Pathological and Clinicopathological Features of Canine and Feline Bladder Disease

Home , Bladder stone, Interstitial cystitis

Pathological and clinicopathological features of canine and feline bladder disease

Dr Emily Enid Jones Bachelor of Veterinary Science

0000-0001-6462-1234

A thesis submitted for the degree of Doctor of Philosophy at The University of Queensland in 2020

School of Veterinary Science

Abstract

Dogs and cats commonly present to veterinary hospitals with urinary bladder disease, but despite their clinical importance and comparative potential to human diseases, bladder diseases in Australian dogs and cats are under investigated. In veterinary pathology, insufficient levels of diagnostic agreement can occur, and this is influenced by sample quality as well as the pathologist’s own experience, training, and cognitive biases. Logistic regression is a statistical technique which, when applied to veterinary histopathology, could improve pathologist agreement. Thus, there were two overarching goals of this thesis - to investigate the pathology and comparative potential of canine and feline urinary bladder disease in Australia, and to explore the utility of logistic regression modelling in improving inter-pathologist agreement.

This project conducted a retrospective evaluation of pathology cases of canine and feline urinary bladder tissue from the veterinary pathology archives of the University of Queensland School of Veterinary Science and Murdoch University’s School of Veterinary and Life Sciences, with prospective sampling from veterinary clinics and a commercial veterinary pathology service in South East Queensland. The demographics of the dataset were examined using proportionate morbidity and logistic regression to identify associations between animal factors and the diagnosis. Secondly, a comprehensive histological evaluation was undertaken of every sample, with logistic regression modelling performed to identify associations between histological variables and diagnosis. Thirdly, a subset of canine and feline diseased and normal bladder tissue samples was tested for biomarker expression using immunohistochemistry and polymerase chain reaction. This combined approach tested if retrospective samples were of sufficient quality, and when validated provided quantity as well as cellular location of the target biomarkers. To further investigate the comparative potential of feline idiopathic cystitis (FIC), a systematic review was conducted on biomarkers in bladder pain syndrome (BPS) compared to FIC, following the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. Finally, to investigate the role of logistic regression modelling in veterinary pathologist agreement, the modelling of histological variables was used to formulate a predictive probability tool which we then tested on four pathologists evaluating the same set of twenty-five slides, with diagnostic agreement evaluated using the Fleiss kappa statistic.

The main findings from the demographic analysis were a higher risk of bladder neoplasia in dogs compared to cats, increasing risk for bladder neoplasia with age, and decreased risk for cystitis in neutered animals. Next, logistic regression modelling on the histology dataset of canine and feline urinary bladder tissue from Eastern and Western Australia identified six significant variables that 2

were associated with the diagnosis – urothelial ulceration, urothelial inflammation, neutrophilic submucosal inflammation, submucosal lymphoid aggregates, amount of submucosal haemorrhage, and species. These six variables were used to create a predictive probability tool for bladder disease diagnosis. The pathologist agreement study revealed a good level of agreement between the four pathologists when diagnosing neoplastic lesions, but poor to fair agreement for cystitis, urolithiasis and normal bladder tissue. Agreement between pathologists did improve when signalment and clinical history was provided, with mixed results on inter-pathologist agreement when the predictive probability tool was used. However, the predictive tool did prove valuable in increasing the agreement of the study pathologists’ diagnosis with the reference diagnosis. There were multiple other confounders at play in this experiment such as variable digital slide quality and different interpretations of the study instructions. A systematic review on biomarkers in bladder pain syndrome revealed that nerve growth factor is the most likely urine biomarker to be useful in the diagnosis of human BPS. The aim of this review had been to compare biomarkers in BPS to those in FIC, however an unexpected variability in the study parameters meant we could not fulfil this goal. A final laboratory-based investigation of biomarkers of canine and feline bladder diseases revealed two findings - that archived formalin fixed, paraffin embedded tissues are not good samples for PCR experiments, and secondly that tight junction protein-1 may be a promising tissue biomarker for differentiating between some urinary bladder diseases in dogs and cats.

In conclusion, this thesis has undertaken a comprehensive analysis of the pathogenesis and comparative potential of canine and feline bladder diseases and is the first to apply logistic regression modelling to veterinary histopathology diagnosis and to improving inter-pathologist agreement. Logistic regression modelling is a promising tool for veterinary pathology. Dogs and cats are potentially good comparative models for human bladder diseases; however, inconsistent case definitions in human research complicates veterinary and medical field alignment. Finally, a collaborative multicentre approach would be invaluable to collect high quality prospective samples of feline idiopathic cystitis cases to allow further investigation into this disease. In summary, the comprehensive approach utilised in this thesis has provided valuable information on bladder disease in cats and dogs and sets a foundation for further work in this field.

Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

Publications included in this thesis

No publications included.

Submitted manuscripts included in this thesis

1. E. Jones, J. Alawneh, M. Thompson, R. Allavena. The association between case signalment and disease diagnosis in canine and feline bladder disease, submitted to Journal of Diagnostic Investigation on 6th April 2020 - incorporated as Chapter 2.

The concept and design of the study was formulated by EJ (70%) with the assistance of RA (10%), JA (10%) and all co-authors (10%). Protocol and search strategy were developed by EJ (80%) with the assistance of JA (20%). EJ was responsible for data management (100%), and data analyses (80%) with assistance from JA (20%), and the interpretation of results (70%, in consultation with JA 20% and all co-authors 10%). EJ was responsible for drafting the manuscript (100%). EJ was responsible for revision of the final version of the manuscript (80%), considering the comments and suggestions of RA (10%) and all co-authors (10%)

Other publications during candidature

Conference abstracts

No conference abstracts.

Poster presentations

1. E. Jones, J. Alawneh, C. Palmieri, K. Jackson, M. Thompson, R. Allavena. Histological features of Australian canine and feline bladder disease. American College of Veterinary Pathologists/ Society of Toxicologic Pathology Annual Meeting. Washington, DC, USA. 3rd - 7th November 2018.

2. E. Jones, J. Alawneh, C. Palmieri, K. Jackson, M. Thompson, R. Allavena. Canine and feline bladder disease – Pathological features and a novel qualitative analysis of pathology reports (same topic as the 2017 one below, but a revised poster). Australian Veterinary Association Annual Conference, Brisbane, Australia. 14th May 2018.

3. E. Jones, J. Alawneh, C. Palmieri, K. Jackson, M. Thompson, R. Allavena (presenter). Canine and feline bladder disease – Pathological features and a novel qualitative analysis of pathology reports. American College of Veterinary Pathologists/ Society of Toxicologic Pathology Annual Meeting. Vancouver, Canada. 4th - 8th November 2017. 6

Book chapters

No book chapters.

Contributions by others to the thesis

Significant contributions were made to the thesis by Solomon Woldeyohannes (Chapter 4, assistance with statistical analysis and critically revising chapter drafts; Chapter 5, assistance with meta-analysis and critically revising chapter drafts), Leeanne McMichael (Chapter 6, guidance with PCR lab work, interpreting of results and critically revising chapter drafts), and Ameh James (Chapter 6, primer design and guidance with PCR work).

Statement of parts of the thesis submitted to qualify for the award of another degree

No works submitted towards another degree have been included in this thesis.

Research involving human or animal subjects

Animal ethics approval for usage of bladder biopsies collected from animals already undergoing diagnostic procedures was granted by the University of Queensland Animal Ethics Committee (Approval number: ANRFA/SVS/259/16), Appendix 1: Copy of Animal Ethics Approval Certificate ANRFA/SVS/259/16 by UQ Animal Ethics Committee.

Acknowledgments

I would like to express my sincere thanks and gratitude to everybody who has been involved in getting this project to completion. Thank you to my supervisory team - even though with five of you it was challenging to get a consensus at times, you all provided valuable input to the project. Rachel, thank you for your invaluable guidance and support on this project and in helping me become the researcher I am today. Chiara and Karen, thank you for your input on study design, your help with immunohistochemistry and clinical pathology, and your valuable critiques of chapter and paper drafts. Mary, thank you for your kindness and support throughout my project. I would always look forward to your input on chapter drafts as you provide a unique perspective, and your proofreading is second to none! John, thank you for your insights into the epidemiological and statistical components of this project.

I am indebted to my wonderful family (Graeme, Kit, Spike and the Jones, Abreu and Drinkwater families) and friends for all their love, friendship, support, encouragement, snacks, and coffee during this project and also throughout my life. I feel lucky to be surrounded by such incredible people.

Thank you to Lee McMichael and Ameh James for the high level of skill and patience you contributed to my biomarker chapter, and to Solomon Woldeyohannes for your gratefully received statistical help and commitment to this project at short notice. I would also like to thank the UQ pathology and laboratory technicians, Jo Gordon, Mick Cobbin, Graham Panzram, Brian Bynon, Tina Maguire and Lana Bradshaw, for their technical and practical assistance, as well as the UQ School of Veterinary Science finance and administration staff.

I would also like to acknowledge the following South East Queensland veterinary clinics and institutions for contributing to the sample pool - UQVets, Veterinary Specialist Services, Manly Road Veterinary Hospital, SuperVets, Morningside General Practice, The Cat Clinic, the UQ Clinical Studies Centre and Andrea Schaffer-White and Independent Veterinary Pathology.

Financial support

This research was supported by the Dr David Galbraith Postgraduate Research Scholarship in Companion Animal Health.

Funding for this project was gratefully received from the John and Mary Kibble Bequest for Companion Animal Research and the University of Queensland School of Veterinary Science Donor/Bequest Companion Animal Research funding scheme.

