Figure 5.5c) and the inflamamotry markers CRP (p=0.005, Figure 5.5d) and ESR

(p=0.02, Figure 5.5e) were significantly higher in patients who had not yet received treatment Rx = treatment. Symbol and line colour reflect individual patients disease classification (colour coded as in previous figures)

246

Figure 5-6: Influence of treatment on growth factors

Figure 5.6a-d:When patients with active vasculitis had their treatment status considered there were significant differences in growth factor levels between patients who were treatment naïve (Rx naïve) and those who were on immunosuppressive therapy (on Rx) when they flared. Figure 5.6a: VEGF was significantly higher in 12 treatment naive patients compared to 7 on immunosuppression (p=0.03) Figure

5.6b Ang-1 was not significantly different between 10 treatment naive patients compared to 4 on immunosuppression (p=0.1), however this comparison was underpowered Figure 5.6c Ang-2 was significantly higher in 12 treatment naive patients compared to 13 on immunosuppression (p=0.002) .

Figure 5.6d: there were no signficant differences in Tβ4 between treatment naïve and those on immunosuppression. Symbol and line colour reflect individual patients disease classification (colour coded as in previous figures)

247

5.4 Discussion

Growth factors regulate the angiogenic and vasculogenic pathways that are necessary for vascular repair and thus monitoring of their levels may provide insight into potential reparative response. Like CECs and EMPs, they can be expressed as a consequence of endothelial injury and it was with this in mind that this study was undertaken, to assess their potential as an indicator of active disease in vasculitis.

The preliminary observations made are interesting and warrant further investigation, however it is difficult to draw firm conclusions from them since patient numbers are small. Although VEGF and Ang-2 in particular correlate with a number of markers of disease activity, responses may be influenced by treatment. Thus their use as biomarkers in a highly heterogenous cohort on widespread immunosuppression may not be appropriate. The complexity of the results is perhaps not surprising when it is considered that growth factors form part of extensive regulatory networks and could conceivably be contributing to the pathological processes involved in vascular inflammation as well as promoting repair.

None of the growth factors examined herein correlated with CECs or EMPs. VEGF and

Ang-2 did however correlate with clinical scores of disease activity (BVAS) as well as non specific markers of inflammatory disease activity such as ESR and CRP. This could be indicative of the upstream signalling required to release them: VEGF is produced by monocytes which are integral to the inflammatory process and although

Ang-2 is relatively endothelial specific, it is released from the Weibel-palade bodies in response to inflammatory stimuli. Perhaps these growth factors are upregulated in reponse to inflammatory stimuli that are not sufficient to induce EC detachment

248

(forming CECs). VEGF also correlated with Tβ4 levels (although this did not withstand the rigours of correction for multiple comparisons), corresponding with previously published reports that in animal models high Tβ4 in isolated organs is associated with elevated VEGF (Cha et al., 2003).

The concept that growth factors may be more closely linked with inflammation than as a direct consequence of endothelial injury was further reinforced by the effect of therapy on levels of VEGF, Ang-2 and Ang-1 (although this latter observation was underpowered and lacked statistical significance). Routine maintenance therapy (in the majority of cases corticosteroids, with azathioprine) dampened down the elevations seen in these growth factors when patients had flares of their disease. While it would be easy to suggest that patients on maintenance therapy do not have such extensive disease activity in flare and that is why their response is dampened, CECs and EMPS (which reflect the extensiveness of EC injury and activation) were unaffected by treatment status in patients with active disease. The pathways upstream of VEGF and ANG-2 are complicated, a number of mediators can influence their production and thus it is conceivable that inhibition of several immune pathways by treatment such as corticosteroids and cytotoxic immunosuppressants could impact on growth factor release.

The recently published data from Kumpers et al on Ang-2 levels in adults with AAV, is similar to that described herein, Ang-2 levels were elevated at disease onset, declined with therapy and correlated strongly with BVAS and (in contrast to the data presented herein) CEC number (Kumpers et al., 2009). Their interpretation of the usefulness of

Ang-2 as a biomarker of disease activity in vasculitis is more optimistic than the

249

conclusion reached in this study as they did not observe that immunosuppressive treatment could be influencing the growth factor responses. While this could be due to differences between adult and child physiology, it may be that the inclusion criteria of the patients studied is the cause. Of the 15 patients with active disease studied by

Kumpers et al, 8 were treatment naive and the remainder were off immunosuppressive therapy or on <7.5mg/day prednisolone (exact numbers were not stated). If the majority of patients were in fact not on any immunosuppressive therapy then this interesting observation may well have gone undetected in their cohort. Regardless of the status of patients included in these studies, both remain preliminary as the datasets are small and a larger scale study is necessary to assess the potential of Ang-2 as a biomarker of disease activity, and if indeed its levels are influenced by treatment status.

The implications of the apparent growth factor suppression by treatment in children may not necessarily be a bad thing. VEGF is a well known mediator of vascular permeability and elevations of this could be contributing to oedema formation often associated with inflammation (Mehta and Malik, 2006). Furthermore, as discussed in the introduction to this chapter, newly formed vessels may amplify and target vascular inflammation. Indeed, one line of argument has regarded angiogenic signals as a means of amplifying vasculitic injury (Cid, 2002). In healthy individuals growth factors act in a localised manner to control angiogenesis and vasculogenesis, processes reliant on concentration gradients to direct tightly regulated changes in vessel behaviour.

Therefore large changes in growth factors could in fact be pathological and not indicative of a reparative response but of a “growth factor storm” propagating further disruption of endothelial cells and injury.

250

In conclusion, children with active vasculitis demonstrate significant angiogenic and vasculogenic signals in peripheral blood at a time when their vasculitis is most active.

Although routine clinical testing of growth factors in vasculitis would be a simple clinical test, the influence of treatment, the inconsistent results observed (particularly for Ang-1 and Tβ4) and the complexity associated with growth factor responses and effects suggest that the growth factors assessed in this study may not be an appropriate biomarker of disease activity for clinical monitoring of disease activity. These observations, particularly in relation to the finding of elevated VEGF and Ang-2 in active vasculitis (albeit influenced by treatment), do however, provide impetus for further study of endothelial progenitor cell reponses in relation to vasculitis disease activity. EPCs are the subject of Chapter 6.

251

6 Endothelial Progenitor Cells

6.1. Summary

6.2. Introduction

6.3. Optimisation of flow cytometric for detection of EPCs

6.4. EPCs in vasculitis

6.5. Detection of EPCs using cell culture

6.6. Discussion

252

6.1 Summary

Endothelial progenitor cells (EPCs) are bone marrow derived cells which migrate to sites of endothelial injury and assist in repair. Whether this occurs is via EPCs differentiating down an endothelial lineage or by orchestrating a localised proangiogenic environment is unclear. As has been discussed in Chapter 5, growth factors which promote angiogenesis have been implicated in recruitment of EPCs from the bone marrow and I previously demonstrated elevations of VEGF and Ang-2 in children with active vasculitis. This chapter describes EPCs in children with vasculitis and changes with treatment.

6.2 Introduction

Endothelial repair is integral to maintaining health, however the exact mechanisms by which this occurs remain as yet unclear. As mentioned in the introduction to this thesis, angiogenesis was considered to be the mechanism by which the endothelium repaired itself until relatively recently, when Asahara et al introduced the concept of postnatal vasculogenesis and endothelial progenitor cells (EPCs) (Asahara et al., 1997).

The EPCs described by Asahara et al were a population of CD34+ cells. CD34+ cells are routinely used in stem cell transplantation and can reconstitute all blood cell lineages (Cheng et al., 1996). CD34 is found on mature endothelium and haematopoietic progenitor cells (of myeloid and lymphoid lineages) although is lost during the latters maturation. Vasculogenesis studies in embryonic development, indicate that both the cells forming the vasculature and those that form the cells of the blood are derived from a common progenitor – known as the haemangioblast (Vogeli et 253

al., 2006). Haematopoeitic cells are continually repopulated throughout life and thus the concept that EPCs also remain involved in vascular repair and remodeling is an attractive hypothesis.

While the work of Asahara et al may have been the first description of EPCs, evidence had accumulated for many years previously that circulating cells could repopulate the endothelium. As early as 1932 Hueper and Russell had found capillary like tubule formations in cultures of leukocytes and continuation of their work by Parker in 1934 demonstrated that leukocytes require a base of red blood cells and platelets in order to form these structures (Parker, 1934). By 1963 the concept of fallout endothelialisation had been described, whereby synthetic grafts implanted into animals showed re- endothelialisation which was not derived from infiltrating surrounding endothelium (Shi et al., 1994,Stump et al., 1963). In 1998 Shi et al demonstrated that these circulating cells originated from bone marror as donor derived cells were found to be the predominant endothelial cell on reendothelialised grafts implanted after transplantation in dogs (Shi et al., 1998).

6.2.1 EPCs influence in vivo endothelial repair

Since the discovery of EPCs by Asahara et al numerous studies involving animal models of acute vascular injury, transplantation and atherosclerosis have been described, indicating a role for EPCs in reparative responses. The exact phenotype of

EPCs is a controversial subject and as yet no specific combination of markers conclusively defines the EPC population (Hirschi et al., 2008). Most in-vivo studies where EPCs have been perfused into subjects (animals and human) have used mixed populations of cells which include EPCs; some studies have used CD34+ or whole

254

mononuclear fractions of freshly isolated peripheral blood. In animal models culture modified mononuclear cells have also been used. In a number of studies pharmacological agents such as G-CSF have been used to mobilise bone marrow progenitors. These will now be discussed in more detail.

6.2.1.1 EPCs in animal models

In a series of vein graft transplant studies in transgenic mice, Xu et al demonstrated that circulating cells from the recipient rapidly repopulate the vein graft endothelium, about a third of which are bone marrow derived cells (Xu et al., 2003). In apolipoprotein E−/−

(ApoE-/-) knockout mice, which have reduced circulating EPCs, regeneration of the endothelium is delayed and development of atherosclerotic lesions in the vein graft accelerated (Xu et al., 2003). ApoE1-/- are a well established model of atherosclerosis and develop spontaneous disease, the progression of which can be decreased with the infusion of EPCs from younger mice, not yet exhibiting atherosclerotic changes, but not EPCs from older sick mice (Rauscher et al., 2003). In this study cells isolated from bone marrow and cultured for up to 2 weeks were used. Interestingly other studies have demonstrated that EPC infusion is detrimental in this model when combined with an ischaemic injury possibly due to an inflammatory role. Although EPC infusion improved ischaemia reperfusion injury in 2 studies, it also accelerated atherosclerotic plaque development and instability (Silvestre et al., 2003), which was associated with higher serum IL-6 and MCP-1 (George et al., 2005). However both these latter studies used uncultured cells isolated from bone marrow.

In arterial balloon injury models in rats, mobilization of EPCs using G-CSF enhances

255

repair and reduces associated hyperplasia of the intima and smooth muscle (Kong et al.,

2004a). The same group demonstrated that in a similar model in rabbits, modification of cultured autologous EPCs to overexpress eNOS further enhanced the reparative response and suppression of hyperplasia (Kong et al., 2004b). While in these models of vascular injury, EPCs suppressed hyperplasia, the opposite effect was observed when experimental stents coated with anti-CD34 to promote EPC recruitment were implanted in pigs. While endothelialisation of coated stents was more rapid than in controls, it was also associated with excessive intimal hyperplasia (Rotmans et al., 2005).

In induced myocardial infarction in both mice and rats, transplantation of bone marrow mononuclear cells into the heart improves recovery and function, associated with enhanced angiogenesis (Zhang et al., 2004) with increased IL-10 production and reduced T cell infiltration, indicating a possible anti-inflammatory role for EPCs

(Burchfield et al., 2008). Similarly enhanced angiogenic responses and improved function are also seen in hindlimb ischaemia models by injection of autologous and donor animal bone marrow mononuclear cells (Asahara et al., 1997) and even human circulating PBMCs (into athymic rats) (Iba et al., 2002).

While there is wide variation in the methods employed in these animal studies, particularly relating to the phenotype of cells considered to be EPCs and their isolation and culture (in some studies), it is apparent that EPCs do promote neovascularisation.

The extent of incorporation of EPCs into the newly formed vasculature is extremely variable (0-90%, (Ward et al., 2007) and recently there has been much interest in understanding the paracrine effects of these cells. Some in vivo studies have implied that EPC recruitment to the site of injury is a transient response and hypothesise that the

256

major role of EPCs is to provide a localized pro-angiogenic environment facilitating repair (Fazel et al., 2006).

6.2.1.2 EPCs in human clinical trials

Despite many unanswered questions relating to function, phenotype and possible detrimental effects of EPC infusions, there have been a small number of groundbreaking studies that have taken EPC therapy into humans. The majority of studies have involved patients with myocardical infarction and while safety has not been an issue in the studies performed, efficacy has been questionable (Jujo et al., 2008).

The majority of studies undertaken have used infusions of freshly isolated mononuclear cells, from either blood (Assmus et al., 2002,Wollert et al., 2004) or bone marrow

(Lunde et al., 2008,Janssens et al., 2006) rather than a selected population. A small number of studies have examined the effects of isolated CD34+(Balogh et al., 2007) or

CD133+ populations (Bartunek et al., 2005). Overall it appears that localized infusion into the coronary arteries or myocardium during bypass surgery of either mononuclear cells, CD133+ or CD34+ cells can result in a mild to moderate improvement in cardiac function and/or reduction in infarct size, although these are inconsistent findings, with some studies indicating no improvement (Jujo et al., 2008). This is in contrast to the promising and significant differences seen in the aforementioned animal models.

However, the small human studies so far performed have been underpowered, and the methodologies variable, particularly the timing of EPC infusions post injury; the harvesting of cells and types of cell populations infused (Jujo et al., 2008). Thus it is not surprising that the results obtained so far are inconsistent. Only a longer clinical trial or metanalaysis of similar trials with common treatment protocols, will establish

257

whether this is truly a therapeutic option for the future.

Clinical trials of G-CSF mobilisation of autologous bone marrow stem cells have so far demonstrated no efficacy, and concerns have been raised over the risk associated with mobilization of inflammatory cells using this approach (Ward et al., 2007). In the context of vasculitis, G-CSF has been implicated in experimental models of glomerulonephritis; mice in which GM-CSF or G-CSF has been knocked out are protected from a neutrophil driven glomerulonephritis (although not T-cell driven disease) (Kitching et al., 2002) and when rats with glomerulonephritis were treated with

G-CSF, the condition was exacerbated (Futamura et al., 2002).

6.2.2 Detecting EPCs

Alongside the studies that implicate EPCs in endothelial reparative responses, there has been much interest in assessing their levels and functional capacity in different human diseases. In various conditions it has been demonstrated that EPCs may be impaired or reduced in number, which in the context of vasculitis could conceivably be contributing to the disease. As mentioned above, the exact phenotype of EPCs is unclear, however there are several methodologies by which putative EPCs have been enumerated, and in many studies these do appear to correlate with parameters relating to endothelial health.

To date putative EPCs have been enumerated using cell culture techniques and flow cytometry. Both techniques will now be described in detail.

6.2.2.1 Cultured EPCs

There are a number of different protocols described that use cell culture techniques to evaluate EPCs. A number of studies have evaluated EPC colonies, which can be

258

derived from PBMCs or CD34+ enriched populations. Colonies are clusters of cells, consisting of a central core of small rounded cells surrounded by spindle-like outgrowths (examples of which can be seen in section 2.6) and are formed after 5-7 days of culture in media optimised for endothelial cell growth. Cells are grown on an attachment scaffold such as fibronectin or collagen, and protocols may or may not involve a pre-plating step, whereby the non-adherent population is removed after 1-2 days and replated (in order to remove possible contamination by mature endothelial cells (Hill et al., 2003)).

Although colonies derived using different protocols look similar (Section 2.6, Figure

2.4ab), their phenotype and cellular composition may well vary. This is highlighted by two large studies from Hill et al and Xiao et al which have compared EPC colonies with cardiovascular risk factors. The EPC colonies in these studies were grown using different culture media - Hill et al used Medium 199 (Gibco) supplemented with 20% fetal calf serum (FCS) and antiobiotics, whilst Xiao et al used the commercially available endothelial optimised culture media EGM-2. While Hill et al counted colonies derived from non-adherent pre-plated cells, Xiao et al did not pre-plate their

EPCs. Hill et al, demonstrated an inverse correlation with Framingham risk factors of cardiovascular disease and a positive correlation with flow mediated dilatation (FMD)

(Hill et al., 2003), whilst Xiao et al in contrast observed no correlation with cardiovascular risk factors (Xiao et al., 2007b).

The methodology described by Hill et al, has led to a commercially available colony counting media – EndocultTM which has been widely used to study putative EPC colony forming units (known henceforth as Hill-CFU). The colonies formed consist of a

259

central core of haematopoeitic cells, which includes myeloid progenitor cells, monocytes, and T lymphocytes and macrophage/monocyte derived outgrowths (Yoder et al., 2007). While the outgrowths have been shown to express endothelial markers, produce nitric oxide and angiogenic growth factors they also have the ability to ingest bacteria – reflective of their myeloid lineage. Moreover, when implanted in vivo they do not incorporate in de novo vessels, they have low proliferative capacity and if continued in culture after colony formation do not form endothelial-like monolayers, thus it is hypothesized that these colonies are not endothelial progenitors per se but proangiogenic cells of a myeloid lineage (Yoder et al., 2007). Regardless of which cell population Hill-CFUs represent, it is apparent that they are a reflection of endothelial health (Hill et al., 2003), in a more recent study by Werner et al in individuals with coronary artery disease increased levels of Hill-CFUs were accompanied by decreasing risk of death from cardiovascular cause (Werner et al., 2005).

Whilst Hill-CFUs may reflect endothelial health, the colonies described by Xiao et al which did not reflect cardiovascular risk may in fact contain the elusive EPC population. Longer term continuation (20+ days) of (non pre-plated) adherent colony cultures in EGM-2 can lead to the development of endothelial-like monolayers, known as late outgrowth EPCs (Lin et al., 2000,Ingram et al., 2004), although not in all individuals. In contrast to Hill-CFUs this population of cells is highly proliferative, do not express haematopoetic markers such as CD45, CD14, or CD115 and do not ingest bacteria. More importantly when implanted in mice this population spontaneously forms blood vessels that incorporate with murine vessels (Yoder et al., 2007).

Using cell sorted umbilical cord blood, it has been demonstrated that late outgrowth

260

EPCs derive from a CD14 negative (Gulati et al., 2003) small CD34+ CD45 negative subset that expresses VEGFR2 but not CD133 (Timmermans et al., 2007), whereas the early colonies that form prior to late-outgrowth EPCs (the population examined by Xiao et al) are derived from a haematopoeitic source including CD34+CD45+ (Timmermans et al., 2007) and CD14+ cells (Gulati et al., 2003).

Other cell culture techniques that have been used to examine EPC number and function, include the counting of cells staining for both DiI-acetylated low-density lipoprotein and

Ulex europaeus (endothelial markers) after 4-5 days in EGM-2 culture; the assessment of cultured cells ability to adhere to endothelial cells (such as HUVEC); or their incorporation into HUVEC tubules in extracellular matrix formulations (MatrigelTM).

As with colony cultures a number of protocols are described and the exact cell populations which orchestrates these observations is unknown. Most experiments relating to EPCs adherence/incorporation with HUVEC have used the aforementioned adherent cell colonies cultured in EGM-2. As described above, despite the study by

Xiao et al demonstrating that these colonies do not reflect cardiovascular risk factors, this is the culture which can go on to form late-outgrowth EPCs.

The major restriction relating to EPC cell culture assays in children is the large volumes of blood required to obtain enough PBMCs for culture. As described in the methods chapter the initial cultures require a minimum of 5 x 106 cells per well for the Hill-CFU assay (preferably in duplicate) and 5 x 106 cells per well for culturing of outgrowth monolayers or for MatrigelTM. To perform each assay in triplicate would require 30 x

106 cells or a minimum of 30mls of blood (conceivably more in an immunosuppressed patient). Thus for the purposes of this thesis it was decided to focus on development of

261

flow cytometric assays for detection of EPCs in vasculitis patients and to examine the culture methods in healthy adults to determine if it was possible to reduce blood volume required for these assays to ultimately apply to the vasculitis cohort.

6.2.2.2 Flow cytometry of EPCs

The flow cytometric markers used for EPC enumeration are based on the hypothesis that a putative circulating endothelial progenitor (adult angioblast) would share a combination of cell surface markers found on both embryonic angioblasts and haematopoeitic cells (Asahara et al., 1997). The most commonly used are combinations of CD34 (found on angioblasts, haematopoietic progenitors and mature endothelium),

VEGFR2 (found on angioblasts and mature endothelium) and CD133 (found on angioblasts and haematopoietic progenitors), with or without other endothelial markers such as CD31, CD144 or CD146 (Hirschi et al., 2008). Some groups have defined

EPCs as cells which stain for CD34, VEGFR2 and CD133 (Kondo et al., 2004)while others have used combinations of only 2 markers as these represent a larger cell population: CD34+CD133+ (Allanore et al., 2007), CD34+VEGFR2+ (George et al.,

2006), CD34+CD144+ (Redondo et al., 2008), CD133+VEGFR2+ (Brunner et al.,

2008) and CD133+CD144+(Rehman et al., 2004) have all been called EPCs. While it is apparent that these populations do indeed correlate with a wide number of conditions associated with vascular injury or cardiovascular disease, it is unclear whether they reflect a true EPC population, and use of these different EPC definitions means that direct comparison between studies is problematic.

When CD34+VEGFR2+CD133+ cells are isolated and cultured they do not form colonies or late outgrowth endothelial-like cells (Case et al., 2007) and neither do they

262

correlate with Hill-CFUs in healthy individuals (George et al., 2006). However, as colonies have been shown to primarily consist of haematopoeitic cells, it is unsurprising that these methods are not comparable and that isolated populations do not form colonies alone. The work of Timmermans et al indicates that late outgrowth EPCs may derive from the CD34+ VEGFR2+ population and so this combination of cell surface markers may be indicative of a true EPC. However as these markers are both expressed on mature endothelium they cannot be used alone to define EPC populations and further phenotypic clarification is still required (Timmermans et al., 2007). For the purposes of this thesis, as both in vivo and in vitro evidence indicate that the CD34+ population contains true EPCs, subpopulations of CD34+ cells were examined using flow cytometry.

To date putative EPCs have been enumerated using flow cytometry from either whole blood or isolated PBMCs. While using whole blood has the benefit that only small volumes of sample are necessary, the assay must be performed on the day of sampling; requires lysis which may remove fragile cell populations; and has a much higher level of non specific antibody binding. In contrast isolated PBMCs can be stored (-80oC) allowing batch analysis, and as well as having overall less background noise neutrophils and red blood cells are removed without the need for lysis, thereby further increasing the signal to noise ratio. Putative EPCs are rare events in the circulation and therefore careful consideration of the flow cytometric methodology is required to optimize their detection. Optimisation of flow cytometry for rare event analysis has been described in detail in Chapters 3 and 4, the principles of which remain the same when considering

EPCs.

263

6.2.3 EPCs in health and disease

While the exact phenotype of EPCs remains unclear, the field of EPC research has expanded rapidly since the early publications and much work has been done examining these putative EPC populations in health and disease. As previously described care must be taken when comparing different studies as there is much heterogeneity in methods employed including cell surface markers examined. In general, EPC numbers

(as defined by both colony formation and by flow cytometry) appear to be lower in conditions associated with vascular risk and dysfunction including rheumatoid arthritis

(Egan et al., 2008a), SLE (Moonen et al., 2007), acute stroke (Chu et al., 2008), diabetes mellitus (Egan et al., 2008b), hypertension (Umemura et al., 2008) and in the presence of cardiovascular risk factors such as smoking (Kondo et al., 2004). EPCs can be mobilized using GM-CSF (granulocyte macrophage colony stimulating factor)(Takahashi et al., 1999), on exercise (Laufs et al., 2005) and in response to acute vascular injury such as acute myocardial infarction (Shintani et al., 2001) or in burns victims (Fox et al., 2008).

EPC number and function declines with age (Heiss et al., 2005,Vasa et al., 2001) and children have higher levels of EPCs and haematopoeitic stem cells (CD34+) than adults

(Jie et al., 2008). This observation may explain in part why EPC therapy in young healthy animal models has proved more effective than in older, sick human patients.

6.2.4 EPCs in vasculitis

In adults with PR3-ANCA+ WG the number of Hill-CFUs is lower in active disease than in remission or in healthy controls (Holmen et al., 2005). In a recent study in

AAV, Hill-CFUs were again found to be decreased in patients compared to healthy

264

controls, although not compared to patients on hemodialysis (without vasculitis) or with peripheral arterial occlusive disease (Zavada et al., 2008) and in contrast to the work of

Holmen et al, in this study remission inducing treatment did not increase Hill-CFUs

(perhaps due to the more heterogeneic population studied). In a recent flow cytometric study, de Groot et al demonstrated that adults with active AAV have comparable haematopoietic progenitor cells (CD34+) to healthy controls, however treatment increased numbers (de Groot et al., 2007). As discussed above changes in CD34+ cells are not necessarily representative of an EPC response, but could indicate repopulation of a number of haematopoetic lineages (perhaps not surprising after cytotoxic immunosuppression).

EPCs have only been examined in children with KD, although the study performed used a bead based isolation protocol, from which no EPCs were observed in healthy controls and only 11 out of 20 KD patients had measurable EPCs. It was therefore concluded that EPCs were higher in KD both in the acute and aubacute phase compared to healthy child controls (Nakatani et al., 2003).

6.3 Optimisation of flow cytometry for detection of

EPCs

6.3.1 Optimisation of antibody staining

3 colour flow cytometry, as described in section 2.6.2 was used to evaluate expression of combinations of the most commonly studied EPC markers: VEGFR2, CD133 and

CD34, with a variety of other cell surface markers. These included the constitutive

265

endothelial markers CD144 and CD146 (used in CEC enumeration), the monocyte marker, the pan leukocyte marker CD45 and the T cell marker CD3.

As EPCs are found in low levels in the circulation, there were a number of important considerations relating to rare event analysis which had to be taken into account when setting up this assay. Non-specific antibody binding was minimized via use of FC- receptor blocking agents. Non stained samples and fluorochrome matched isotype controls were used to define positivity. Antibodies were carefully titrated to determine optimum concentrations. In the case of cell surface markers found on the endothelium

(VEGFR2, CD144,CD146) the antibodies were titrated using HUVEC. All other antibodies were titrated using umbilical cord PBMCs as cord blood contains higher numbers of haematopoeitic progenitor cells than adult blood (Watt and Contreras,

2005).

6.3.2 Sample storage - fresh/frozen PBMCS vs whole blood

To further reduce non-specific antibody binding, isolated PBMCs as opposed to whole blood could be used – removing the need for red cell lysis. Another advantage of this approach is that it would allow sample collection and storage (-80oC) while the techniques were being optimized. In order to determine if EPC enumeration was dramatically affected by the freeze-thaw process, or was considerably different between whole blood samples and isolated PBMCs, blood was obtained from 5 healthy adult controls and the percentage of mononuclear cells which expressed CD34, CD133 and

VEGFR2 was determined (Figure 6.1a-d). As indicated in Figure 6.1 there were no significant differences in the proportion of mononuclear cells expressing CD34, CD133 or the combined triple positive cells, when whole blood, fresh PBMCs and frozen

266

PBMCs were compared. While there were some wide differences in values observed in individual patients, there were no overall directional trends observed. Interestingly

VEGFR2 staining was higher (although not significantly) in whole blood compared to

PBMCs (frozen or fresh, Figure 6.1c) however this did not influence the triple positive population of interest and was presumably due to non-specific binding of this antibody.

A larger study may have revealed significant differences, however as all samples from patients and controls were to be treated the same, the study examining EPCs in children with vasculitis was performed using isolated PBMCs, frozen and stored at -80oC.

Storage was frequently necessary as it was often not practical to assay EPCs on the same day as sampling, and this also allowed convenient batch analysis.

6.3.3 Flow cytometry – instrument settings and gates

6.3.3.1 FSC/SSC of CD34+CD133+VEGFR2+ cells

PBMCs isolated from umbilical cord blood were used to determine the FSC/SSC profile of cells expressing CD34, CD133 and VEGFR2 in order to determine the most appropriate gating strategy for enumeration of EPCs in isolated PBMCs. Previous studies have used FSC/SSC gates for viable cells based on the lymphocyte population alone (Vasa et al., 2001), an “extended lymphocyte gate” (Laufs et al., 2005) or both lymphocytes and monocytes combined (Rustemeyer et al., 2006) and therefore it was decided to investigate which strategy would include all putative EPCs

As Figure 6.2 indicates, when cells positive for CD34, VEGFR2 or CD133 were back- gated their FSC/SSC profiles fell within a similar region to lymphocytes (Figure 6.2a-f).

However it was notable that both the CD34 and CD133 populations extended beyond

267

Figure 6-1: % of mononuclear cells expressing EPC markers in whole blood, fresh or frozen isolated PBMCs

Figure 6.1a: Blood was obtained from 5 healthy adult controls. The percentage of mononuclear cells

(monocytes and lymphocytes expressing of CD34 (figure 6.1a), CD133 (Figure 6.1b), VEGFR2 (Figure

6.1c) and the combination of all three markers (figure 6.1d) was determined in whole blood, fresh

PBMCs and frozen PBMCs. There were no overall significant differences between fresh and frozen

PBMCs for expression for these markers and only VEGFR2 appeared higher (although not significantly) in whole blood from PBMCs (paired analysis vs fresh p=0.12, vs frozen p=0.14)

268

the lymphocyte population (highlighted in Figure 6.2g and h). Thus an “extended lymphocyte gate” might be most appropriate to detect these cells.

However this may be difficult to reproduce between different samples, particularly in vasculitis when the populations gated on may be activated or apoptotic which can influence their FSC/SSC profile (Darzynkiewicz et al., 1992). Of particular concern was defining the upper limit of the gate, between the lymphocyte and monocyte populations. Thus it was decided that a gate encompassing all PBMCs allowing EPCs to be expressed as a percentage of PBMCs would be a more standardized measure in this cohort (bold outline in Figure 6.2f).

6.3.3.2 CD34+ cell gating

Using a large viable cell gate has disadvantages, however, as cells from different populations may have slightly different staining intensities and there is a greater risk of non specific binding. As can be seen in Figures 6.2a and b CD34+ and CD133+

PBMCs isolated from cord blood formed discrete populations. The same was true for the CD34+ population in adult blood (data not shown). In some individuals while the

VEGFR2 staining was a much greater percentage of the PBMC gate than the percentage of CD34+ cells, there was much variability and discrete populations were not observed.

It was decided that in order to allow identification of the different populations of CD34+ cells, PBMCS could be gated first on their FSC SSC and then on CD34 positivity, to further reduce background noise. The staining of other putative EPC markers can then be examined in this population as demonstrated in Figure 6.3a-d.

269

Figure 6-2: FSC SSC characteristics of CD34+, CD133+ and VEGFR+ cells

Figure 6.2a-h: PBMCs isolated from umbilical cord blood were stained for CD34, CD133 and VEGFR2.

Cells positive for these markers (Figure 6.2a-c) were gated on, and the FSC/SSC characteristics of these populations determined by back-gating (Figure 6.2d-f). Figures 6.2g - h indicates the combined regions that these populations fall in (thin lines) as well as FSC SSC gate for EPC enumeration, both an extended lymphocyte gate (thick line, Figure 6.2g) and a broad mononuclear cell gate (thick line, Figure

6.2h).

270

Figure 6-3: EPC gating strategy

Figure 6.3: PBMCs were obtained from umbilical cord blood and stained for CD34, CD133 and

VEGFR2. Figure 6.3a: PBMCs were gated initially on their FSC SSC characteristics (R1); a large mononuclear cell gate incorporating the regions of high staining for CD34 and CD133 as described in

Figure 6.2. Figure 6.3b: Cells were then gated on CD34 as this marker formed a discrete cell population

(R2). Figure 6.3c: R1 and R2 were combined in order to determine the proportion of CD34 cells expressing CD133 and VEGFR2. The triple positive population could then be calculated. Figure 6.3d demonstrates the low nonspecific binding of the isotype control for CD34. CD34/VEGFR2/CD133 EPC count for this individual would be (0.69-0.04)% x 5.58/100 = 0.036%

271

6.3.3.3 How best to express EPC numbers?

In the majority of published studies EPCs have been enumerated as a percentage of a viable cell gate as described above. However, as previously discussed this will give varying results dependent on the size of the gate, making it difficult to compare different studies. Of perhaps more relevance in the vasculitis cohort though is the possibility that the total number of cells within the gated population is changed in disease – for example there may be increased white cell count at disease onset which may decrease with immunosuppressive cytotoxic treatment and thus using a percentage of this gate may no longer be representative of absolute EPC number (giving falsely elevated results in patients with low mononuclear cells).