Keywords veterinary, pathology, epidemiology, urinary bladder, logistic regression, feline, canine, comparative, agreement, biomarker

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 070709, Veterinary Pathology, 70%

ANZSRC code: 070704, Veterinary Epidemiology, 30%

Fields of Research (FoR) Classification

FoR code: 0707, Veterinary Sciences, 100%

Table of Contents

Abstract ...... 2

Declaration by author ...... 4

Publications included in this thesis ...... 5

Submitted manuscripts included in this thesis ...... 6

Other publications during candidature ...... 6

Book chapters ...... 7

Contributions by others to the thesis ...... 7

Statement of parts of the thesis submitted to qualify for the award of another degree ...... 8

Research involving human or animal subjects ...... 8

Acknowledgments ...... 9

Financial support ...... 10

Keywords ...... 10

Australian and New Zealand Standard Research Classifications (ANZSRC) ...... 11

Fields of Research (FoR) Classification ...... 11

Table of Contents ...... 12

List of Figures ...... 19

List of Tables ...... 24

List of abbreviations used in the thesis ...... 27

Chapter 1 Introduction, literature review and scope of the study...... 30

Introduction ...... 30

Variation between veterinary pathologists ...... 31

A novel role for logistic regression modelling ...... 33

Canine and feline urinary bladder disease ...... 35

Infectious bladder disease ...... 38

1.5.1 Aetiology and risk factors ...... 38

1.5.2 Subclinical bacteriuria ...... 40

1.5.3 Pathogenesis ...... 40

Inflammatory bladder diseases ...... 41

1.6.1 Clinical signs and epidemiology ...... 43

1.6.2 Types of BPS/FIC ...... 46

1.6.3 Pathogenesis ...... 46

Bladder disease diagnosis and the role of histopathology ...... 47

1.7.1 Bladder pain syndrome and feline idiopathic cystitis ...... 47

1.7.2 Urolithiasis ...... 50

Conclusions ...... 51

Chapter 2 Relationship of signalment data to disease outcome in canine and feline bladder pathology records in South East Queensland, Australia ...... 53

Introduction ...... 53

Hypothesis and objectives ...... 54

Material and Methods ...... 54

2.3.1 Study Population ...... 54

2.3.2 Data collection and case definitions ...... 55 13

2.3.3 Statistical analysis ...... 59

Results ...... 61

2.4.1 Data summary ...... 61

2.4.2 Signalment descriptive data ...... 64

2.4.3 Biopsy as a diagnostic tool ...... 65

2.4.4 Proportionate morbidity and proportionate morbidity ratios ...... 65

2.4.5 Multivariable models ...... 69

Discussion ...... 71

Conclusion ...... 74

Chapter 3 A novel role for logistic regression modelling – predicting diagnosis of canine and feline urinary bladder disease based on histological features ...... 76

Introduction ...... 76

Hypothesis and objectives ...... 78

Methods ...... 78

3.3.1 Study population and data collection ...... 78

3.3.2 Histological analysis ...... 81

3.3.3 Logistic regression ...... 83

Results ...... 84

3.4.1 Graphic representation of the six significant variables ...... 90

Discussion ...... 94

Conclusion ...... 99

Chapter 4 Concordance of pathologist bladder biopsy assessment with and without the use 14

of a predictive tool...... 100

Introduction ...... 100

Hypothesis and objectives ...... 102

Materials and Methods ...... 103

4.3.1 Sample size ...... 103

4.3.2 Building the predictive tool ...... 104

4.3.3 Pathologists ...... 105

4.3.4 Statistical analysis ...... 107

Results ...... 109

4.4.1 Statistical analysis of inter-pathologist agreement of test pathologists...... 112

4.4.2 Statistical analysis of agreement between the test pathologists and the reference diagnosis...... 116

Discussion ...... 118

Conclusion ...... 121

Chapter 5 Systematic review Biomarkers of bladder pain syndrome and feline interstitial cystitis 122

Introduction ...... 122

5.1.1 Biomarkers in BPS ...... 124

5.1.2 Biomarkers in FIC ...... 125

5.1.3 The importance of this review ...... 126

Hypothesis and objectives ...... 127

Methods ...... 127

5.3.1 Criteria for considering studies for this review and outcome measures ...... 127

5.3.2 Screening of studies ...... 129

5.3.3 Assessment of risk of bias in included studies ...... 130

5.3.4 Unit of measurement issues ...... 131

5.3.5 Dealing with missing data ...... 132

5.3.6 Data synthesis and meta-analysis ...... 132

Results ...... 133

5.4.1 Description of studies ...... 133

5.4.2 Risk of bias in included studies ...... 138

5.4.3 Meta-Analysis ...... 141

Discussion ...... 148

Conclusion ...... 150

Chapter 6 Biomarker investigation of canine and feline bladder disease using a combined approach of polymerase chain reaction and immunohistochemistry...... 151

Introduction/background ...... 151

6.1.1 Biomarkers in human bladder disease ...... 152

6.1.2 Canine and feline bladder disease biomarkers ...... 155

6.1.3 Internal controls ...... 160

6.1.4 Conclusions ...... 161

Hypothesis and objectives ...... 162

Methods ...... 162

6.3.1 Sample size ...... 162 16

6.3.2 Samples ...... 163

6.3.3 Polymerase Chain Reaction ...... 165

6.3.4 Immunohistochemistry ...... 174

Results ...... 177

6.4.1 Canine and feline protein homology ...... 177

6.4.2 Optimising biomarker primers using synthetic controls ...... 178

6.4.3 Reference gene optimisation on sample extracts ...... 179

6.4.4 Real time PCR ...... 183

6.4.5 Immunohistochemistry scoring ...... 186

6.4.6 Immunohistochemistry statistical analysis ...... 190

Discussion ...... 191

Conclusion ...... 195

Chapter 7 Discussion ...... 196

Chapter 8 References ...... 202

Appendix 1: Copy of Animal Ethics Approval Certificate ANRFA/SVS/259/16 by UQ Animal Ethics Committee...... 227

Appendix 2: Crude and adjusted odds ratio of biopsy sampling method stratified by disease cases and non-cases...... 231

Appendix 3: All animal, sampling and diagnostic histological variables measured on each bladder slide and used in the logistic regression modelling process ...... 232

Appendix 4: Raw data from analysis of histological features of all bladder specimens in the dataset...... 235

Appendix 5: Significant predicted probabilities (P≤0.05) and 95% confidence intervals for diagnostic outcomes by histopathology finding patterns for canine cases submitted to 17

UQVLS between 1994 and 2016 and collected from MUSVLS and southeast Queensland veterinary clinics from 2016-2019...... 249

Appendix 6: Bladder histology photomicrographs of cases used in the pathologist agreement study, Chapter 4...... 255

Appendix 7: A sample of the Microsoft Excel worksheet from the pathologist agreement study...... 279

Appendix 8: Database search strings for systematic review...... 283

8.1.1 Bladder pain syndrome database search strings ...... 284

8.1.2 Feline idiopathic cystitis database search strings ...... 285

Appendix 9: List of included studies ...... 288

Appendix 10: List of excluded studies...... 290

Appendix 11: Characteristics of excluded studies...... 294

Appendix 12: Meta-analysis for all biomarkers for which a meta-analysis was possible (more than one study for that biomarker, n=5)...... 299

Appendix 13: Summary of results from the Review Manager analysis of all 45 outcomes (biomarkers)...... 300

Appendix 14: Biomarker control genes synthesised by Bioneer Pacific...... 303

Appendix 15: Protein BLAST homology reports, displaying the top protein sequence for feline and canine TJP-1, and E-cadherin...... 304

Appendix 16: Nanodrop results for extracted RNA and reverse transcribed cDNA...... 306

List of Figures

Figure 1-1: Normal bladder wall with the GAG layer. B. Chronic inflammation leads to defects in the GAG and urothelial layer. Adapted from (74)...... 37

Figure 2-1: Flow chart for selection of pathology records of dogs and cats with bladder tissue submitted to the University of Queensland Veterinary Laboratory Service between January 1994 and March 2016. UQVLS – The University of Queensland Veterinary Laboratory Service...... 56

Figure 3-1: Flow chart for selection of canine and feline urinary bladder histology slides submitted to the University of Queensland Veterinary Laboratory Service between January 1994 and March 2016, selected slides from the Murdoch University School of Veterinary and Life Sciences pathology archives, and prospective samples obtained from local veterinary clinics and a veterinary pathology company in the Brisbane region. MUSVLS – Murdoch University School of Veterinary and Life Sciences; UQVLS – The University of Queensland Veterinary Laboratory Service...... 80

Figure 3-2: The final logistic regression model developed in Stata (382)...... 87

Figure 3-3: Probability for bladder outcomes by species (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _F = feline; _C = canine...... 91

Figure 3-4: Probability for bladder outcomes by presence of urothelial inflammatory infiltrate (95% CI). Probability for bladder outcomes by type of submucosal inflammation (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _N = no urothelial inflammation; _Y = urothelial inflammation was present...... 91

Figure 3-5: Probability for bladder outcomes by presence or absence of urothelial ulceration (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _N = no urothelial ulceration; _Y = yes, urothelial ulceration was present...... 92

Figure 3-6: Probability for bladder outcomes by type of submucosal inflammation (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _MON = mononuclear (non-neutrophilic) inflammation (lymphocytes, plasma cells and macrophages as the primary inflammatory cell type); _NEU = neutrophilic inflammation was the primary inflammatory cell type...... 92

Figure 3-7: Probability for bladder outcomes by amount of submucosal haemorrhage (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _N = no submucosal haemorrhage; _mild = mild amount of submucosal haemorrhage; _mod = moderate amount of submucosal haemorrhage...... 93

Figure 3-8: Probability for bladder outcomes by presence or absence of submucosal lymphoid aggregates (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _N = no submucosal lymphoid aggregates; _Y = yes, submucosal lymphoid aggregates were present...... 93

Figure 4-1: Sequence of assessment of the 25-slide set by each pathologist...... 105

Figure 4-2: Bar plot showing the inter-pathologist agreement kappa statistics with 95% CI for the three conditions of diagnosing bladder syndrome in canine and feline patients ...... 115

Figure 4-3: Bar plot showing classification accuracy for the four pathologists’ assessment of bladder tissue in canine and feline patients against the reference diagnosis...... 117

Figure 5-1: Flow diagram for inclusion of FIC studies...... 135

Figure 5-2: Flow diagram for inclusion of BPS studies...... 136

Figure 5-3: Review Manager meta-analysis of NGF urine measurements ...... 144

Figure 5-4: CCL2 meta-analysis...... 145

Figure 5-5: MIF meta-analysis...... 145

Figure 5-6: CXCL10 meta-analysis...... 146

Figure 5-7: IL-6 meta-analysis...... 146

Figure 5-8: IL-6 HIC only (row 1.8.1 and 1.8.3) ...... 147

Figure 6-1: Experiment 17, optimised PCR for TJP-1, using TJP2_modified primer set, 10µM concentration...... 178

Figure 6-2: Evaluation of GAPDH and RPS7 in some of the RNAlater samples (#51, 52 and 53)...... 179

Figure 6-3: Experiment 21, showing GAPDH on the left and RPS7 on the right, on the same set of samples...... 180

Figure 6-4: Experiment 22, PCR for GAPDH on canine FFPE samples, with positive bands in the positive control well (sample #1) and sample #6 in well 8 (asterisk)...... 181

Figure 6-5: Experiment 22, PCR for GAPDH on canine FFPE samples, showing positive bands in the positive control well sample #1, and sample #50 in well 14 (asterisk)...... 181

Figure 6-6: Experiment 22, PCR for GAPDH on feline FFPE samples, showing positive bands for samples #53 (positive control, well 2), and samples #1 and #19 (wells 3 and 14, asterisks)...... 182

Figure 6-7: Experiment 22, PCR for GAPDH on feline FFPE samples, showing positive bands for samples #53 (positive control, well 2), and samples # 31 and #39 (wells 7 and 11, asterisks)...... 182

Figure 6-8: Experiment 29, showing three clear amplification curves for samples 51, 53 and 54, while the remainder samples only have curves consistent with primer dimers...... 183

Figure 6-9: Melt curve from experiment 28, GAPDH on all samples, showing melt peaks in samples 51, 53 and 54, with very slight but inadequate peaks in samples 52 (at 70 -d(RFU)/dT)and sample 1...... 184

Figure 6-10: Real time PCR amplification curve for experiment 27, showing a positive curve only for the well containing the synthetic control gene...... 185

Figure 6-11: The melt peak for experiment 27, showing a single melt peak for the well containing the synthetic control gene...... 185

Figure 6-12: TJP-1 (left side) and E-cadherin (right side) immunohistochemistry. 1 and 2: TJP negative and positive controls, normal feline bladder. 3 and 4: E-cadherin negative and positive controls, normal feline bladder. 5 TJP-1 and 6 E-cadherin: Normal canine bladder showing faint staining (sample #41), with 30% of the cells exhibiting positive immunolabelling. 7 TJP-1 and 8 E- cadherin: Canine UC (sample #17), showing positive immunolabelling of the normal urothelium but variable immunolabelling within the neoplasm. 9: TJP-1, feline idiopathic cystitis (sample #28), showing scant urothelium remaining due to the disease process. 10: E-cadherin, feline urinary tract infection and urethral obstruction (sample #1), showing loss of urothelium during processing due to the submucosal haemorrhage...... 187

Figure 6-13: TJP-1 (1-3) and E-cadherin (4-6) immunohistochemistry, normal feline bladder. 1: TJP- 1 positive control. 2:TJP-1, negative control by omission of primary antibody. 3: TJP-1, negative control using species-matched IgG. 4: E cadherin positive control. 2: E cadherin, negative control by omission of primary antibody. 3: E cadherin, negative control using species-matched IgG...... 188

Figure 6-14: Chart of immunohistochemistry results, depicting the percentage of urothelial cells in each category that had positive immunolabelling for the target proteins, stratified b disease, then biomarker then species. SD standard deviation...... 190

Figure 7-1: A simple example of the type of decision tree that can be formulated by using logistic regression modelling...... 200

Figure 8-1: Case 02/0691A, canine, urothelial carcinoma...... 255

Figure 8-2: 04/1068A, canine, urothelial carcinoma...... 256

Figure 8-3: Case 05/1429A, feline, urothelial carcinoma...... 257

Figure 8-4: Case 07/2196, canine, mild cystitis...... 258

Figure 8-5: Case 08/1399D, feline, cystitis secondary to lower motor neuron bladder...... 259

Figure 8-6: Case 08/1885J, canine, urolithiasis...... 260

Figure 8-7: Case 10/0259B, canine, bladder wall leiomyosarcoma...... 261

Figure 8-8: Case 11/0220A, canine, urolithiasis...... 262

Figure 8-9: Case 12/0516, feline, suspected FIC due to clinical urethral obstruction, no uroliths on necropsy, and lack of urine culture results...... 263

Figure 8-10: Case 13/0197A, feline, suspected FIC due to clinical signs of stranguria, no uroliths on necropsy, and lack of urine culture results...... 264

Figure 8-11: Case 14/0707A, feline, suspected FIC due to clinical signs of urethral obstruction, no uroliths on necropsy, and lack of urine culture results...... 265

Figure 8-12: Case 14/0993B, canine, chronic follicular cystitis...... 266

Figure 8-13: Case 15/0893E, canine, normal bladder...... 267 22

Figure 8-14: Case 15-017E, feline, metastatic epitheliotropic lymphoma...... 268

Figure 8-15: 16/0034I, canine, normal bladder...... 269

Figure 8-16: Case 17/0311, canine, normal bladder...... 270

Figure 8-17: Case 17/0398, canine, normal bladder with submucosal lymphoid follicles...... 271

Figure 8-18: Case 17/0503A, feline, urolithiasis with urethral obstruction and bladder rupture. ... 272

Figure 8-19: Case 17/1335A, feline, normal bladder...... 273

Figure 8-20: Case 17/1336A, feline, normal bladder...... 274

Figure 8-21: Case 17/1337A, feline, urolithiasis + urinary tract infection (positive bacterial culture)...... 275

Figure 8-22: Case 18/0073A, canine, urolithiasis...... 276

Figure 8-23: Case 19/0528A, feline, suspected FIC due to urethral obstruction with no uroliths at necropsy and no urine culture results...... 277

Figure 8-24: Case 19/0729A, canine, urolithiasis and urinary tract infection (positive urine culture)...... 278

List of Tables

Table 1-1: Types of urinary bladder disease affecting people, cats and dogs (68, 69, 80)...... 36

Table 1-2: The pathogens most frequently isolated from urinary tract infections in humans, dogs and cats (24, 212, 243, 244, 260, 305, 313, 372, 448)...... 39

Table 1-3: Histological changes of bladder pain syndrome and feline idiopathic cystitis...... 50

Table 2-1: Reasons for bladder tissue collection...... 54

Table 2-2: Types of bladder neoplasia in the dataset ...... 58

Table 2-3: Breakdown of pathology records where the diagnosis for the bladder tissue was assigned to the ‘other’ category (n=46)...... 59

Table 2-4: Count data for canine and feline records with bladder tissue submitted to UQVLS January 1994 to March 2016...... 63

Table 2-5: Total number of records to number of cases, proportionate morbidity (PM) and proportionate morbidity ratios (PMR) of canine and feline neoplasia records as confirmed by histology findings...... 66

Table 2-6: Total number of records to number of cases, proportionate morbidity (PM) and proportionate morbidity ratios (PMR) of canine and feline cystitis records as confirmed by histology findings...... 67

Table 2-7: Total number of records to number of cases, proportionate morbidity (PM) and proportionate morbidity ratios (PMR) of canine and feline urolithiasis records as confirmed by clinical history and post-mortem findings...... 68

Table 3-1: Count data of all reviewed histology slides by species, following slide review...... 85

Table 3-2: Original and revised diagnoses of cases that had the diagnosis changed during the slide review process ...... 85 24

Table 3-3: Breakdown of pathology records where the diagnosis for the bladder tissue was assigned to the ‘other’ category (n = 47) ...... 86

Table 3-4: Neoplasia types ...... 87

Table 3-5: Results of the final multivariate multinomial logistic regression model showing animal or histological factors associated with diagnoses (cystitis, neoplasia and urolithiasis) compared with the reference category of normal/other diagnoses combined...... 89

Table 4-1: Histological criteria to be assessed by the pathologists in worksheets one and two...... 106

Table 4-2: Column headings and potential answers for the pathologists participating in the agreement study...... 107

Table 4-3: The count data from all study pathologists, P1-P4...... 109

Table 4-4: Diagnoses for every slide with the reference diagnosis, grouped by pathologist (P1-P4), under three different conditions (First, Hx and Tool)...... 110

Table 4-5: Inter-pathologist agreement kappa statistics for the three conditions of diagnosing bladder syndrome in canine and feline patients...... 114

Table 4-6: Classification accuracy and kappa statistics for pathologist agreement with the reference diagnosis for the three conditions of diagnosing bladder syndrome in canine and feline patients. . 116

Table 5-1: Characteristics of included studies ...... 137

Table 5-2: Reasons for study exclusion ...... 138

Table 5-3: Human and feline data availability by outcome (biomarker) ...... 140

Table 6-1: Potential biomarkers for canine and feline bladder disease...... 158

Table 6-2: Samples for PCR and IHC ...... 164

Table 6-3: Original and modified biomarker primers, designed from the National Center for Biotechnology Information nucleotide database...... 169

Table 6-4: Reference genes ...... 173

Table 6-5: Dako autostainer immunohistochemistry protocol after primary antibody incubation. . 176

Table 6-6: Antibodies used for immunohistochemistry ...... 177

Table 6-7: Grading results showing percentage of urothelial cells with positive membranous staining for the tight junction protein-1 (TJP-1) and E-cadherin, by species and diagnosis...... 189

List of abbreviations used in the thesis

ACVP American College of Veterinary Pathologists

Amp Amplicon

APF Antiproliferative factor

ATP Adenosine triphosphate

BLAST Basic Local Alignment Search Tool bp Base pairs

BRAF Gene that encodes the B-raf protein

CI Confidence interval

DNA Deoxyribonucleic acid

EGF Epidermal growth factor

ELISA Enzyme-linked immunosorbent assay

ESSIC International Society for the Study of BPS

FFPE Formalin fixed, paraffin embedded

FIC Feline idiopathic cystitis

FLUTS Feline lower urinary tract signs

FN-1 Fibronectin-1

GAG Glycosaminoglycan

GAPDH Glyceraldehyde-3-phosphatedehydrogenase

GIT Gastrointestinal tract

HB-EGF Heparin-binding EGF-like growth factor

HIC Hunner-type IC/BPS

H&E Haematoxylin and eosin

IC/BPS Interstitial cystitis/bladder pain syndrome, referred to in this work as BPS

ICSI Interstitial Cystitis Symptom Index

ICPI Interstitial Cystitis Problem Index

IHC Immunohistochemistry

IL Interleukin

INHAND International Harmonization of Nomenclature and Diagnostic Criteria for Lesions in Rats and Mice

ISH In situ hybridisation

ITPIG Investigative Toxicologic Pathology Interest Group

LUT Lower urinary tract

MICE Multiple imputation by chained equations

MUSVLS Murdoch University School of Veterinary and Life Sciences

NGF Nerve growth factor

NHIC Non-Hunner-type IC/BPS

NIDDK National Institute for Diabetes and Digestive and Kidney Diseases

OAB Overactive bladder

OR Odds ratio

PCR Polymerase chain reaction

PM Proportionate morbidity

PMR Proportionate morbidity ratio 28

PUF Pain Urgency and Frequency score

RFU Relative fluorescence units

RNA Ribonucleic acid

RPS7 Ribosomal protein S7

RT Reverse transcription

SB Subclinical bacteriuria

SE Standard error

TFF2 Trefoil factor 2

THP Tamm-Horsfall protein

TJP-1 Tight junction protein-1, also known as zona occludens-1

UC Urothelial carcinoma, previously transitional cell carcinoma/TCC

UPEC Uropathogenic Escherichia coli

UPK3A Uroplakin IIIa

UQ The University of Queensland

UQVLS The University of Queensland Veterinary Laboratory Service

UTI Urinary tract infection

µL Microlitre

µM Micromolar oC Degrees Celsius

Chapter 1 Introduction, literature review and scope of the study.