Absolute enumeration of EPCs using counting beads in whole blood would be one way of avoiding this potential pitfall. However as a number of precious samples had already been stored prior to establishing the assay, and rare event analysis in whole blood requires extensive optimization (due to increased background noise and the possibility of lysis removing some of the cells of interest) it was decided to investigate this phenomena concurrently to the aforementioned widely used methodology. This was done by estimating absolute EPC number and evaluating an alternative enumeration strategy which may reduce the effect of this potential error as described below

(summarized in Table 6.1).

Estimating absolute EPC number

As research bloods were taken from vasculitis patients in the study at the same time as routine clinical tests, full blood count as determined by the hospital laboratory could be used to estimate absolute EPC number. The dual platform ISHAGE (International

272

Society of Hematotherapy and Graft Engineering) protocol for enumeration of CD34+ cells is routinely performed in hospital laboratories. This utilizes the leukocyte count determined using an automated haematology analyzer and the % of nucleated white blood cells (CD45+, including CD45dim cells) expressing CD34, to determine absolute

CD34 count in a sample, either of peripheral blood, bone marrow or leukapheresis product (Gajkowska et al., 2006). It is important to note that applying this principle to the patient samples in his study (as these have undergone density gradient separation to isolate PBMCs) does not give an exact measure of CD34+ cells, but can be used as a crude tool to estimate EPC levels and determine if trends observed using a percentage of the mononuclear gate are reflected when the absolute number of leukocytes is taken into account. Thus validating the use of a percentage of a viable cell gate as an expression of EPC number in this cohort.

Alternative gating

Expressing EPCs as a percentage of the large mononuclear cell gate, gives a very small value (in healthy adult controls values were <0.1%, Figure 6.1d), which as discussed above could be influenced by changes in cell numbers. Another approach which has been utilized in T cell biology when examining rare populations is to express the population of interest as a percentage of a much more defined, relevant population. For example, the IL-17 producing T cell subset (Th17 cells) is expressed as a percentage of

CD4+ T cells (Nistala et al., 2008), not as an absolute number or as a percentage of the larger total T cell population. As the CD34+ population can reconstitute all blood cell lineages (Cheng et al., 1996) and is thought to contain the “true EPC” population, changes in expression of EPC markers in this cell population, may be an informative indicator of a reparative responses - reflecting changes in haematopoetic progenitor

273

differentiation. As can be seen in Figure 6.3, while cells expressing CD34, CD133 and

VEGFR2 are only 0.036% of the mononuclear cell gate, they make up 5.68% of the

CD34 population. The advantages and disadvantages associated with each of the aforementioned methodologies for EPC enumeration is summarised in Table 6.1.

Table 6-1: How best to enumerate and express EPC number

Method Advantages Disadvantages Only a small volume of blood Much background noise required (<1ml) Whole blood – absolute Lysis is necessary - may enumeration, using counting Can get an absolute value for remove fragile cell beads or ISHAGE dual EPCs per ml of blood populations platform protocols

(Gajkowska et al., 2006) Samples must be processed quickly, cannot be stored PBMCs can be isolated and Variable gating strategy has stored for batch analysis been described in literature

PBMC – expressing EPCs as If same gate is used, studies Changes in total % of viable cell gate. can be easily compared mononuclear cell number with therapeutic intervention could influence results PBMCs can be isolated and This methodology has not stored for batch analysis been described in the literature for EPC Gate is less variable enumeration PBMCs – expressing EPCs

as a % of CD34 cells Changes in total mononuclear cell number will have less of an affect on population of interest

For the purposes of this thesis EPCs were enumerated from PBMCs isolated from patient using the two enumeration methodologies described in the above table.

Absolute EPC number was also estimated from the values obtained in PBMCs to determine if either of these two enumeration methodologies mirrored the trends observed in absolute number.

274

6.4 EPCs in vasculitis

6.4.1 Patients and Methods

6.4.1.1 Patients

15 children with primary systemic vasculitis were studied. These patients were the same individuals described in Chapters 3, 4 and 5, allowing EPCs to be compared with

CECs, MPs and growth factors involved in angiogenic/vasculogenic responses. The patients studied included 6 males and 9 females of median age 9.7 years (range 1.0.-

15.3 years). The vasculitis type was polyarteritis nodosa (PAN) n=6, Wegener’s granulomatosis (WG) n=4, Kawasaki disease (KD) n=1, unclassified n=3. Disease activity was assessed using a modified paediatric BVAS as previously described. 18 healthy child controls (4 males, 14 female, median age 8.4, range 1.5-14.6), 4 febrile controls (1 male, 3 females with a variety of autoimmune or other inflammatory disorders without the presence of vasculitis, median age 7.8, range 1.5-12) and 5 healthy adult controls were also recruited (1 male, 4 females, age range 24-30 years). As with the vasculitis patients, the control individuals were the same as those studied in

Chapters 3 and 4. Blood was processed immediately after collection as described in

Chapter 2.6.

6.4.1.2 Methods

FC was performed as described previously and in section 2.6.2. In order to allow enough flow cytometric cell events for accurate rare event analysis, the number of markers examined was varied so that in individuals where only small numbers of cells were obtained, meaningful data could still be acquired (Flow cytometric staining plans are listed in Appendix 3). Putative EPCs were enumerated as a % of total PBMC population and as a % of CD34+ cells to determine which was the most appropriate

275

measure in this cohort. Absolute EPC number was estimated from whole blood counts obtained from the hospital laboratory, to determine if trends observed in individual patients were reflective of absolute EPC number.

6.4.1.3 Statistics

The Mann-Whitney U test or Wilcoxon signed rank test (paired samples) were used to compare EPCs between patient groups (active vs inactive vasculitis – paired samples) and controls (unpaired analysis). As only a small number of patients were studied – the majority of the active-inactive vasculitis data was paired and therefore there was not enough data to exclude paired events for comparative unpaired analysis between active and inactive disease. Spearman rank correlation coefficiants were used to determine correlation. All statistical analyses were performed in SPSS version 15.0. Statistical significance was set at p<0.05. In all data graphs patient classification was denoted using colour coding as indicated below:

6.4.2 Results

6.4.2.1 CD34+ cells

Kruskall Wallis testing indicated no overall significant differences in the percentage of

PBMCs expressing CD34 between test groups (p=0.3). When individual analysis was performed the percentage of PBMCs expressing CD34, was not significantly different between 14 patients with active vasculitis (median value 0.198%) and 18 healthy child 276

controls (median value 0.161%) or 4 febrile controls (median value 0.18%). The percentage of PBMCs expressing CD34 was however significantly lower in 5 healthy adult controls (median value 0.04%) compared to 18 healthy children (p=0.03) or 14 children with active vasculitis (p=0.04). There were no significant differences between

9 patients with inactive vasculitis (median value 0.11%) and any of the control groups.

(Figure 6.4a).

In 9 individual patients, there were no significant differences in the percentage of

PBMCs expressing CD34 between active vasculitis and remission (median time to remission 4 months) (Figure 6.4b). When 10 individual patients were followed up longitudinally from disease onset and initiation of treatment for up to 15 months there were no significant differences in CD34+ cells between onset and median followup of 3 and 12 months (Figure 6.4c).

In order to determine if the % of PBMCs expressing CD34 was comparable to the absolute numbers of CD34+ cells the changes in absolute CD34+ cell number was estimated as described above in section 6.6.3. (Figure 6.5a and b). Line colour reflects individual patients disease classification and the different line styles

(solid/dashed/dotted) represent individual patients. Notably, whilst a number of patients’ changes in absolute CD34 number reflected the changes in % of PBMCs expressing CD34, in some individuals the trends went in different directions (3/10, although only 1 – the red dashed line was considerably different), indicating that % of

PBMCs may not be the most appropriate way of expressing EPC count in this cohort.

At median followup of 3 and 12 months, there were no significant differences observed for either of these measures of CD34 count from pretreatment levels.

277

Figure 6-4:CD34+ cells in vasculitis

Figure 6.4a: Kruskall Wallis testing indicated no overall significant differences between test groups

(p=0.3). Individual analysis revealed there were no significant differences in the % of PBMCs expressing CD34 between 15 patients with active vasculitis and 4 febrile controls or 18 healthy child controls. However 5 adults had significantly lower CD34% compared to 15 children with active vasculitis (p=0.04) and 18 healthy child controls (p=0.03). There were no significant differences between 9 patients in remission and these controls groups. Figure 6.4b: In 9 individual patients, there were no significant difference in CD34+ PBMCs during active or inactive disease, median time to remission 4 months (p=0.08).

278

Figure 6-5: CD34+ cells in vasculitis – responses to treatment

Figure 6.5a-b: Changes in the % of PBMCs expressing CD34 over time from disease onset was compared to changes in estimated absolute numbers of CD34+ cells. Line colour reflects individual patients disease classification (as before) and the different line styles (solid/dashed/dotted) represent individual patients. Of the 10 patients followed longitudinally, the changes observed in 7 individuals were the same using both enumeration methods. At median followup of 3 and 2 months, there were no significant differences using either technique when compared to pretreatment levels.

279

6.4.2.2 CD34+CD133+ cells

PBMCs expressing both CD34 and CD133, could be indicative of an early progenitor population (although not necessarily one of endothelial lineage). Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing both CD34 and CD133 between test groups (p=0.13). Individual analysis, however revealed that the percentage of CD34+CD133+ PBMCs were significantly higher in 15 patients with active vasculitis (median value 0.099%) compared to 18 child controls (median value,0.037%, p=0.017), although not 4 febrile controls (median value

0.056%) or 5 healthy adult controls (median value 0.05%). There were no significant differences between 9 patients in remission (median value 0.057%) and these control groups, nor was there any difference using unpaired analysis (although paired data was included in this due to small numbers) between active and inactive vasculitis (p=0.08,

Figure 6.6a). In 9 individual patients, there appeared to be a trend towards a higher % of PBMCs expressing CD34+CD133+ during active disease compared to remission, although this had not reached statistical significance (p=0.054, median time to remission

4 months) (Figure 6.6b).

When the CD34+CD133+ population was expressed as the percentage of CD34+ cell expressing CD133, Kruskall Wallis testing indicated an overall significant difference between populations (p=0.006). Individual analysis revealed that the percentage of

CD34+ cells expressing CD133 was significantly higher in patients with active vasculitis (median value 53.5%) compared to 18 child controls (median value 28%, p=0.005), although not 4 febrile controls (median value 39%) or 4 healthy adult controls

(median value 27.5%). There was no significant difference between 9 patients in remission (median value 39%) and these control groups. Unpaired analysis (again

280

paired data was included in this due to small numbers) between active and inactive vasculitis revealed significantly higher CD133+CD34+ cells in patients with active disease (p=0.03, Figure 6.6c). This finding was also true when paired analysis in 9 individual patients was performed (p=0.04, median time to remission 4 months).

In order to determine if the % of PBMCs expressing CD34 and CD133 was comparable to the proportion of CD34+ PBMCs expressing CD133 or the absolute numbers of

CD34+ CD133+ cells, the changes observed in individual patients using these methods of expressing EPC are shown in figure 6.7a-c. Of the 10 patients followed up from disease onset, the changes in the % of PBMCs expressing CD34 and CD133 were mirrored in 7/10 individuals by the changes in the % of CD34+ expressing CD133 and

8/10 individuals by changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 9/10 individuals by the proportion of

CD34+ cells expressing CD133.

When 10 patients were followed up from disease onset, the only significant difference observed was between pretreatment levels and median followup of 12 months for the absolute estimate of CD34+CD133+ cells. There were no other significant differences in CD34+CD133+ between onset and median followup of 3 and 12 months for any of the three ways of expressing cell numbers (Figure 6.5c). P values for percentage of

PBMCs were p=0.25 and p= 0.22, for the percentage of CD34+ cells p = 0.054 and p =

0.16 and for the estimate of absolute number p=0.1, p=0.03 respectively.

281

Figure 6-6: CD34+CD133+ cells in vasculitis

Figure 6.6a: Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34 and CD133 between test groups (p=0.13). Individual analysis however revealed that

CD133+CD34+ PBMCs were significantly higher in 15 patients with active disease compared to 18 healthy child controls (p=0.017) although not 4 febrile controls or 5 healthy adult controls. There were no significant differences between 9 patients in remission and these controls groups. Figure 6.6b: In 9 individual patients, there were no significant difference in CD34+CD133 PBMCs during active or inactive disease, median time to remission 4 months

(p=0.054). Figure 6.6c: Kruskall Wallis testing indicated significantly different proportions of CD34+ cells expressing CD133 between test groups (p=0.006). Individual analysis revealed that the CD133+ proportion of

CD34+ PBMCs was significantly higher in 15 patients with active disease compared to 18 healthy child controls

(p=0.005) and 9 patients in remission (p=0.03) although not 4 febrile controls or 18 healthy adult controls. There were no significant differences between 9 patients in remission and these controls groups. Figure 6.6d: Paired analysis in 9 individual patients, showed that the CD133+ proportion of CD34+ PBMCs was significantly higher during active compared to remission (p=0.04, median time to remission 4 months).

282

Figure 6-7: CD34+CD133+ cells in vasculitis – responses to treatment

Figure 6.7a-c: Changes in the % of PBMCs expressing CD34 and CD133 (Figure 6.7a), the proportion of CD34 cells expressing CD133 (Figure 6.7b) and the estimated absolute numbers of CD34+CD133+ cells (Figure 6.7c) were observed over time from disease onset in individual patients. Line colour reflects individual patients disease classification (as before) and the different line styles

(solid/dashed/dotted) represent individual patients. Of the 10 patients followed longitudinally, the changes in the % of PBMCs expressing CD34 and CD133 were mirrored by 7/10 individuals in the changes in the proportion of CD34+ expressing CD133 and 8/10 individuals in changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 9/10 individuals by changes in the proportion of CD34+ cells expressing CD133. When values at median followup of 3 and

12 months, were compared to disease onset, a significant difference was only observed at 12 months when the estimate of absolute number was considered (p=0.03).

283

6.4.2.3 CD34+CD133+VEGFR2+ cells

PBMCs which express CD34, VEGFR2 and CD133, has been the most widely published description of EPCs. In this study Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34, VEGFR2 and CD133 between test groups (p=0.19). Individual analysis, however revealed that the percentage of CD34+VEGFR2+CD133+ PBMCs was significantly higher in 15 patients with active vasculitis (median value 0.027%) compared to 18 child controls

(median value 0.016%, p=0.04), although not 4 febrile controls (median value 0.027%) or 5 healthy adult controls (median value 0.017%). There were no significant differences between 9 patients in remission (median value 0.023%) and these control groups nor was there any difference using unpaired analysis (although paired data was included in this due to small numbers) between active and inactive vasculitis (p=0.3,

Figure 6.8a). In 9 individual patients, paired analysis also revealed no significant difference in the percentage of CD34+VEGFR2+CD133+ PBMCs between active vasculitis and remission (p=0.4, median time to remission 4 months, Figure 6.8b).

When the CD34+VEGFR2+CD133+ population was expressed as the percentage of

CD34+ cell expressing VEGFR2 and CD133, Kruskall Wallis testing failed to reach significance (p=0.052). Individual analysis however revealed that the proportion of

CD34+ cells expressing VEGFR2 and CD133 was significantly higher in patients with active vasculitis (median value 28.5%) compared to 18 child controls (median value

11%, p=0.007), although not 4 febrile controls (median value 12%) or 5 healthy adult controls (median value 20%). There was no significant difference between 9 patients in remission (median value 15%) and these control groups. Unpaired analysis (again paired data was included due to small numbers) between active and inactive

284

Figure 6-8: CD34+VEGFR2+CD133+ cells in vasculitis

Figure 6.8a: Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34, VEGFR2 and CD133 between test groups (p=0.2). Individual analysis however revealed that

CD34+VEGFR2+CD133+ PBMCs were significantly higher in 15 patients with active disease compared to 18 healthy child controls (p=0.04) although not 4 febrile controls or 5 healthy adult controls. There were no significant differences between 9 patients in remission and these control groups. Figure 6.8b: In 9 individual patients, there were no significant difference in CD34+VEGFR2+CD133+ PBMCs during active or inactive disease, median time to remission 4 months (p=0.4). Figure 6.8c: Kruskall Wallis testing indicated a significantly higher proportion of

CD34+ cells expressing VEGFR2 and CD133 between test groups (p=0.05). Individual analysis revealed that the

VEGFR2+CD133+ proportion of CD34+ PBMCs was significantly higher in 15 patients with active disease compared to 18 healthy child controls (p=0.004) although not 9 patients in remission, 4 febrile controls or 18 healthy adult controls. There were also no significant differences between 9 patients in remission and any control group.

Figure 6.8d: Paired analysis in 9 individual patients, showed that the VEGFR2+CD133+ proportion of CD34+

PBMCs was not significantly different between active vasculitis and remission (p=0.5, median time to remission 4 months).

285

Vasculitis revealed no significant differences (p=0.3, Figure 6.8c). This finding was also true when paired analysis in 9 individual patients was performed (p=0.5, median time to remission 4 months, Figure 6.8d).

In order to determine if the % of PBMCs expressing CD34, VEGFR2 and CD133 was comparable to the proportion of CD34+ PBMCs expressing VEGFR2 and CD133 or the absolute numbers of CD34+ VEGFR2+CD133+ cells, the changes observed in individual patients using the methods of expressing EPC are shown in Figure 6.9a-c.

Of the 10 patients followed up form disease onset, the changes in the % of PBMCs expressing CD34, VEGFR2 and CD133 were mirrored in 5/10 individuals by the changes in the % of CD34+ expressing VEGFR2 and CD133 and 7/10 individuals by changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 6/10 individuals by the proportion of CD34+ cells expressing

VEGFR2 and CD133.

When 10 patients were followed up from disease onset, no significant differences were observed between pretreatment levels and median followup of 3 or 12 months for for any of the three ways of expressing cell numbers (Figure 6.9a-c). P values for the percentage of PBMCs were p = 0.6 and p = 0.3, for the percentage of CD34+ cells were p=0.2 and p=0.3 and for the estimate of absolute number p=0.1, p=0.15 respectively.

6.4.2.4 CD34+ VEGFR2+ cells

As discussed in the introduction to this chapter some groups consider only the

CD34+VEGFR2+ population when describing EPCs using flow cytometry. Using

Kruskall Wallis testing revealed that the percentage of PBMCs expressing CD34 and

VEGFR2 was not significantly different between test groups (p=0.5). 286

Figure 6-9: CD34+VEGFR2+CD133+ cells in vasculitis

Figure 6.9a-c: Changes in the % of PBMCs expressing CD34,VEGFR2 and CD133 (Figure 6.9a), the proportion of

CD34 cells expressing VEGFR2 and CD133 (Figure 6.7b) and the estimated absolute numbers of

CD34+VEGFR2+CD133+ cells (Figure 6.7c) were observed over time from disease onset in individual patients.

Line colour reflects individual patients disease classification (as before) and the different line styles

(solid/dashed/dotted) represent individual patients. Of the 10 patients followed longitudinally, the changes in the % of PBMCs expressing CD34, VEGFR2 and CD133 were mirrored by 5/10 individuals in the changes in the proportion of CD34+ expressing CD133 and 7/10 individuals in changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 6/10 individuals by changes in the proportion of

CD34+ cells expressing VEGFR2 and CD133. When values at median followup of 3 and 12 months, were compared to disease onset, no significant differences were observed for any of the methods of enumeration.

287

On individual analysis no significant differences were observed between 15 patients with active vasculitis (median value 0.04%) compared to 18 child controls (median value 0.05%), 4 febrile controls (median value 0.09%) or 5 healthy adult controls

(median value 0.03%). There was no significant difference between 9 patients in remission (median value 0.05%) and these control groups. Unpaired analysis between active vasculitis and patients in remission also revealed no significant differences

(Figure 6.10a). A finding repeated when data pairing was considered in the analysis

(Figure 6.10b).

When CD34+VEGFR2+ cells were expressed as the proportion of CD34+ cells expressing VEGFR2, Kruskall Wallis testing also revealed no overall significant differences between tesr groups (p=0.3). Individual analysis again revealed no significant differences between patients with active vasculitis (median value 43%) or those in remission (median value 18%) and any of the control groups (median values adults = 21%, healthy children = 29%, febrile controls = 37%) . Unpaired analysis between active vasculitis and patients in remission also revealed no significant differences, however when paired analysis was performed although not yet significant there appeared to be a trend with 7/9 patients having lower levels in disease remission

(p=0.07).

In order to determine if the % of PBMCs expressing CD34 and VEGFR2 was comparable to the proportion of CD34+ PBMCs expressing VEGFR2 or the absolute numbers of CD34+ VEGFR2+ cells, the changes observed in individual patients using the methods of expressing EPCs are shown in Figure 6.11a-c. Of the 10 patients followed up form disease onset, the changes in the % of PBMCs expressing CD34 and

288

Figure 6-10: CD34+VEGFR2+ cells in vasculitis

Figure 6.10a: Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34 and VEGFR2 between test groups (p=0.5). Individual analysis also revealed no significant differences between 14 patients with active vasculiits and 18 healthy child controls, 4 febrile controls or 5 healthy adult controls. There were no significant differences between 9 patients in remission and these control groups.

Figure 6.10b: In 9 individual patients, there was a trend towards a decrease in CD34+VEGFR2+ PBMCs during remission, however this had not reached significance (p=0.07, median time to remission 4 months). Figure 6.10c:

Kruskall Wallis testing indicated no overall significant difference in the proportion of CD34+ cells expressing

VEGFR2 between test groups (p=0.3). Individual analysis also revealed no significant differences between 14 patients with active vasculiits and 18 healthy child controls, 4 febrile controls or 5 healthy adult controls. There were no significant differences between 9 patients in remission and these control groups. Figure 6.10: Paired analysis in 9 individual patients, showed that the VEGFR2+ proportion of CD34+ PBMCs expressing VEGFR2 was not significantly different between active vasculitis and remission (p=0.15, median time to remission 4 months).

289

VEGFR2 were mirrored in 6/10 individuals by the changes in the % of CD34+ expressing VEGFR2 and in 8/10 individuals by changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 8/10 individuals by the proportion of CD34+ cells expressing VEGFR2. When 10 patients were followed up from disease onset, no significant differences were observed between pretreatment levels and median followup of 3 or 12 months for any of the three ways of expressing cell numbers (Figure 6.11a-c). P values for the percentage of PBMCs were p = 0.5 and p = 0.3, for the percentage of CD34+ cells were p = 0.1 and p= 0.5 and for the estimate of absolute number p=0.2, p=0.15 respectively.

6.4.2.5 CD34+CD144+ cells

Another putative EPC combination described in a number of studies is CD34+CD144+ cells. Using Kruskall Wallis testing revealed that the percentage of PBMCs expressing

CD34 and CD144 was not significantly different between test groups (p=0.13).

However, individual analysis revealed that 14 patients with active vasculitis (median value 0.05%) had significantly higher CD34+CD144+ PBMCs compared to 17 child controls (median value 0.019%, p=0.013), although not 4 febrile controls (median value

0.09%) or 5 healthy adult controls (median value 0.03%). There was no significant difference between 8 patients in remission (median value 0.05%) and these control groups. Unpaired analysis between active vasculitis and patients in remission also revealed no significant differences, a finding repeated when data pairing was considered in the analysis (Figure 6.12b). When CD34+CD144+ cells were expressed as the proportion of CD34+ cells expressing CD144, Kruskall Wallis testing also revealed no overall significant differences between test groups (p=0.17).

290

Figure 6-11: CD34+VEGFR2+cells in vasculitis – responses to treatment

Figure 6.11a-c: Changes in the % of PBMCs expressing CD34 and VEGFR2 (Figure 6.11a), the proportion of CD34 cells expressing VEGFR2 (Figure 6.11b) and the estimated absolute numbers of

CD34+VEGFR2 cells (Figure 6.11c) were observed over time from disease onset in individual patients.

Line colour reflects individual patients disease classification (as before) and the different line styles

(solid/dashed/dotted) represent individual patients. Of the 10 patients followed longitudinally, the changes in the % of PBMCs expressing CD34, and VEGFR2 were mirrored by 6/10 individuals in the changes in the proportion of CD34+ expressing VEGFR2 an in 8/10 individuals in changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 8/10 individuals by changes in the proportion of CD34+ cells expressing VEGFR2 and CD133. When values at median followup of 3 and 12 months, were compared to disease onset, no significant differences were observed for any of the methods of enumeration. 291

Individual analysis revealed no significant differences between patients with active vasculitis (median value 38%) and any of the control groups (median values adults =

35%, healthy children = 34%, febrile controls = 9%) or those in remission (median value 16%) (Figure 6.12c). Unpaired analysis between active vasculitis and patients in remission revealed a significantly higher proportion of CD144+ cells during active disease (p=0.026), however when paired analysis was performed on 9 individual patients, the data was no longer significant (p=0.19, Figure 6.12d).

In order to determine if the % of PBMCs expressing CD34 and CD144 was comparable to the proportion of CD34+ PBMCs expressing CD144 or the absolute numbers of

CD34+ CD144+ cells, the changes observed in individual patients using the methods of expressing EPCs are shown in Figure 6.13a-c. Of the 9 patients followed up form disease onset, the changes in the % of PBMCs expressing CD34 and CD144 were mirrored in 6/9 individuals by the changes in the % of CD34+ expressing CD144 and in

6/9 individuals by changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 7/9 individuals by the proportion of

CD34+ cells expressing CD144.

When 9 patients were followed up from disease onset, no significant differences were observed between pretreatment levels and median followup of 3 or 12 months for any of the three ways of expressing cell numbers (Figure 6.13a-c). P values for the percentage of PBMCs were p = 0.6 and p = 0.7, for the percentage of CD34+ cells were p = 0.5 and p= 0.8 and for the estimate of absolute number p=0.3, p=0.8 respectively.

292

Figure 6-12: CD34+CD144+ cells in vasculitis

Figure 6.12a: Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34 and CD144 between test groups (p=0.13). However individual analysis revealed that 14 patients with active vasculitis had significantly higher CD34+CD144+ cells compared to 17 healthy child controls (p=0.013), although not 4 febrile controls or 5 healthy adult controls. There were no significant differences between 9 patients in remission and these control groups. Figure 6.12b: In 8 individual patients, there was no significant difference between active vasculitis and remission (p=0.07, median time to remission 4 months). Figure 6.12c: Kruskall Wallis testing indicated no overall significant difference in the proportion of CD34+ cells expressing CD144 between test groups (p=0.17). However individual analysis revealed that 14 patients with active vasculitis had significantly higher CD34+CD144+ cells compared to 8 patients in remission, although not 17 healthy child controls, 4 febrile controls or 5 healthy adult control. Figure 6.10: In 8 individual patients, there was no significant difference between active vasculitis and remission (p=0.19, median time to remission 4 months).

293

Figure 6-13: CD34+CD144+ cell in vascultis – response to treatment

Figure 6.13a-c: Changes in the % of PBMCs expressing CD34 and CD144 (Figure 6.13a), the proportion of CD34 cells expressing CD144 (Figure 6.13b) and the estimated absolute numbers of CD34+CD144+ cells (Figure 6.13c) were observed over time from disease onset in individual patients. Line colour reflects individual patients disease classification (as before) and the different line styles (solid/dashed/dotted) represent individual patients. Of the 9 patients followed longitudinally, the changes in the % of PBMCs expressing CD34 and CD144 were mirrored by 6/9 individuals in the changes in the proportion of CD34+ expressing VEGFR2 an in 6/9 individuals in changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 7/9 individuals by changes in the proportion of CD34+ cells expressing VEGFR2 and CD133. When values at median followup of 3 and 12 months, were compared to disease onset, no significant differences were observed for any of the methods of enumeration.

294

6.4.2.6 CD34+CD144+CD133+ cells

As CD34 and CD144 are both found on mature ECs, further discrimination is necessary and thus this population was examined in combination with the progenitor marker

CD133. Using Kruskall Wallis testing revealed that the percentage of PBMCs expressing CD34 and CD144 was not significantly different between test groups

(p=0.17). Individual analysis also revealed no significant differences between 14 patients with active vasculitis (median value 0.03%) and any of the control groups (17 healthy children, median value = 0.01%, 4 febrile controls, median value 0.01%, 5 healthy adult controls, median value 0.02%). 8 patients in remission (median value

0.01%) appeared to have lower CD34+CD144+CD133+ cells than patients with active vasculitis (unpaired analysis, p=0.07) and healthy child controls although this had not reached significance (p=0.051, Figure 6.14a). When paired analysis was performed between active and inactive vasculitis in 8 individual patients there was no significant difference observed (p=0.1, Figure 6.14b).

When CD34+CD144+CD133+ cells were expressed as the proportion of CD34+ cells expressing CD144 and CD133, Kruskall Wallis testing also revealed no overall significant differences between test groups (p=0.12). Individual analysis revealed no significant differences between patients with active vasculitis (median value 18.6%) and any of the control groups (median values adults = 17%, healthy children = 14%, febrile controls = 9%). There was also no significant difference between patients in remission

(median value 8.5%) and these control groups, although values appeared lower in patients in remission compared to healthy child controls, this failed to reach significance

(p=0.07, Figure 6.14c). Unpaired analysis between active and inactive disease revealed patients with active disease had significantly higher CD34+CD144+CD133+ cells than

295

Figure 6-14 :CD34+CD144+CD133+ cells in vasculitis

Figure 6.14a: Kruskall Wallis testing indicated no overall significant differences in the percentage of PBMCs expressing CD34, CD144 and CD133 between test groups (p=0.17). Individual analysis also revealed no significant difference between 14 patients with active vasculitis and 17 healthy child controls, 4 febrile controls or 5 healthy adult controls. There were no significant differences between 8 patients in remission and these control groups.

Figure 6.14b: In 8 individual patients, there was no significant difference between active vasculitis and remission

(p=0.1, median time to remission 4 months). Figure 6.14c: Kruskall Wallis testing indicated no overall significant difference in the proportion of CD34+ cells expressing CD144 and CD133 between test groups (p=0.12). However individual analysis revealed that 14 patients with active vasculitis had significantly higher CD34+CD144+CD133+ cells compared to 8 patients in remission, although not 17 healthy child controls, 4 febrile controls or 5 healthy adult control. There were no significant differences between 8 patients in remission and these control groups Figure

6.14d: In 8 individual patients, there was no significant difference between active vasculitis and remission (p=0.1, median time to remission 4 months).

296

patients in remission (p=0.023), although when paired analysis was performed on 8 individual patients, the data was no longer significant (p=0.1, Figure 6.14d).

In order to determine if the % of PBMCs expressing CD34, CD144 and CD133 was comparable to the proportion of CD34+ PBMCs expressing CD144 and CD133 or the absolute numbers of CD34+ CD144+CD133+ cells, the changes observed in individual patients using these methods of expressing EPCs are shown in Figure 6.13a-c. Of the 9 patients followed up from disease onset, the changes in the % of PBMCs expressing

CD34 CD144 and CD133 were mirrored in 5/9 individuals by the changes in the % of

CD34+ expressing CD144 and CD133 and in 8/9 individuals by changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in

6/9 individuals by the proportion of CD34+ cells expressing CD144 and CD133 (Figure

6.15a-c)

When 9 patients were followed up from disease onset, no significant differences were observed between pretreatment levels and median followup of 3 or 12 months for any of the three ways of expressing cell numbers (Figure 6.15a-c). P values for the percentage of PBMCs were p = 0.5 and p = 0.9, for the percentage of CD34+ cells were p = 0.3 and p= 0.25 and for the estimate of absolute number p=0.4, p=0.4 respectively.

6.4.2.7 Other EPC related CD34 subpopulations

In a small number of individuals, the EPC markers CD34, CD133, VEGFR2 and

CD144 were examined in combination with CD14, CD45, CD3 and CD146 to further establish the origins of CD34+ progenitor cells.

297

Figure 6-15: CD34+CD144+CD133+ cells in vasculitis – response to treatment

Figure 6.15a-c: Changes in the % of PBMCs expressing CD34 CD144 and CD133 (Figure 6.15a), the proportion of

CD34 cells expressing CD144 and CD133 (Figure 6.15b) and the estimated absolute numbers of

CD34+CD144+CD133+ cells (Figure 6.15c) were observed over time from disease onset in individual patients.