Introduction

Throughout my career as a veterinary clinician and more recently as a trainee veterinary anatomic pathologist, I have noticed variation in opinions and, occasionally, varying diagnoses provided by different pathologists on the same microscope slide.

Problem 1: Variation in opinions and diagnoses provided by different pathologists on the same microscope slide – there appears to be a lack of standardisation in the field.

a. Logistic regression is an objective statistical technique that has been used to predict disease occurrence – can we utilise this method help to standardise veterinary histopathology?

In clinical practice, urinary bladder diseases are common in dogs and cats, however I noticed that there is little published on bladder disease in an Australian canine and feline population.

Problem 2: Paucity of published literature on canine and feline urinary bladder disease in an Australian population.

b. Further information on pathogenesis is always helpful, particularly for the common but currently unexplained enigma of feline idiopathic cystitis

i. Investigate all canine and feline urinary bladder diseases by conducting a highly detailed histological evaluation of all bladder conditions found in dogs and cats in an Australian population.

c. FIC is a good model for human bladder pain syndrome – can we investigate this further?

i. Systematic review on biomarkers

ii. Lab work on biomarker expression in bladder samples (with the caveat that it is very difficult to obtain feline urinary bladder tissue)

For this PhD project, the decision was made to combine these two problems. Therefore, the overall PhD goals (and therefore chapters) were to:

1. Explore canine and feline urinary bladder disease in Australia using histopathology (Chapter 2 and Chapter 3).

2. Use this histological data to evaluate the utility of logistic regression modelling in histopathology; Can logistic regression modelling help to standardise veterinary pathology? (Chapter 3).

3. Test the logistic regression findings on professional veterinary pathologists currently working in diagnostics; Does incorporating findings from logistic regression analysis improve inter-observer agreement when evaluating bladder histology? (Chapter 4).

4. Conduct a systematic review on biomarkers in bladder pain syndrome in humans, as well as feline idiopathic cystitis in cats, to explore the current biomarkers of BPS and look at just how comparable BPS and FIC are to each other (Chapter 5).

5. Utilise laboratory techniques to interrogate one of the BPS biomarkers identified from the systematic review in both canine and feline bladder tissue, particularly tissue from cats with FIC (Chapter 6).

Variation between veterinary pathologists

The discipline of anatomic pathology involves the gross or histological morphologic assessment of lesions within tissues, with the goal of providing a diagnosis and prognosis to the clinician and subsequently the patient or pet. Inter-observer variability is a common problem in both human and veterinary pathology (93, 418), and research-heavy toxicological pathology groups have researched this in detail (350). Inter-observer variability is a potential hazard to human and animal health in the form of diagnostic discrepancies, potentially resulting in inappropriate subsequent therapeutics and disease management (445, 454).

Variation between pathologists has long been considered a problem across the human and veterinary medical fields, with the potential subjectiveness of histological assessment leading to the recommendation that second pathology review become standard practice (10, 83, 410). Veterinary pathologists face the unique challenge of dealing with histological specimens from a plethora of species, with many pathologists having certain areas of interest or specialisation. The large number of potential species and diseases means it is virtually impossible to have seen examples of every possible disease. This may predispose veterinary pathologists to low levels of cognitive biases

towards some diseases, depending on their area of interest or work practice. Cohen’s kappa statistic measures the level of agreement between two or more observers and the likelihood that agreement would occur by chance alone (90). A Cohen’s kappa statistic of greater than 60% is generally deemed to be an adequate level of agreement (306), however 60% agreement may not be adequate for the life or death consequences of disease misdiagnosis. No large cross-disciplinary studies have been identified by this author, however some examples of levels of inter-observer agreement in human and veterinary pathology include 40% for canine and feline intestinal biopsies (445), 60% for human pathologists evaluating pulmonary neoplasms (301), and 60% for human pathologists evaluating breast cancer response to chemotherapy (454).

There are many causes for inter-pathologist variation including level of experience (429), sample quality and processing (444), and the organ system being examined. For example, agreement between pathologists is low when assessing gastrointestinal tract (GIT) biopsies in cats and dogs due to the difficulty in interpreting the degree of resident leukocytes compared to pathological inflammation, the focal nature of some gastrointestinal diseases and limited GIT tissue sampling from live patients, and the numerous, sometimes inconsistent, systems or grading schemes for GIT evaluation (298, 444). A further limitation to consistent disease diagnosis is a lack of standardization of pathology reporting, where pathologists may agree that an abnormality is present, but disagree on the specific disease diagnosed, or even the best name for that entity when multiple terms are used (445). The literature suggests that clearly defined histological criteria improves pathologist agreement regarding the assessment and interpretation of histological changes (141, 445), and efforts in this area have been made, particularly in the toxicologic veterinary pathology fields. Examples include the development of the INHAND (International Harmonization of Nomenclature and Diagnostic Criteria for Lesions in Rats and Mice) guides from 2013-2018 (274), and the work of the Investigative Toxicologic Pathology Interest Group (ITPIG). However, these guidelines are limited to laboratory animal species and toxicologic changes, so do not encompass the full range of lesions that are encountered in general pathology. In general veterinary pathology, standardisation guidelines are common for neoplastic diseases (145, 200) and there have been recent advances in standardising renal and intestinal lesions (81, 93), however to date, there are no standardised veterinary histological criteria for assessment of naturally occurring urinary bladder diseases of cats and dogs.

Pathologists make a diagnosis through the recognition of one or more histological findings, ideally combined with the patient’s clinical history (341). It is well known that pathologists are influenced by cognitive bias (6, 128) – a facet of the discipline that has both benefits and limitations (156).

Examples of cognitive bias include (a) confirmation bias – looking for evidence of a favoured hypothesis; (b) diagnostic drift – slight variations in scoring values over the course of a study; (c) tunnel vision - also known as anchoring, the tendency to rely too heavily on the first information presented; (d) avoidance of extreme scoring ranges; as well as (e) availability bias – referring to what most easily comes to mind, which means less familiar diseases are forgotten (6, 272). There has been much research into medical diagnostic cognitive biases, with up to 71% of studies showing an association between cognitive biases and treatment or patient management errors (360). The blinding of pathologists for research studies has been a source of controversy. Unblinding can lead to cognitive bias, but the pathologist diagnostic process by definition incorporates elements of animal signalment and clinical information, therefore blinding of pathologists essentially prevents them from making an informed assessment of the observed lesions (88, 376).

Disagreement between pathologists evaluating histological specimens has prompted recent advances in computer learning and computer-based image analysis (6). When assessing a microscope slide, pathologists incorporate the observed changes with their knowledge obtained via extensive training and experience to formulate an ordered differential list (24). The diagnosis with the highest probability is then reported as the final diagnosis. Similarly, mathematical models also evaluate information and then provide probabilities of an outcome occurring. In this context, a ‘model’ is a mathematical representation of a biological system, designed using computer software to help the user understand that system (404). First, the concepts are identified, then mathematical formulas are chosen that best suit the characteristics of that system (404). Next, the model is applied to the data, then the final step in modelling a system is validation of the model – that is, testing the model against the biological data, to see if the model provided an accurate representation of the biology (404).

Mathematical models for the pathologists’ diagnostic processes do exist, however they only exist in human pathology and very few have been validated (418). There is a gap for researchers to explore the utility of mathematical models in the veterinary pathology diagnostic process.

A novel role for logistic regression modelling

Logistic regression is a statistical method used to explain the relationship between data categories and outcomes (108). Note: the word ‘outcome’ in this discussion refers to the disease diagnosed. The output of logistic regression is an odds ratio which reflects the likelihood that an event/disease occurs, depending on dichotomous explanatory variables (e.g. what is the odds of a patient developing diabetes if they are overweight or not overweight) (379). In a stepwise procedure, logistic regression

is used to build the best possible model that explains the relationship between the outcome/s of interest and the explanatory variables in the dataset (108, 379). To do this, the model uses logit (log-odds) transformation to predict the probability of the outcome/s occurring in every possible combination of explanatory variables, thus is a sort of simulation (108, 379). Logistic regression modelling allows prediction of outcome probability from a combination of continuous and discrete independent variables (35).

Logistic regression multivariate models have the potential to predict disease risk in both the human and veterinary medical fields by providing an objective probability of the disease occurring given the combination of variables (158). These models analyse user-entered data from a (ideally) large number of cases, or a ‘reference dataset’. The model then formulates probabilities for the diagnosis/outcome of a new case, based on the reference dataset. Predictive models have been used to estimate disease probability in some human medical fields such as cardiac exercise testing, and to estimate the risk of cardiac disease given multiple test results and patient factors (35, 293, 451). Lung cancer is another area of human medicine that has utilised predictive models based on logistic regression (100). In the veterinary field, predictive logistic regression models have been used in genetics (27), ultrasonography (296, 297), surgery and surgical prognosis development (150, 344), predicting disease outbreaks (369), and even evaluating associations between animal factors and the presence of subclinical bacteriuria in cats (288). However, their use in veterinary histopathology has been limited thus far to occasional studies in wildlife (144). Glueckert and colleagues used logistic regression modelling to identify potential predictors/risk factors for skunks testing positive to a new Amdoparvovirus, with variables including age, sex, geographic zone, signs of illness, body condition score, and time period (144). Predictive logistic regression models have potential to assist decision making in veterinary histopathological diagnostics, and to improve pathologist and veterinarian awareness of the risk of a disease occurring in an individual patient. In summary, pathologists incorporate a wide range of knowledge and experience to provide the most likely diagnosis based on the microscopic changes they observe. An alternative to this manual decision-making process, such as logistic regression modelling, may be able to bypass cognitive bias and improve inter-pathologist agreement and subsequently disease diagnosis in patients.

In summary, logistic regression is an interesting method for the analysis of histological data. This project will conduct a detailed histological examination of canine and feline bladder diseases, and interrogate this data using logistic regression modelling. The results from the model will then be used to build a diagnostic tool for pathologists to use when assessing bladder samples. This tool will be

tested on pathologists and the results analysed to investigate whether this method had any effect on inter-pathologist agreement.

Canine and feline urinary bladder disease

The remainder of this literature review explores the current knowledge of selected urinary bladder disorders of dogs and cats, with a focus on pathogenesis and comparative aspects. Many similarities exist between bladder conditions that occur in humans as well as in cats or dogs, demonstrating potential for ongoing comparative and translational research of these conditions using companion animal models. Comparative medicine refers to the study of naturally occurring diseases in nonhuman species as a model for understanding human diseases (204). Briefly, translational medicine is a branch of biomedicine that focuses on translating research findings in animals to improving human health (85). Traditional laboratory animals (rats, mice, rabbits) were used to research human diseases, however dogs and cats are being increasingly recognised as better models for human diseases as they share our environment and are exposed to similar stressors and external stimuli (94, 174). In addition, compared to rodents the longer lifespan of dogs and cats allows the contraction of more age-related disorders, and also a longer time to explore disease progress and explore the safety and efficacy of novel therapies (174).

Many bladder conditions affecting cats and dogs also occur in people, including urothelial carcinomas (UC, previously termed transitional cell carcinoma), urinary tract infection (UTI), and non-infectious cystitis. Urinary bladder neoplasms account for 1.5-2% of all canine neoplasms, while the proportionate morbidity (or prevalence) of neoplasia cases in cats is much lower (76). Bladder neoplasia is more common in older male cats than in females (446) and in dogs is more common in older female neutered dogs with Scottish terriers overrepresented (204). Feline idiopathic cystitis (FIC) is a spontaneously occurring, non-infectious bladder disease of cats that is believed to have many comparable properties to bladder pain syndrome (BPS) in people, however the pathogenesis of these two diseases has not been fully elucidated. Table 1-1 depicts a summary of types of inflammatory disease in people, dogs and cats, with the diseases in bold face the focus of this thesis. Herein, this review will focus on the comparative aspects of inflammatory and infectious urinary bladder diseases of cats and dogs.

Table 1-1: Types of urinary bladder disease affecting people, cats and dogs (68, 69, 80).

Disease Humans Cats Dogs

Cystitis Bacterial* Yes Yes Yes Papillary-polypoid Yes Yes Yes Follicular Yes Rare Yes Interstitial/idiopathic Yes Yes No Eosinophilic Yes Not documented Yes Emphysematous Yes Yes Yes Gangrenous Yes Rare Rare Haemorrhagic Yes Yes Yes (cyclophosphamide) Viral Rare Rare Rare Radiation (iatrogenic) Yes Yes Yes Chemical (iatrogenic) Yes Rare Rare Malakoplakia Rare Rare Rare Fungal Rare Rare Rare Schistosomiasis Yes No Yes

Malformations Bladder duplication Rare Not documented Rare Other malformations Rare Rare Rare

Benign polypoid and Yes Rare Rare papillary lesions

Neoplastic Urothelial carcinoma Yes Yes Yes Other neoplasms Yes Yes Yes

Urolithiasis Rare Yes Yes

Other disorders Dilation of the bladder Yes Yes Yes (obstruction or neuromuscular disease) Haematoma (trauma) Yes Yes Yes *See Table 1-2: The pathogens most frequently isolated from urinary tract infections in humans, dogs and cats (25, 212, 243, 244, 260, 305, 313, 372, 448).

The urinary system plays a vital role in the excretion of waste products and regulation of electrolytes, with the urinary bladder responsible for storage of urine prior to voiding (107). As shown diagrammatically in Figure 1-1 the bladder is comprised of four histologically distinct layers (2, 70): 1) The urothelium, a protective mucosa composed of three to seven layers of stratified columnar epithelial cells (transitional epithelium), covered by a luminal layer of mucin comprised of sulphonated glycosaminoglycans (GAGs) and glycoproteins (37). This GAG layer is important in preventing bacterial adherence and providing physical protection to the urothelium (37); 2) The lamina propria and submucosa (the term submucosa will herein be used to refer to both layers), comprised of inner dense and outer loose collagenous connective tissue containing afferent nerves, blood vessels and interstitial cells; 3) A smooth muscle wall (muscularis) comprising inner and outer longitudinal layers and a middle circular layer; and 4) The serosa (20, 70, 174, 286, 348). The submucosal layer is thought to play an important role in bladder filling and compliance, and adaptation to altered bladder size, while the muscularis serves to prevent overdistension (14).

Figure 1-1: Normal bladder wall with the GAG layer. B. Chronic inflammation leads to defects in the GAG and urothelial layer. Adapted from (74).

The bladder wall is home to a number of resident leukocytes, primarily lymphocytes and mast cells, with mast cells playing a controversial role in some inflammatory conditions (262, 263). Afferent nerves in the submucosa detect biochemical changes that signal bladder filling and fullness (174) and travel with both parasympathetic and sympathetic nerves (107). As the bladder fills the urothelium is stretched, resulting in the release of neurotransmitters, primarily adenosine triphosphate (ATP) and acetylcholine, increasing firing rates of the afferent nerves (174). In people, the activity of these 37

nerves has been shown to be altered by bladder pressure, bladder wall tension and mental stress, as is postulated to play a role in inflammatory bladder disorders (54, 72, 225).

The urinary bladder has a fairly consistent response to injury regardless of the inciting cause - in the acute stages of injury, changes include vascular dilation and congestion with leakage and accumulation of leukocytes and plasma constituents in the bladder wall, as well as erythema and haemorrhage in the urothelial layer with moderate to severe oedema within the mucosa and submucosa (69, 135, 153). The first inflammatory cells to be recruited are neutrophils into the urothelial layer, with expansion into the submucosa with severe acute inflammation (69). In acute injury the urothelium undergoes hyperplasia and hypertrophy (69, 135). With ongoing insult, the urothelium may become metaplastic, and may become ulcerated particularly in the case of urocystoliths, while it is common in bacterial cystitis to get erosion of the urothelium with fibrinous exudate (69, 153). Inflammatory cell type shifts to a more lymphocytic or lymphoplasmacytic infiltrate with prolonged inflammation (69), and may also include eosinophil and macrophages (153). In chronic inflammation, submucosal haemorrhage and oedema may persist, and granulation tissue and fibrosis may develop in the bladder wall (153, 232). Finally, development of lymphoid follicles in the submucosa is a common response to chronic bladder injury in people, and is present in 35% of patients with urinary tract infection (69, 361).

Infectious bladder disease

1.5.1 Aetiology and risk factors

Bacterial urinary bladder infection, often simply termed urinary tract infection (UTI), is common in humans and dogs. The term cystitis refers to inflammation of the bladder, which can occur in the presence or apparent absence of infectious organisms. Urinary tract infections are more common in women than in men, with 25-30% of 20-40 year old women having a history of treatment for UTI (212), and an estimated 40% of all women having a UTI during their lifetime (137, 372). In dogs, UTIs are more common in females than males, presumably due to a shorter urethral length (23), and are thought to affect 14% of dogs at some point during their life (240). In a South American study, bacterial urinary tract infection made up 34% of lower urinary tract disorders in dogs presenting with lower urinary tract clinical signs such as haematuria, stranguria, pollakiuria, or urinary incontinence (285). UTIs are relatively rare in cats, with a reported lifetime incidence of 3-19% (25, 111, 420).

In people, bacteria are thought to be involved in the development of a granulomatous bladder inflammation termed malakoplakia (69, 356), which has also been diagnosed very rarely in cats and 38

dogs (31, 356). In addition to bacteria, other organisms may infect the bladder including Schistosomes, fungal or viral organisms in people (69), and fungi and parasites (Capillaria spp.) in cats and dogs (25, 179, 240, 425). Viruses have also been detected in feline urine, including Herpesvirus (259) and Calicivirus (222). This section will herein focus only on bacterial urinary tract infection as clinical infection with other pathogen types in Australia is very rare (Table 1-2).

Clinical risk factors for bacterial UTI in humans include pregnancy, diabetes mellitus, spinal cord injury, and family history of UTI (372). Risk factors for UTI in dogs include having urothelial carcinoma (55% [47/85] dogs with UC had at least one positive urine culture during one study)(47), the presence of skin folds around the vulva (105), advancing age (397), sex and neuter status (spayed females are at higher risk) and breed (305, 368), as well as urinary catheterisation, urinary incontinence and antibiotic or corticosteroid medications (80). Risk factors for UTIs in cats include advanced age (greater than ten years), urinary catheterisation, perineal urethrostomy, hyperthyroidism, diabetes mellitus, and chronic kidney disease (244, 460).

Table 1-2: The pathogens most frequently isolated from urinary tract infections in humans, dogs and cats (25, 212, 243, 244, 260, 305, 313, 372, 448).