Line colour reflects individual patients disease classification (as before) and the different line styles

(solid/dashed/dotted) represent individual patients. Of the 9 patients followed longitudinally, the changes in the % of

PBMCs expressing CD34,CD144 and CD133 were mirrored by 5/9 individuals in the changes in the proportion of

CD34+ expressing CD144 and CD133 in 8/9 individuals in changes in the estimate of absolute number. The changes in the estimate of absolute number were mirrored in 6/9 individuals by changes in the proportion of CD34+ cells expressing CD144 and CD133. When values at median followup of 3 and 12 months, were compared to disease onset, no significant differences were observed for any of the methods of enumeration. 298

The discrete CD34+ population gated on was frequently negative for CD45 or had only dim CD45 staining (Figure 6.16a). Only a small proportion of CD34 cells expressed

CD14 (Figure 6.16b) or the EC marker CD146 (Figure 6.16c). CD146 as often at >1% in the blood, the majority of cells expressing it however were CD3+ T cells (Figure

6.16d). In the small number of patients who were examined for CD34+CD14+CD133+

PBMCs, 6 patients with active vasculitis had a significantly higher percentage of this monocytic lineage cell expressing progenitor markers, compared to 16 healthy child controls (p=0.035) (Figure 6.17a). There were no significant differences for

CD14+CD34+ (figure 6.17b) or CD146+CD34+ (Figure 6.17c) compared to healthy child controls. Figure 6.17a-c only expresses these cell populations as percentage of

PBMCs, other ways of expressing cell populations were also not significant.

299

Figure 6-16:Further CD34+ staining

Figure 6.16a and b: The discrete CD34+ population, was frequently negative for CD45 (Figure

6.16a) and CD14 (Figure 6.16b) or only exhibited dim staining of either of these markers. Figure

6.16c: The discrete CD34 population was also predominantly negative for the endothelial marker

CD146. The majority of CD146+ cells were CD3 positive T cells. (Figure 6.16d).

300

Figure 6-17: CD34+CD14+CD146+ cells in vasculitis

Figure 6.17a: The % of PBMCs expressing CD34, CD14 and CD133 were significantly higher in 6 patients with active vasculitis compared to 16 healthy child controls, although not 4 healthy adult controls (not enough data for comparison with febrile controls (n=2)). 6 patients in remission were not significantly different from these control groups Figure 6.17b: The % of PBMCs expressing

CD34 and CD14 were not significantly different between 6 patients with active vasculitis or 6 patients in remission compared to16 healthy child controls or 4 healthy adult controls (not enough data for comparison with febrile controls (n=2)). Figure 6.17c: The % of PBMCs expressing CD34 and CD146 were not significantly different between 4 patients with active disease or 4 patients in remission compared to 4 healthy child controls.

301

6.4.3 Correlation of EPCs with disease activity

Spearman Rank correlation coefficiant analysis was performed between the different combinations of EPCs markers (measured as % of PBMCs and % of the CD34+ population) and the traditional markers of active disease; BVAS, CRP, ESR, platelet count, albumin, white cell count as well as the previously demonstrated novel markers of endothelial injury and cellular activation CEC count, EMPs and PMPs and the 4 growth factors assessed in Chapter 5, in patients with vasculitis. Tables 6.1 and 6.2 summarises the results. Table 6.3 shows the Holmn-Bonferroni critical values to correct for multiple comparisons.

When EPCs were expressed as a percentage of PBMCs, correlations were observed between CD34+CD133+ PBMCs and white cell count (r = 0.444, p = 0.012),

CD34+VEGFR2+ PBMCs and E-selectin +ve MPs (r = -0.516, p = 0.007) and total MP count (r = -0.460, p = 0.018) and CD34+ VEGFR2+CD133+ PBMCs and E-selectin

+ve MPs (r = -0.496, p = 0.010). When EPCs were expressed as a percentage of

CD34+ cells, correlations were observed between CD34+CD133+ PBMCs and white cell count (r = 0.364, p=0.044), CD34+ VEGFR2+ CD133+ PBMCs and CD144 +ve

MPs (r = -0.233, p = 0.022*), CD34+VEGFR2+ PBMCs and CD144 +ve MPs (r = -

0.556, p = 0.003), CD14+ MMPs (r = -0.418, p = 0.034) and total MP population (r = -

0.529, p = 0.005). CD34+CD144+ PBMCs correlated with levels of Ang-1 (r = 0.464, p = 0.046). Interestingly all significant correlations for both methods of expressing putative EPC populations with MPs were negative.

When correction for multiple comparisons were taken into account non of the correlations observed remained significant (Table 6.3).

302

Table 6-2: Spearman rank correlation coefficiants of putative EPC populations – expressed as percentage of PBMCs.

Marker of % of % of PBMCs % of PBMCs % of % of % of PBMCs disease PBMCs expressing expressing PBMCs PBMCs expressing activity/ expressing CD34/CD133 CD34/CD133/ expressing expressing CD34/CD133/ growth CD34 VEGFR2 CD34/ CD34/ CD144 factors VEGFR2 /CD144 BVAS r = 0.18 r = 0.266 r = 0.152 r = 0.080 r = 0.131 r = 0.164 p = 0.39 p = 0.149 p = 0.42 p = 0.68 p = 0.515 p = 0.40 n = 30 n = 31 n = 30 n = 30 n = 27 n = 28 ESR r = 0.229 r = 0.296 r = 0.221 r = 0.186 r = 0.118 r = 0.152 p = 0.231 p = 0.112 p = 0.25 p = 0.34 p = 0.57 p = 0.42 n = 29 n = 29 n = 29 n = 29 n = 26 n = 27 CRP r = 0.335 r = 0.325 r = 0.090 r = 0.177 r = 0.025 r = 0.177 p = 0.075 p = 0.080 p = 0.64 p = 0.58 p = 0.90 p = 0.38 n = 29 n = 30 n = 29 n = 29 n = 26 n = 27 WCC r = 0.274 r = 0.444 r = 0.256 r = 0.168 r = -0.035 r = 0.331 p = 0.113 p = 0.012* p = 0.17 p = 0.37 p = 0.863 p = 0.085 n = 30 n = 31 n = 30 n = 30 n = 27 n = 28 Platelet r =-0.04 r = 0.118 r = 0.053 r = -0.121 r = -0.093 r = 0.088 count p = 0.83 p = 0.53 p = 0.79 p = 0.53 p = 0.63 p = 0.67 n = 29 n = 30 n = 29 n = 29 n = 27 n = 24 Albumin r = -0.160 r = -0.213 r = -0.162 r = -0.011 r = -0.024 r = -0.165 p = 0.436 p = 0.29 p = 0.43 p = 0.96 p = 0.65 p = 0.44 n = 26 n = 27 n = 26 n = 26 n = 23 n = 24 CEC count r = 0.209 r = 0.340 r = 0.232 r = 0.160 r = -0.026 r = -0.014 p = 0.34 p = 0.113 p = 0.29 p = 0.47 p = 0.91 p = 0.96 n = 23 n = 23 n = 23 n = 23 n = 20 n = 20 EMP r = 0.083 r = 0.122 r = -0.258 r = -0.236 r = -0.174 r = -0.110 (CD144+) p = 0.68 p = 0.545 p = 0.20 p = 0.25 p = 0.43 p = 0.61 n = 27 n = 27 n = 26 n = 26 n = 23 n = 24 EMP r = 0.003 r = -0.084 r = -0.233 r = -0.234 r = 0.167 r = -0.220 (CD105+) p = 0.988 p = 0.68 p = 0.25 p = 0.25 p = 0.45 p = 0.92 n = 27 n = 27 n = 26 n = 26 n = 23 n = 24 EMP r = -0.184 r = -0.261 r = -0.496 r = -0.516 r = -0.303 r = -0.186 (eselectin+) p = 0.359 p = 0.188 p = 0.010* p = 0.007* p = 0.159 p = 0.39 n = 27 n = 27 n = 26 n = 26 n = 23 n = 24 PMP r = -0.816 r = -0.127 r = -0.120 r = -0.342 r = -0.060 r = -0.007 (CD42a+) p = 0.354 p = 0.528 p = 0.56 p = 0.087 p = 0.79 p = 0.97 n = 27 n = 27 n = 26 n = 26 n = 23 n = 24 MMP r = -0.014 r = -0.009 r = -0.359 r = -0.352 r = -0.044 r = -0.220 (CD14+) p = 0.95 p = 0.963 p = 0.072 p = 0.078 p = 0.84 p = 0.30 n = 27 n = 27 n = 26 n = 26 n = 23 n = 24 Total MP r = 0.217 r = -0.007 r = -0.296 r = -0.460 r = 0.093 r = -0.257 p = 0.32 p = 0.96 p = 0.141 p = 0.018* p = 0.67 p = 0.23 n = 23 n = 27 n = 26 n = 26 n = 23 n = 24 VEGF r = 0.1 r = -0.029 r = -0.068 r = -0.068 r = 0.039 r = 0.299 p = 0.62 p = 0.89 p = 0.75 p = 0.748 p = 0.86 p = 0.17 n = 26 n = 26 n = 25 n = 25 n = 22 n = 23 ANG-1 r = -0.357 r = -0.196 r = -0.050 r = -0.050 r = 0.027 r = 0.012 p = 0.112 p = 0.394 p = 0.83 p = 0.830 p = 0.91 p = 0.96 n = 21 n = 21 n = 21 n = 21 n = 19 n = 19 ANG-2 r = -0.101 r = -0.003 r = -0.204 r = -0.204 r = -0.276 r = -0.117 p = 0.61 p = 0.989 p = 0.31 p = 0.308 p = 0.19 p = 0.58 n = 28 n = 28 n = 27 n = 27 n = 24 n = 25 TB4 r = -0.205 r = -0.015 r = -0.228 r = 0.1 r = -0.382 r = 0.042 p = 0.387 p = 0.95 p = 0.35 p = 0.63 p = 0.14 p = 0.87 n = 20 n = 20 n = 19 n = 25 n = 16 n = 17

Significant correlations highlighted in blue

303

Table 6-3: Spearman rank correlation coefficiants of putative EPC populations – expressed as percentage of CD34+ PBMCs.

Marker of % of CD34+ % of CD34+ % of CD34+ % of CD34+ % of CD34+ disease cells expressing cells expressing cells expressing cells expressing cells expressing activity/ CD133 CD133/VEGFR2 VEGFR2 CD144 CD133/CD144 growth factors BVAS r = 0.139 r = 0.013 r = 0.013 r = 0.263 r = 0.177 p = 0.456 p = 0.95 p = 0.94 p = 0.17 p = 0.36 n = 31 n = 30 n = 30 n = 27 n = 28 ESR r = 0.285 r = 0.051 r = 0.104 r = 0.139 r = 0.129 p = 0.126 p = 0.79 p = 0.59 p = 0.49 p = 0.52 n = 29 n = 29 n = 29 n = 26 n = 27 CRP r = 0.113 r = -0.116 r = -0.104 r = 0.236 r = 0.063 p = 0.554 p = 0.55 p = 0.59 p = 0.24 p = 0.75 n = 30 n = 29 n = 29 n = 26 n = 27 WCC r = 0.364 r = -0.105 r = 0.216 r = -0.259 r = -0.133 p = 0.044* p = 0.58 p = 0.25 p = 0.18 p = 0.50 n = 31 n = 30 n = 30 n = 27 n = 28 Platelet count r = 0.290 r = -0.001 r = -0.164 r = -0.181 r = -0.076 p = 0.12 p = 0.99 p = 0.40 p = 0.34 p = 0.70 n = 30 n = 29 n = 29 n = 27 n = 24 Albumin r = -0.103 r = 0.048 r = 0.043 r = -0.277 r = -0.210 p = 0.61 p = 0.81 p = 0.83 p = 0.19 p = 0.32 n = 27 n = 26 n = 26 n = 23 n = 24 CEC count r = 0.146 r = -0.174 r = -0.282 r = -0.276 r = -0.229 p = 0.51 p = 0.42 p = 0.19 p = 0.24 p = 0.33 n = 23 n = 23 n = 23 n = 20 n = 20 EMP r = -0.004 r = -0.233 r = -0.556 r = -0.239 r = -0.262 (CD144+) p = 0.9 p = 0.022* p = 0.003 p = 0.26 p = 0.22 n = 27 n = 26 n = 26 n = 24 n = 24 EMP r = -0.168 r = -0.149 r = -0.245 r = 0.169 r = -0.140 (CD105+) p = 0.40 p = 0.47 p = 0.23 p = 0.43 p = 0.52 n = 27 n = 26 n = 26 n = 24 n = 24 EMP r = -0.301 r = -0.340 r = -0.351 r = 0.054 r = 0.012 (eselectin+) p = 0.13 p = 0.089 p = 0.078 p = 0.80 p = 0.96 n = 27 n = 26 n = 26 n = 24 n = 24 PMP r = -0.107 r = 0.048 r = -0.257 r = -0.041 r = -0.055 (CD42a+) p = 0.59 p = 0.82 p = 0.20 p = 0.85 p = 0.79 n = 27 n = 26 n = 26 n = 24 n = 24 MMP r = 0.007 r = -0.26 r = -0.418 r = -0.074 r = -0.377 (CD14+) p = 0.98 p = 0.2 p = 0.034 p = 0.26 p = 0.069 n = 27 n = 26 n = 26 n = 24 n = 23 Total MP r = 0.004 r = -0.201 r = -0.529 r = -0.179 r = -0.336 p = 0.98 p = 0.32 p = 0.005* p = 0.40 p = 0.11 n = 27 n = 26 n = 26 n = 24 n = 23 VEGF r = -0.245 r = -0.245 r = -0.139 r = 0.333 r = 0.328 p = 0.23 p = 0.24 p = 0.51 p = 0.12 p = 0.099 n = 26 n = 25 n = 25 n = 23 n = 23 ANG-1 r = -0.083 r = 0.204 r = 0.284 r = 0.464 r = 0.489 p = 0.72 p = 0.28 p = 0.21 p = 0.046 p = 0.034 n = 21 n = 21 n = 21 n = 19 n = 19 ANG-2 r = 0.310 r = 0.252 r = 0.098 r = 0.228 r = 0.227 p = 0.109 p = 0.21 p = 0.63 p = 0.27 p = 0.27 n = 28 n = 27 n = 27 n = 24 n = 25 TB4 r = -0.180 r = -0.170 r = -0.205 r = 0.145 r = -0.042 p = 0.45 p = 0.23 p = 0.40 p = 0.58 p = 0.87 n = 20 n = 19 n = 25 n = 17 n = 17

Significant correlations highlighted in blue

304

Table 6-4: Holm-Bonferroni critical values for significant correlations between EPCs and markers of disease activity Test P value Adjusted critical value

(Holm-Bonferroni)

EMP (eselectin+) vs % of

PBMCs expressing CD34 0.007 0.00049

EMP (eselectin+) vs % of

PBMCs expressing CD34 0.01 0.000495

WCC vs % of PBMCs expressing CD34/CD133 0.012 0.0005

Total MP vs % of PBMCs expressing CD34/CD133 0.018 0.000505

EMP (CD144+) vs % of CD34+ cells expressing VEGFR2 0.003 0.000588

Total MP vs % of CD34+ cells expressing VEGFR2 0.005 0.000595

EMP (CD144+) vs % of CD34+ cells expressing

CD133/VEGFR2 0.022 0.000602

MMP vs % of CD34+ cells expressing VEGFR2 0.034 0.00061

ANG-1 vs % of CD34+ cells expressing CD133/CD144 0.034 0.000617

WCC vs % of CD34+ cells expressing CD133 0.044 0.000625

ANG-1 vs % of CD34+ cells expressing CD144 0.046 0.000633

305

6.5 Detection of EPCs using cell culture

While the main focus of this chapter has been on using flow cytometry to detect populations of EPCs, culture methodologies may give some insight into the functional capacity of EPCs, and although time limitations (and blood volume limitations as previously mentioned) of this project did not permit extensive assessment of the vasculitis cohort, the preliminary observations relating to methodology performed in 4 adult controls and 2 patients bears mention.

As discussed in the introduction to this chapter, many groups have used a colony counting methodology to assess EPCs in vitro. While the non-adherent population isolated after 1-2 days in culture gives rise to a colony phenotype which correlates with endothelial function (Hill et al., 2003) and mortality (Werner et al., 2005), as yet the colonies derived from the non-adherent cells have not shown such a clear cut relationship (Xiao et al., 2007b), yet this population contains a cell population which can go on to form endothelial-like monolayers after 2-3 weeks in culture (notably the non-adherent population has not been documented to do this). The aim of these preliminary studies was to develop a methodology where it would be possible to analyse both adherent and non-adherent colonies as well as using the non-adherent population in matrigel tubule experiments, based on techniques previously established in our group but minimizing the volume of blood necessary. As has been apparent throughout this study, it is frequently a problem to obtain greater than 10-15mls total volume of research blood from small children. In the following series of preliminary experiments blood was obtained from 4 healthy adult controls and 2 children with vasculitis (one individual had PAN, the other an unclassified small vessel vasculitis, both were on immunosuppressive treatment and disease was in remission).

306

6.5.1 Colony counting

The protocol for enumeration of colonies derived from the non-adherent culture population as described in the section 2.6.3 was based on the protocol described by Hill et al (Hill et al., 2003). Briefly 2.5 x 106 cells were added per well on a 12 well fibronectin coated plate and cultured for 48hrs in RPMI with 20% FCS and antibiotics.

After 48hrs the supernatant was removed and the non-adherent cells contained in it were counted and replated at a concentration of 1 x106 cells per well on a second 12 well fibronectin coated plate and cultured in EPC media for 7 days (section 2.6.3). The remaining adherent populations on the original culture plate were also cultured concurrently in EPC mix to see if the two populations differed considerably when colonies were counted.

In 5 individuals (2 patients, 3 adult controls) each initial culture well gave enough non- adherent cells for 1 well of second stage non-adherent culture. In 1 healthy control, however, 3 wells only gave sufficient non-adherent cells for 2 wells. Thus in order to get a minimum of 2 non-adherent cultures for enumeration of colonies, at least 3 initial wells would be necessary using this methodology (a total of 7.5x106 cells, approximately 7-10mls of blood). Figure 6.18a-d demonstrates the typical phenotype of both the non-adherent and adherent colonies observed in this experiment and the number of colonies counted for both the patients and controls. As can be seen from

Figure 6.18, more colonies were formed in non-adherent cell cultures and both patients had less colonies than 3 of 4 healthy adults studied.

307

Figure 6-18: EPC colony culture

Figure 6.18a and b: When PBMCs were cultured in EPC mix the non-adherent colonies were often large (formed after 5-9 days) (figure 6.18a) and in 6 individuals (4 healthy adult controls and 2 vasculitis patients – one with PAN, one with an unclassified small vessel vasculitis, both on immunosuppressive treatment) colony count at day 9 ranged from 0-86 colonies per well. Figure

6.18c and d: When the non-adherent population was removed after an initial 48 hours in culture and replated, colonies formed after 5-7 days in culture (figure 6.18c) and in 6 individuals (4 healthy adult controls and 2 vasculitis patients as before) colony count at day 7 ranged from 0-42 colonies per well.

308

6.5.2 Matrigel

When endothelial cells such as HUVEC are grown on the extracellular matrix scaffold -

MatrigelTM, they form tubule networks (Figure 6.19a). In this system, the ability of

EPCs to incorporate and influence EC growth (and repair) can be assessed. As mentioned above, culturing of PBMCs in EGM-2 can lead to the development of late- outgrowth EC-like monolayers. Thus earlier cultures must include the putative EPC population, thus EGM-2 cultures in which colonies have formed (herein described as

EPC enriched) are used in these in vitro assays.

In order to generate sufficient adherent cells for matrigel assays (as described in section

2.6.4) 5 x 106 cells per well were cultured on fibronectin coated 12 well plates in EGM-

2 for 9 days. Cells from these EPC enriched cultures were then lifted and 40,000 cells were combined with 40,000 HUVEC in EGM-2 on a MatrigelTM coated 12 well plate and cultured for a further 18 hours. In all of the 6 individuals studied (4 adult controls,

2 patients), 1 well of initial culture was more than sufficient to generate enough cells for

1 well of MatrigelTM culture. However even in the small numbers studied it was immediately apparent that the cultures obtained from patients grew poorly compared to the adults. After 9 days, all of the healthy adult cultures were confluent with cells, making it impossible to distinguish colonies (Figure 6.19b). In both patients, however, there were much fewer cells and neither culture appeared confluent, individual colonies could be readily counted (Figure 6.19d). Thus in order to generate enough cells for a

MatrigelTM assay using this methodology, at least 2 wells would be necessary (10 x 106 cells, approximately 10-15mls of blood).

The extent of incorporation and the influence that EPC enriched cultures can have on

309

endothelial cells, can be readily assessed in this system. EPC enriched culture cells can be distinguished from HUVEC by using a metabolic dye such as diI labeled-acetylated low density lipoprotein as described in section 2.6.5 (Figure 6.19d) to determine the extent of incorporation. Growth characteristics such as total tubule length and number of branch point in a random field can also be readily determined (highlighted in Figure

6.19a).

The changes in growth characteristics attributable to EPC enriched cells were determined by comparing mixed EPC enriched/HUVEC cultures to a control well containing only 40,000 HUVEC. This is an important comparison as it controls for variations in HUVEC growth between different experiments. Figures 6.20a-c demonstrate the preliminary results obtained in the individuals studied. In this series of experiments 5 out of the 6 EPC enriched cultures studied (3 healthy adults and 2 patients) exerted a slightly negative effect on HUVEC tubule length (Figure 6.20a) and number of branch points (Figure 6.20b) when compared to the control well of HUVEC alone. The range of tubule length (expressed as a ratio of HUVEC cultured with EPC enriched cells to HUVEC alone) was 0.75 – 1.25 in healthy adults and 0.3-0.75 in the two patients. When branch points were considered the range (as before expressed as a ratio) was 0.4-1.25 in healthy adults and 0.4-0.5 in the two patients. Incorporation of

EPC enriched cultures was low in all cultures (5-14 cells per random field, Figure

6.20c).

310

Figure 6-19: Matrigel EPC culture

Figure 6.19a: When HUVEC were cultured on MatrigelTM, they formed tubule networks. Growth of these tubules were assessed by counting of branch points (arrowed) and measuring total tubule length within a random field. Figure 6.19b: To assess how EPC enriched cells could influence

HUVEC tubule formation, PBMCs were cultured for 9 days in EGM-2 before combining with

HUVEC in MatrigelTM. All of the EPC enriched cultures from healthy adults were confluent with cells making it impossible to count colonies (figure 6.19b). In the 2 patients studied both had much fewer cells (Figure 6.19c). Figure 6.19d: DiI-acetylated LDL could be used to stain cultured

PBMCs to allow them to be distinguished from the HUVEC they incorporated into.

311

Figure 6-20: Matrigel assay results

Figure 6.20a-c: 40,000 PBMCs cultured for 9 days in EGM-2 were lifted and mixed with 40,000 HUVEC and cultured on MatrigelTM for 18 hours. These were compared to HUVEC cultured alone. Figure

6.20a: In 6 individuals (4 healthy adult controls and 2 patients on treatment) the ratio of total tubule length between mixed cultures and HUVEC alone ranged from 0.3 to 1.25. Figure 6.20b: In 6 individuals (4 healthy adult controls and 2 patients on treatment) the ratio of total branch points between mixed cultures and HUVEC alone ranged from 0.4 to 1.2. Figure 6.20c: Incorporation of cells was low in all cultures, ranging from 2-14 cells per random field.

312

6.6 Discussion

EPCs have been enumerated in a number of different ways and as yet no consensus has been reached as to which culture techniques or flow cytometric protocols best describes this population of cells. In this chapter flow cytometry was used to investigate the different CD34+ populations within which EPCs are thought to reside and some preliminary data relating to optimization of culture methodologies for detection of EPCs was also presented. The latter are important, but were not applied widely to the paediatric vasculitis cohort because the blood volumes required were too great to obtain from children concurrently with the other tests described in this thesis.

Using flow cytometry the proportion of PBMCs expressing a number of combinations of EPC markers (CD34+CD133+, CD34+CD133+VEGFR2+ and CD34+CD144+) were significantly higher in patients with active disease compared to healthy child controls. Two different combinations of cell surface markers which other groups have used to define EPCs (CD34+VEGFR2+ and CD34+CD144+CD133+) were not. 10 individual patients were studied from disease onset and no significant differences were observed in any of the combinations of cell surface markers between active and inactive disease (although CD133+CD34+ was close to significance with p=0.054) or at median followup of 3 or 12 months from disease onset. This is undoubtedly influenced by the small numbers studied.

As described earlier in this chapter, concerns were raised as to whether enumeration of putative EPCs as a proportion of a viable cell gate (a protocol which is frequently described within EPC literature) was appropriate in this cohort. While this is an acceptable definition of an EPC for some studies, and has been widely published, in the

313

context of vasculitis, immunosuppressive treatment might be skewing the population by reducing the absolute number of cells within the gate. In order to determine if this was a major limitation of this study, putative EPC populations were enumerated in 3 different ways. Firstly using the widely published percentage of the broad PBMC gate, secondly as a percentage of a smaller more defined population (CD34+ cells). Both of these were compared to an estimate of absolute number in individual patients followed up from disease onset. In order to estimate the absolute number of EPCs, the leukocyte count derived from the routine bloods taken in the vasculits patients was utilized in a methodology adapted from the dual platform ISHAGE protocol for enumeration of

CD34+ cells (Gajkowska et al., 2006).

For each combination of putative EPC markers, when these 3 enumeration methods were compared, the majority of changes in EPC number over time in individual patients was comparable. However in a small number of patients, there were obvious differences; for example where EPCs enumerated as a percentage of PBMCs had increased, absolute number may have declined. Enumerating EPCS as a proportion of

CD34+ cells did not mirror the changes occurring in the estimate of absolute number any better than enumerating them as a percentage of the PBMC gate. However when differences between active disease and controls groups for this method of enumeration were examined, p values for the groups where a significant difference was observed were smaller (for CD34+CD133+ and CD34+VEGFR2+CD133+ populations). The exception to this was the CD34+CD144+ population which was significantly different between active disease and controls when expressed as a percentage of PBMCs, but not when expressed as a percentage of CD34+ cells.

314

While this could be consided over-interpretation of the available data, it raises an interesting question relating to what is the most appropriate way to quantify an EPC population. When considered alone absolute number seems to a more intuitive measure, however when the relationship between EPCs and haematopeotic cells is considered matters become more complicated. The measure of EPCs which has been shown to most closely reflect endothelial function (non-adherent colonies) is actually a measure of the interaction between haematopoeitic cells and presumanbly EPCs (the majority of cells making up these colonies staining postiviely for CD3 or CD14).

Therefore the ratio (or percentage) of EPCs within this population may actually be more informative. For future studies, it would be important to consider both, thus optimization of whole blood flow cytometry from which both an absolute number and a

% of the mononuclear cells expressing these markers could be derived would be an obvious future direction of this work. It would also allow a much smaller amount of blood to be used, allowing culture work to be carried out in parallel on these samples.

Despite questions relating to the best way to enumerate EPCs, significant increases were observed during active disease in CD34+VEGFR2+CD133+ and CD133+CD34+ populations in the two methods that could be compared with healthy child controls.

This is comparable with the study by Nakatani et al who observed increased EPC number in patients with active KD compared to healthy controls, with highest levels being found in patients with coronary artery lesions (Nakatani et al., 2003). The increase in EPCs number seen in these children with vasculitis also corresponds with the findings from Chapter 5, where it was determined that growth factors associated with vasculogenesis are also elevated in active disase, particularly at disease onset, and decline with remission inducing therapy.

315

When the putative EPC populations observed in this chapter with correlated with other markers of disease activity, including growth factors, no striking correlations were observed, however numbers were small (>30 events) and a larger study population may be needed, also the large number of tests performed meant the significant correlations observed did not withstand multiple comparison corrections. Interestingly the majority of EPC markers studied actually correlated negatively with endothelial microparticles, perhaps this is indicative of EPCs being targeting towards activated endothelium, hence being at lower levels.

In adults however, the situation is in contrast to that observed in children. Adults with active AAV have comparable levels of CD34+ PBMCs to healthy controls, which increase on treatment (de groot et al., 2007). This interesting observation indicates that perhaps children have a different response to vasculitic injury than adults. Further study will however be necessary to clarify this. While increases in EPC number may well be relevant to attempted reparative responses, the function of these EPCs in children, could conceivably be impaired. Studies which have utilized colony formation assays in adults with active AAV have demonstrated decreased formation of colonies which may

(Holmen et al., 2005) or may not improve with treatment (Zavada et al., 2008) .

Considering the underwhelming differences in the different EPC populations between disease and the various controls, and the lack of consistency within the literature as to what constitutes an EPC, EPC functional assays in this cohort may indeed give more informative data and is an avenue of considerable future interest. Using the EPC culture methodologies that have been previously established within the group (Arrigoni

F & Haliday N, personal communication), it was immediately apparent that the volume

316

of blood necessary for generating informative data was prohibitive in the vasculitis cohort, particularly if any of the other assays described in this thesis were to be carried out in parallel. Thus further optimization would be necessary. Currently performing assays to assess non-adherent colonies and MatrigelTM would require a minimum of 15-

20mls of blood (to obtain 17.5x106 PBMCs in total). This is too much blood to obtain from small children, especially when a further 5-10mls is required for the other assays described in this thesis. The methodologies for these assays described in this chapter were preliminarily investigated because they are based on techniques previously published which have provided the most informative data relating to endothelial health

(in the case of the non-adherent colonies which correlate with FMD) or in the case of the EPC enriched population which can integrate into endothelium (the matrigel experiments). The first obvious optimization would be to combine the two above mentioned culture assays in order to generate one culture from which both the non- adherent culture and the adherent cells for MatrigelTM could be derived. This would require extensive investigation into the effects that slight changes to culture media, number of cells plated and protocols may make however and time restraints meant that this could not be done within the remit of this thesis.

In order to examine both adherent and non-adherent colonies, one possible protocol would be to establish the EGM-2 matrigel cultures, removing the non-adherent cells at day 2. In order to ascertain whether changes to methodology (including media used, number of cells initially plated) influences the colonies formed, cultures using the different protocols must be grown in parallel and compared. Phenotypic characterization using flow cytometry would be one way of doing this, however comparing colony counts between cultures would also be necessary. While this may be

317

time consuming, if the end result was a robust assay, which could use a small volume of blood (<10mls) to enumerate both the non-adherent colonies which correlate with endothelial function and to investigate the interaction of EPC cultures with HUVEC the effort would be rewarded.

In conclusion, this chapter has described the different CD34+ subpopulations that may constitute a population of cells which are involved in vascular repair. A number of these populations are elevated at disease onset, corresponding with a time when growth factors involved in their recruitment are also elevated. Thus children with vasculitis in contrast to adults with the disease may be instigating a reparative response, whether this is of functional significance remains unknown and is an avenue of future work.

318

7 Discussion and concluding remarks

7.1. Discussion

7.2. Concluding remarks

319

Discussion

The vasculitides are a rare and heterogeneous group of disorders that are characterized by the presence of inflammation in blood vessel walls. As has been discussed previously, robust classification of the vasculitides has proven difficult; to date there are no specific diagnostic tests and patients frequently exhibit complex and overlapping symptoms. Consequently the most commonly used classification criteria (CHCC) is based on the size of the vessels predominantly affected. In relation to children however, these criteria have now been formally shown to have poor sensitivity and specificity

(Ozen, Brogan et al, manuscript in preparation 2009), highlighting the need to consider children differently to adults. A number of the vasculitides are also associated with the risk of disease recurrence and patients require careful monitoring long term to allow detection of flaring disease.

It was with these concepts in mind that this thesis was undertaken, to investigate assays which could indicate the presence of active systemic vasculitis irrespective of causal pathology and disease classification and would be of relevance in adults, but more importantly in children, where small sample volumes are an issue. Although causal pathology remains poorly defined in many conditions, one common downstream consequence of inappropriate inflammation within the vasculature, which may be relevant to virtually all of the vasculitides, is injury to the endothelium. This concept underlies all of the work described within this thesis. Moreover, the damaged endothelium releases factors into the blood, thus allowing testing in a relatively non invasive manner from small volumes of peripheral blood.

320

The first putative biomarkers of disease activity described within this thesis were circulating endothelial cells (CECs, Chapter 3). At the time of establishing the assay it was apparent that two techniques were being interchangeably used within the literature for enumeration of CECs, namely flow cytometry (FC) and immunomagnetic bead extraction (IBE). It had even been reported that the two techniques gave comparable results in adult controls and patients with breast cancer or acute coronary syndromes

(Goon et al., 2006). Thus initially both techniques were utilised within the GOSH cohort to determine which was the most appropriate method for examining CECs in this population (Section 3.5). In contrast to the study by Goon et al, it was soon apparent that IBE and FC did not give comparable results in children with vasculitis, although the

IBE data obtained was comparable with previous studies in adults with vasculitis which had utilized the IBE methodology (Woywodt et al., 2003b).