Humans (%) Dogs (%, where reported) Cats (%) Escherichia coli (E. coli) (77) E. coli (52.5) E. coli (37-60%) Staphylococcus saprophyticus (<10) Staphylococcus species (13.6) Enterococcus faecalis (27) Enterococcus faecalis (<10) Enterococcus species (13.3) Staphylococcus species (20) Proteus mirabilis (<10) Streptococcus species Proteus species (4.7) Klebsiella pneumoniae (<10) Proteus species Enterobacter species (2.4) Pseudomonas aeruginosa (<10) Klebsiella species Pseudomonas aeruginosa (1.6) Pseudomonas species Streptococcus bovis (1.6) Klebsiella pneumoniae (0.8)

Escherichia coli (E. coli) is a particularly important bladder pathogen in people, dogs and cats as there is a uropathogenic strain (UPEC) that targets the lower urinary tract. UPEC in cats and dogs shows a high rate of antimicrobial resistance (163, 299), while the canine strains have zoonotic potential, posing a human health risk (299, 456).

Of young cats showing lower urinary tract clinical signs (FLUTS) 3% (49, 62, 280) to 33% have evidence of a urinary tract infection (117, 211). In contrast, 35% of cats with positive urine culture (n=155) did not show any lower urinary tract clinical signs, i.e. had bacterial colonisation of the bladder without active infection (subclinical bacteriuria, SB), which may suggest a falsely high

prevalence of positive urine bacterial culture in cats (280).

Dogs can rarely develop a polypoid cystitis that may be associated with chronic bacterial infection (281). A recent Brazilian study reported on the prevalence of urinary tract lesions in dogs, and found that 40% of lower urinary tract lesions were compatible with cystitis (n=60) (359). Although this study did not confirm the presence of organisms in these patients, since non-infectious cystitis has not yet been documented in dogs, it is assumed that their aetiology was bacterial.

1.5.2 Subclinical bacteriuria

Subclinical bacteriuria (SB), is the bacterial colonisation of the urinary tract without LUT clinical signs (435), also termed asymptomatic bacteriuria in people. There has been much discussion on how to differentiate between SB and active urinary tract infection, however information is lacking on the role of asymptomatic bacteriuria in people (60). In women suffering recurrent UTI, asymptomatic bacteriuria was found to be protective against UTI recurrence, with Escherichia coli and Enterococcus faecalis the two most commonly isolated pathogens from asymptomatic bacteria- colonised urine (60).

In dogs, bacterial growth upon urine culture has been identified in up to 12% of patients (428), with increased rates in dogs with concurrent health problems such as diabetes mellitus or obesity (261, 284). A microbiome analysis of the canine lower urinary tract has revealed a number of organisms that seem to be resident organisms in the urine of normal dogs (no lower urinary tract clinical signs and a negative bacterial urine culture), indicative of a urinary bladder-specific microbiome (57). This finding suggests that further caution is needed when interpreting low-growth urine culture results (57). Subclinical bacteriuria has been detected in cystocentesis-obtained urine samples in 6.1% to 13% of cats greater than six years of age with no lower urinary tract signs (288, 339, 442), suggesting this could be a normal baseline level for SB in cats. Female sex and hepatic disease were found to be significant risk factors for subclinical bacteriuria in a recent study of 179 cats (288). Further studies are needed to evaluate the clinical significance of SB in cats (339).

1.5.3 Pathogenesis

The pathogenesis of UTIs in people, dogs and cats relies on the ability of the infecting organism to overcome the host’s defences. Bacteria typically enter the urinary bladder via retrograde urethral flow, where deficient host defences can result in colonisation of the bladder wall and infection (415).

Host defences of the urinary bladder include physical barriers, normal anatomy, mucosal defence mechanisms and the composition of urine itself (25, 244, 280, 415). Physical barriers include urothelium, urethral length and peristalsis, high pressure zones within the urethra, longitudinal folds in the proximal urethra that can trap bacteria, and bladder filling and emptying (415). These mechanisms are disrupted by urothelial trauma, stagnation of urine (obstruction), and incomplete bladder emptying (neurological) (80). Anatomical defences include the round, smooth lining of the bladder, ureteral openings into the bladder neck proximal to the urethral sphincter, and a completely occluding urethral sphincter (415). Mucosal defence mechanisms include normal urethral flora (impacted by systemic antibiotic administration), GAG mucin layer, mucosal antimicrobial properties such as the release of microbial peptides such as lipocalin-2 in response to bacterial presence (59), and the local immune response, particularly immunoglobulin A and IL-6 (432). Urine composition provides a further layer of defence due to its high concentration and high osmolality from urea, organic acids, and antimicrobial peptides (415). In people, given the strong association of history of UTI, as well as increased risk of recurrent UTI and pyelonephritis, in first-degree female relatives, this combination of host defences is postulated to be genetically heritable (362). This could also be true for animals.

In addition to host factors, bacterial factors also play a role in the development of urinary tract infections. Adhesins are bacterial characteristics that encourage adhesion to mucosal surfaces, such as pili or components of the bacterial wall or capsule (415). Some bacteria, such as uropathogenic Escherichia coli, have specific bacterial factors to infect a host including fimbriae, flagella, adhesins, siderophores, endotoxin, polysaccharide coatings and biofilm formation (166, 299).

Inflammatory bladder diseases

Inflammatory processes within the body are complex and they involve a plethora of inflammatory mediators including cytokines, histamine, complement factors and neuropeptides, mostly produced by leukocytes, such as eosinophils and mast cells, and other components of inflamed or damaged tissue (457). The pathways and mechanisms of inflammatory responses are similar in all mammals; thus, companion animals can serve as valuable disease models for human inflammatory diseases (174, 223, 354). Although elements of inflammation are associated with bladder neoplasia, particularly lymphocyte infiltration (68), this review will focus on non-neoplastic diseases of the bladder. There are many potential causes of urinary bladder inflammation including congenital, chemical, toxic, radiation, ischemic, hypoxia, iatrogenic, parasites, viruses, and fungal causes as well as the influence of metabolic disorders such as diabetes mellitus (69), however an in-depth discussion of all of these 41

causes was deemed outside the scope of this review. Herein, we focus on the more common non- neoplastic inflammatory bladder disorders in dogs and cats - urolithiasis and feline idiopathic cystitis (FIC) respectively, as well as the human bladder pain syndrome for which FIC is commonly cited as a naturally occurring animal model (174). It is important to note that the term cystitis can have two meanings - the clinical definition of cystitis typically refers to patients with urinary tract infection (397), while the pathological diagnosis of cystitis refers to inflammation of the bladder, in the form of cellular infiltrates combined with vascular changes such as haemorrhage and oedema (80). I have endeavoured to use the pathological definition in my subsequent research chapters, however the research on prevalence and risk factors is often referring to the clinical definition.

Cats are well recognised as being predisposed to lower urinary tract disease. One disease of major interest is feline idiopathic/interstitial cystitis (FIC) which has many comparable features to human bladder pain syndrome/interstitial cystitis (IC/BPS). The terms idiopathic cystitis and interstitial cystitis have been used interchangeably in both the human and veterinary literature; the term ‘bladder pain syndrome’ (BPS) will be used to refer to the human disease in this discussion (400, 417), while idiopathic cystitis will be used to discuss the feline condition. Feline idiopathic cystitis is a non- infectious, inflammatory disease of the urinary bladder, and is one of a group of disorders labelled feline lower urinary tract disease (FLUTD). The prevalence of FLUTD in the general cat population has been found to be 1.5% to 4.6% in the following studies: 1.5-1.7% (258) and 4.6% (198) of cats in the United States, 3% of Norwegian cats (259), 3.3% of cats in the United Kingdom (282) and 2.2% of cats in Thailand (337). Of cats with FLUTD, 54-69% of these have been found to have FIC (98, 199, 337, 400, 460). Of FIC cats, 80% are male, with 20-55% found to have urethral obstruction (98, 142, 204). Potential causes of urethral obstruction are urethral inflammation leading to muscular spasms, neurological dysfunction, urethral plug formation [more urolith matrix than mineral, (316)], and obstruction with a urolith (12-30% of obstructed cats), most commonly struvite (98, 142, 316). Therefore, the prevalence of urolithiasis in cats is similar to dogs, at up to 0.5% (calculated by the author from the above percentages). Besides FIC, other less common causes of FLUTD include bacterial, fungal or parasitic infection, urethral plugs, uroliths, anatomic abnormalities, trauma, neurological disease and iatrogenic causes (259, 316, 460).

A higher proportion of non-infectious cystitis is reported in people in contrast to dogs and cats. Non- infectious cystitis is characterized as a purely inflammatory bladder wall disease and can include interstitial cystitis/bladder pain syndrome (BPS, also termed sterile cystitis) and overactive bladder (OAB) which involves either neurogenic or idiopathic detrusor (muscularis) overactivity (15). Non-

infectious cystitis may be induced by some medications such as cyclophosphamide chemotherapy (69), the presence of a foreign body, or secondary to an autoimmune response (151) or upper urinary tract obstruction (235). Uroliths (i.e. bladder stones) are relatively uncommon in people, constituting only 5% of all urinary tract stones (109), with most bladder stones forming secondary to urinary stasis from benign prostatic hyperplasia or neurogenic bladder disease (405). In women, uroliths are commonly associated with foreign bodies such as suture material (383). The most common mineral composition of human uroliths is uric acid (50%), with struvite stones being associated with bacterial urinary tract infection (237).

The prevalence of canine urolithiasis is thought to be up to 0.5% of dogs (427). One and a half to three percent of dogs admitted into veterinary care are diagnosed with urolithiasis (bladder stones) (318), with one South American study finding that urolithiasis was diagnosed in 20% of dogs presenting with lower urinary tract clinical signs (285). Uroliths in cats and dogs may be associated with a non-infectious cystitis (inflammation of the bladder wall in the absence of cultured microorganisms) (74, 193). However, up to 50% of canine bladder stones are thought to be infection related (387), a phenomenon that is much more common in dogs than in cats (169). Bacterial cystitis may lead to urolithiasis, particularly struvite stone formation secondary to urease-producing bacteria (314, 366). Conversely, uroliths can impair bladder defences, facilitating the development of urinary tract infection (314). Urinalysis parameters in dogs with calcium oxalate urolithiasis compared to controls (dogs without bladder stones and no lower urinary tract clinical signs), have shown no difference in the number of dogs with bacteriuria in each group (6%, 2/34 in both groups) (193). It is important to note that urine sampling method was not disclosed so the presence of bacteria could be the result of urethral or vaginal flora contamination if a voided urine sample was used (193). In summary, while non-infectious cystitis in people is primarily associated with BPS and OAB, non- infectious cystitis reported in dogs and cats is primarily associated with urolithiasis or feline idiopathic cystitis.

1.6.1 Clinical signs and epidemiology

The nomenclature and diagnosis of FIC has long been a source of confusion for veterinarians and veterinary pathologists. The definition of FIC is inconsistent throughout the literature, particularly with regard to the presence of urethral plugs and whether such should exclude a diagnosis of FIC. The most commonly accepted clinical definition of FIC is the presence of chronic, waxing and waning clinical signs of irritative voiding (pollakiuria/increased frequency, stranguria/painful urination, and dysuria) with an absence of neoplasia or bacteriuria, with or without the presence of uroliths or 43

urethral plugs (97, 214, 215, 256). Uroliths and urethral plugs are thought to be a sequela of underlying urinary bladder and/or metabolic abnormalities and are not specific to FIC (316).

Bladder pain syndrome in people is currently defined as chronic, debilitating, hyperaemic inflammation and pain of the urinary bladder of more than six months duration, in the absence of any other bladder pathology (99, 151, 255, 417). Pain on bladder filling is a defining feature of BPS (92, 417), along with other symptoms such as nocturia (needing to urinate during the night), urinary urgency and increased frequency (92). There is a large variation in clinical presentation, quality of life, and cystoscopic and biopsy findings in BPS, which has caused inconsistency in diagnosis, treatment and prognosis (417).

The welfare of cats with FIC is severely compromised, with a mortality rate of 12.5% that increases to 26% in cats with obstructive FIC, primarily due to elective euthanasia following repeated obstructive or non-obstructive episodes (98, 142). The prevalence of feline lower urinary tract signs in the general cat population is 2-4.6% (171, 198, 258, 259), with around two thirds of these cases attributed to FIC (62, 98, 110, 231, 232), therefore one in every 200 cats faces possible euthanasia due to the current lack of knowledge regarding the pathogenesis of FIC.

Cats with FIC can present with varying combinations of the following clinical signs: dysuria (77%), stranguria and periuria (70%), pollakiuria (78%) and macroscopic haematuria (71%) (55, 98). Episode duration for FIC is 2-90 days (median 6.5 days), with 78% of cats experiencing more than one episode and 50% having a mean inter-episode interval of less than six months (98). In summary, episodes of FIC can occur often, with each episode lasting around seven days. Male cats are predisposed to potentially fatal urethral obstruction due to their narrow urethra (315). In addition, 67% of obstructed cats had crystalluria without urolithiasis, suggesting their obstruction was caused not by uroliths but by other causes such as protein/mucus plugs or urethral spasm (98).

Clinical risk factors for FIC include higher body weight, younger age (less than 10 years) (74), higher number of cats in the household, lower activity level, less hunting behaviour, higher body condition score, lower water intake, using a litter box, and having less access to the outside (98, 168). Magnesium ammonium phosphate (struvite) crystals have been found in the urine of up to 58% of cats with FIC (98, 193); however, it is generally accepted that crystals are commonly present in feline urine, with crystalluria documented in 41/99 (41%) healthy, middle-aged to older cats during health screening (104, 321). Further, struvite is the mineral most frequently detected in feline urethral plugs thus it may play a role in FIC, however limited information is available about this potential

pathogenesis (312, 319). Currently, urinary crystal formation is thought to be exacerbated by the inflammation caused by FIC, and it is not believed to be involved in the primary pathogenesis of FIC (74). In summary, most cats with FIC are most likely to be overweight, male, young adults that live indoors.

Both FIC and urolithiasis can result in urethral obstruction in cats, with one study of 45 cats with urethral obstruction identifying uroliths in 29%, urethral plugs in 18%, and no cause identifiable (suggesting idiopathic cystitis) in 53% of these patients (142). Uroliths and their sequelae are deemed among the most important urinary tract problems of companion animals (80). Clinical signs of urolithiasis include dysuria, haematuria, pollakiura, and sometimes inappropriate urination (40). Clinical risk factors for urolithiasis in dogs and cats are many and include: age, breed, sex, hydration, altered urine pH, genetic alterations in protein or calcium metabolism, altered renal protein excretion (urate and cysteine), urinary tract infection with urease-producing bacteria (struvite stones), and liver disease or portosystemic shunts (urate stones) (169, 257, 319). Struvite uroliths are the most common urolith in dogs, while calcium oxalate stones are the most common in cats (319). See section 1.6.3.2.

The prevalence of BPS in women is thought to be around 0.5%, although may be up to 12% (353), with up to 0.04% of men being affected by BPS (18, 92, 151). Prevalence increases from 0.5% up to 1.4% in women who have a close relative with BPS, suggesting a potential hereditary component to the disease (92, 431). Neither crystalluria nor urolithiasis appear to be a features of BPS. Many individuals with BPS experience non-urinary tract problems such as other chronic pain disorders (175), endometriosis (399), psychiatric disorders (271), drug hypersensitivity (331) and irritable bowel syndrome, asthma and allergies (432). It has been hypothesised that both BPS and FIC comprise part of a broader, systemic problem involving the brain, nerves and other organ systems (51, 52, 192). Adverse events occurring early in life have been linked to chronic pain disorders and are frequently identified in BPS patients, prompting the description of BPS as a functional somatic syndrome (106, 430). Along the same lines, it has been hypothesised that FIC development could reflect a combination of genetics and foetal/early life stressors, similar to some chronic human syndromes including BPS (48). These factors are thought to result in an overactive sympathetic nervous system, leading to FIC as well as other comorbidities such as intestinal, behavioural, dermatologic, epithelial, neurologic, endocrine, or immune disorders (52). Naturally occurring FIC represents a more comparable bladder disease to BPS than any other inflammatory bladder disease model (440).

1.6.2 Types of BPS/FIC

There are two types of BPS, Hunner-type BPS (classic, previously ‘ulcerative’) or non-Hunner (nonulcerative) type, with or without Hunner’s lesions respectively (151). The two subtypes of BPS have different histological features, discussed further in section 1.7 (195, 267). It has been hypothesised that the Hunner and non-Hunner subtypes of BPS could in fact be separate diseases (230, 335). Histologically, patients with Hunner’s lesions have a markedly worse combination of severe, transmural bladder inflammation as well as muscularis fibrosis and mastocytosis (183). On the other hand, patients without Hunner’s lesions experienced the same clinical symptoms but histologically had mostly normal urothelium and only mild inflammation, with the primary changes being mucosal rupture and suburothelial haemorrhages (183). It is thought that FIC is most similar to non-Hunner BPS (50, 271).

1.6.3 Pathogenesis

1.6.3.1 Bladder pain syndrome/feline idiopathic cystitis

The precise aetiology and pathogenesis of BPS and FIC are unknown but are hypothesised to involve a multitude of factors including increased bladder wall permeability, neurological abnormalities, stress and environmental factors, genetic factors, and concurrent diseases such as autoimmune, infectious or systemic disease processes. The literature currently suggests that bladder pain syndrome and feline idiopathic cystitis may be part of a multi-organ disorder involving physiology, genetics and early life environment (48, 52). Neurological abnormalities are one of the major hypothesised features of BPS and FIC. One proposed mechanism involves neurogenic inflammation with increased C-fibre neuron sensitivity, and upregulation of the stress response system (SRS), indicating increased sympathetic activation (72) due to environmental stressors (439, 440). Increased immunoreactivity for substance P, a neurokinin neurotransmitter, and increased density of neurokinin-1 receptors has been associated with BPS and FIC (54, 173, 291). In addition, larger dorsal root ganglion cell bodies and abnormal neuropeptide profiles and hyperexcitability have been found in FIC cats (365), suggesting a role for anatomical and physiological changes in sensory neurons. Finally, altered acetylcholine synthesis and release in the oesophageal mucosa has been identified in FIC cats, suggesting that systemic cholinergic changes may be involved in altering systemic sensory function and visceral hyperalgesia (38). A more recent found altered serotonin release and therefore acetylcholine release from the mucosa of FIC cat bladder tissue compared to healthy controls (172). Neurogenic changes, such as upregulation of neuro-inflammatory mediators, have been observed in

BPS patients compared to controls (358). Nerve growth factor (NGF), a neurotrophic factor, has been found to be elevated in BPS patients compared to controls, but NGF has not been evaluated in FIC (177, 255). The role of these neurotransmitters is not completely known, however the release of these inflammatory mediators, such as substance P, by the sensory nerves results in local inflammation and hyperalgesia, and in some cases triggering an inflammatory cascade that can initiate mast cell degranulation and stimulate other nearby nerves (165, 291, 325).

1.6.3.2 Urolithiasis

The most important factors for the formation of uroliths are increased or decreased urinary pH and reduced water intake which can lead to urine supersaturation - the essential precursor to urolith formation (80). Other factors leading to urolith formation are many, and include the following: urinary tract infection with urease-producing bacteria (struvite stones), increased urine calcium levels causing formation of calcium oxalate stones, typically occurring in the form of increased bone resorption of calcium, increased gastrointestinal absorption, hypercalcemia, or increased renal loss of calcium (221, 257). One serious complication of urolithiasis is the obstruction of the urethra and progressive distension of the bladder, leading to eventual degeneration and necrosis of the bladder wall (367).

Bladder disease diagnosis and the role of histopathology

1.7.1 Bladder pain syndrome and feline idiopathic cystitis

Urinary tract infections are typically diagnosed by the presence of lower urinary tract signs and the presence of pyuria or bacteriuria (166). A positive urine culture result confirms the diagnosis, provides the offending organism and sensitivity testing advises the most appropriate antimicrobial for treatment (166). Histological changes in UTI can be broad, ranging from only hyperaemia and denudation of urothelium, to severe haemorrhagic, fibrinosuppurative inflammation (66). Other microscopic changes include vascular congestion, urothelial haemorrhage, submucosal haemorrhage, submucosal oedema, and ulcerated urothelium replaced by fibrin, neutrophils and bacteria (66, 69). More chronic cases may develop hyperplastic urothelium (sometimes with atypical or reactive urothelial cells), submucosal fibrosis, and infiltration of mononuclear inflammatory cells such as lymphocytes and plasma cells with the development of submucosal lymphoid follicles in some chronic cases (66, 69). Eventually, squamous metaplasia of the urothelium may occur (69).