Subsequently a number of experiments were performed in order to determine why the

FC protocol, as described by Goon et al, was not appropriate for enumeration of CECs in children with vasculitis. The following conclusions were reached: firstly that size matters – whilst Goon et al had gated on a mononuclear cell FSC SSC region, light microscopy had indicated that CECs were of comparable sizes to HUVEC and PAEC which exhibited a different FSC SSC profile. Thus a much larger gate was necessary to include this; secondly, factors that are inherent to FC also limited CEC recovery. These included sensitivity of detection of rare events and the ability of the flow cytometer to resolve individual events. Most importantly it was found that the lysis necessary for removal of red cells significantly limited recovery of CECs (Clarke et al., 2008).

The published findings from this series of experiments have added to what is currently

321

an ongoing area of debate within the CEC literature – how best to quantify CECs

(Shantsila and Lip, 2008). Currently new methodologies utilizing IBE and automated microscopy (Rowand et al., 2007) or IBE enriched FC (Widemann et al., 2008) have now been described which may provide a more automated approach to CEC enumeration, bringing it closer to clinical utilization. It is also likely that an approach to

FC based around previously published protocols for detection of apoptotic and necrotic cells may also provide another methodology by which FC could be utilized to detect

CECs in patients with vasculitis and would be an obvious future direction of this study

(discussed in Chapter 3). However, as demonstrated herein any techniques developed for CEC enumeration need to be carefully validated in individuals where CECs may be fragile in nature. While data was not presented in this thesis to indicate why CECs may be less robust on handling than in other conditions, Woywodt et al had already convincingly demonstrated in adults with AAV that 84% of cells obtained from patients with active disease stained positive for annexin V/propidium iodide indicating a necrotic phenotype (Woywodt et al., 2003b). Whilst repeating this work in children would be an important next step, the limiting blood volumes and rarity of patients meant this was not possible during this project.

Despite the methodological controversies IBE enumeration of CECs proved robust in this particular cohort (Section 3.7) and demonstrated significantly higher CECs in patients with active vasculitis than in all control groups studied (healthy and disease control children, and adults) as well as compared to patients in remission. Of considerable interest was the indication that CECs were elevated during active disease irrespective of treatment status, declined with remission inducing therapy and were highest in individuals with more widespread disease. CECs also correlated strongly

322

with clinical scores of disease activity but not with inflammatory markers such as ESR and CRP (which are influenced by treatment). Therefore in this context CECs are a particularly useful biomarker of disease activity, although the current methodology limits use to a research tool.

In a number of individuals, when CECs were enumerated, small ULEX-bright staining particles were frequently seen alongside the larger CECs, which could be indicative of fragile CECs fragmenting or possibly of endothelial MPs. Chapter 4 therefore went on to describe MPs in the same group of patients. Because FC was utilized to enumerate

MPs, several different populations could easily be studied within this project and these included MPs of endothelial, platelet, neutrophil, monocyte and lymphocyte origins. As indicated in Chapter 3, FC for rare event analysis requires careful optimization and thus this chapter also contained a significant proportion of optimization work, particularly as

MPs are at the lower limits of detection of the flow cytometer. A major step in optimising the technique was limiting the background noise, which had been shown to adversely influence accuracy of detection of latex counting beads and thus presumably

MPs themselves. As the MPs of interest were AnV positive it was decided to utilize the

AnV staining, by using a fluoresence threshold, below which events were not counted, thereby removing the majority of electrical noise and any debris in the fluidics, increasing the recovery of MP events. Although this greatly improved sensitivity

(Figure 4.7), the major downside to this approach was that post-hoc analysis of AnV negative populations could not be performed if required at a later date.

The utilization of a threshold to reduce background noise has been widely used by

Jimenez et al, although not by other groups, which makes comparison with the other

323

studies performed in vasculitis difficult. Despite this the results obtained were similar to other studies. The previous study from our group indicated higher EMPs (CD105+ and E-selectin+) during active disease compared to patients in remission and controls

(Brogan et al., 2004a). In the present study CD144+ EMPs were also significantly higher in patients with active vasculitis compared to controls and declined when individual patients entered remission, although CD105+ EMPs and E-selectin + EMPs were not. Interestingly in this study CD105+ EMPs were significantly higher in patients in remission compared to controls. Although this contrasts with the previous study from within our group it could be explained by the higher proportion of KD patients within the previously studied cohort - preliminary data indicates that patients with acute KD may have higher EMPs than other vasculitis patients (Figure 4.27). As before PMPs were not significantly different between patients and controls, although they did correlate inversely with markers of disease activity (white cell count, BVAS,

ESR and CRP) indicating that PMPs may indeed be lower during active disease. MPs of monocytic origin (CD14+) were also found to be elevated during active disease.

The differences in MP numbers between active disease and remission or the control groups were not particularly striking even when significant differences were observed and it may be that changes in individual patients provide more informative data than broad studies within a heterogenous population. Whilst MPs may not be as robust a biomarker of disease activity as CEC, they may provide more informative data relating to cellular activation in individual patients, which could be of particular importance when predicting whether a patient is likely to enter disease flare. Further prospective studies in larger cohorts will be required to truly clarify whether MPs could provide a way of monitoring patients cellular activation profiles.

324

Another interesting finding in this study (although preliminary) was that in a number of individual patients with active disease prior to treatment MPs were low (particularly apparent for EMPs and PMPs), suggesting that in active untreated disease there may be sequestering of MPs to activated cells within the blood or to the endothelium. As MPs of a number of different phenotypes have been demonstrated to influence cell function

(discussed in detail in the introduction Chapter 4), in this context they could be contributing to the pathogenesis of the disease and thus future studies concurrently investigating MP-cell complexes would also be an interesting direction to take this study.

Whilst EMPs and CECs reflect the activation and injury of the endothelium, it is possible that reparative responses maybe underway in an attempt to repair the damaged vessels. Thus the latter chapters of this thesis focused on possible repair mechanisms, firstly by measuring growth factors associated with angiogenesis and vasculogenesis - the signals of repair and secondly by enumerating endothelial progenitor cells, the cells believed to play a key role in orchestrating repair. As a downstream consequence of endothelial injury, these could also be utilized as biomarkers of disease activity.

However as the results from these studies indicate, this highly controversial and complex field is likely influenced by treatment and thus although their use as biomarkers is limited, the data presented herein raises some interesting questions relating to the long term potential for damage repair in vasculitis.

In Chapter 5, the relationship between the growth factors VEGF, Ang-1, Ang-2 and Tβ4 and disease activity was considered. Whilst Tβ4 was not significantly different between patients and controls, children (controls and patients) had higher levels than adults – a

325

previously unpublished finding, perhaps indicative of ongoing growth. When patients with active disease prior to treatment were considered it was apparent that with the exception of Tβ4, growth factors were elevated, when compared to controls and patients in remission. In individual patients, levels declined with treatment, however whether this is due to patients being in remission or due to therapy is unclear. Growth factor levels appear to be influenced by treatment and preliminary data demonstrated that when patients had an active disease flare while on routine maintenance therapy

(corticosteroids, with azathioprine) their levels of VEGF, Ang-1 and Ang-2 were lower when compared to treatment naive patients (Figure 5.6). This finding could thus limit growth factors usefulness as biomarkers of disease activity and is in contrast to the conclusions reached in the recent publication by Kumpers et al, whereby Ang-2 was considered to be a good biomarker of disease activity in adults with AAV (Kumpers et al., 2009). However as discussed in Chapter 5, the influence of treatment status was not studied in depth as in this study.

Regardless of the usefulness of growth factors as putative biomarkers, these studies indicate that angiogenic and vasculogenic pathways are activated in patients with active disease and therefore there may be an attempted reparative response occurring.

Conversely, when one considers the tightly regulated systems that control angiogenesis and vasculogenesis, these elevations in several interlinked growth factors (a “growth factor storm”) could also be of pathological consequence, propagating further dysfunction and injury to the endothelium. Rather than viewing growth factors as targets for influencing repair, their double edged sword action could indeed make them novel therapeutic targets. In this context a recent publication has indicated that PDGF receptor inhibition using imatinib (GleevecTM, Novartis) may be effective in inhibiting

326

the cell responses necessary for the development of intimal hyperplasia characteristic of giant cell arteritis and TD using an in vitro system. In vivo data is eagerly awaited

(Lozano et al., 2008). While an attractive hypothesis, this avenue of research would require extensive in vivo justification before clinical trials could be performed, particularly in children.

While Chapter 5 indicated there may indeed be changes in reparative responses, Chapter

6 went on to investigate the controversial topic of endothelial progenitor cells (EPCs).

As discussed previously while this cell population has received much interest, as yet there is no consensus relating to cell surface marker expression or culture methodology for their detection. As with the other FC techniques described in this thesis, technical difficulties associated with rare event analysis had to be considered, as did the gating strategy and way of enumerating EPCs. What was apparent from the data obtained however, was that EPCs (defined as cells expressing CD34, CD133 and VEGFR2) and

“early” progenitor cells not necessarily of endothelial lineage (expressing CD34 and

CD133) were elevated at disease onset when compared to controls and declined with treatment in individual patients. This is in striking contrast to the data published in adults (de Groot et al., 2007), whereby treatment increased EPC number, further hilighting the need to consider children as different from adults, particular when considering vascular growth and repair.

Whilst EPC number may be elevated, function cannot be determined in FC assays and thus the latter part of the chapter described some preliminary data in order to establish culture assays. While time constraints and the low blood volumes available meant that patients could not be studied concurrently with the other biomarkers in this thesis, even

327

preliminary data in 2 individuals on maintenance therapy indicated that cultured cell growth may be impaired when compared to healthy adults (Figures 6.18 and 6.19).

Extensive study is required to determine whether EPC function as determined using these culture assays is impaired and whether this relates to EPC number. However this is an avenue of ongoing work within the group, which on the basis of the preliminary data described in this thesis has now been funded by the Arthritis Research Campaign.

Because of the nature of this study, one recurrent theme throughout related to the statistical interpretation of data, particularly when multiple comparisons were performed simultaneously. The purpose of tests such as the Bonferroni is to correct for the familywise error that occurs when multiple statistical tests are performed simultaneously. A test is considered significant when the likelyhood of it occurring by chance is less than 5%. Conversely this means that the likelyhood of a rare event occurring randomly is thus 5% and therefore when multiple comparisons are performed

5% of these could be occurring by chance. Corrections such as the Holm Bonferroni have been developed to attempt to correct this. However there remains much debate to the valididity of these rigid statistical tests particularly when relating to heterogenous patient populations such as those studied herein. While interesting observations were made relating to the putative biomarkers studied in this project in order to clarify whether these are in fact truly significant effects rather than false positives resulting from multiple comparisons further study in more patients of those interesting observations is required. This is work is ongoing within the group.

The term biomarker was defined as an objectively measured characteristic which could be utilized as a clinical outcome or endpoint in trials. Currently clinical outcomes

328

which can be utilized in clinical trials of vasculitis are limited and include parameters such as mortality, although in the short term and in children particularly this is not a useful outcome as it is so low. In the three years of this study only one individual died, this individual had had preexisting TD for several years prior to enrolling in the study and at the time of study was considered to have inactive vasculitis, being well into the late “fibrotic” phase of this disease.

Thus biomarkers such as those described herein must be utilized in the future to allow more realistic clinical endpoints in trials. Alongside blood biomarkers, other possible endpoints could include more longterm responses to disease such as the assessment of structural changes in arteries and overall vascular function, using non invasive techniques such as assessment of flow mediated dilatation, pulse wave velocity or intima media thickness. These changes ultimately would reflect the balance between extent of injury and the ability to repair and may not be as influenced by treatment as some of the rapidly changing blood biomarkers described herein. Assessment of structure and function is an ongoing avenue of work within the group.

Concluding remarks

This study has been underlined by the common pathology of all the vasculitiides – endothelial injury. By focusing on this pivotal consequence of innappropriate and sustained inflammation in the vessel wall, a number of putative biomarkers which have potential to be utilised clinically to assess disease activity, (particularly that which may not have overt clinical symptoms) have been assessed. Of those studied CECs have proven the most robust to date, however methodological limitations restrict their use

329

clinically to a research tool. During this study it also became apparent that children may have differences relating to vasculogenic responses when compared to adults and that therapy necessary to promote disease remission may be influencing pathways involved in reparitive responses and thus could impact on the long term health of the vasculature.

330

Appendix 1 - Classification and definitions of Vasculitis

Chapel Hill Consensus Criteria (Jennette et al., 1994)

Large vessel vasculitis

Giant cell Granulomatous arteritis of the aorta and its major branches, with a (temporal) arteritis predilection for the extracranial branches of the carotid artery. Often involves the temporal artery. Usually occurs in patients older than 50 and often is associated with polymyalgia rheumatica.

Takayasu arteritis Granulomatous inflammation of the aorta and its major branches. Usually occurs in patients younger than 50.

Medium-sized vessel vasculitis

Polyarteritis Necrotizing inflammation of medium-sized or small arteries without nodosa (classic glomerulonephritis or vasculitis in arterioles, capillaries, or venules. PAN)

Kawasaki disease Arteritis involving large, medium-sized, and small arteries, and associated with muco-cutaneous lymph node syndrome. Coronary arteries are often involved. Aorta and veins may be involved. Usually occurs in children. Small vessel vasculitis

Wegener’s Granulomatous inflammation involving the respiratory tract, and necrotizing granulomatosis vasculitis affecting small to medium-sized vessels (e.g., capillaries, venules, arterioles, and arteries). Necrotizing glomerulonephritis is common.

Churg-Strauss Eosinophil-rich and granulomatous inflammation involving the respiratory syndrome tract, and necrotizing vasculitis affecting small to medium-sized vessels, and associated with asthma and eosinophilia.

Microscopic Necrotizing vasculitis, with few or no immune deposits, affecting small polyangiitis vessels (i.e., capillaries, venules, or arterioles). Necrotizing arteritis involving (microscopic small and medium- sized arteries may be present. Necrotizing polyarteritis) glomerulonephritis is very common. Pulmonary capillaritis often occurs.

Henoch-Schönlein Vasculitis, with IgA-dominant immune deposits, affecting small vessels (i.e., purpura capillaries, venules, or arterioles). Typically involves skin, gut, and glomeruli, and is associated with arthralgias or arthritis.

Essential Vasculitis, with cryoglobulin immune deposits, affecting small vessels (i.e., cryoglobulinemic capillaries, venules, or arterioles), and associated with cryoglobulins in serum. vasculitis Skin and glomeruli are often involved.

Cutaneous Isolated cutaneous leukocytoclastic angiitis without systemic vasculitis or leukocytoclastic glomerulonephritis. angiitis

331

American College of Rheumatology (ACR) Criteria, 1990

Polyarteritis nodosa (Lightfoot, Jr. et al., 1990) modified for children (Brogan et al., 2002). The presence of 3 or more of the criteria defines a patient as having PAN. Arteriography or biopsy are mandatory criteria  Weight loss of 5% or more (not due to diet or other factors) or failure to thrive  Livedo reticularis  Testicular pain or tenderness  Myalgias, weakness, or leg tenderness  Mononeuropathy or polyneuropathy  Systemic hypertension  Elevated blood urea nitrogen or creatinine  Hepatitis B virus - presence of hepatitis B surface antigen or antibody in serum  Arteriographic abnormality - aneurysms or occlusions of the visceral arteries, not caused by arteriosclerosis, fibromuscular dysplasia, or other non- inflammatory causes  Biopsy of small- or medium-sized artery containing granulocytes/mononuclear leukocytes in vessel wall

Wegener’s Granulomatosis (Leavitt et al., 1990) Diagnosis requires the presence of two of the four criteria. The presence of any two or more criteria has a sensitivity of 88.2% and a specificity of 92.0%.  Nasal or oral inflammation – ulcers, nasal discharge  Abnormal-appearing chest radiograph - Nodules, fixed infiltrates or cavities  Abnormal urinary sediment - Microhematuria (>5 RBC/hpf) or RBC casts (RBC- red blood cell)  Granulomatous inflammation within artery wall

Takayasu’s disease (Arend et al., 1990) The presence of 3 or more of these 6 criteria gives a sensitivity of 90.5% and a specificity of 97.8%.  Onset at less than 40 years of age  Claudication of an extremity  Decreased brachial artery pulse  > 10mm Hg difference in blood pressure between arms  A bruit over the subclavian arteries or the aorta  Arteriographic evidence of narrowing or occlusion of the entire aorta, its primary branches, or large arteries in the proximal upper or lower extremities

332

Hypersensitivity vasculitis (Calabrese et al., 1990) For classification, hypersensitivity vasculitis is defined if at least three of these criteria are present. The presence of any three or more criteria yields a sensitivity of 71.0% and specificity of 83.9%.  Age at onset >16 yr (not relevant for children)  Medication at disease onset (agents that may have been factor in disease development)  Palpable purpura – elevated rash over one of more areas of skin – does not blanch with pressure  Maculopapular rash – flat raised lesions  Biopsy (including arteriole and venule) - showing granulocytes in perivascular of extravascular location

Henoch-Schönlein purpura (Mills et al., 1990) For classification, Henöch- Schonlein purpura is defined if at least two of these criteria are present. The presence of any two or more criteria yields a sensitivity of 87.1% and specificity of 87.7%.  Palpable purpura  Age ischemia, usually including bloody diarrhoea  Wall granulocytes on biopsy

333

EULAR/PReS endorsed consensus criteria for the classification of childhood vasculitides (Ozen et al., 2006)

Predominantly large vessel vasculitis

Takayasu arteritis Predominantly medium sized vessel vasculitis

Childhood polyarteritis nodosa Cutaneous polyarteritis Kawasaki disease

Predominantly small vessel vasculitis

Wegener’s granulomatosis Granulomatous Churg-Strauss syndrome

Microscopic polyangiitis Henoch-Schönlein purpura Non-Granulomatous Isolated cutaneous leucocytoclastic vasculitis Hypocomplementic urticarial vasculitis

Other vasculitides

Behçet disease Vasculitis secondary to infection (including hepatitis B associatedpolyarteritis nodosa), malignancies, and drugs, including hypersensitivity vasculitis Vasculitis associated with connective tissue diseases Isolated vasculitis of the central nervous system Cogan syndrome Unclassified

Classification criteria:

Henoch-Schönlein purpura

Palpable purpura (mandatory criterion) in the presence of at least one of the following four features:  Diffuse abdominal pain  Any biopsy showing predominant IgA deposition  Arthritis (acute any joint) or arthralgia  Renal involvement (any haematuria and/or proteinuria)

334

Kawasaki disease Fever persisting for at least five days (mandatory criterion) plus four of thefollowing five features:  Changes in peripheral extremities or perineal area  Polymorphous exanthema  Bilateral conjunctival injection  Changes of lips and oral cavity: injection of oral and pharyngeal mucosa  Cervical lymphadenopathy In the presence of coronary artery involvement (detected on echocardiography) and fever, fewer than four of the remaining five criteria are sufficient (the exact number of criteria

Polyarteritis nodosa A systemic illness characterised by the presence of either a biopsy showing small and mid-size artery necrotising vasculitis OR angiographic abnormalities* (aneurysms or occlusions) (mandatory criteria), plus at least two of the following:  Skin involvement (livedo reticularis, tender subcutaneous nodules,other vasculitic lesions)  Myalgia or muscle tenderness  Systemic hypertension, relative to childhood normative data  Mononeuropathy or polyneuropathy  Abnormal urine analysis and/or impaired renal function (Glomerular filtration rate of less than 50% normal for age)  Testicular pain or tenderness  Signs or symptoms suggesting vasculitis of any other major organ system (gastrointestinal, cardiac, pulmonary, or central nervous system) *Should include conventional angiography if magnetic resonance angiography is negative.

Wegener’s granulomatosis Three of the following six features should be present:  Abnormal urinalysis* (Haematuria and/or significant proteinuria.)  Granulomatous inflammation on biopsy_  Nasal sinus inflammation  Subglottic, tracheal, or endobronchial stenosis  Abnormal chest x ray or CT  PR3 ANCA or C-ANCA staining *If a kidney biopsy is done it characteristically shows necrotising pauciimmune glomerulonephritis.

Takayasu’s arteritis Angiographic abnormalities (conventional, CT, or MR) of the aorta or its main branches (mandatory criterion), plus at least one of the following four features:  Decreased peripheral artery pulse(s) and/or claudication of extremities  Blood pressure difference >10 mm Hg  Bruits over aorta and/or its major branches  Hypertension (related to childhood normative data)  CT, computed tomography; MR, magnetic resonance.

335

Appendix 2 - The Birmingham Vasculitis Activity Score

(Luqmani et al., 1994)

Tick box only if abnormality is newly present or worsening within the previous 4 weeks and ascribable to vasculitis (changes from adult BVAS are in bold)

1. SYSTEMIC 3 (maximum total) none 0 malaise 1 myalgia 1 arthralgia/arthritis 1 fever (<38.5°C) 1 fever (>38.5°C) 2 wt loss (1-2 kg) within past month 2 wt loss (>2 kg) within past month 3

2. CUTANEOUS 6 (maximum total) none 0 infarct 2 purpura 2 other skin vasculitis 2 ulcer 4 gangrene 6 multiple digit gangrene 6

3. MUCOUS MEMBRANES/EYES 6 (maximum total) none 0 mouth ulcers 1 genital ulcers 1 conjunctivitis 1 episcleritis 2 uveitis 6 retinal exudates 6 retinal haemorrhage 6

4. ENT 6 (maximum total) nil 0 nasal discharge/obstruction 2 sinusitis 2 epistaxis 4 crusting 4 aural discharge 4 otitis media 4 new deafness 6 hoarseness/laryngitis 2 subglottic involvement 6

5. CHEST 6 (maximum total) none 0 dyspnoea or wheeze 2 nodules or fibrosis 2 pleural effusion/pleurisy 4 infiltrate 4 haemoptysis/haemorrhage 4

336

massive haemoptysis 6

6. CARDIOVASCULAR 6 (maximum total) none 0 bruits 2 new loss of pulses 4 aortic incompetence 4 pericarditis 4 new coronary artery lesions 4 new myocardial infarct 6 CCF/cardiomyopathy 6

7. ABDOMINAL 9 (maximum total) none 0 abdominal pain 3 bloody diarrhoea 6 gall bladder perforation 9 gut infarction 9 pancreatitis 9

8. RENAL 12 (maximum total) none 0 hypertension (systolic >95th centile) 4 proteinuria (>1 + or >20mmol/mg) 4 haematuria (> 1 + or >10 rbc/ml) 8 cGFR 50-80 mls/min/1.73m2 8 cGFR 15-49 mls/min/1.73m2 10 cGFR <15 mls/min/1.73m2 12 rise in creatinine >10% 12

9. NERVOUS SYSTEM 9 (maximum total) none 0 organic confusion/dementia 3 seizures (not hypertensive) 9 stroke 9 cord lesion 9 peripheral neuropathy 6 motor mononeuritis multiplex 9

MAXIMUM SCORE 63

337

Appendix 3 – Flow cytometry plans

Microparticles tube plan

22 tube plan – samples >200ul 12 tube plan – samples 1. Unstained MPs <200ul 2. AnV (beads) 1. Unstained MPs 3. AnV + IgG PE (BD) 2. AnV (beads) 4. AnV + IgG PE (Serotec) 3. AnV + IgG PE (BD) 5. AnV + IgG PE (ebiosciences) 4. AnV + IgG PE (Serotec) 6. AnV + IgG PE (Biolegend) 5. AnV + IgG PE (ebiosciences) 7. AnV + IgG CY5 (BD) 6. AnV + IgG PerCP (BD) 8. AnV + IgG PerCP (BD) 7. AnV + E-selectin PE (BD) 9. AnV + E-selectin PE (BD) 8. AnV + CD144 PE (BD) 10. AnV + P-selectin PE (BD) 9. AnV + CD105 PE (BD) 11. AnV + CD15 PE (Biolegend) 10. AnV + CD42a PerCP (BD) 12. AnV + CD144 PE (BD) 11. AnV + CD14 PE (BD) 13. AnV + CD105 PE (BD) 12. AnV (beads) 14. AnV + CD66b PE (BD) 15. AnV + CD42a PerCP (BD) 16. AnV + ICAM-1 CY5 (BD) 17. AnV + CD11b CY5 (BD) 18. AnV + L-selectin PE (BD) 19. AnV + TF PE (BD) 20. AnV + CD14 PE (BD) 21. AnV + CD3 CY5 PE (BD) 22. AnV (beads)

338

EPC tube plan for frozen PBMCs

23 tube plan 12 tube plan 1. Unstained PBMCs 1. Unstained PBMCs 2. VEGFR2 isotype control 2. VEGFR2 isotype control 3. VEGFR2 3. VEGFR2 4. CD144 isotype controls 4. CD144 isotype controls 5. CD144 5. CD144 6. CD45 isotype control 6. CD133 PE control 7. CD45 7. CD133 8. CD14 isotype contro*l 8. CD34 controls 9. CD14* 9. CD34 10. CD133 PE control 10. VEGFR2/CD133/CD34 11. CD133 11. CD144/CD133/CD34 12. CD146 PE control 13. CD146 14. CD34 controls 15. CD34 16. CD3 control* 17. CD3* 18. VEGFR2/CD133/CD34 19. CD144/CD133/CD34 20. CD14/CD133/CD34* 21. CD144/CD146/CD34* 22. CD45/CD146/CD34 23. CD45/CD146/CD3*

The 23 tube plan was used when samples have large numbers of PBMCs

The 16 tube plan (same as the 23 plan, excluding tubes with *) was used for samples with between 5x106 and 10x106 frozen PBMCs

The 11 well tube plan was used for samples with less than 5x106 frozen PBMCs

339

Appendix 4 – MP generation

MPs were generated from paternt cells as described in section 4.3.2.

Neutrophil MP generation – activation of parent cells

340

Neutrophil MP generation – phenotype of MPs

341

MPs generated from TNFα stimulated PBMCS - phenotype

Nb. PBMC samples isolated were lymphoprep often had platelet contamination, hence the high CD42a and P-selectin expression in this individual.

342

MPs generated from TNFα stimulated endothelial cells - activation of parent cells

343

MPs generated from TNFα stimulated endothelial cells

– phenotype

344

Appendix 5 – Standard graphs for ELISAs

Angiopoietin-1 standard graph

y = 0.00039x + 0.03021 1.8

1.6

1.4

1.2

1 OD 0.8

0.6

0.4

0.2

0 0 1000 2000 3000 4000 5000 Conc (pg/ml)

Angiopoietin-2 standard graph y = 0.00043x + 0.02116 1.4

1.2

0.8

OD 0.6

0.4

0.2

0 0 500 1000 1500 2000 2500 3000 3500 Conc (pg/ml)

345

VEGF standard graph y = 0.00102x + 0.07065 2.5

1.5 OD

0.5

0 0 500 1000 1500 2000 2500 Conc (pg/ml)

Tβ4 standard graph 3

2.5

1.5 OD

0.5

0 0.0001 0.001 0.01 0.1 1

log10 concentration

346

Appendix 6 – Changes in CECs and MPs with disease activity in individual patients

Patient 1 – A 13 year old male with PAN, patient was not treatment naïve at onset of study and was followed from a flare (t=0).

347

Patient 2 – A 9 year old male with systemic vasculitis and colitis. Patient was treatment naïve at onset of study.

348

Patient 3 – A 10 year old male with PAN, patient was treatment naïve at start of study, at first presentation of disease. Notably this individual entered a period of disease flare at 23 months from disease onset.

349

Patient 4 - A 12 year old female with ANCA+ WG, this patient was treatment naïve at onset of study at disease presentation. Changes in this individual relating to CECs and

MPS were highlighted in Figures 3.19 and 4.24. Notably this individual entered a period of systemic flare 12 months into their disease.

350

Patient 5 - An 11 year old female with WG. This patient was treatment naïve at onset of study at disease presentation. Changes in this individuals CECs and MPs were highlighted in figures 3.20 and 4.25. Notably this patient had an episode of localized granulomatous flare 28 months into their disease (orbital).

351

Patient 6 – A 2 year old female with PAN who suffered a stroke at disease onset. Patient was treatment naïve at onset of study at presentation of disease.

352

Patient 7 - A 6 year old male with PAN, this patient was treatment naïve at onset of study at disease presentation.

353

Patient 8 – A 15 year old male with WG, patient was treatment naïve at onset of study.

15 months into treatment patient exhibited localized nasal symptoms. By 19 months symptoms were more systemic, although CECs remained low.

354

Patient 9 – a 6 year old female with PAN. Patient was treatment naïve at onset of study at disease onset.

355

Patient 10 – a 1 year old female with PAN. Patient was treatment naïve at onset of study at disease onset.

356

Patient 11 – a 10 year female with PAN, presented with gastrointestinal inflammation.

Patient was undergoing immunosuppression for this at onset of study.

357

Patient 12 – a 14 year old female with WG. Treatment had been instigated prior to study, although patient was observed at first presentation of disease.

358

Patient 13 - a 12 year old male with PAN, patient was followed from a disease flare.

359

References

Aase K, von E G, Li X, Ponten A, Thoren P, Cao R, Cao Y, Olofsson B, Gebre-Medhin S, Pekny M, Alitalo K, Betsholtz C, Eriksson U. Vascular endothelial growth factor- B-deficient mice display an atrial conduction defect. Circulation 2001; (104): 358- 364.

Abdulahad W H, Stegeman C A, Limburg P C, Kallenberg C G. Skewed distribution of Th17 lymphocytes in patients with Wegener's granulomatosis in remission. Arthritis Rheum 2008; (58): 2196-2205.

Abdulahad W H, Stegeman C A, van der Geld Y M, Doornbos-van der M B, Limburg P C, Kallenberg C G. Functional defect of circulating regulatory CD4+ T cells in patients with Wegener's granulomatosis in remission. Arthritis Rheum 2007; (56): 2080-2091.

Adams T E, Huntington J A. Thrombin-cofactor interactions: structural insights into regulatory mechanisms. Arterioscler Thromb Vasc Biol 2006; (26): 1738-1745.

Agarwal R, Brunelli S M, Williams K, Mitchell M D, Feldman H I, Umscheid C A. Gadolinium-based contrast agents and nephrogenic systemic fibrosis: a systematic review and meta-analysis. Nephrol Dial Transplant 2008.

Akikusa J D, Schneider R, Harvey E A, Hebert D, Thorner P S, Laxer R M, Silverman E D. Clinical features and outcome of pediatric Wegener's granulomatosis. Arthritis Rheum 2007; (57): 837-844.

Ala Y, Palluy O, Favero J, Bonne C, Modat G, Dornand J. Hypoxia/reoxygenation stimulates endothelial cells to promote interleukin-1 and interleukin-6 production. Effects of free radical scavengers. Agents Actions 1992; (37): 134-139.

Albert D, Clarkin C, Komoroski J, Brensinger C M, Berlin J A. Wegener's granulomatosis: Possible role of environmental agents in its pathogenesis. Arthritis Rheum 2004; (51): 656-664.

Allanore Y, Batteux F, Avouac J, Assous N, Weill B, Kahan A. Levels of circulating endothelial progenitor cells in systemic sclerosis. Clin Exp Rheumatol 2007; (25): 60-66.

Allt G, Lawrenson J G. Pericytes: cell biology and pathology. Cells Tissues Organs 2001; (169): 1-11.

Amoli M M, Calvino M C, Garcia-Porrua C, Llorca J, Ollier W E, Gonzalez-Gay M A. Interleukin 1beta gene polymorphism association with severe renal manifestations and renal sequelae in Henoch-Schonlein purpura. J Rheumatol 2004a; (31): 295-298.

360

Amoli M M, Garcia-Porrua C, Calvino M C, Ollier W E, Gonzalez-Gay M A. Lack of association between endothelial nitric oxide synthase polymorphisms and Henoch- Schonlein purpura. J Rheumatol 2004b; (31): 299-301.

Amoli M M, Martin J, Miranda-Filloy J A, Garcia-Porrua C, Ollier W E, Gonzalez-Gay M A. Lack of association between interleukin-6 promoter polymorphism at position - 174 and Henoch-Schonlein pur pura. Clin Exp Rheumatol 2007; (25): S6-S9.

Amoli M M, Thomson W, Hajeer A H, Calvino M C, Garcia-Porrua C, Ollier W E, Gonzalez-Gay M A. Interleukin 1 receptor antagonist gene polymorphism is associated with severe renal involvement and renal sequelae in Henoch-Schonlein purpura. J Rheumatol 2002; (29): 1404-1407.

Ara J, Mirapeix E, Arrizabalaga P, Rodriguez R, Ascaso C, Abellana R, Font J, Darnell A. Circulating soluble adhesion molecules in ANCA-associated vasculitis. Nephrol Dial Transplant 2001; (16): 276-285.

Arend W P, Michel B A, Bloch D A, Hunder G G, Calabrese L H, Edworthy S M, Fauci A S, Leavitt R Y, Lie J T, Lightfoot R W, Jr., . The American College of Rheumatology 1990 criteria for the classification of Takayasu arteritis. Arthritis Rheum 1990; (33): 1129-1134.

Asahara T, Murohara T, Sullivan A, Silver M, van der Z R, Li T, Witzenbichler B, Schatteman G, Isner J M. Isolation of putative progenitor endothelial cells for angiogenesis. Science 1997; (275): 964-967.

Assmus B, Schachinger V, Teupe C, Britten M, Lehmann R, Dobert N, Grunwald F, Aicher A, Urbich C, Martin H, Hoelzer D, Dimmeler S, Zeiher A M. Transplantation of Progenitor Cells and Regeneration Enhancement in Acute Myocardial Infarction (TOPCARE-AMI). Circulation 2002; (106): 3009-3017.

Aupeix K, Hugel B, Martin T, Bischoff P, Lill H, Pasquali J L, Freyssinet J M. The significance of shed membrane particles during programmed cell death in vitro, and in vivo, in HIV-1 infection. J Clin Invest 1997; (99): 1546-1554.

Bacon P A, Stevens R J, Carruthers D M, Young S P, Kitas G D. Accelerated atherogenesis in autoimmune rheumatic diseases. Autoimmun Rev 2002; (1): 338- 347.

Ball S G, Shuttleworth C A, Kielty C M. Vascular endothelial growth factor can signal through platelet-derived growth factor receptors. J Cell Biol 2007; (177): 489-500.

Balogh L, Czuriga I, Hunyadi J, Galuska L, Kristof E, Edes I. [Effects of autologous bone marrow derived CD34+ stem cells on the left ventricular function following myocardial infarction]. Orv Hetil 2007; (148): 243-249.

Bansal P J, Tobin M C. Neonatal microscopic polyangiitis secondary to transfer of maternal myeloperoxidase-antineutrophil cytoplasmic antibody resulting in neonatal pulmonary hemorrhage and renal involvement. Ann Allergy Asthma Immunol 2004; (93): 398-401.

361

Barron K, DeCunto C, Montalvo J, Orson F, Lewis D. Abnormalities of immunoregulation in Kawasaki syndrome. J Rheumatol 1988; (15): 1243-1249.

Barry O P, Pratico D, Lawson J A, FitzGerald G A. Transcellular activation of platelets and endothelial cells by bioactive lipids in platelet microparticles. J Clin Invest 1997; (99): 2118-2127.

Barry O P, Pratico D, Savani R C, FitzGerald G A. Modulation of monocyte-endothelial cell interactions by platelet microparticles. J Clin Invest 1998; (102): 136-144.

Bartunek J, Vanderheyden M, Vandekerckhove B, Mansour S, De B B, De B P, Van H, I, Lootens N, Heyndrickx G, Wijns W. Intracoronary injection of CD133-positive enriched bone marrow progenitor cells promotes cardiac recovery after recent myocardial infarction: feasibility and safety. Circulation 2005; (112): I178-I183.

Becker B F, Heindl B, Kupatt C, Zahler S. Endothelial function and hemostasis. Z Kardiol 2000; (89): 160-167.

Belhocine T, Blockmans D, Hustinx R, Vandevivere J, Mortelmans L. Imaging of large vessel vasculitis with (18)FDG PET: illusion or reality? A critical review of the literature data. Eur J Nucl Med Mol Imaging 2003; (30): 1305-1313.

Belizna C, Tervaert J W. Specificity, pathogenecity, and clinical value of antiendothelial cell antibodies. Semin Arthritis Rheum 1997; (27): 98-109.

Bellomo D, Headrick J P, Silins G U, Paterson C A, Thomas P S, Gartside M, Mould A, Cahill M M, Tonks I D, Grimmond S M, Townson S, Wells C, Little M, Cummings M C, Hayward N K, Kay G F. Mice lacking the vascular endothelial growth factor-B gene (Vegfb) have smaller hearts, dysfunctional coronary vasculature, and impaired recovery from cardiac ischemia. Circ Res 2000; (86): E29-E35.

Belostotsky V M, Shah V, Dillon M J. Clinical features in 17 paediatric patients with Wegener granulomatosis. Pediatr Nephrol 2002; (17): 754-761.

Bely M, Apathy A, Beke-Martos E. Cardiac changes in rheumatoid arthritis. Acta Morphol Hung 1992; (40): 149-186.

Benoist C, Mathis D. Autoimmunity. The pathogen connection. Nature 1998; (394): 227-228.

Bernal-Mizrachi L, Jy W, Jimenez J J, Pastor J, Mauro L M, Horstman L L, de M E, Ahn Y S. High levels of circulating endothelial microparticles in patients with acute coronary syndromes. Am Heart J 2003; (145): 962-970.

Berse B, Brown L F, van de W L, Dvorak H F, Senger D R. Vascular permeability factor (vascular endothelial growth factor) gene is expressed differentially in normal tissues, macrophages, and tumors. Mol Biol Cell 1992; (3): 211-220.

Bertolini F, Shaked Y, Mancuso P, Kerbel R S. The multifaceted circulating endothelial cell in cancer: towards marker and target identification. Nat Rev Cancer 2006; (6): 835-845. 362

Besbas N, Ozen S, Saatci U, Topaloglu R, Tinaztepe K, Bakkaloglu A. Renal involvement in polyarteritis nodosa: evaluation of 26 Turkish children. Pediatr Nephrol 2000; (14): 325-327.

Bettelli E, Korn T, Kuchroo V K. Th17: the third member of the effector T cell trilogy. Curr Opin Immunol 2007; (19): 652-657.

Bevers E M, Comfurius P, van Rijn J L, Hemker H C, Zwaal R F. Generation of prothrombin-converting activity and the exposure of phosphatidylserine at the outer surface of platelets. Eur J Biochem 1982; (122): 429-436. bid Hussein M N, Boing A N, Biro E, Hoek F J, Vogel G M, Meuleman D G, Sturk A, Nieuwland R. Phospholipid composition of in vitro endothelial microparticles and their in vivo thrombogenic properties. Thromb Res 2008; (121): 865-871. bid Hussein M N, Meesters E W, Osmanovic N, Romijn F P, Nieuwland R, Sturk A. Antigenic characterization of endothelial cell-derived microparticles and their detection ex vivo. J Thromb Haemost 2003; (1): 2434-2443.

Biezeveld M H, Geissler J, Weverling G J, Kuipers I M, Lam J, Ottenkamp J, Kuijpers T W. Polymorphisms in the mannose-binding lectin gene as determinants of age- defined risk of coronary artery lesions in Kawasaki disease. Arthritis Rheum 2006; (54): 369-376.

Biomarkers definitions working group. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther 2001; (69): 89-95.

Bjerkeli V, Damas J K, Fevang B, Holter J C, Aukrust P, Froland S S. Increased expression of fractalkine (CX3CL1) and its receptor, CX3CR1, in Wegener's granulomatosis--possible role in vascular inflammation. Rheumatology (Oxford) 2007; (46): 1422-1427.

Bland J M, Altman D G. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; (1): 307-310.

Blank M, Krause I, Goldkorn T, Praprotnik S, Livneh A, Langevitz P, Kaganovsky E, Morgenstern S, Cohen S, Barak V, Eldor A, Weksler B, Shoenfeld Y. Monoclonal anti-endothelial cell antibodies from a patient with Takayasu arteritis activate endothelial cells from large vessels. Arthritis Rheum 1999; (42): 1421-1432.

Blann A D, Woywodt A, Bertolini F, Bull T M, Buyon J P, Clancy R M, Haubitz M, Hebbel R P, Lip G Y, Mancuso P, Sampol J, Solovey A, gnat-George F. Circulating endothelial cells. Biomarker of vascular disease. Thromb Haemost 2005; (93): 228- 235.

Boehme M W, Schmitt W H, Youinou P, Stremmel W R, Gross W L. Clinical relevance of elevated serum thrombomodulin and soluble E-selectin in patients with Wegener's granulomatosis and other systemic vasculitides. Am J Med 1996; (101): 387-394.

363

Bompais H, Chagraoui J, Canron X, Crisan M, Liu X H, Anjo A, Tolla-Le P C, Leboeuf M, Charbord P, Bikfalvi A, Uzan G. Human endothelial cells derived from circulating progenitors display specific functional properties compared with mature vessel wall endothelial cells. Blood 2004; (103): 2577-2584.

Boos C J, Lane D A, Karpha M, Beevers D G, Haynes R, Lip G Y. Circulating endothelial cells, arterial stiffness, and cardiovascular risk stratification in hypertension. Chest 2007; (132): 1540-1547.

Booth A, Harper L, Hammad T, Bacon P, Griffith M, Levy J, Savage C, Pusey C, Jayne D. Prospective study of TNFalpha blockade with infliximab in anti-neutrophil cytoplasmic antibody-associated systemic vasculitis. J Am Soc Nephrol 2004a; (15): 717-721.

Booth A D, Jayne D R, Kharbanda R K, McEniery C M, Mackenzie I S, Brown J, Wilkinson I B. Infliximab improves endothelial dysfunction in systemic vasculitis: a model of vascular inflammation. Circulation 2004b; (109): 1718-1723.

Bordron A, Dueymes M, Levy Y, Jamin C, Leroy J P, Piette J C, Shoenfeld Y, Youinou P Y. The binding of some human antiendothelial cell antibodies induces endothelial cell apoptosis. J Clin Invest 1998; (101): 2029-2035.

Bosch X, Guilabert A, Font J. Antineutrophil cytoplasmic antibodies. Lancet 2006; (368): 404-418. bou-Raya A, bou-Raya S. Inflammation: a pivotal link between autoimmune diseases and atherosclerosis. Autoimmun Rev 2006; (5): 331-337.

Boyer D, Vargas S O, Slattery D, Rivera-Sanchez Y M, Colin A A. Churg-Strauss syndrome in children: a clinical and pathologic review. Pediatrics 2006; (118): e914- e920.

Braun M G, Csernok E, Gross W L, Muller-Hermelink H K. Proteinase 3, the target antigen of anticytoplasmic antibodies circulating in Wegener's granulomatosis. Immunolocalization in normal and pathologic tissues. Am J Pathol 1991; (139): 831- 838.

Breunis W B, Biezeveld M H, Geissler J, Ottenkamp J, Kuipers I M, Lam J, Hutchinson A, Welch R, Chanock S J, Kuijpers T W. Vascular endothelial growth factor gene haplotypes in Kawasaki disease. Arthritis Rheum 2006; (54): 1588-1594.

Brogan P A. What's new in the aetiopathogenesis of vasculitis? Pediatr Nephrol 2007; (22): 1083-1094.

Brogan P A, Davies R, Gordon I, Dillon M J. Renal angiography in children with polyarteritis nodosa. Pediatr Nephrol 2002; (17): 277-283.

Brogan P A, Dillon M J. Vasculitis from the pediatric perspective. Curr Rheumatol Rep 2000b; (2): 411-416.

364

Brogan P A, Dillon M J. The use of immunosuppressive and cytotoxic drugs in non- malignant disease. Arch Dis Child 2000a; (83): 259-264.

Brogan P A, Shah V, Brachet C, Harnden A, Mant D, Klein N, Dillon M J. Endothelial and platelet microparticles in vasculitis of the young. Arthritis Rheum 2004a; (50): 927-936.

Brogan P A, Shah V, Clarke L A, Dillon M J, Klein N. T cell activation profiles in Kawasaki syndrome. Clin Exp Immunol 2008; (151): 267-274.

Brogan P A, Shah V, Klein N, Dillon M J. Vbeta-restricted T cell adherence to endothelial cells: a mechanism for superantigen-dependent vascular injury. Arthritis Rheum 2004b; (50): 589-597.

Brogan R J, Eleftheriou D, Gnanapragasam J, Klein N J, Brogan P A. Infliximab for the treatment of intravenous immunoglobulin resistant Kawasaki disease complicated by coronary artery aneurysms: a case report. Pediatr Rheumatol Online J 2009; (7): 3.

Bronstein D E, Dille A N, Austin J P, Williams C M, Palinkas L A, Burns J C. Relationship of climate, ethnicity and socioeconomic status to Kawasaki disease in San Diego County, 1994 through 1998. Pediatr Infect Dis J 2000; (19): 1087-1091.

Brunner M, Thurnher D, Heiduschka G, Grasl M C, Brostjan C, Erovic B M. Elevated levels of circulating endothelial progenitor cells in head and neck cancer patients. J Surg Oncol 2008.

Bunting S, Gryglewski R, Moncada S, Vane J R. Arterial walls generate from prostaglandin endoperoxides a substance (prostaglandin X) which relaxes strips of mesenteric and coeliac ateries and inhibits platelet aggregation. Prostaglandins 1976; (12): 897-913.

Burchfield J S, Iwasaki M, Koyanagi M, Urbich C, Rosenthal N, Zeiher A M, Dimmeler S. Interleukin-10 from transplanted bone marrow mononuclear cells contributes to cardiac protection after myocardial infarction. Circ Res 2008; (103): 203-211.

Burns J C, Mason W H, Hauger S B, Janai H, Bastian J F, Wohrley J D, Balfour I, Shen C A, Michel E D, Shulman S T, Melish M E. Infliximab treatment for refractory Kawasaki syndrome. J Pediatr 2005a; (146): 662-667.

Burns J C, Shimizu C, Gonzalez E, Kulkarni H, Patel S, Shike H, Sundel R S, Newburger J W, Ahuja S K. Genetic variations in the receptor-ligand pair CCR5 and CCL3L1 are important determinants of susceptibility to Kawasaki disease. J Infect Dis 2005b; (192): 344-349.

Cabrera G E, Citera G, Gutierrez M, Scopelitis E, Espinoza L R. Digital vasculitis following allergic desensitization treatment. J Rheumatol 1993; (20): 1970-1972.

Cacoub P, Costedoat-Chalumeau N, Lidove O, Alric L. Cryoglobulinemia vasculitis. Curr Opin Rheumatol 2002; (14): 29-35.

365

Calabrese L H, Michel B A, Bloch D A, Arend W P, Edworthy S M, Fauci A S, Fries J F, Hunder G G, Leavitt R Y, Lie J T, . The American College of Rheumatology 1990 criteria for the classification of hypersensitivity vasculitis. Arthritis Rheum 1990; (33): 1108-1113.

Camerer E, Kolsto A B, Prydz H. Cell biology of tissue factor, the principal initiator of blood coagulation. Thromb Res 1996; (81): 1-41.

Carmeliet P, Moons L, Luttun A, Vincenti V, Compernolle V, De M M, Wu Y, Bono F, Devy L, Beck H, Scholz D, Acker T, DiPalma T, Dewerchin M, Noel A, Stalmans I, Barra A, Blacher S, Vandendriessche T, Ponten A, Eriksson U, Plate K H, Foidart J M, Schaper W, Charnock-Jones D S, Hicklin D J, Herbert J M, Collen D, Persico M G. Synergism between vascular endothelial growth factor and placental growth factor contributes to angiogenesis and plasma extravasation in pathological conditions. Nat Med 2001; (7): 575-583.

Carmeliet P, Ng Y S, Nuyens D, Theilmeier G, Brusselmans K, Cornelissen I, Ehler E, Kakkar V V, Stalmans I, Mattot V, Perriard J C, Dewerchin M, Flameng W, Nagy A, Lupu F, Moons L, Collen D, D'Amore P A, Shima D T. Impaired myocardial angiogenesis and ischemic cardiomyopathy in mice lacking the vascular endothelial growth factor isoforms VEGF164 and VEGF188. Nat Med 1999; (5): 495-502.

Case J, Mead L E, Bessler W K, Prater D, White H A, Saadatzadeh M R, Bhavsar J R, Yoder M C, Haneline L S, Ingram D A. Human CD34+AC133+VEGFR-2+ cells are not endothelial progenitor cells but distinct, primitive hematopoietic progenitors. Exp Hematol 2007; (35): 1109-1118.

Castaman G, Yu-Feng L, Battistin E, Rodeghiero F. Characterization of a novel bleeding disorder with isolated prolonged bleeding time and deficiency of platelet microvesicle generation. Br J Haematol 1997; (96): 458-463.

Cha H J, Jeong M J, Kleinman H K. Role of thymosin beta4 in tumor metastasis and angiogenesis. J Natl Cancer Inst 2003; (95): 1674-1680.

Chamberlain M A. Behcet's syndrome in 32 patients in Yorkshire. Ann Rheum Dis 1977; (36): 491-499.

Chang H K, Jang W C, Park S B, Han S M, Nam Y H, Lee S S, Kim J U, Lee H S. Association between interleukin 6 gene polymorphisms and Behcet's disease in Korean people. Ann Rheum Dis 2005; (64): 339-340.

Cheng J, Baumhueter S, Cacalano G, Carver-Moore K, Thibodeaux H, Thomas R, Broxmeyer H E, Cooper S, Hague N, Moore M, Lasky L A. Hematopoietic defects in mice lacking the sialomucin CD34. Blood 1996; (87): 479-490.

Cheung Y F, Brogan P A, Pilla C B, Dillon M J, Redington A N. Arterial distensibility in children and teenagers: normal evolution and the effect of childhood vasculitis. Arch Dis Child 2002; (87): 348-351.

366

Cheung Y F, Huang G Y, Chen S B, Liu X Q, Xi L, Liang X C, Huang M R, Chen S, Huang L S, Liu X Q, Chan K W, Lau Y L. Inflammatory gene polymorphisms and susceptibility to kawasaki disease and its arterial sequelae. Pediatrics 2008; (122): e608-e614.

Chi W, Zhu X, Yang P, Liu X, Lin X, Zhou H, Huang X, Kijlstra A. Upregulated IL-23 and IL-17 in Behcet patients with active uveitis. Invest Ophthalmol Vis Sci 2008; (49): 3058-3064.

Chironi G, Pagnoux C, Simon A, Pasquinelli-Balice M, Del-Pino M, Gariepy J, Guillevin L. Increased prevalence of subclinical atherosclerosis in patients with small-vessel vasculitis. Heart 2007; (93): 96-99.

Christensson M, Pettersson E, Sundqvist K G, Christensson B. T cell activation in patients with ANCA-associated vasculitis: inefficient immune suppression by therapy. Clin Nephrol 2000; (54): 435-442.

Chu K, Jung K H, Lee S T, Park H K, Sinn D I, Kim J M, Kim D H, Kim J H, Kim S J, Song E C, Kim M, Lee S K, Roh J K. Circulating endothelial progenitor cells as a new marker of endothelial dysfunction or repair in acute stroke. Stroke 2008; (39): 1441-1447.

Chun J K, Lee T J, Choi K M, Lee K H, Kim D S. Elevated anti-alpha-enolase antibody levels in Kawasaki disease. Scand J Rheumatol 2008; (37): 48-52.

Cid M C. Endothelial cell biology, perivascular inflammation, and vasculitis. Cleve Clin J Med 2002; (69 Suppl 2): SII45-SII49.

Cines D B, Pollak E S, Buck C A, Loscalzo J, Zimmerman G A, McEver R P, Pober J S, Wick T M, Konkle B A, Schwartz B S, Barnathan E S, McCrae K R, Hug B A, Schmidt A M, Stern D M. Endothelial cells in physiology and in the pathophysiology of vascular disorders. Blood 1998; (91): 3527-3561.

Clarke L A, Shah V, Arrigoni F, Eleftheriou D, Hong Y, Halcox J, Klein N, Brogan P A. Quantitative detection of circulating endothelial cells in vasculitis: comparison of flow cytometry and immunomagnetic bead extraction. J Thromb Haemost 2008; (6): 1025-1032.

Claudy A. Pathogenesis of leukocytoclastic vasculitis. Eur J Dermatol 1998; (8): 75-79.

Clayton A R, Prue R L, Harper L, Drayson M T, Savage C O. Dendritic cell uptake of human apoptotic and necrotic neutrophils inhibits CD40, CD80, and CD86 expression and reduces allogeneic T cell responses: relevance to systemic vasculitis. Arthritis Rheum 2003; (48): 2362-2374.

Coll-Vinent B, Cebrian M, Cid M C, Font C, Esparza J, Juan M, Yague J, Urbano- Marquez A, Grau J M. Dynamic pattern of endothelial cell adhesion molecule expression in muscle and perineural vessels from patients with classic polyarteritis nodosa. Arthritis Rheum 1998; (41): 435-444.

367

Coll-Vinent B, Grau J M, Lopez-Soto A, Oristrell J, Font C, Bosch X, Mirapeix E, Urbano-Marquez A, Cid M C. Circulating soluble adhesion molecules in patients with classical polyarteritis nodosa. Br J Rheumatol 1997; (36): 1178-1183.

Combes V, gnat-George F, Mutin M, Sampol J. A new flow cytometry method of platelet-derived microvesicle quantitation in plasma. Thromb Haemost 1997; (77): 220.

Combes V, Simon A C, Grau G E, Arnoux D, Camoin L, Sabatier F, Mutin M, Sanmarco M, Sampol J, gnat-George F. In vitro generation of endothelial microparticles and possible prothrombotic activity in patients with lupus anticoagulant. J Clin Invest 1999; (104): 93-102.

Csernok E, Ernst M, Schmitt W, Bainton D F, Gross W L. Activated neutrophils express proteinase 3 on their plasma membrane in vitro and in vivo. Clin Exp Immunol 1994; (95): 244-250.

Cunningham M A, Huang X R, Dowling J P, Tipping P G, Holdsworth S R. Prominence of cell-mediated immunity effectors in "pauci-immune" glomerulonephritis. J Am Soc Nephrol 1999; (10): 499-506.

D'Cruz D, Direskeneli H, Khamashta M, Hughes G R. Lymphocyte activation markers and von Willebrand factor antigen in Wegener's granulomatosis: potential markers for disease activity. J Rheumatol 1999; (26): 103-109.

Daleke D L. Regulation of transbilayer plasma membrane phospholipid asymmetry. J Lipid Res 2003; (44): 233-242.

Dalli J, Norling L V, Renshaw D, Cooper D, Leung K Y, Perretti M. Annexin 1 mediates the rapid anti-inflammatory effects of neutrophil-derived microparticles. Blood 2008; (112): 2512-2519.

Daly C, Wong V, Burova E, Wei Y, Zabski S, Griffiths J, Lai K M, Lin H C, Ioffe E, Yancopoulos G D, Rudge J S. Angiopoietin-1 modulates endothelial cell function and gene expression via the transcription factor FKHR (FOXO1). Genes Dev 2004; (18): 1060-1071.

Damianovich M, Gilburd B, George J, Del P N, Afek A, Goldberg I, Kopolovic Y, Roth D, Barkai G, Meroni P L, Shoenfeld Y. Pathogenic role of anti-endothelial cell antibodies in vasculitis. An idiotypic experimental model. J Immunol 1996; (156): 4946-4951.

Daniel L, Fakhouri F, Joly D, Mouthon L, Nusbaum P, Grunfeld J P, Schifferli J, Guillevin L, Lesavre P, Halbwachs-Mecarelli L. Increase of circulating neutrophil and platelet microparticles during acute vasculitis and hemodialysis. Kidney Int 2006; (69): 1416-1423.

Darzynkiewicz Z, Bruno S, Del B G, Gorczyca W, Hotz M A, Lassota P, Traganos F. Features of apoptotic cells measured by flow cytometry. Cytometry 1992; (13): 795- 808.

368

Darzynkiewicz Z, Juan G, Li X, Gorczyca W, Murakami T, Traganos F. Cytometry in cell necrobiology: analysis of apoptosis and accidental cell death (necrosis). Cytometry 1997; (27): 1-20. de Groot K, Goldberg C, Bahlmann F H, Woywodt A, Haller H, Fliser D, Haubitz M. Vascular endothelial damage and repair in antineutrophil cytoplasmic antibody- associated vasculitis. Arthritis Rheum 2007; (56): 3847-3853. de la Fuente F R, Grana G J. Anticardiolipin antibodies and polyarteritis nodosa. Lupus 1994; (3): 523-524. de B M, Meyer O, Hakim J, Pasquier C. Antibodies to proteinase-3 mediate expression of intercellular adhesion molecule-1 (ICAM-1, CD 54). Br J Rheumatol 1997a; (36): 839-846. de B M, Ollivier V, Meyer O, Babin-Chevaye C, Khechai F, de P D, Hakim J, Pasquier C. Induction of interleukin-1 and subsequent tissue factor expression by antiproteinase 3 antibodies in human umbilical vein endothelial cells. Arthritis Rheum 1997b; (40): 2030-2038.

DeBusk L M, Hallahan D E, Lin P C. Akt is a major angiogenic mediator downstream of the Ang1/Tie2 signaling pathway. Exp Cell Res 2004; (298): 167-177.

Del Buono E A, Flint A. Diagnostic usefulness of nasal biopsy in Wegener's granulomatosis. Hum Pathol 1991; (22): 107-110.

Del P N, Colombo G, Fracchiolla N, Moronetti L M, Ingegnoli F, Maglione W, Comina D P, Vitali C, Fantini F, Cortelezzi A. Circulating endothelial cells as a marker of ongoing vascular disease in systemic sclerosis. Arthritis Rheum 2004; (50): 1296- 1304.

Del P N, Guidali L, Sironi M, Shoenfeld Y, Mantovani A, Tincani A, Balestrieri G, Radice A, Sinico R A, Meroni P L. Anti-endothelial cell IgG antibodies from patients with Wegener's granulomatosis bind to human endothelial cells in vitro and induce adhesion molecule expression and cytokine secretion. Arthritis Rheum 1996; (39): 758-766.

Delorme B, Basire A, Gentile C, Sabatier F, Monsonis F, Desouches C, Blot-Chabaud M, Uzan G, Sampol J, gnat-George F. Presence of endothelial progenitor cells, distinct from mature endothelial cells, within human CD146+ blood cells. Thromb Haemost 2005; (94): 1270-1279.

Densmore J C, Signorino P R, Ou J, Hatoum O A, Rowe J J, Shi Y, Kaul S, Jones D W, Sabina R E, Pritchard K A, Jr., Guice K S, Oldham K T. Endothelium-derived microparticles induce endothelial dysfunction and acute lung injury. Shock 2006; (26): 464-471.

Dhawan V, Mahajan N, Jain S. Role of C-C chemokines in Takayasu's arteritis disease. Int J Cardiol 2006; (112): 105-111.

369

Diamant M, Tushuizen M E, Sturk A, Nieuwland R. Cellular microparticles: new players in the field of vascular disease? Eur J Clin Invest 2004; (34): 392-401.

Dillon M J. Childhood vasculitis. Lupus 1998; (7): 259-265.

Dinc A, Takafuta T, Jiang D, Melikoglu M, Saruhan-Direskeneli G, Shapiro S S. Anti- endothelial cell antibodies in Behcet's disease. Clin Exp Rheumatol 2003; (21): S27- S30.

Dolman K M, Stegeman C A, van de Wiel B A, Hack C E, von dem Borne A E, Kallenberg C G, Goldschmeding R. Relevance of classic anti-neutrophil cytoplasmic autoantibody (C-ANCA)-mediated inhibition of proteinase 3-alpha 1-antitrypsin complexation to disease activity in Wegener's granulomatosis. Clin Exp Immunol 1993; (93): 405-410.

Du X. Signaling and regulation of the platelet glycoprotein Ib-IX-V complex. Curr Opin Hematol 2007; (14): 262-269.

Duff S E, Li C, Garland J M, Kumar S. CD105 is important for angiogenesis: evidence and potential applications. FASEB J 2003; (17): 984-992.

Dustin M L, Rothlein R, Bhan A K, Dinarello C A, Springer T A. Induction by IL 1 and interferon-gamma: tissue distribution, biochemistry, and function of a natural adherence molecule (ICAM-1). J Immunol 1986; (137): 245-254.

Egan C G, Caporali F, Garcia-Gonzalez E, Galeazzi M, Sorrentino V. Endothelial progenitor cells and colony-forming units in rheumatoid arthritis: association with clinical characteristics. Rheumatology (Oxford) 2008a; (47): 1484-1488.

Egan C G, Lavery R, Caporali F, Fondelli C, Laghi-Pasini F, Dotta F, Sorrentino V. Generalised reduction of putative endothelial progenitors and CXCR4-positive peripheral blood cells in type 2 diabetes. Diabetologia 2008b; (51): 1296-1305.

El-Zammar O A, Katzenstein A L. Pathological diagnosis of granulomatous lung disease: a review. Histopathology 2007; (50): 289-310.

Eleftheriou D, Melo M, Marks SD, Tullus K, Cleary G, Dolezalova P, Ozen S, Pilkington C, Woo P, Klein N, Dillon MJ, Brogan PA. Biologic therapy in primary systemic vasculitis of the young. Rheumatology.(Oxford) . 2009. Ref Type: In Press

Elzouki A N, Segelmark M, Wieslander J, Eriksson S. Strong link between the alpha 1- antitrypsin PiZ allele and Wegener's granulomatosis. J Intern Med 1994; (236): 543- 548.

Emeis J J. Regulation of the acute release of tissue-type plasminogen activator from the endothelium by coagulation activation products. Ann N Y Acad Sci 1992; (667): 249-258.

Engelmann B, Luther T, Muller I. Intravascular tissue factor pathway--a model for rapid initiation of coagulation within the blood vessel. Thromb Haemost 2003; (89): 3-8. 370

Enjeti A K, Lincz L F, Seldon M. Detection and measurement of microparticles: an evolving research tool for vascular biology. Semin Thromb Hemost 2007; (33): 771- 779.

Erdbruegger U, Grossheim M, Hertel B, Wyss K, Kirsch T, Woywodt A, Haller H, Haubitz M. Diagnostic role of endothelial microparticles in vasculitis. Rheumatology (Oxford) 2008.

Erdbruegger U, Haubitz M, Woywodt A. Circulating endothelial cells: a novel marker of endothelial damage. Clin Chim Acta 2006; (373): 17-26.

Fadok V A, Savill J S, Haslett C, Bratton D L, Doherty D E, Campbell P A, Henson P M. Different populations of macrophages use either the vitronectin receptor or the phosphatidylserine receptor to recognize and remove apoptotic cells. J Immunol 1992; (149): 4029-4035.

Faure V, Dou L, Sabatier F, Cerini C, Sampol J, Berland Y, Brunet P, gnat-George F. Elevation of circulating endothelial microparticles in patients with chronic renal failure. J Thromb Haemost 2006; (4): 566-573.

Fazel S, Cimini M, Chen L, Li S, Angoulvant D, Fedak P, Verma S, Weisel R D, Keating A, Li R K. Cardioprotective c-kit+ cells are from the bone marrow and regulate the myocardial balance of angiogenic cytokines. J Clin Invest 2006; (116): 1865-1877.

Feletou M, Vanhoutte P M. Endothelial dysfunction: a multifaceted disorder (The Wiggers Award Lecture). Am J Physiol Heart Circ Physiol 2006; (291): H985-1002.

Feng D, Nagy J A, Pyne K, Dvorak H F, Dvorak A M. Neutrophils emigrate from venules by a transendothelial cell pathway in response to FMLP. J Exp Med 1998; (187): 903-915.

Ferrara N, Kerbel R S. Angiogenesis as a therapeutic target. Nature 2005; (438): 967- 974.

Ferrara N, Winer J, Burton T. Aortic smooth muscle cells express and secrete vascular endothelial growth factor. Growth Factors 1991; (5): 141-148.

Fiedler U, Augustin H G. Angiopoietins: a link between angiogenesis and inflammation. Trends Immunol 2006; (27): 552-558.

Fiedler U, Reiss Y, Scharpfenecker M, Grunow V, Koidl S, Thurston G, Gale N W, Witzenrath M, Rosseau S, Suttorp N, Sobke A, Herrmann M, Preissner K T, Vajkoczy P, Augustin H G. Angiopoietin-2 sensitizes endothelial cells to TNF-alpha and has a crucial role in the induction of inflammation. Nat Med 2006; (12): 235- 239.

Finn A, Strobel S, Levin M, Klein N. Endotoxin-induced neutrophil adherence to endothelium: relationship to CD11b/CD18 and L-selectin expression and matrix disruption. Ann N Y Acad Sci 1994; (725): 173-182.

371

Fong G H, Rossant J, Gertsenstein M, Breitman M L. Role of the Flt-1 receptor tyrosine kinase in regulating the assembly of vascular endothelium. Nature 1995; (376): 66- 70.

Fox A, Smythe J, Fisher N, Tyler M P, McGrouther D A, Watt S M, Harris A L. Mobilization of endothelial progenitor cells into the circulation in burned patients. Br J Surg 2008; (95): 244-251.

Franssen C F, Stegeman C A, Kallenberg C G, Gans R O, de Jong P E, Hoorntje S J, Tervaert J W. Antiproteinase 3- and antimyeloperoxidase-associated vasculitis. Kidney Int 2000; (57): 2195-2206.

Fries J F, Hunder G G, Bloch D A, Michel B A, Arend W P, Calabrese L H, Fauci A S, Leavitt R Y, Lie J T, Lightfoot R W, Jr., . The American College of Rheumatology 1990 criteria for the classification of vasculitis. Summary. Arthritis Rheum 1990; (33): 1135-1136.

Fritzsching B, Schwer B, Kartenbeck J, Pedal A, Horejsi V, Ott M. Release and intercellular transfer of cell surface CD81 via microparticles. J Immunol 2002; (169): 5531-5537.

Fujieda M, Oishi N, Kurashige T. Antibodies to endothelial cells in Kawasaki disease lyse endothelial cells without cytokine pretreatment. Clin Exp Immunol 1997; (107): 120-126.

Furie B, Furie B C. Mechanisms of thrombus formation. N Engl J Med 2008; (359): 938-949.

Furie B, Furie B C. Cancer-associated thrombosis. Blood Cells Mol Dis 2006; (36): 177-181.

Furui J, Ishii M, Ikeda H, Muta H, Egami K, Sugahara Y, Himeno W, Akagi T, Kato H, Matsuishi T. Soluble forms of the selectin family in children with Kawasaki disease: prediction for coronary artery lesions. Acta Paediatr 2002; (91): 1183-1188.

Furuno K, Yuge T, Kusuhara K, Takada H, Nishio H, Khajoee V, Ohno T, Hara T. CD25+CD4+ regulatory T cells in patients with Kawasaki disease. J Pediatr 2004; (145): 385-390.

Futamura A, Izumino K, Sugawara H, Nakagawa Y, Inoue H, Takata M. Effects of leukocytosis and macrophage activation on anti-Thy 1.1 glomerulonephritis in the rat. Scand J Urol Nephrol 2002; (36): 435-442.

Gajkowska A, Oldak T, Jastrzewska M, Machaj E K, Walewski J, Kraszewska E, Pojda Z. Flow cytometric enumeration of CD34+ hematopoietic stem and progenitor cells in leukapheresis product and bone marrow for clinical transplantation: a comparison of three methods. Folia Histochem Cytobiol 2006; (44): 53-60.

Gale N W, Dominguez M G, Noguera I, Pan L, Hughes V, Valenzuela D M, Murphy A J, Adams N C, Lin H C, Holash J, Thurston G, Yancopoulos G D.

372

Haploinsufficiency of delta-like 4 ligand results in embryonic lethality due to major defects in arterial and vascular development. Proc Natl Acad Sci U S A 2004; (101): 15949-15954.

Gale N W, Thurston G, Hackett S F, Renard R, Wang Q, McClain J, Martin C, Witte C, Witte M H, Jackson D, Suri C, Campochiaro P A, Wiegand S J, Yancopoulos G D. Angiopoietin-2 is required for postnatal angiogenesis and lymphatic patterning, and only the latter role is rescued by Angiopoietin-1. Dev Cell 2002; (3): 411-423.

Ganong WF. Dynamics of blood and lymph flow. In: Review of medical physiology. New York: McGraw-Hill, 2001; 20: 556-573.

Ganter M T, Cohen M J, Brohi K, Chesebro B B, Staudenmayer K L, Rahn P, Christiaans S C, Bir N D, Pittet J F. Angiopoietin-2, marker and mediator of endothelial activation with prognostic significance early after trauma? Ann Surg 2008; (247): 320-326.

Gardner-Medwin J M, Dolezalova P, Cummins C, Southwood T R. Incidence of Henoch-Schonlein purpura, Kawasaki disease, and rare vasculitides in children of different ethnic origins. Lancet 2002; (360): 1197-1202.

Gaudry M, Bregerie O, Andrieu V, El B J, Pocidalo M A, Hakim J. Intracellular pool of vascular endothelial growth factor in human neutrophils. Blood 1997; (90): 4153- 4161.

Gayraud M, Guillevin L, le T P, Cohen P, Lhote F, Casassus P, Jarrousse B. Long-term followup of polyarteritis nodosa, microscopic polyangiitis, and Churg-Strauss syndrome: analysis of four prospective trials including 278 patients. Arthritis Rheum 2001; (44): 666-675.

Gearing A J, Newman W. Circulating adhesion molecules in disease. Immunol Today 1993; (14): 506-512.

Gencik M, Meller S, Borgmann S, Fricke H. Proteinase 3 gene polymorphisms and Wegener's granulomatosis. Kidney Int 2000; (58): 2473-2477.

Genta M S, Genta R M, Gabay C. Systemic rheumatoid vasculitis: a review. Semin Arthritis Rheum 2006; (36): 88-98.

George F, Brisson C, Poncelet P, Laurent J C, Massot O, Arnoux D, Ambrosi P, Klein- Soyer C, Cazenave J P, Sampol J. Rapid isolation of human endothelial cells from whole blood using S-Endo1 monoclonal antibody coupled to immuno-magnetic beads: demonstration of endothelial injury after angioplasty. Thromb Haemost 1992; (67): 147-153.

George F, Brouqui P, Boffa M C, Mutin M, Drancourt M, Brisson C, Raoult D, Sampol J. Demonstration of Rickettsia conorii-induced endothelial injury in vivo by measuring circulating endothelial cells, thrombomodulin, and von Willebrand factor in patients with Mediterranean spotted fever. Blood 1993; (82): 2109-2116.

373

George J, Afek A, Abashidze A, Shmilovich H, Deutsch V, Kopolovich J, Miller H, Keren G. Transfer of endothelial progenitor and bone marrow cells influences atherosclerotic plaque size and composition in apolipoprotein E knockout mice. Arterioscler Thromb Vasc Biol 2005; (25): 2636-2641.

George J, Shmilovich H, Deutsch V, Miller H, Keren G, Roth A. Comparative analysis of methods for assessment of circulating endothelial progenitor cells. Tissue Eng 2006; (12): 331-335.

Gilligan H M, Bredy B, Brady H R, Hebert M J, Slayter H S, Xu Y, Rauch J, Shia M A, Koh J S, Levine J S. Antineutrophil cytoplasmic autoantibodies interact with primary granule constituents on the surface of apoptotic neutrophils in the absence of neutrophil priming. J Exp Med 1996; (184): 2231-2241.

Goldsby RA, Kindt TJ, Osborne BA, Kuby J. The complement system. In: Immunology. New York: W.H Freeman, 2003b; 5th: 299-318.

Goldsby RA, Kindt TJ, Osborne BA, Kuby J. Cells and organs of the immune system. In: Immunology. New York: W.H Freeman, 2003a; 5th: 24-56.

Goon P K, Boos C J, Stonelake P S, Blann A D, Lip G Y. Detection and quantification of mature circulating endothelial cells using flow cytometry and immunomagnetic beads: a methodological comparison. Thromb Haemost 2006; (96): 45-52.

Gotway M B, Araoz P A, Macedo T A, Stanson A W, Higgins C B, Ring E J, Dawn S K, Webb W R, Leung J W, Reddy G P. Imaging findings in Takayasu's arteritis. AJR Am J Roentgenol 2005; (184): 1945-1950.

Grau G E, Roux-Lombard P, Gysler C, Lambert C, Lambert P H, Dayer J M, Guillevin L. Serum cytokine changes in systemic vasculitis. Immunology 1989; (68): 196-198.

Guillevin L, Dorner T. Vasculitis: mechanisms involved and clinical manifestations. Arthritis Res Ther 2007; (9 Suppl 2): S9.

Guilpain P, Servettaz A, Goulvestre C, Barrieu S, Borderie D, Chereau C, Kavian N, Pagnoux C, Guillevin L, Weill B, Mouthon L, Batteux F. Pathogenic effects of antimyeloperoxidase antibodies in patients with microscopic polyangiitis. Arthritis Rheum 2007; (56): 2455-2463.

Gulati R, Jevremovic D, Peterson T E, Chatterjee S, Shah V, Vile R G, Simari R D. Diverse origin and function of cells with endothelial phenotype obtained from adult human blood. Circ Res 2003; (93): 1023-1025.

Guo D Q, Wu L W, Dunbar J D, Ozes O N, Mayo L D, Kessler K M, Gustin J A, Baerwald M R, Jaffe E A, Warren R S, Donner D B. Tumor necrosis factor employs a protein-tyrosine phosphatase to inhibit activation of KDR and vascular endothelial cell growth factor-induced endothelial cell proliferation. J Biol Chem 2000; (275): 11216-11221.

374

Guo R F, Ward P A. Role of C5a in inflammatory responses. Annu Rev Immunol 2005; (23): 821-852.

Hagen E C, Andrassy K, Csernok E, Daha M R, Gaskin G, Gross W L, Hansen B, Heigl Z, Hermans J, Jayne D, Kallenberg C G, Lesavre P, Lockwood C M, Ludemann J, Mascart-Lemone F, Mirapeix E, Pusey C D, Rasmussen N, Sinico R A, Tzioufas A, Wieslander J, Wiik A, van der Woude F J. Development and standardization of solid phase assays for the detection of anti-neutrophil cytoplasmic antibodies (ANCA). A report on the second phase of an international cooperative study on the standardization of ANCA assays. J Immunol Methods 1996; (196): 1-15.

Hamilton K K, Hattori R, Esmon C T, Sims P J. Complement proteins C5b-9 induce vesiculation of the endothelial plasma membrane and expose catalytic surface for assembly of the prothrombinase enzyme complex. J Biol Chem 1990; (265): 3809- 3814.

Hamzaoui K, Hamzaoui A, Houman H. CD4+CD25+ regulatory T cells in patients with Behcet's disease. Clin Exp Rheumatol 2006; (24): S71-S78.

Hannappel E. beta-Thymosins. Ann N Y Acad Sci 2007; (1112): 21-37.

Harashima S, Horiuchi T, Hatta N, Morita C, Higuchi M, Sawabe T, Tsukamoto H, Tahira T, Hayashi K, Fujita S, Niho Y. Outside-to-inside signal through the membrane TNF-alpha induces E-selectin (CD62E) expression on activated human CD4+ T cells. J Immunol 2001; (166): 130-136.

Harnden A, Alves B, Sheikh A. Rising incidence of Kawasaki disease in England: analysis of hospital admission data. BMJ 2002; (324): 1424-1425.

Harper L, Cockwell P, Adu D, Savage C O. Neutrophil priming and apoptosis in antineutrophil cytoplasmic autoantibody-associated vasculitis. Kidney Int 2001; (59): 1729-1738.

Harper L, Ren Y, Savill J, Adu D, Savage C O. Antineutrophil cytoplasmic antibodies induce reactive oxygen-dependent dysregulation of primed neutrophil apoptosis and clearance by macrophages. Am J Pathol 2000; (157): 211-220.

Harper L, Williams J M, Savage C O. The importance of resolution of inflammation in the pathogenesis of ANCA-associated vasculitis. Biochem Soc Trans 2004; (32): 502-506.

Hattar K, Sibelius U, Bickenbach A, Csernok E, Seeger W, Grimminger F. Subthreshold concentrations of anti-proteinase 3 antibodies (c-ANCA) specifically prime human neutrophils for fMLP-induced leukotriene synthesis and chemotaxis. J Leukoc Biol 2001; (69): 89-97.

Heiss C, Keymel S, Niesler U, Ziemann J, Kelm M, Kalka C. Impaired progenitor cell activity in age-related endothelial dysfunction. J Am Coll Cardiol 2005; (45): 1441- 1448.

375

Hellstrom M, Gerhardt H, Kalen M, Li X, Eriksson U, Wolburg H, Betsholtz C. Lack of pericytes leads to endothelial hyperplasia and abnormal vascular morphogenesis. J Cell Biol 2001; (153): 543-553.

Hellstrom M, Kalen M, Lindahl P, Abramsson A, Betsholtz C. Role of PDGF-B and PDGFR-beta in recruitment of vascular smooth muscle cells and pericytes during embryonic blood vessel formation in the mouse. Development 1999; (126): 3047- 3055.

Henes J C, Muller M, Krieger J, Balletshofer B, Pfannenberg A C, Kanz L, Kotter I. [18F] FDG-PET/CT as a new and sensitive imaging method for the diagnosis of large vessel vasculitis. Clin Exp Rheumatol 2008; (26): S47-S52.

Hergesell O, Andrassy K, Nawroth P. Elevated levels of markers of endothelial cell damage and markers of activated coagulation in patients with systemic necrotizing vasculitis. Thromb Haemost 1996; (75): 892-898.

Hergesell O, Egbring R, Andrassy K. Presence of anticardiolipin antibodies discriminates between Wegener's granulomatosis and microscopic polyarteritis. Adv Exp Med Biol 1993; (336): 393-396.

Hill J M, Zalos G, Halcox J P, Schenke W H, Waclawiw M A, Quyyumi A A, Finkel T. Circulating endothelial progenitor cells, vascular function, and cardiovascular risk. N Engl J Med 2003; (348): 593-600.

Hirschi K K, Ingram D A, Yoder M C. Assessing identity, phenotype, and fate of endothelial progenitor cells. Arterioscler Thromb Vasc Biol 2008; (28): 1584-1595.

Hladovec J, Prerovsky I. Endothelial lesion in hypertension. Cor Vasa 1989; (31): 51- 54.

Hladovec J, Prerovsky I, Stanek V, Fabian J. Circulating endothelial cells in acute myocardial infarction and angina pectoris. Klin Wochenschr 1978; (56): 1033-1036.

Hoffman M, Monroe D M. Coagulation 2006: a modern view of hemostasis. Hematol Oncol Clin North Am 2007; (21): 1-11.

Holmen C, Elsheikh E, Stenvinkel P, Qureshi A R, Pettersson E, Jalkanen S, Sumitran- Holgersson S. Circulating inflammatory endothelial cells contribute to endothelial progenitor cell dysfunction in patients with vasculitis and kidney involvement. J Am Soc Nephrol 2005; (16): 3110-3120.

Holthofer H, Virtanen I, Kariniemi A L, Hormia M, Linder E, Miettinen A. Ulex europaeus I lectin as a marker for vascular endothelium in human tissues. Lab Invest 1982; (47): 60-66.

Hong Y M, Jin H S, Park I S, Hong S J. Association of the matrix metalloproteinase-3 (- 439C/G) promoter polymorphism with Kawasaki disease in Korean children. Heart Vessels 2008; (23): 341-347.

376

Horstman L L, Jy W, Jimenez J J, Bidot C, Ahn Y S. New horizons in the analysis of circulating cell-derived microparticles. Keio J Med 2004; (53): 210-230.

House S J, Potier M, Bisaillon J, Singer H A, Trebak M. The non-excitable smooth muscle: calcium signaling and phenotypic switching during vascular disease. Pflugers Arch 2008; (456): 769-785.

Hsia J, Sztein M B, Naylor P H, Simon G L, Goldstein A L, Hayden F G. Modulation of thymosin alpha 1 and thymosin beta 4 levels and peripheral blood mononuclear cell subsets during experimental rhinovirus colds. Lymphokine Res 1989; (8): 383-391.

Huang D, Giscombe R, Zhou Y, Lefvert A K. Polymorphisms in CTLA-4 but not tumor necrosis factor-alpha or interleukin 1beta genes are associated with Wegener's granulomatosis. J Rheumatol 2000; (27): 397-401.

Huang F Y, Chang T Y, Chen M R, Hsu C H, Lee H C, Lin S P, Kao H A, Chiu N C, Chi H, Liu T Y, Liu H F, Dang C W, Chu C C, Lin M, Sung T C, Lee Y J. Genetic variations of HLA-DRB1 and susceptibility to Kawasaki disease in Taiwanese children. Hum Immunol 2007; (68): 69-74.

Huang F Y, Chang T Y, Chen M R, Lee H C, Chi H, Chiu N C, Hsu C H, Lin S P, Kao H A, Chen W F, Chan H W, Liu H F, Chu C C, Lin M, Lee Y J. Lack of association of the vascular endothelial growth factor gene polymorphisms with Kawasaki disease in Taiwanese children. J Clin Immunol 2008; (28): 322-328.

Huber J, Vales A, Mitulovic G, Blumer M, Schmid R, Witztum J L, Binder B R, Leitinger N. Oxidized membrane vesicles and blebs from apoptotic cells contain biologically active oxidized phospholipids that induce monocyte-endothelial interactions. Arterioscler Thromb Vasc Biol 2002; (22): 101-107.

Hugel B, Martinez M C, Kunzelmann C, Freyssinet J M. Membrane microparticles: two sides of the coin. Physiology (Bethesda ) 2005; (20): 22-27.

Hughes D P, Marron M B, Brindle N P. The antiinflammatory endothelial tyrosine kinase Tie2 interacts with a novel nuclear factor-kappaB inhibitor ABIN-2. Circ Res 2003; (92): 630-636.

Hunting C B, Noort W A, Zwaginga J J. Circulating endothelial (progenitor) cells reflect the state of the endothelium: vascular injury, repair and neovascularization. Vox Sang 2005; (88): 1-9.

Hur J, Yoon C H, Kim H S, Choi J H, Kang H J, Hwang K K, Oh B H, Lee M M, Park Y B. Characterization of two types of endothelial progenitor cells and their different contributions to neovasculogenesis. Arterioscler Thromb Vasc Biol 2004; (24): 288- 293.

Hurwitz A A, Lyman W D, Guida M P, Calderon T M, Berman J W. Tumor necrosis factor alpha induces adhesion molecule expression on human fetal astrocytes. J Exp Med 1992; (176): 1631-1636.

377

Huugen D, Xiao H, van E A, Falk R J, Peutz-Kootstra C J, Buurman W A, Tervaert J W, Jennette J C, Heeringa P. Aggravation of anti-myeloperoxidase antibody-induced glomerulonephritis by bacterial lipopolysaccharide: role of tumor necrosis factor- alpha. Am J Pathol 2005; (167): 47-58.

Iba O, Matsubara H, Nozawa Y, Fujiyama S, Amano K, Mori Y, Kojima H, Iwasaka T. Angiogenesis by implantation of peripheral blood mononuclear cells and platelets into ischemic limbs. Circulation 2002; (106): 2019-2025.

Idil A, Gurler A, Boyvat A, Caliskan D, Ozdemir O, Isik A, Tuncbilek A, Kocyigit P, Calikoglu E. The prevalence of Behcet's disease above the age of 10 years. The results of a pilot study conducted at the Park Primary Health Care Center in Ankara, Turkey. Ophthalmic Epidemiol 2002; (9): 325-331.

Ingram D A, Mead L E, Tanaka H, Meade V, Fenoglio A, Mortell K, Pollok K, Ferkowicz M J, Gilley D, Yoder M C. Identification of a novel hierarchy of endothelial progenitor cells using human peripheral and umbilical cord blood. Blood 2004; (104): 2752-2760.

Ishiwata N, Takio K, Katayama M, Watanabe K, Titani K, Ikeda Y, Handa M. Alternatively spliced isoform of P-selectin is present in vivo as a soluble molecule. J Biol Chem 1994; (269): 23708-23715.

Ito N, Nishi K, Kawahara S, Okamura Y, Hirota T, Rand S, Fechner G, Brinkmann B. Difference in the ability of blood group-specific lectins and monoclonal antibodies to recognize the ABH antigens in human tissues. Histochem J 1990; (22): 604-614.

Jamin C, Dugue C, Alard J E, Jousse S, Saraux A, Guillevin L, Piette J C, Youinou P. Induction of endothelial cell apoptosis by the binding of anti-endothelial cell antibodies to Hsp60 in vasculitis-associated systemic autoimmune diseases. Arthritis Rheum 2005; (52): 4028-4038.

Janowska-Wieczorek A, Wysoczynski M, Kijowski J, Marquez-Curtis L, Machalinski B, Ratajczak J, Ratajczak M Z. Microvesicles derived from activated platelets induce metastasis and angiogenesis in lung cancer. Int J Cancer 2005; (113): 752-760.

Janssens S, Dubois C, Bogaert J, Theunissen K, Deroose C, Desmet W, Kalantzi M, Herbots L, Sinnaeve P, Dens J, Maertens J, Rademakers F, Dymarkowski S, Gheysens O, Van C J, Bormans G, Nuyts J, Belmans A, Mortelmans L, Boogaerts M, Van de W F. Autologous bone marrow-derived stem-cell transfer in patients with ST-segment elevation myocardial infarction: double-blind, randomised controlled trial. Lancet 2006; (367): 113-121.

Jaroszeski M J, Radcliff G. Fundamentals of flow cytometry. Mol Biotechnol 1999; (11): 37-53.

Jayne D. Update on the European Vasculitis Study Group trials. Curr Opin Rheumatol 2001; (13): 48-55.

378

Jayne D R, Gaskin G, Rasmussen N, Abramowicz D, Ferrario F, Guillevin L, Mirapeix E, Savage C O, Sinico R A, Stegeman C A, Westman K W, van der Woude F J, de Lind van Wijngaarden RA, Pusey C D. Randomized trial of plasma exchange or high-dosage methylprednisolone as adjunctive therapy for severe renal vasculitis. J Am Soc Nephrol 2007; (18): 2180-2188.

Jelkmann W. Pitfalls in the measurement of circulating vascular endothelial growth factor. Clin Chem 2001; (47): 617-623.

Jennette J C, Falk R J, Andrassy K, Bacon P A, Churg J, Gross W L, Hagen E C, Hoffman G S, Hunder G G, Kallenberg C G, . Nomenclature of systemic vasculitides. Proposal of an international consensus conference. Arthritis Rheum 1994; (37): 187-192.

Jie K E, Goossens M H, van O O, Lilien M R, Verhaar M C. Circulating endothelial progenitor cell levels are higher during childhood than in adult life. Atherosclerosis 2008.

Jimenez J J, Jy W, Mauro L M, Horstman L L, Ahn Y S. Elevated endothelial microparticles in thrombotic thrombocytopenic purpura: findings from brain and renal microvascular cell culture and patients with active disease. Br J Haematol 2001; (112): 81-90.

Jimenez J J, Jy W, Mauro L M, Soderland C, Horstman L L, Ahn Y S. Endothelial cells release phenotypically and quantitatively distinct microparticles in activation and apoptosis. Thromb Res 2003; (109): 175-180.

Johnston B, Butcher E C. Chemokines in rapid leukocyte adhesion triggering and migration. Semin Immunol 2002; (14): 83-92.

Joop K, Berckmans R J, Nieuwland R, Berkhout J, Romijn F P, Hack C E, Sturk A. Microparticles from patients with multiple organ dysfunction syndrome and sepsis support coagulation through multiple mechanisms. Thromb Haemost 2001; (85): 810-820.

Jujo K, Ii M, Losordo D W. Endothelial progenitor cells in neovascularization of infarcted myocardium. J Mol Cell Cardiol 2008.

Jy W, Horstman L L, Jimenez J J, Ahn Y S, Biro E, Nieuwland R, Sturk A, gnat-George F, Sabatier F, Camoin-Jau L, Sampol J, Hugel B, Zobairi F, Freyssinet J M, Nomura S, Shet A S, Key N S, Hebbel R P. Measuring circulating cell-derived microparticles. J Thromb Haemost 2004a; (2): 1842-1851.

Jy W, Mao W W, Horstman L, Tao J, Ahn Y S. Platelet microparticles bind, activate and aggregate neutrophils in vitro. Blood Cells Mol Dis 1995; (21): 217-231.

Jy W, Minagar A, Jimenez J J, Sheremata W A, Mauro L M, Horstman L L, Bidot C, Ahn Y S. Endothelial microparticles (EMP) bind and activate monocytes: elevated EMP-monocyte conjugates in multiple sclerosis. Front Biosci 2004b; (9): 3137-3144.

379

Kallenberg C G. Pathogenesis of PR3-ANCA associated vasculitis. J Autoimmun 2008; (30): 29-36.

Kamesh L, Heward J M, Williams J M, Gough S C, Savage C O, Harper L. Mannose- binding lectin gene polymorphisms in a cohort study of ANCA-associated small vessel vasculitis. Rheumatology (Oxford) 2007; (46): 1076-1078.

Kanazawa S, Nomura S, Kuwana M, Muramatsu M, Yamaguchi K, Fukuhara S. Monocyte-derived microparticles may be a sign of vascular complication in patients with lung cancer. Lung Cancer 2003; (39): 145-149.

Karamysheva A F. Mechanisms of angiogenesis. Biochemistry (Mosc ) 2008; (73): 751- 762.

Karkkainen M J, Haiko P, Sainio K, Partanen J, Taipale J, Petrova T V, Jeltsch M, Jackson D G, Talikka M, Rauvala H, Betsholtz C, Alitalo K. Vascular endothelial growth factor C is required for sprouting of the first lymphatic vessels from embryonic veins. Nat Immunol 2004; (5): 74-80.

Kawasaki T. [Acute febrile mucocutaneous syndrome with lymphoid involvement with specific desquamation of the fingers and toes in children]. Arerugi 1967; (16): 178- 222.

Kawasaki T, Kitsukawa T, Bekku Y, Matsuda Y, Sanbo M, Yagi T, Fujisawa H. A requirement for neuropilin-1 in embryonic vessel formation. Development 1999; (126): 4895-4902.

Kerr G S, Hallahan C W, Giordano J, Leavitt R Y, Fauci A S, Rottem M, Hoffman G S. Takayasu arteritis. Ann Intern Med 1994; (120): 919-929.

Khajoee V, Kariyazono H, Ohno T, Ihara K, Mizuno Y, Kusuhara K, Hara T. Inducible and endothelial constitutive nitric oxide synthase gene polymorphisms in Kawasaki disease. Pediatr Int 2003; (45): 130-134.

Khan S S, Solomon M A, McCoy J P, Jr. Detection of circulating endothelial cells and endothelial progenitor cells by flow cytometry. Cytometry B Clin Cytom 2005; (64): 1-8.

Kim H K, Song K S, Park Y S, Kang Y H, Lee Y J, Lee K R, Kim H K, Ryu K W, Bae J M, Kim S. Elevated levels of circulating platelet microparticles, VEGF, IL-6 and RANTES in patients with gastric cancer: possible role of a metastasis predictor. Eur J Cancer 2003; (39): 184-191.

Kim I, Yilmaz O H, Morrison S J. CD144 (VE-cadherin) is transiently expressed by fetal liver hematopoietic stem cells. Blood 2005; (106): 903-905.

King W J, Brooks C J, Holder R, Hughes P, Adu D, Savage C O. T lymphocyte responses to anti-neutrophil cytoplasmic autoantibody (ANCA) antigens are present in patients with ANCA-associated systemic vasculitis and persist during disease remission. Clin Exp Immunol 1998; (112): 539-546.

380

Kinoshita M, Shiraishi T, Koga T, Ayabe M, Rikimaru T, Oizumi K. Churg-Strauss syndrome after corticosteroid withdrawal in an asthmatic patient treated with pranlukast. J Allergy Clin Immunol 1999; (103): 534-535.

Kitching A R, Ru H, X, Turner A L, Tipping P G, Dunn A R, Holdsworth S R. The requirement for granulocyte-macrophage colony-stimulating factor and granulocyte colony-stimulating factor in leukocyte-mediated immune glomerular injury. J Am Soc Nephrol 2002; (13): 350-358.

Koga H, Sugiyama S, Kugiyama K, Watanabe K, Fukushima H, Tanaka T, Sakamoto T, Yoshimura M, Jinnouchi H, Ogawa H. Elevated levels of VE-cadherin-positive endothelial microparticles in patients with type 2 diabetes mellitus and coronary artery disease. J Am Coll Cardiol 2005; (45): 1622-1630.

Kolls J K, Linden A. Interleukin-17 family members and inflammation. Immunity 2004; (21): 467-476.

Kondo T, Hayashi M, Takeshita K, Numaguchi Y, Kobayashi K, Iino S, Inden Y, Murohara T. Smoking cessation rapidly increases circulating progenitor cells in peripheral blood in chronic smokers. Arterioscler Thromb Vasc Biol 2004; (24): 1442-1447.

Kong D, Melo L G, Gnecchi M, Zhang L, Mostoslavsky G, Liew C C, Pratt R E, Dzau V J. Cytokine-induced mobilization of circulating endothelial progenitor cells enhances repair of injured arteries. Circulation 2004a; (110): 2039-2046.

Kong D, Melo L G, Mangi A A, Zhang L, Lopez-Ilasaca M, Perrella M A, Liew C C, Pratt R E, Dzau V J. Enhanced inhibition of neointimal hyperplasia by genetically engineered endothelial progenitor cells. Circulation 2004b; (109): 1769-1775.

Krebs L T, Xue Y, Norton C R, Shutter J R, Maguire M, Sundberg J P, Gallahan D, Closson V, Kitajewski J, Callahan R, Smith G H, Stark K L, Gridley T. Notch signaling is essential for vascular morphogenesis in mice. Genes Dev 2000; (14): 1343-1352.

Krishnaswamy G, Kelley J, Yerra L, Smith J K, Chi D S. Human endothelium as a source of multifunctional cytokines: molecular regulation and possible role in human disease. J Interferon Cytokine Res 1999; (19): 91-104.

Krzewski K, Strominger J L. The killer's kiss: the many functions of NK cell immunological synapses. Curr Opin Cell Biol 2008; (20): 597-605.

Kumpers P, Erdbrugger U, Grossheim M, Meyer G P, Hiss M, Gwinner W, Haller H, Haubitz M. Endothelial microparticles as a diagnostic aid in Churg-Strauss vasculitis-induced cardiomyopathy. Clin Exp Rheumatol 2008; (26): S86-S89.

Kumpers P, Hellpap J, David S, Horn R, Leitolf H, Haller H, Haubitz M. Circulating angiopoietin-2 is a marker and potential mediator of endothelial cell detachment in ANCA-associated vasculitis with renal involvement. Nephrol Dial Transplant 2009.

381

Kutlay S, Calayoglu R, Boyvat A, Turkcapar N, Sengul S, Keven K, Nergizoglu G. Circulating endothelial cells: a disease activity marker in Behcet's vasculitis? Rheumatol Int 2008; (29): 159-162.

Lane S E, Watts R A, Bentham G, Innes N J, Scott D G. Are environmental factors important in primary systemic vasculitis? A case-control study. Arthritis Rheum 2003; (48): 814-823.

Lanneau D, Brunet M, Frisan E, Solary E, Fontenay M, Garrido C. Heat shock proteins: essential proteins for apoptosis regulation. J Cell Mol Med 2008; (12): 743-761.

Laufs U, Urhausen A, Werner N, Scharhag J, Heitz A, Kissner G, Bohm M, Kindermann W, Nickenig G. Running exercise of different duration and intensity: effect on endothelial progenitor cells in healthy subjects. Eur J Cardiovasc Prev Rehabil 2005; (12): 407-414.

Leavitt R Y, Fauci A S, Bloch D A, Michel B A, Hunder G G, Arend W P, Calabrese L H, Fries J F, Lie J T, Lightfoot R W, Jr., . The American College of Rheumatology 1990 criteria for the classification of Wegener's granulomatosis. Arthritis Rheum 1990; (33): 1101-1107.

Lee K W, Lip G Y, Blann A D. Plasma angiopoietin-1, angiopoietin-2, angiopoietin receptor tie-2, and vascular endothelial growth factor levels in acute coronary syndromes. Circulation 2004; (110): 2355-2360.

Lee S, Chen T T, Barber C L, Jordan M C, Murdock J, Desai S, Ferrara N, Nagy A, Roos K P, Iruela-Arispe M L. Autocrine VEGF signaling is required for vascular homeostasis. Cell 2007; (130): 691-703.

Ley K, Laudanna C, Cybulsky M I, Nourshargh S. Getting to the site of inflammation: the leukocyte adhesion cascade updated. Nat Rev Immunol 2007; (7): 678-689.

Lhotta K, Vogel W, Meisl T, Buxbaum M, Neyer U, Sandholzer C, Konig P. Alpha 1- antitrypsin phenotypes in patients with anti-neutrophil cytoplasmic antibody-positive vasculitis. Clin Sci (Lond) 1994; (87): 693-695.

Li J G, McLaughlin R F, Wellings S R. Endothelial cells in blood smears. Calif Med 1958; (89): 22.

Lightfoot R W, Jr., Michel B A, Bloch D A, Hunder G G, Zvaifler N J, McShane D J, Arend W P, Calabrese L H, Leavitt R Y, Lie J T, . The American College of Rheumatology 1990 criteria for the classification of polyarteritis nodosa. Arthritis Rheum 1990; (33): 1088-1093.

Lim H S, Lip G Y, Blann A D. Angiopoietin-1 and angiopoietin-2 in diabetes mellitus: relationship to VEGF, glycaemic control, endothelial damage/dysfunction and atherosclerosis. Atherosclerosis 2005; (180): 113-118.

382

Lin M T, Tsao L Y, Cheng M L, Chang Y J, Chiu H Y, Chen H N, Kuo S F, Chiou S J. Absence of hypercoagulability in acute Kawasaki disease. Pediatr Int 2005; (47): 126-131.

Lin Y, Weisdorf D J, Solovey A, Hebbel R P. Origins of circulating endothelial cells and endothelial outgrowth from blood. J Clin Invest 2000; (105): 71-77.

Little M A, Smyth C L, Yadav R, Ambrose L, Cook H T, Nourshargh S, Pusey C D. Antineutrophil cytoplasm antibodies directed against myeloperoxidase augment leukocyte-microvascular interactions in vivo. Blood 2005; (106): 2050-2058.

Liu S M, Liu H C. Identification of megakaryocytic cells by Ulex Europaeus Agglutinin I (UEA 1) in B5 fixed, decalcified, paraffin-embedded specimens. Zhonghua Yi Xue Za Zhi (Taipei) 1990; (45): 75-82.

Lockwood C M, Hale G, Waldman H, Jayne D R. Remission induction in Behcet's disease following lymphocyte depletion by the anti-CD52 antibody CAMPATH 1-H. Rheumatology (Oxford) 2003; (42): 1539-1544.

Loppnow H, Werdan K, Buerke M. Vascular cells contribute to atherosclerosis by cytokine- and innate-immunity-related inflammatory mechanisms. Innate Immun 2008; (14): 63-87.

Lozano E, Segarra M, Garcia-Martinez A, Hernandez-Rodriguez J, Cid M C. Imatinib mesylate inhibits in vitro and ex vivo biological responses related to vascular occlusion in giant cell arteritis. Ann Rheum Dis 2008; (67): 1581-1588.

Ludviksson B R, Sneller M C, Chua K S, Talar-Williams C, Langford C A, Ehrhardt R O, Fauci A S, Strober W. Active Wegener's granulomatosis is associated with HLA- DR+ CD4+ T cells exhibiting an unbalanced Th1-type T cell cytokine pattern: reversal with IL-10. J Immunol 1998; (160): 3602-3609.

Lum H, Malik A B. Regulation of vascular endothelial barrier function. Am J Physiol 1994; (267): L223-L241.

Lunde K, Solheim S, Forfang K, Arnesen H, Brinch L, Bjornerheim R, Ragnarsson A, Egeland T, Endresen K, Ilebekk A, Mangschau A, Aakhus S. Anterior myocardial infarction with acute percutaneous coronary intervention and intracoronary injection of autologous mononuclear bone marrow cells: safety, clinical outcome, and serial changes in left ventricular function during 12-months' follow-up. J Am Coll Cardiol 2008; (51): 674-676.

Luqmani R A, Bacon P A, Moots R J, Janssen B A, Pall A, Emery P, Savage C, Adu D. Birmingham Vasculitis Activity Score (BVAS) in systemic necrotizing vasculitis. QJM 1994; (87): 671-678.

Lwaleed B A, Bass P S. Tissue factor pathway inhibitor: structure, biology and involvement in disease. J Pathol 2006; (208): 327-339.

383

Maes C, Stockmans I, Moermans K, Van L R, Smets N, Carmeliet P, Bouillon R, Carmeliet G. Soluble VEGF isoforms are essential for establishing epiphyseal vascularization and regulating chondrocyte development and survival. J Clin Invest 2004; (113): 188-199.

Maisonpierre P C, Suri C, Jones P F, Bartunkova S, Wiegand S J, Radziejewski C, Compton D, McClain J, Aldrich T H, Papadopoulos N, Daly T J, Davis S, Sato T N, Yancopoulos G D. Angiopoietin-2, a natural antagonist for Tie2 that disrupts in vivo angiogenesis. Science 1997; (277): 55-60.

Majka M, Kijowski J, Lesko E, Gozdizk J, Zupanska B, Ratajczak M Z. Evidence that platelet-derived microvesicles may transfer platelet-specific immunoreactive antigens to the surface of endothelial cells and CD34+ hematopoietic stem/ progenitor cells-- implication for the pathogenesis of immune thrombocytopenias. Folia Histochem Cytobiol 2007; (45): 27-32.

Mancuso P, Burlini A, Pruneri G, Goldhirsch A, Martinelli G, Bertolini F. Resting and activated endothelial cells are increased in the peripheral blood of cancer patients. Blood 2001; (97): 3658-3661.

Manzi S, Meilahn E N, Rairie J E, Conte C G, Medsger T A, Jr., Jansen-McWilliams L, D'Agostino R B, Kuller L H. Age-specific incidence rates of myocardial infarction and angina in women with systemic lupus erythematosus: comparison with the Framingham Study. Am J Epidemiol 1997; (145): 408-415.

Marceau F, Grassi J, Frobert Y, Bergeron C, Poubelle P E. Effects of experimental conditions on the production of interleukin-1 alpha and -1 beta by human endothelial cells cultured in vitro. Int J Immunopharmacol 1992; (14): 525-534.

Martin J, Paco L, Ruiz M P, Lopez-Nevot M A, Garcia-Porrua C, Amoli M M, Calvino M C, Ollier W E, Gonzalez-Gay M A. Inducible nitric oxide synthase polymorphism is associated with susceptibility to Henoch-Schonlein purpura in northwestern Spain. J Rheumatol 2005; (32): 1081-1085.

Matsumoto N, Nomura S, Kamihata H, Kimura Y, Iwasaka T. Association of platelet- derived microparticles with C-C chemokines on vascular complication in patients with acute myocardial infarction. Clin Appl Thromb Hemost 2002; (8): 279-286.

Mayet W J, Schwarting A, Orth T, Duchmann R, Meyer Zum Buschenfelde K H. Antibodies to proteinase 3 mediate expression of vascular cell adhesion molecule-1 (VCAM-1). Clin Exp Immunol 1996; (103): 259-267.

McCarthy M I, Abecasis G R, Cardon L R, Goldstein D B, Little J, Ioannidis J P, Hirschhorn J N. Genome-wide association studies for complex traits: consensus, uncertainty and challenges. Nat Rev Genet 2008; (9): 356-369.

McClung J A, Naseer N, Saleem M, Rossi G P, Weiss M B, Abraham N G, Kappas A. Circulating endothelial cells are elevated in patients with type 2 diabetes mellitus independently of HbA(1)c. Diabetologia 2005; (48): 345-350.

384

Mehta D, Malik A B. Signaling mechanisms regulating endothelial permeability. Physiol Rev 2006; (86): 279-367.

Merten M, Thiagarajan P. P-selectin in arterial thrombosis. Z Kardiol 2004; (93): 855- 863.

Mesri M, Altieri D C. Endothelial cell activation by leukocyte microparticles. J Immunol 1998; (161): 4382-4387.

Mezentsev A, Merks R M, O'Riordan E, Chen J, Mendelev N, Goligorsky M S, Brodsky S V. Endothelial microparticles affect angiogenesis in vitro: role of oxidative stress. Am J Physiol Heart Circ Physiol 2005; (289): H1106-H1114.

Mills J A, Michel B A, Bloch D A, Calabrese L H, Hunder G G, Arend W P, Edworthy S M, Fauci A S, Leavitt R Y, Lie J T, . The American College of Rheumatology 1990 criteria for the classification of Henoch-Schonlein purpura. Arthritis Rheum 1990; (33): 1114-1121.

Minagar A, Jy W, Jimenez J J, Sheremata W A, Mauro L M, Mao W W, Horstman L L, Ahn Y S. Elevated plasma endothelial microparticles in multiple sclerosis. Neurology 2001; (56): 1319-1324.

Miquerol L, Langille B L, Nagy A. Embryonic development is disrupted by modest increases in vascular endothelial growth factor gene expression. Development 2000; (127): 3941-3946.

Mitani Y, Sawada H, Hayakawa H, Aoki K, Ohashi H, Matsumura M, Kuroe K, Shimpo H, Nakano M, Komada Y. Elevated levels of high-sensitivity C-reactive protein and serum amyloid-A late after Kawasaki disease: association between inflammation and late coronary sequelae in Kawasaki disease. Circulation 2005; (111): 38-43.

Miyazaki Y, Nomura S, Miyake T, Kagawa H, Kitada C, Taniguchi H, Komiyama Y, Fujimura Y, Ikeda Y, Fukuhara S. High shear stress can initiate both platelet aggregation and shedding of procoagulant containing microparticles. Blood 1996; (88): 3456-3464.

Moonen J R, de L K, van S, X, Kallenberg C G, van Luyn M J, Bijl M, Harmsen M C. Reduced number and impaired function of circulating progenitor cells in patients with systemic lupus erythematosus. Arthritis Res Ther 2007; (9): R84.

Morel O, Toti F, Bakouboula B, Grunebaum L, Freyssinet J M. Procoagulant microparticles: 'criminal partners' in atherothrombosis and deleterious cellular exchanges. Pathophysiol Haemost Thromb 2006; (35): 15-22.

Morelli S, Perrone C, Paroli M. Recurrent cerebral infarctions in polyarteritis nodosa with circulating antiphospholipid antibodies and mitral valve disease. Lupus 1998; (7): 51-52.

385

Mourdjeva M, Kyurkchiev D, Mandinova A, Altankova I, Kehayov I, Kyurkchiev S. Dynamics of membrane translocation of phosphatidylserine during apoptosis detected by a monoclonal antibody. Apoptosis 2005; (10): 209-217.

Mu H, Ohashi R, Yang H, Wang X, Li M, Lin P, Yao Q, Chen C. Thymosin beta10 inhibits cell migration and capillary-like tube formation of human coronary artery endothelial cells. Cell Motil Cytoskeleton 2006; (63): 222-230.

Mukhtyar C, Flossmann O, Hellmich B, Bacon P, Cid M, Cohen-Tervaert J W, Gross W L, Guillevin L, Jayne D, Mahr A, Merkel P A, Raspe H, Scott D, Witter J, Yazici H, Luqmani R A. Outcomes from studies of antineutrophil cytoplasm antibody associated vasculitis: a systematic review by the European League Against Rheumatism systemic vasculitis task force. Ann Rheum Dis 2008; (67): 1004-1010.

Mukhtyar C B, Flossmann O, Luqmani R A. Clinical and biological assessment in systemic necrotizing vasculitides. Clin Exp Rheumatol 2006; (24): S92-S99.

Muller Kobold A C, Kallenberg C G, Tervaert J W. Leucocyte membrane expression of proteinase 3 correlates with disease activity in patients with Wegener's granulomatosis. Br J Rheumatol 1998a; (37): 901-907.

Muller Kobold A C, Mesander G, Stegeman C A, Kallenberg C G, Tervaert J W. Are circulating neutrophils intravascularly activated in patients with anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitides? Clin Exp Immunol 1998b; (114): 491-499.

Muller A M, Hermanns M I, Skrzynski C, Nesslinger M, Muller K M, Kirkpatrick C J. Expression of the endothelial markers PECAM-1, vWf, and CD34 in vivo and in vitro. Exp Mol Pathol 2002; (72): 221-229.

Muller I, Klocke A, Alex M, Kotzsch M, Luther T, Morgenstern E, Zieseniss S, Zahler S, Preissner K, Engelmann B. Intravascular tissue factor initiates coagulation via circulating microvesicles and platelets. FASEB J 2003; (17): 476-478.

Mutin M, Canavy I, Blann A, Bory M, Sampol J, gnat-George F. Direct evidence of endothelial injury in acute myocardial infarction and unstable angina by demonstration of circulating endothelial cells. Blood 1999; (93): 2951-2958.

Mutunga M, Fulton B, Bullock R, Batchelor A, Gascoigne A, Gillespie J I, Baudouin S V. Circulating endothelial cells in patients with septic shock. Am J Respir Crit Care Med 2001; (163): 195-200.

Nadar S K, Lip G Y, Lee K W, Blann A D. Circulating endothelial cells in acute ischaemic stroke. Thromb Haemost 2005; (94): 707-712.

Nakamura Y, Yanagawa H, Harada K, Kato H, Kawasaki T. Mortality among persons with a history of Kawasaki disease in Japan: the fifth look. Arch Pediatr Adolesc Med 2002; (156): 162-165.

386

Nakatani K, Takeshita S, Tsujimoto H, Kawamura Y, Tokutomi T, Sekine I. Circulating endothelial cells in Kawasaki disease. Clin Exp Immunol 2003; (131): 536-540.

Narita T, Numao H. Lectin binding patterns in normal, metaplastic, and neoplastic gastric mucosa. J Histochem Cytochem 1992; (40): 681-687.

Navarro M, Cervera R, Font J, Reverter J C, Monteagudo J, Escolar G, Lopez-Soto A, Ordinas A, Ingelmo M. Anti-endothelial cell antibodies in systemic autoimmune diseases: prevalence and clinical significance. Lupus 1997; (6): 521-526.

Newburger J W, Takahashi M, Burns J C, Beiser A S, Chung K J, Duffy C E, Glode M P, Mason W H, Reddy V, Sanders S P, . The treatment of Kawasaki syndrome with intravenous gamma globulin. N Engl J Med 1986; (315): 341-347.

Newburger J W, Taubert K A, Shulman S T, Rowley A H, Gewitz M H, Takahashi M, McCrindle B W. Summary and abstracts of the Seventh International Kawasaki Disease Symposium: December 4-7, 2001, Hakone, Japan. Pediatr Res 2003; (53): 153-157.

Nistala K, Moncrieffe H, Newton K R, Varsani H, Hunter P, Wedderburn L R. Interleukin-17-producing T cells are enriched in the joints of children with arthritis, but have a reciprocal relationship to regulatory T cell numbers. Arthritis Rheum 2008; (58): 875-887.

Nowack R, Schwalbe K, Flores-Suarez L F, Yard B, van der Woude F J. Upregulation of CD14 and CD18 on monocytes In vitro by antineutrophil cytoplasmic autoantibodies. J Am Soc Nephrol 2000; (11): 1639-1646.

Oh J H, Han J W, Lee S J, Lee K Y, Suh B K, Koh D K, Lee J S, Oh C K, Kim T G, Choi H B. Polymorphisms of human leukocyte antigen genes in korean children with Kawasaki disease. Pediatr Cardiol 2008; (29): 402-408.

Ohlsson S, Wieslander J, Segelmark M. Circulating cytokine profile in anti-neutrophilic cytoplasmatic autoantibody-associated vasculitis: prediction of outcome? Mediators Inflamm 2004; (13): 275-283.

Ohno T, Igarashi H, Inoue K, Akazawa K, Joho K, Hara T. Serum vascular endothelial growth factor: a new predictive indicator for the occurrence of coronary artery lesions in Kawasaki disease. Eur J Pediatr 2000; (159): 424-429.

Olfat M, Al-Mayouf S M. Cogan's syndrome in childhood. Rheumatol Int 2001; (20): 246-249.

Olsson A K, Dimberg A, Kreuger J, Claesson-Welsh L. VEGF receptor signalling - in control of vascular function. Nat Rev Mol Cell Biol 2006; (7): 359-371.

Omoto S, Nomura S, Shouzu A, Nishikawa M, Fukuhara S, Iwasaka T. Detection of monocyte-derived microparticles in patients with Type II diabetes mellitus. Diabetologia 2002; (45): 550-555.

387

Onouchi Y, Gunji T, Burns J C, Shimizu C, Newburger J W, Yashiro M, Nakamura Y, Yanagawa H, Wakui K, Fukushima Y, Kishi F, Hamamoto K, Terai M, Sato Y, Ouchi K, Saji T, Nariai A, Kaburagi Y, Yoshikawa T, Suzuki K, Tanaka T, Nagai T, Cho H, Fujino A, Sekine A, Nakamichi R, Tsunoda T, Kawasaki T, Nakamura Y, Hata A. ITPKC functional polymorphism associated with Kawasaki disease susceptibility and formation of coronary artery aneurysms. Nat Genet 2008; (40): 35- 42.

Osterud B. The role of platelets in decrypting monocyte tissue factor. Semin Hematol 2001; (38): 2-5.

Ozen S, Anton J, Arisoy N, Bakkaloglu A, Besbas N, Brogan P, Garcia-Consuegra J, Dolezalova P, Dressler F, Duzova A, Ferriani V P, Hilario M O, Ibanez-Rubio M, Kasapcopur O, Kuis W, Lehman T J, Nemcova D, Nielsen S, Oliveira S K, Schikler K, Sztajnbok F, Terreri M T, Zulian F, Woo P. Juvenile polyarteritis: results of a multicenter survey of 110 children. J Pediatr 2004; (145): 517-522.

Ozen S, Ruperto N, Dillon M J, Bagga A, Barron K, Davin J C, Kawasaki T, Lindsley C, Petty R E, Prieur A M, Ravelli A, Woo P. EULAR/PReS endorsed consensus criteria for the classification of childhood vasculitides. Ann Rheum Dis 2006; (65): 936-941.

Ozkaya O, Soylemezoglu O, Gonen S, Misirlioglu M, Tuncer S, Kalman S, Buyan N, Hasanoglu E. Renin-angiotensin system gene polymorphisms: association with susceptibility to Henoch-Schonlein purpura and renal involvement. Clin Rheumatol 2006; (25): 861-865.

Pagnoux C, Cohen P, Guillevin L. Vasculitides secondary to infections. Clin Exp Rheumatol 2006; (24): S71-S81.

Pai P, Stevenson A, Mason H, Bell G M. Occupational hydrocarbon exposure and nephrotoxicity: a cohort study and literature review. Postgrad Med J 1998; (74): 225- 228.

Palit A, Inamadar A C. Vasculitis: approach to diagnosis and therapy. Indian J Dermatol Venereol Leprol 2006; (72): 334-345.

Parikh S M, Mammoto T, Schultz A, Yuan H T, Christiani D, Karumanchi S A, Sukhatme V P. Excess circulating angiopoietin-2 may contribute to pulmonary vascular leak in sepsis in humans. PLoS Med 2006; (3): e46.

Park J A, Shin K S, Kim Y W. Polymorphism of matrix metalloproteinase-3 promoter gene as a risk factor for coronary artery lesions in Kawasaki disease. J Korean Med Sci 2005a; (20): 607-611.

Park K S, Min K, Nam J H, Bang D, Lee E S, Lee S. Association of HYPA haplotype in the mannose-binding lectin gene-2 with Behcet's disease. Tissue Antigens 2005b; (65): 260-265.

388

Park M C, Park Y B, Jung S Y, Lee K H, Lee S K. Anti-endothelial cell antibodies and antiphospholipid antibodies in Takayasu's arteritis: correlations of their titers and isotype distributions with disease activity. Clin Exp Rheumatol 2006; (24): S10-S16.

Parker R. The ability of isolated blood cells to form organised vessels in vitro. J Exp Med 1934; (60): 351-362.

Patel J V, Lim H S, Varughese G I, Hughes E A, Lip G Y. Angiopoietin-2 levels as a biomarker of cardiovascular risk in patients with hypertension. Ann Med 2008; (40): 215-222.

Patry Y C, Trewick D C, Gregoire M, Audrain M A, Moreau A M, Muller J Y, Meflah K, Esnault V L. Rats injected with syngenic rat apoptotic neutrophils develop antineutrophil cytoplasmic antibodies. J Am Soc Nephrol 2001; (12): 1764-1768.

Pelajo C F, de Oliveira S K, Rodrigues M C, Torres J M. Cutaneous vasculitis as a paraneoplastic syndrome in childhood. Acta Reumatol Port 2007; (32): 181-183.

Pendergraft W F, Alcorta D A, Segelmark M, Yang J J, Tuttle R, Jennette J C, Falk R J, Preston G A. ANCA antigens, proteinase 3 and myeloperoxidase, are not expressed in endothelial cells. Kidney Int 2000; (57): 1981-1990.

Pendergraft W F, III, Preston G A, Shah R R, Tropsha A, Carter C W, Jr., Jennette J C, Falk R J. Autoimmunity is triggered by cPR-3(105-201), a protein complementary to human autoantigen proteinase-3. Nat Med 2004; (10): 72-79.

Penfield J G. Nephrogenic systemic fibrosis and the use of gadolinium-based contrast agents. Pediatr Nephrol 2008; (23): 2121-2129.

Percivalle E, Revello M G, Vago L, Morini F, Gerna G. Circulating endothelial giant cells permissive for human cytomegalovirus (HCMV) are detected in disseminated HCMV infections with organ involvement. J Clin Invest 1993; (92): 663-670.

Pertovaara L, Kaipainen A, Mustonen T, Orpana A, Ferrara N, Saksela O, Alitalo K. Vascular endothelial growth factor is induced in response to transforming growth factor-beta in fibroblastic and epithelial cells. J Biol Chem 1994; (269): 6271-6274.

Petri B, Bixel M G. Molecular events during leukocyte diapedesis. FEBS J 2006; (273): 4399-4407.

Pfister H, Ollert M, Frohlich L F, Quintanilla-Martinez L, Colby T V, Specks U, Jenne D E. Antineutrophil cytoplasmic autoantibodies against the murine homolog of proteinase 3 (Wegener autoantigen) are pathogenic in vivo. Blood 2004; (104): 1411- 1418.

Piccin A, Murphy W G, Smith O P. Circulating microparticles: pathophysiology and clinical implications. Blood Rev 2007; (21): 157-171.

Pigott R, Dillon L P, Hemingway I H, Gearing A J. Soluble forms of E-selectin, ICAM- 1 and VCAM-1 are present in the supernatants of cytokine activated cultured endothelial cells. Biochem Biophys Res Commun 1992; (187): 584-589. 389

Pipitone N, Versari A, Salvarani C. Role of imaging studies in the diagnosis and followup of large-vessel vasculitis: an update. Rheumatology (Oxford) 2008; (47): 403- 408.

Plavina T, Hincapie M, Wakshull E, Subramanyam M, Hancock W S. Increased plasma concentrations of cytoskeletal and Ca2+-binding proteins and their peptides in psoriasis patients. Clin Chem 2008; (54): 1805-1814.

Pober J S, Gimbrone M A, Jr., Lapierre L A, Mendrick D L, Fiers W, Rothlein R, Springer T A. Overlapping patterns of activation of human endothelial cells by interleukin 1, tumor necrosis factor, and immune interferon. J Immunol 1986; (137): 1893-1896.

Pooley R J, Jr., Peterson L, Finn W G, Kroft S H. Cytomegalovirus-infected cells in routinely prepared peripheral blood films of immunosuppressed patients. Am J Clin Pathol 1999; (112): 108-112.

Porat Y, Porozov S, Belkin D, Shimoni D, Fisher Y, Belleli A, Czeiger D, Silverman W F, Belkin M, Battler A, Fulga V, Savion N. Isolation of an adult blood-derived progenitor cell population capable of differentiation into angiogenic, myocardial and neural lineages. Br J Haematol 2006; (135): 703-714.

Porges A J, Redecha P B, Kimberly W T, Csernok E, Gross W L, Kimberly R P. Anti- neutrophil cytoplasmic antibodies engage and activate human neutrophils via Fc gamma RIIa. J Immunol 1994; (153): 1271-1280.

Praprotnik S, Blank M, Meroni P L, Rozman B, Eldor A, Shoenfeld Y. Classification of anti-endothelial cell antibodies into antibodies against microvascular and macrovascular endothelial cells: the pathogenic and diagnostic implications. Arthritis Rheum 2001; (44): 1484-1494.

Preston R A, Jy W, Jimenez J J, Mauro L M, Horstman L L, Valle M, Aime G, Ahn Y S. Effects of severe hypertension on endothelial and platelet microparticles. Hypertension 2003; (41): 211-217.

Puechal X. Antineutrophil cytoplasmic antibody-associated vasculitides. Joint Bone Spine 2007; (74): 427-435.

Racanelli V, Prete M, Minoia C, Favoino E, Perosa F. Rheumatic disorders as paraneoplastic syndromes. Autoimmun Rev 2008; (7): 352-358.

Rajagopalan S, Somers E C, Brook R D, Kehrer C, Pfenninger D, Lewis E, Chakrabarti A, Richardson B C, Shelden E, McCune W J, Kaplan M J. Endothelial cell apoptosis in systemic lupus erythematosus: a common pathway for abnormal vascular function and thrombosis propensity. Blood 2004; (103): 3677-3683.

Rajavashisth T B, Andalibi A, Territo M C, Berliner J A, Navab M, Fogelman A M, Lusis A J. Induction of endothelial cell expression of granulocyte and macrophage colony-stimulating factors by modified low-density lipoproteins. Nature 1990; (344): 254-257.

390

Ralston D R, Marsh C B, Lowe M P, Wewers M D. Antineutrophil cytoplasmic antibodies induce monocyte IL-8 release. Role of surface proteinase-3, alpha1- antitrypsin, and Fcgamma receptors. J Clin Invest 1997; (100): 1416-1424.

Randolph G J, Luther T, Albrecht S, Magdolen V, Muller W A. Role of tissue factor in adhesion of mononuclear phagocytes to and trafficking through endothelium in vitro. Blood 1998; (92): 4167-4177.

Rang H P, Dale M M, Ritter J M. The vascular system. In: Pharmacology. Edinburgh: Churchill Livingstone, 1999; 4th: 278-300.

Rao R M, Yang L, Garcia-Cardena G, Luscinskas F W. Endothelial-dependent mechanisms of leukocyte recruitment to the vascular wall. Circ Res 2007; (101): 234-247.

Rauova L, Gilburd B, Zurgil N, Blank M, Guegas L L, Brickman C M, Cebecauer L, Deutsch M, Wiik A, Shoenfeld Y. Induction of biologically active antineutrophil cytoplasmic antibodies by immunization with human apoptotic polymorphonuclear leukocytes. Clin Immunol 2002; (103): 69-78.

Rauscher F M, Goldschmidt-Clermont P J, Davis B H, Wang T, Gregg D, Ramaswami P, Pippen A M, Annex B H, Dong C, Taylor D A. Aging, progenitor cell exhaustion, and atherosclerosis. Circulation 2003; (108): 457-463.

Redondo S, Hristov M, Gordillo-Moscoso A A, Ruiz E, Weber C, Tejerina T. High- reproducible flow cytometric endothelial progenitor cell determination in human peripheral blood as CD34+/CD144+/CD3- lymphocyte sub-population. J Immunol Methods 2008; (335): 21-27.

Rees L, Webb N, Brogan P. Vasculitis. In: Paediatric nephrology. Oxford: Oxford university press, 2007; 1: 297-316.

Rehman J, Li J, Parvathaneni L, Karlsson G, Panchal V R, Temm C J, Mahenthiran J, March K L. Exercise acutely increases circulating endothelial progenitor cells and monocyte-/macrophage-derived angiogenic cells. J Am Coll Cardiol 2004; (43): 2314-2318.

Rehse M A, Corpuz S, Heimfeld S, Minie M, Yachimiak D. Use of fluorescence threshold triggering and high-speed flow cytometry for rare event detection. Cytometry 1995; (22): 317-322.

Reynolds N. Vasculitis in Behcet's syndrome: evidence-based review. Curr Opin Rheumatol 2008; (20): 347-352.

Rotmans J I, Heyligers J M, Verhagen H J, Velema E, Nagtegaal M M, de Kleijn D P, de Groot F G, Stroes E S, Pasterkamp G. In vivo cell seeding with anti-CD34 antibodies successfully accelerates endothelialization but stimulates intimal hyperplasia in porcine arteriovenous expanded polytetrafluoroethylene grafts. Circulation 2005; (112): 12-18.

391

Rowand J L, Martin G, Doyle G V, Miller M C, Pierce M S, Connelly M C, Rao C, Terstappen L W. Endothelial cells in peripheral blood of healthy subjects and patients with metastatic carcinomas. Cytometry A 2007; (71): 105-113.

Rozmyslowicz T, Majka M, Kijowski J, Murphy S L, Conover D O, Poncz M, Ratajczak J, Gaulton G N, Ratajczak M Z. Platelet- and megakaryocyte-derived microparticles transfer CXCR4 receptor to CXCR4-null cells and make them susceptible to infection by X4-HIV. AIDS 2003; (17): 33-42.

Rueda B, Perez-Armengol C, Lopez-Lopez S, Garcia-Porrua C, Martin J, Gonzalez-Gay M A. Association between functional haplotypes of vascular endothelial growth factor and renal complications in Henoch-Schonlein purpura. J Rheumatol 2006; (33): 69-73.

Ruggeri Z M. Von Willebrand factor: looking back and looking forward. Thromb Haemost 2007; (98): 55-62.

Ruhrberg C, Gerhardt H, Golding M, Watson R, Ioannidou S, Fujisawa H, Betsholtz C, Shima D T. Spatially restricted patterning cues provided by heparin-binding VEGF- A control blood vessel branching morphogenesis. Genes Dev 2002; (16): 2684-2698.

Rustemeyer P, Wittkowski W, Jurk K, Koller A. Optimized flow cytometric analysis of endothelial progenitor cells in peripheral blood. J Immunoassay Immunochem 2006; (27): 77-88.

Rutgers A, Slot M, van P P, van B, V, Heeringa P, Tervaert J W. Coexistence of anti- glomerular basement membrane antibodies and myeloperoxidase-ANCAs in crescentic glomerulonephritis. Am J Kidney Dis 2005; (46): 253-262.

Salvarani C, Cantini F, Boiardi L, Hunder G G. Laboratory investigations useful in giant cell arteritis and Takayasu's arteritis. Clin Exp Rheumatol 2003; (21): S23-S28.

Salvarani C, Macchioni P, Mantovani W, Rossi F, Veneziani M, Boiardi L, Lodi L, Portioli I. Extraarticular manifestations of rheumatoid arthritis and HLA antigens in northern Italy. J Rheumatol 1992; (19): 242-246.

Sanchez-Elsner T, Botella L M, Velasco B, Langa C, Bernabeu C. Endoglin expression is regulated by transcriptional cooperation between the hypoxia and transforming growth factor-beta pathways. J Biol Chem 2002; (277): 43799-43808.

Sapet C, Simoncini S, Loriod B, Puthier D, Sampol J, Nguyen C, gnat-George F, Anfosso F. Thrombin-induced endothelial microparticle generation: identification of a novel pathway involving ROCK-II activation by caspase-2. Blood 2006; (108): 1868-1876.

Sato T N, Tozawa Y, Deutsch U, Wolburg-Buchholz K, Fujiwara Y, Gendron-Maguire M, Gridley T, Wolburg H, Risau W, Qin Y. Distinct roles of the receptor tyrosine kinases Tie-1 and Tie-2 in blood vessel formation. Nature 1995; (376): 70-74.

392

Sato Y, Rifkin D B. Inhibition of endothelial cell movement by pericytes and smooth muscle cells: activation of a latent transforming growth factor-beta 1-like molecule by plasmin during co-culture. J Cell Biol 1989; (109): 309-315.

Saulsbury F T. Henoch-Schonlein purpura in children. Report of 100 patients and review of the literature. Medicine (Baltimore) 1999; (78): 395-409.

Savage C O, Pottinger B E, Gaskin G, Lockwood C M, Pusey C D, Pearson J D. Vascular damage in Wegener's granulomatosis and microscopic polyarteritis: presence of anti-endothelial cell antibodies and their relation to anti-neutrophil cytoplasm antibodies. Clin Exp Immunol 1991; (85): 14-19.

Scanu A, Molnarfi N, Brandt K J, Gruaz L, Dayer J M, Burger D. Stimulated T cells generate microparticles, which mimic cellular contact activation of human monocytes: differential regulation of pro- and anti-inflammatory cytokine production by high-density lipoproteins. J Leukoc Biol 2008; (83): 921-927.

Schattner A. Consequence or coincidence? The occurrence, pathogenesis and significance of autoimmune manifestations after viral vaccines. Vaccine 2005; (23): 3876-3886.

Schleiffenbaum B, Spertini O, Tedder T F. Soluble L-selectin is present in human plasma at high levels and retains functional activity. J Cell Biol 1992; (119): 229- 238.

Schlieben D J, Korbet S M, Kimura R E, Schwartz M M, Lewis E J. Pulmonary-renal syndrome in a newborn with placental transmission of ANCAs. Am J Kidney Dis 2005; (45): 758-761.

Schwarting A, Hagen D, Odenthal M, Brockmann H, Dienes H P, Wandel E, Rumpelt H J, Zum Buschenfelde K H, Galle P R, Mayet W. Proteinase-3 mRNA expressed by glomerular epithelial cells correlates with crescent formation in Wegener's granulomatosis. Kidney Int 2000; (57): 2412-2422.

Sebastian J K, Mahr A D, Ahmed S S, Stone J H, Romay-Penabad Z, Davis J C, Hoffman G S, McCune W J, St Clair E W, Specks U, Spiera R, Pierangeli S, Merkel P A. Antiendothelial cell antibodies in patients with Wegener's granulomatosis: prevalence and correlation with disease activity and manifestations. J Rheumatol 2007; (34): 1027-1031.

Seko Y, Imai Y, Suzuki S, Kamijukkoku S, Hayasaki K, Sakomura Y, Tobe K, Kadowaki T, Maekawa H, Takahashi N, Yazaki Y. Serum levels of vascular endothelial growth factor in patients with acute myocardial infarction undergoing reperfusion therapy. Clin Sci (Lond) 1997; (92): 453-454.

Shaker O, Ay El-Deen M A, El H H, Grace B D, El S H, Abdel H A. The role of heat shock protein 60, vascular endothelial growth factor and antiphospholipid antibodies in Behcet disease. Br J Dermatol 2007; (156): 32-37.

393

Shalaby F, Rossant J, Yamaguchi T P, Gertsenstein M, Wu X F, Breitman M L, Schuh A C. Failure of blood-island formation and vasculogenesis in Flk-1-deficient mice. Nature 1995; (376): 62-66.

Shanberge J N. Accidental occurrence of endothelial cells in peripheral blood smears. Am J Clin Pathol 1955; (23): 460-464.

Shantsila E, Lip G Y. Circulating endothelial cells in health and disease: how do we best quantify them? J Thromb Haemost 2008; (6): 1021-1024.

Sherer Y, Pagnoux C, Chironi G, Simon A, Guillevin L, Mouthon L, Gilburd B, Shoenfeld Y. Carotid artery intima-media thickness, heat shock proteins and oxidized LDL autoantibodies in systemic necrotizing vasculitis. Rheumatol Int 2008.

Shi Q, Rafii S, Wu M H, Wijelath E S, Yu C, Ishida A, Fujita Y, Kothari S, Mohle R, Sauvage L R, Moore M A, Storb R F, Hammond W P. Evidence for circulating bone marrow-derived endothelial cells. Blood 1998; (92): 362-367.

Shi Q, Wu M H, Hayashida N, Wechezak A R, Clowes A W, Sauvage L R. Proof of fallout endothelialization of impervious Dacron grafts in the aorta and inferior vena cava of the dog. J Vasc Surg 1994; (20): 546-556.

Shim Y H, Kim H S, Sohn S, Hong Y M. Insertion/deletion polymorphism of angiotensin converting enzyme gene in Kawasaki disease. J Korean Med Sci 2006; (21): 208-211.

Shintani S, Murohara T, Ikeda H, Ueno T, Honma T, Katoh A, Sasaki K, Shimada T, Oike Y, Imaizumi T. Mobilization of endothelial progenitor cells in patients with acute myocardial infarction. Circulation 2001; (103): 2776-2779.

Shitrit D, Shitrit A B, Starobin D, Izbicki G, Belenky A, Kaufman N, Kramer M R. Large vessel aneurysms in Wegener's granulomatosis. J Vasc Surg 2002; (36): 856- 858.

Shoenfeld Y. Classification of anti-endothelial cell antibodies into antibodies against microvascular and macrovascular endothelial cells: the pathogenic and diagnostic implications. Cleve Clin J Med 2002; (69 Suppl 2): SII65-SII67.

Shulman S T, De I J, Hirsch R. Kawasaki disease. Pediatr Clin North Am 1995; (42): 1205-1222.

Sica A, Matsushima K, Van D J, Wang J M, Polentarutti N, Dejana E, Colotta F, Mantovani A. IL-1 transcriptionally activates the neutrophil chemotactic factor/IL-8 gene in endothelial cells. Immunology 1990a; (69): 548-553.

Sica A, Wang J M, Colotta F, Dejana E, Mantovani A, Oppenheim J J, Larsen C G, Zachariae C O, Matsushima K. Monocyte chemotactic and activating factor gene expression induced in endothelial cells by IL-1 and tumor necrosis factor. J Immunol 1990b; (144): 3034-3038.

394

Silvestre J S, Gojova A, Brun V, Potteaux S, Esposito B, Duriez M, Clergue M, Le Ricousse-Roussanne S, Barateau V, Merval R, Groux H, Tobelem G, Levy B, Tedgui A, Mallat Z. Transplantation of bone marrow-derived mononuclear cells in ischemic apolipoprotein E-knockout mice accelerates atherosclerosis without altering plaque composition. Circulation 2003; (108): 2839-2842.

Siner J M, Bhandari V, Engle K M, Elias J A, Siegel M D. Elevated serum angiopoietin 2 levels are associated with increased mortality in sepsis. Shock 2008.

Singh A J, Meyer R D, Band H, Rahimi N. The carboxyl terminus of VEGFR-2 is required for PKC-mediated down-regulation. Mol Biol Cell 2005; (16): 2106-2118.

Sireci G, Dieli F, Salerno A. T cells recognize an immunodominant epitope of heat shock protein 65 in Kawasaki disease. Mol Med 2000; (6): 581-590.

Skubitz K, Balke J, Ball E. Report on the CD15 cluster work. In: Leucocyte Typing IV: White Cell Differentiation Antigens. (Eds.Knapp W, Dörken B, Gilks W). New York: Oxford University Press, 1989; 800-805.

Smart N, Risebro C A, Melville A A, Moses K, Schwartz R J, Chien K R, Riley P R. Thymosin beta4 induces adult epicardial progenitor mobilization and neovascularization. Nature 2007a; (445): 177-182.

Smart N, Rossdeutsch A, Riley P R. Thymosin beta4 and angiogenesis: modes of action and therapeutic potential. Angiogenesis 2007b; (10): 229-241.

Sohn M H, Hur M W, Kim D S. Interleukin 6 gene promoter polymorphism is not associated with Kawasaki disease. Genes Immun 2001; (2): 357-362.

Solovey A, Gui L, Ramakrishnan S, Steinberg M H, Hebbel R P. Sickle cell anemia as a possible state of enhanced anti-apoptotic tone: survival effect of vascular endothelial growth factor on circulating and unanchored endothelial cells. Blood 1999; (93): 3824-3830.

Solovey A, Lin Y, Browne P, Choong S, Wayner E, Hebbel R P. Circulating activated endothelial cells in sickle cell anemia. N Engl J Med 1997; (337): 1584-1590.

Somer T, Finegold S M. Vasculitides associated with infections, immunization, and antimicrobial drugs. Clin Infect Dis 1995; (20): 1010-1036.

Spriewald B M, Witzke O, Wassmuth R, Wenzel R R, Arnold M L, Philipp T, Kalden J R. Distinct tumour necrosis factor alpha, interferon gamma, interleukin 10, and cytotoxic T cell antigen 4 gene polymorphisms in disease occurrence and end stage renal disease in Wegener's granulomatosis. Ann Rheum Dis 2005; (64): 457-461.

Sproson E L, Jones N S, Al-Deiri M, Lanyon P. Lessons learnt in the management of Wegener's Granulomatosis: long-term follow-up of 60 patients. Rhinology 2007; (45): 63-67.

St Clair E W, McCallum R M. Cogan's syndrome. Curr Opin Rheumatol 1999; (11): 47- 52. 395

Stegeman C A, Tervaert J W, Huitema M G, de Jong P E, Kallenberg C G. Serum levels of soluble adhesion molecules intercellular adhesion molecule 1, vascular cell adhesion molecule 1, and E-selectin in patients with Wegener's granulomatosis. Relationship to disease activity and relevance during followup. Arthritis Rheum 1994; (37): 1228-1235.

Stegmayr B G, Gothefors L, Malmer B, Muller Wiefel D E, Nilsson K, Sundelin B. Wegener granulomatosis in children and young adults. A case study of ten patients. Pediatr Nephrol 2000; (14): 208-213.

Steppich B, Mattisek C, Sobczyk D, Kastrati A, Schomig A, Ott I. Tissue factor pathway inhibitor on circulating microparticles in acute myocardial infarction. Thromb Haemost 2005; (93): 35-39.

Stone J H, Hoffman G S, Merkel P A, Min Y I, Uhlfelder M L, Hellmann D B, Specks U, Allen N B, Davis J C, Spiera R F, Calabrese L H, Wigley F M, Maiden N, Valente R M, Niles J L, Fye K H, McCune J W, St Clair E W, Luqmani R A. A disease-specific activity index for Wegener's granulomatosis: modification of the Birmingham Vasculitis Activity Score. International Network for the Study of the Systemic Vasculitides (INSSYS). Arthritis Rheum 2001; (44): 912-920.

Stump M, ORDAN G L, Jr., EBAKEY M E, ALPERT B. Endothelium grown from circulating blood on isolated intravascular dacron hub. Am J Pathol 1963; (43): 361- 367.

Suri C, Jones P F, Patan S, Bartunkova S, Maisonpierre P C, Davis S, Sato T N, Yancopoulos G D. Requisite role of angiopoietin-1, a ligand for the TIE2 receptor, during embryonic angiogenesis. Cell 1996; (87): 1171-1180.

Suzuki A, Kamiya T, Kuwahara N, Ono Y, Kohata T, Takahashi O, Kimura K, Takamiya M. Coronary arterial lesions of Kawasaki disease: cardiac catheterization findings of 1100 cases. Pediatr Cardiol 1986; (7): 3-9.

Suzuki A, Ohosone Y, Obana M, Mita S, Matsuoka Y, Irimajiri S, Fukuda J. Cause of death in 81 autopsied patients with rheumatoid arthritis. J Rheumatol 1994; (21): 33- 36.

Swerlick R A, Lee K H, Wick T M, Lawley T J. Human dermal microvascular endothelial but not human umbilical vein endothelial cells express CD36 in vivo and in vitro. J Immunol 1992; (148): 78-83.

Takahashi H, Shibuya M. The vascular endothelial growth factor (VEGF)/VEGF receptor system and its role under physiological and pathological conditions. Clin Sci (Lond) 2005; (109): 227-241.

Takahashi T, Kalka C, Masuda H, Chen D, Silver M, Kearney M, Magner M, Isner J M, Asahara T. Ischemia- and cytokine-induced mobilization of bone marrow-derived endothelial progenitor cells for neovascularization. Nat Med 1999; (5): 434-438.

396

Takeshita S, Dobashi H, Nakatani K, Koike Y, Tsujimoto H, Hirayama K, Kawamura Y, Mori K, Sekine I, Yoshioka S. Circulating soluble selectins in Kawasaki disease. Clin Exp Immunol 1997; (108): 446-450.

Takeshita S, Kawamura Y, Takabayashi H, Yoshida N, Nonoyama S. Imbalance in the production between vascular endothelial growth factor and endostatin in Kawasaki disease. Clin Exp Immunol 2005; (139): 575-579.

Tammela T, Enholm B, Alitalo K, Paavonen K. The biology of vascular endothelial growth factors. Cardiovasc Res 2005; (65): 550-563.

Tammela T, Zarkada G, Wallgard E, Murtomaki A, Suchting S, Wirzenius M, Waltari M, Hellstrom M, Schomber T, Peltonen R, Freitas C, Duarte A, Isoniemi H, Laakkonen P, Christofori G, Yla-Herttuala S, Shibuya M, Pytowski B, Eichmann A, Betsholtz C, Alitalo K. Blocking VEGFR-3 suppresses angiogenic sprouting and vascular network formation. Nature 2008; (454): 656-660.

Tanaka T, Yamakawa N, Koike N, Suzuki J, Mizuno F, Usui M. Behcet's disease and antibody titers to various heat-shock protein 60s. Ocul Immunol Inflamm 1999; (7): 69-74.

Tannenbaum S H, Finko R, Cines D B. Antibody and immune complexes induce tissue factor production by human endothelial cells. J Immunol 1986; (137): 1532-1537.

Tans G, Rosing J, Thomassen M C, Heeb M J, Zwaal R F, Griffin J H. Comparison of anticoagulant and procoagulant activities of stimulated platelets and platelet-derived microparticles. Blood 1991; (77): 2641-2648.

Terai M, Kohno Y, Namba M, Umemiya T, Niwa K, Nakajima H, Mikata A. Class II major histocompatibility antigen expression on coronary arterial endothelium in a patient with Kawasaki disease. Hum Pathol 1990; (21): 231-234.

Terai M, Yasukawa K, Narumoto S, Tateno S, Oana S, Kohno Y. Vascular endothelial growth factor in acute Kawasaki disease. Am J Cardiol 1999; (83): 337-339.

Tervaert J W, Stegeman C A, Kallenberg C G. Silicon exposure and vasculitis. Curr Opin Rheumatol 1998; (10): 12-17.

Thompson R D, Wakelin M W, Larbi K Y, Dewar A, Asimakopoulos G, Horton M A, Nakada M T, Nourshargh S. Divergent effects of platelet-endothelial cell adhesion molecule-1 and beta 3 integrin blockade on leukocyte transmigration in vivo. J Immunol 2000; (165): 426-434.

Thurston G, Kitajewski J. VEGF and Delta-Notch: interacting signalling pathways in tumour angiogenesis. Br J Cancer 2008.

Till S H, Amos R S. Long-term follow-up of juvenile-onset cutaneous polyarteritis nodosa associated with streptococcal infection. Br J Rheumatol 1997; (36): 909-911.

Timmermans F, Van H F, De S M, Raedt R, Plasschaert F, De Buyzere M L, Gillebert T C, Plum J, Vandekerckhove B. Endothelial outgrowth cells are not derived from 397

CD133+ cells or CD45+ hematopoietic precursors. Arterioscler Thromb Vasc Biol 2007; (27): 1572-1579.

Tomer Y, Gilburd B, Blank M, Lider O, Hershkoviz R, Fishman P, Zigelman R, Meroni P L, Wiik A, Shoenfeld Y. Characterization of biologically active antineutrophil cytoplasmic antibodies induced in mice. Pathogenetic role in experimental vasculitis. Arthritis Rheum 1995; (38): 1375-1381.

Topaloglu R, Sungur A, Baskin E, Besbas N, Saatci U, Bakkaloglu A. Vascular endothelial growth factor in Henoch-Schonlein purpura. J Rheumatol 2001; (28): 2269-2273.

Torgerson T R. Regulatory T cells in human autoimmune diseases. Springer Semin Immunopathol 2006; (28): 63-76.

Torheim E A, Yndestad A, Bjerkeli V, Halvorsen B, Aukrust P, Froland S S. Increased expression of chemokines in patients with Wegener's granulomatosis - modulating effects of methylprednisolone in vitro. Clin Exp Immunol 2005; (140): 376-383.

Toti F, Satta N, Fressinaud E, Meyer D, Freyssinet J M. Scott syndrome, characterized by impaired transmembrane migration of procoagulant phosphatidylserine and hemorrhagic complications, is an inherited disorder. Blood 1996; (87): 1409-1415.

Tripathi P, Tripathi P, Kashyap L, Singh V. The role of nitric oxide in inflammatory reactions. FEMS Immunol Med Microbiol 2007; (51): 443-452.

Tsan M F, Gao B. Heat shock protein and innate immunity. Cell Mol Immunol 2004; (1): 274-279.

Tse W Y, Williams J, Pall A, Wilkes M, Savage C O, Adu D. Antineutrophil cytoplasm antibody-induced neutrophil nitric oxide production is nitric oxide synthase independent. Kidney Int 2001; (59): 593-600.

Tullus K, Brennan E, Hamilton G, Lord R, McLaren C A, Marks S D, Roebuck D J. Renovascular hypertension in children. Lancet 2008; (371): 1453-1463.

Turgut S, Turgut G, Atalay E O, Atalay A. Angiotensin-converting enzyme I/D polymorphism in Behcet's disease. Med Princ Pract 2005; (14): 213-216.

Turkoz Y, Evereklioglu C, Ozkiris A, Mistik S, Borlu M, Ozerol I H, Duygulu F, Ilhan O. Serum levels of soluble P-selectin are increased and associated with disease activity in patients with Behcet's syndrome. Mediators Inflamm 2005; (2005): 237- 241.

Umemura T, Soga J, Hidaka T, Takemoto H, Nakamura S, Jitsuiki D, Nishioka K, Goto C, Teragawa H, Yoshizumi M, Chayama K, Higashi Y. Aging and hypertension are independent risk factors for reduced number of circulating endothelial progenitor cells. Am J Hypertens 2008; (21): 1203-1209.

398

Uyttendaele H, Ho J, Rossant J, Kitajewski J. Vascular patterning defects associated with expression of activated Notch4 in embryonic endothelium. Proc Natl Acad Sci U S A 2001; (98): 5643-5648.

Vaglio A, Martorana D, Maggiore U, Grasselli C, Zanetti A, Pesci A, Garini G, Manganelli P, Bottero P, Tumiati B, Sinico R A, Savi M, Buzio C, Neri T M. HLA- DRB4 as a genetic risk factor for Churg-Strauss syndrome. Arthritis Rheum 2007; (56): 3159-3166.

Vallien G, Langley R, Jennings S, Specian R, Granger D N. Expression of endothelial cell adhesion molecules in neovascularized tissue. Microcirculation 2000; (7): 249- 258.

Van de Wouwer M., Collen D, Conway E M. Thrombomodulin-protein C-EPCR system: integrated to regulate coagulation and inflammation. Arterioscler Thromb Vasc Biol 2004; (24): 1374-1383. van der Geld Y M, Hellmark T, Selga D, Heeringa P, Huitema M G, Limburg P C, Kallenberg C G. Rats and mice immunised with chimeric human/mouse proteinase 3 produce autoantibodies to mouse Pr3 and rat granulocytes. Ann Rheum Dis 2007; (66): 1679-1682. van der Geld Y M, Tool A T, Videler J, de H M, Tervaert J W, Stegeman C A, Limburg P C, Kallenberg C G, Roos D. Interference of PR3-ANCA with the enzymatic activity of PR3: differences in patients during active disease or remission of Wegener's granulomatosis. Clin Exp Immunol 2002; (129): 562-570.

Van Paassen P., Cohen Tervaert J W, Heeringa P. Mechanisms of vasculitis: how pauci- immune is ANCA-associated renal vasculitis? Nephron Exp Nephrol 2007; (105): e10-e16.

Van D S, McColl G, Wicks I P. Accelerated atherosclerosis: an extraarticular feature of rheumatoid arthritis? Arthritis Rheum 2002; (46): 862-873.

VanWijk M J, Nieuwland R, Boer K, van der Post J A, Vanbavel E, Sturk A. Microparticle subpopulations are increased in preeclampsia: possible involvement in vascular dysfunction? Am J Obstet Gynecol 2002; (187): 450-456.

Varagunam M, Nwosu A C, Adu D, Garner C, Taylor C M, Michael J, Thompson R A. Little evidence for anti-endothelial-cell antibodies in microscopic polyarteritis and Wegener's granulomatosis. Adv Exp Med Biol 1993; (336): 419-422.

Vasa M, Fichtlscherer S, Aicher A, Adler K, Urbich C, Martin H, Zeiher A M, Dimmeler S. Number and migratory activity of circulating endothelial progenitor cells inversely correlate with risk factors for coronary artery disease. Circ Res 2001; (89): E1-E7.

Venneri M A, De P M, Ponzoni M, Pucci F, Scielzo C, Zonari E, Mazzieri R, Doglioni C, Naldini L. Identification of proangiogenic TIE2-expressing monocytes (TEMs) in human peripheral blood and cancer. Blood 2007; (109): 5276-5285.

399

Verity D H, Wallace G R, Seed P T, Kanawati C A, Ayesh I, Holland-Gladwish J, Stanford M R. Soluble adhesion molecules in Behcet's disease. Ocul Immunol Inflamm 1998; (6): 81-92.

Vestweber D. Adhesion and signaling molecules controlling the transmigration of leukocytes through endothelium. Immunol Rev 2007; (218): 178-196.

Vogeli K M, Jin S W, Martin G R, Stainier D Y. A common progenitor for haematopoietic and endothelial lineages in the zebrafish gastrula. Nature 2006; (443): 337-339.

Wallace G R, Kondeatis E, Vaughan R W, Verity D H, Chen Y, Fortune F, Madanat W, Kanawati C A, Graham E M, Stanford M R. IL-10 genotype analysis in patients with Behcet's disease. Hum Immunol 2007; (68): 122-127.

Walton E.W. Giant-cell granuloma of the respiratory tract (Wegener's granulomatosis). Br Med J 1958; (2): 265-270.

Wang J M, Wang Y, Huang J Y, Yang Z, Chen L, Wang L C, Tang A L, Lou Z F, Tao J. C-Reactive protein-induced endothelial microparticle generation in HUVECs is related to BH4-dependent NO formation. J Vasc Res 2007; (44): 241-248.

Wang L J, Yang Y H, Lin Y T, Chiang B L. Levels of intracellular adhesion molecule-1 and vascular cell adhesion molecule-1 in children with Henoch-Schonlein purpura. J Microbiol Immunol Infect 2006; (39): 109-113.

Ward M R, Stewart D J, Kutryk M J. Endothelial progenitor cell therapy for the treatment of coronary disease, acute MI, and pulmonary arterial hypertension: current perspectives. Catheter Cardiovasc Interv 2007; (70): 983-998.

Warkentin T E, Hayward C P, Boshkov L K, Santos A V, Sheppard J A, Bode A P, Kelton J G. Sera from patients with heparin-induced thrombocytopenia generate platelet-derived microparticles with procoagulant activity: an explanation for the thrombotic complications of heparin-induced thrombocytopenia. Blood 1994; (84): 3691-3699.

Watt S M, Contreras M. Stem cell medicine: umbilical cord blood and its stem cell potential. Semin Fetal Neonatal Med 2005; (10): 209-220.

Wattiaux M J, Kahn M F, Thevenet J P, Sauvezie B, Imbert J C. [Vascular involvement in rheumatoid polyarthritis. Retrospective study of 37 cases of rheumatoid polyarthritis with vascular involvement and review of the literature]. Ann Med Interne (Paris) 1987; (138): 566-587.

Watts R, Harper L, Jayne D, Levy J, Pusey C, Savage C, Scott D G, Williams J. Translational research in autoimmunity: aims of therapy in vasculitis. Rheumatology (Oxford) 2005; (44): 573-576.

400

Watts R A, Gonzalez-Gay M A, Lane S E, Garcia-Porrua C, Bentham G, Scott D G. Geoepidemiology of systemic vasculitis: comparison of the incidence in two regions of Europe. Ann Rheum Dis 2001; (60): 170-172.

Watts R A, Mooney J, Lane S E, Scott D G. Rheumatoid vasculitis: becoming extinct? Rheumatology (Oxford) 2004; (43): 920-923.

Weller F E, Mutchnick M G, Goldstein A L, Naylor P H. Enzyme immunoassay measurement of thymosin beta 4 in human serum. J Biol Response Mod 1988; (7): 91-96.

Werner N, Kosiol S, Schiegl T, Ahlers P, Walenta K, Link A, Bohm M, Nickenig G. Circulating endothelial progenitor cells and cardiovascular outcomes. N Engl J Med 2005; (353): 999-1007.

Weyand C M, Schonberger J, Oppitz U, Hunder N N, Hicok K C, Goronzy J J. Distinct vascular lesions in giant cell arteritis share identical T cell clonotypes. J Exp Med 1994; (179): 951-960.

Whitaker G B, Limberg B J, Rosenbaum J S. Vascular endothelial growth factor receptor-2 and neuropilin-1 form a receptor complex that is responsible for the differential signaling potency of VEGF(165) and VEGF(121). J Biol Chem 2001; (276): 25520-25531.

Widemann A, Sabatier F, Arnaud L, Bonello L, Al-Massarani G, Paganelli F, Poncelet P, gnat-George F. CD146-based immunomagnetic enrichment followed by multiparameter flow cytometry: a new approach to counting circulating endothelial cells. J Thromb Haemost 2008; (6): 869-876.

Wieczorek S, Hellmich B, Arning L, Moosig F, Lamprecht P, Gross W L, Epplen J T. Functionally relevant variations of the interleukin-10 gene associated with antineutrophil cytoplasmic antibody-negative Churg-Strauss syndrome, but not with Wegener's granulomatosis. Arthritis Rheum 2008; (58): 1839-1848.

Wiedmer T, Shattil S J, Cunningham M, Sims P J. Role of calcium and calpain in complement-induced vesiculation of the platelet plasma membrane and in the exposure of the platelet factor Va receptor. Biochemistry 1990; (29): 623-632.

Wiik A. Drug-induced vasculitis. Curr Opin Rheumatol 2008; (20): 35-39.

Wollert K C, Meyer G P, Lotz J, Ringes-Lichtenberg S, Lippolt P, Breidenbach C, Fichtner S, Korte T, Hornig B, Messinger D, Arseniev L, Hertenstein B, Ganser A, Drexler H. Intracoronary autologous bone-marrow cell transfer after myocardial infarction: the BOOST randomised controlled clinical trial. Lancet 2004; (364): 141- 148.

Woolard J, Wang W Y, Bevan H S, Qiu Y, Morbidelli L, Pritchard-Jones R O, Cui T G, Sugiono M, Waine E, Perrin R, Foster R, gby-Bell J, Shields J D, Whittles C E, Mushens R E, Gillatt D A, Ziche M, Harper S J, Bates D O. VEGF165b, an inhibitory vascular endothelial growth factor splice variant: mechanism of action, in

401

vivo effect on angiogenesis and endogenous protein expression. Cancer Res 2004; (64): 7822-7835.

Woywodt A, Blann A D, Kirsch T, Erdbruegger U, Banzet N, Haubitz M, gnat-George F. Isolation and enumeration of circulating endothelial cells by immunomagnetic isolation: proposal of a definition and a consensus protocol. J Thromb Haemost 2006a; (4): 671-677.

Woywodt A, Goldberg C, Kirsch T, de G K, Erdbruegger U, Haller H, Haubitz M. Circulating endothelial cells in relapse and limited granulomatous disease due to ANCA associated vasculitis. Ann Rheum Dis 2006b; (65): 164-168.

Woywodt A, Scheer J, Hambach L, Buchholz S, Ganser A, Haller H, Hertenstein B, Haubitz M. Circulating endothelial cells as a marker of endothelial damage in allogeneic hematopoietic stem cell transplantation. Blood 2004; (103): 3603-3605.

Woywodt A, Schroeder M, Gwinner W, Mengel M, Jaeger M, Schwarz A, Haller H, Haubitz M. Elevated numbers of circulating endothelial cells in renal transplant recipients. Transplantation 2003a; (76): 1-4.

Woywodt A, Streiber F, de G K, Regelsberger H, Haller H, Haubitz M. Circulating endothelial cells as markers for ANCA-associated small-vessel vasculitis. Lancet 2003b; (361): 206-210.

Wu S F, Chang J S, Wan L, Tsai C H, Tsai F J. Association of IL-1Ra gene polymorphism, but no association of IL-1beta and IL-4 gene polymorphisms, with Kawasaki disease. J Clin Lab Anal 2005; (19): 99-102.

Xiao H, Heeringa P, Hu P, Liu Z, Zhao M, Aratani Y, Maeda N, Falk R J, Jennette J C. Antineutrophil cytoplasmic autoantibodies specific for myeloperoxidase cause glomerulonephritis and vasculitis in mice. J Clin Invest 2002; (110): 955-963.

Xiao H, Heeringa P, Liu Z, Huugen D, Hu P, Maeda N, Falk R J, Jennette J C. The role of neutrophils in the induction of glomerulonephritis by anti-myeloperoxidase antibodies. Am J Pathol 2005; (167): 39-45.

Xiao H, Schreiber A, Heeringa P, Falk R J, Jennette J C. Alternative complement pathway in the pathogenesis of disease mediated by anti-neutrophil cytoplasmic autoantibodies. Am J Pathol 2007a; (170): 52-64.

Xiao Q, Kiechl S, Patel S, Oberhollenzer F, Weger S, Mayr A, Metzler B, Reindl M, Hu Y, Willeit J, Xu Q. Endothelial progenitor cells, cardiovascular risk factors, cytokine levels and atherosclerosis--results from a large population-based study. PLoS ONE 2007b; (2): e975.

Xu Q, Zhang Z, Davison F, Hu Y. Circulating progenitor cells regenerate endothelium of vein graft atherosclerosis, which is diminished in ApoE-deficient mice. Circ Res 2003; (93): e76-e86.

402

Yang J, Li C R, Li Y B, Li R X, Sun L B, Huang H J, Wang G B. [The correlation between Kawasaki disease and polymorphisms of Tumor necrosis factor alpha and interleukin-10 gene promoter]. Zhonghua Er Ke Za Zhi 2003; (41): 598-602.

Yang Y H, Chuang Y H, Wang L C, Huang H Y, Gershwin M E, Chiang B L. The immunobiology of Henoch-Schonlein purpura. Autoimmun Rev 2008; (7): 179-184.

Yang Y H, Hung C F, Hsu C R, Wang L C, Chuang Y H, Lin Y T, Chiang B L. A nationwide survey on epidemiological characteristics of childhood Henoch- Schonlein purpura in Taiwan. Rheumatology (Oxford) 2005; (44): 618-622.

Yilmaz D, Kavakli K, Ozkayin N. The elevated markers of hypercoagulability in children with Henoch-Schonlein purpura. Pediatr Hematol Oncol 2005; (22): 41-48.

Yoder M C, Mead L E, Prater D, Krier T R, Mroueh K N, Li F, Krasich R, Temm C J, Prchal J T, Ingram D A. Redefining endothelial progenitor cells via clonal analysis and hematopoietic stem/progenitor cell principals. Blood 2007; (109): 1801-1809.

Yoshioka T, Xu Y X, Yoshida H, Shiraga H, Muraki T, Ito K. Deletion polymorphism of the angiotensin converting enzyme gene predicts persistent proteinuria in Henoch- Schonlein purpura nephritis. Arch Dis Child 1998; (79): 394-399.

Yuan H T, Venkatesha S, Chan B, Deutsch U, Mammoto T, Sukhatme V P, Woolf A S, Karumanchi S A. Activation of the orphan endothelial receptor Tie1 modifies Tie2- mediated intracellular signaling and cell survival. FASEB J 2007; (21): 3171-3183.

Yuan L, Moyon D, Pardanaud L, Breant C, Karkkainen M J, Alitalo K, Eichmann A. Abnormal lymphatic vessel development in neuropilin 2 mutant mice. Development 2002; (129): 4797-4806.

Yun M, Yeh D, Araujo L I, Jang S, Newberg A, Alavi A. F-18 FDG uptake in the large arteries: a new observation. Clin Nucl Med 2001; (26): 314-319.

Yurdakul S, Gunaydin I, Tuzun Y, Tankurt N, Pazarli H, Ozyazgan Y, Yazici H. The prevalence of Behcet's syndrome in a rural area in northern Turkey. J Rheumatol 1988; (15): 820-822.

Zavada J, Kideryova L, Pytlik R, Vankova Z, Tesar V. Circulating endothelial progenitor cells in patients with ANCA-associated vasculitis. Kidney Blood Press Res 2008; (31): 247-254.

Zeiger F, Stephan S, Hoheisel G, Pfeiffer D, Ruehlmann C, Koksch M. P-Selectin expression, platelet aggregates, and platelet-derived microparticle formation are increased in peripheral arterial disease. Blood Coagul Fibrinolysis 2000; (11): 723- 728.

Zhang S, Guo J, Zhang P, Liu Y, Jia Z, Ma K, Li W, Li L, Zhou C. Long-term effects of bone marrow mononuclear cell transplantation on left ventricular function and remodeling in rats. Life Sci 2004; (74): 2853-2864.

403

Zhou Y, Huang D, Paris P L, Sauter C S, Prock K A, Hoffman G S. An analysis of CTLA-4 and proinflammatory cytokine genes in Wegener's granulomatosis. Arthritis Rheum 2004; (50): 2645-2650.

Publications from this thesis

The following are from studies not included in this thesis, that were carried out during period of study and have included techniques described herein.

Brogan P A, Shah V, Clarke L A, Dillon M J, Klein N. T cell activation profiles in Kawasaki syndrome. Clin Exp Immunol 2008; (151): 267-274.

Giles I, Pericleous C, Liu X, Ehsanullah J, Clarke LA, Brogan PA, Newton-West M, Swerlick R, Lambrianides N, Chen P, Latchman D, Isenberg D, Pierangeli S, Rahman A. Thrombin Binding Predicts the Effects of Sequence Changes in a Human Monoclonal Antiphospholipid Antibody on its In-Vivo Biological Actions. J Immunol 2009; (182) 4836-4843

404