Bladder pain syndrome and feline idiopathic cystitis are both painful, chronic, waxing and waning urinary bladder diseases of people and cats respectively. Bladder pain syndrome is currently

diagnosed by excluding all other diseases using imaging and cystoscopy, and sometimes bladder biopsy (129), and there are currently two proposed criteria for BPS diagnosis – the NIDDK criteria (National Institute for Diabetes and Digestive and Kidney Diseases) (143) and ESSIC criteria (International Society for the Study of BPS) (417). Bladder pain syndrome is diagnosed primarily based upon a patient’s clinical history and symptoms, and the use of questionnaires such as the O’Leary–Sant Interstitial Cystitis Symptom Index and Problem Index (ICSI) and (ICPI) and the Pain Urgency and Frequency score (PUF) (307). Other diagnostic tools for BPS include physical examination, urinalysis and urine culture to exclude infection, use of urinary biomarkers such as antiproliferative factor (APF) and Tamm–Horsfall protein (THP) (191, 328), potassium sensitivity test and intravesical anaesthetic challenge, urodynamic studies (controversial), cystoscopy to rule out bladder neoplasia, as well as bladder biopsy with staining for mast cells (129, 417). Bladder biopsy is not routinely used as a purely diagnostic tool as histological changes in BPS have been found to be non-specific, however biopsy has been found to be useful for determining disease severity, and more research into the role of mast cells is required (129).

In cats, techniques routinely used to definitively diagnose FIC are limited to abdominal radiography and ultrasound or exploratory laparotomy, as the small size of cats prevent the use of cystoscopy and cystoscopic biopsy (315). Diagnosis of FIC does involve compiling objective findings such as physical examination, urinalysis and urine culture, and presence of FLUTS in the absence of obvious neoplasia and bacteriuria, currently making FIC a diagnosis of exclusion. As discussed above, BPS primarily involves the symptom of pain upon filling of the bladder. As this and other symptoms are impossible to determine clinically in a cat, histopathology plays a more vital role in the diagnosis of FIC, prompting the theory that cystotomy and biopsy may be helpful in diagnosis of FIC patients.

Histopathology is not frequently used for diagnosis of lower urinary disease in dogs and cats; however, bladder biopsy may be useful in identifying deep bladder wall infections. Further, an upcoming area of research involves the use of in situ hybridisation (ISH), a laboratory technique in which a labelled probe identifies the presence of a target, such as a bacterial organism (450). There is potential to use ISH to identify bacteria within the bladder wall itself, which would be able to differentiate between bacteriuria and true colonisation of bladder tissue (300). This is particularly important for patients with chronic or recurrent urinary tract infections, as the precise location of the bacteria can provide information on the pathogenesis of the infection and inform further treatment options (300).

Bladder biopsy and histology does have a role as an adjunct tool to determine BPS disease severity 48

and to rule out underlying neoplastic disease. Histopathological findings in BPS have been variable, but generally include non-specific changes associated with chronic inflammation in addition to the possible and controversial theory of mast cell involvement (417), summarised in Table 1-3. Historically, BPS patients both with and without Hunner’s lesions have shown histological changes of variable pancystitis, muscularis fibrosis and mastocytosis (7, 183, 195), with some differences between the two subtypes (195, 267). Patients with HBPS have been shown to have focal to diffuse urothelial ulceration, severe lymphocytic inflammation throughout the entire bladder wall including perineural and perivascular infiltrates, and increased numbers of mast cells in both the muscularis and mucosa (127, 195), as well as muscularis fibrosis (183). On the other hand, patients with NHBPS have mostly normal urothelium and only mild mononuclear inflammation with the primary changes being mucosal disruption, sub-urothelial haemorrhages and more prominent fibrosis in the muscularis interstitium (7, 183, 195).

The histological lesions in FIC cases include damaged (or unaffected) urothelium, submucosal oedema, submucosal blood vessel dilation, submucosal haemorrhage, increased submucosal infiltration of mast cells, mild lymphoplasmacytic inflammation in the submucosa (neutrophilic infiltrates are uncommon), fibrosis and increased sensory nerve fibre density determined by special stains (50, 75). The histological findings of FIC are comparable with those described in NHBPS, however the level of mononuclear inflammation in FIC is often more severe than that seen in NHBPS (195, 216). FIC and urolithiasis cases were both found to have elevated bladder wall lymphocyte counts compared to normal bladders, and there was a higher proportion of T-lymphocytes than B- lymphocytes (378). Increased numbers of light-chain restricted B lymphocytes have recently been associated with HBPS (267) but are yet to be evaluated in feline bladder disease. Light-chain restriction suggests clonal expansion of B-lymphocytes, a process typically part of the local immune response but also occurring in autoimmune disorders (267). The increased number of T-cells compared to B-cells in FIC suggests that FIC is less like HBPS, and perhaps more similar to NHBPS. It is important to note that there are no publications on the normal number of resident lymphocytes in the bladder wall of dogs or cats, and very few in human medicine. In humans, one paper reports normal urothelium and submucosa containing up to 42 lymphocytes per field (size unspecified but probably 150x magnification)(78). Another early paper reported ‘small numbers’ of lymphocytes in the lamina propria of normal human bladder, particularly CD4+ lymphocytes (140). There are no equivalent reports for canine and feline urinary bladders.

Table 1-3: Histological changes of bladder pain syndrome and feline idiopathic cystitis.

Histological changes HBPS NHBPS FIC Urothelial denudation/ulceration (75, Yes, 6% Yes, 70% Yes 183, 255, 262) remaining remaining Submucosal oedema (50, 75, 380, Yes Yes Yes 402) Chronic inflammatory infiltrates (75, Yes, more severe Yes Yes, moderate 183, 216, 255, 262, 378, 380) than NH (195) Granulation tissue in the submucosa Yes Yes NR (170, 183, 402, 417)

Increased numbers of mast cells in Yes Yes, mild Yes* the submucosa (75, 183, 216, 290, 378, 402) Detrusor muscle fibrosis (75, 183, Yes Yes, more severe Yes 378, 417) than Hunner-type disease (195) Submucosal blood vessel dilation NR NR Yes (75), haemorrhage (50, 75), vascular reaction (50) Increased sub-urothelial proliferation NR NR Yes in the form of Brunn’s nests (216) *One study reported an increase only in degranulated submucosal mast cells. NR = not reported.

There is a relative paucity of literature on histological features of FIC compared to BPS. The author hypothesises a combination of two reasons for this; 1) The small penile urethral size of male cats prevents urethral cystoscopy and biopsy, therefore the only available method of bladder biopsy is laparotomy and cystotomy (32, 73), and 2) The waxing and waning nature of the disease prohibits many owners from proceeding to laparotomy and bladder biopsy. Thus, this research endeavours to conduct a detailed histological analysis of FIC cases, as part of a broader study on histological features of all canine and feline bladder diseases.

1.7.2 Urolithiasis

The histological changes observed in cases of urolithiasis in cats and dogs are secondary to the physical trauma of the uroliths and include urothelial hyperplasia, urothelial erosion and ulceration, submucosal haemorrhage, congestion and oedema as part of the inflammatory process, perivascular neutrophilic and/or lymphocytic inflammation, and the presence of mineral calculi in the bladder lumen (26, 180, 324, 367). Necrosis of the superficial bladder wall can occur in chronic cases due to 50

the local pressure exerted by the urolith on the bladder wall (26, 367). Urethral obstruction is a serious consequence of urolithiasis, causing increased pressure in the bladder which damages the urothelium, muscularis, and nerves in the bladder wall (26, 389). Increasing numbers of inflammatory cells transmigrate out of the vasculature in response to this insult, along with plasma proteins resulting in oedema (26). Sustained urethral obstruction and high bladder pressure leads to hypoxia and ischemia, with eventual degeneration and necrosis of the bladder wall (26, 367). Urothelial necrosis and submucosal fibrosis may be evident as soon as 10 hours after urethral obstruction (389). If there is intermittent or partial urethral obstruction, then more chronic inflammatory changes are likely to be observed, such as bladder wall oedema, lymphoplasmacytic cell infiltrates, and submucosal fibrosis (80). If the urethral obstruction is not relieved, catastrophic bladder rupture may occur (26).

It is important to note that urethral obstruction can be a feature of urolithiasis and FIC. If a cat has urethral obstruction, then differentiating FIC from urolithiasis becomes a clinical issue, with radiographs or ultrasound required to diagnose or rule out a urolith. In addition, if a case has no submitted clinical history, then the presence or absence of a urolith on postmortem examination would be required to differentiate urolithiasis and FIC.

Conclusions

In conclusion, this literature review consolidated the ideas behind this research and reinforced the aims for this project. It is important to note that the urinary bladder has a fairly consistent response to injury, thus there is substantial crossover in the histological features of various bladder diseases. Firstly, there is ample room for improvement in standardising veterinary pathologist diagnosis on histological samples, and logistic regression modelling is hypothesised to be a useful way to do this. Secondly, there are multiple gaps in the current knowledge surrounding demographic information on urinary bladder diseases in an Australian pet population, as well as histological features of these diseases, particularly FIC. Thirdly, there is immense capacity for comparative medicine and pathology between human bladder diseases and those of dogs and cats. Dogs and cats are spontaneously occurring disease models for urinary tract infection and bladder pain syndrome respectively. Despite the use of these animals as models of human diseases, many of the biomarkers evaluated in humans have not been investigated in dogs or cats.

From this literature review, five research chapters were developed:

1. Demographics analysis of urinary bladder diseases in dogs and cats, that had bladder tissue submitted to the University of Queensland’s pathology service over the last 22 years; 51

2. Conduct a detailed histological assessment on these cases, formulate a logistic regression model to analyse the histology data, then use the logistic regression results to predict probabilities of disease based on significant findings;

3. Test the predictive probabilities derived from the logistic regression model on a statistically significant number of professional veterinary pathologists to evaluate the effect of the predictive probabilities on inter-pathologist agreement;

4. Conduct a systematic review on biomarkers of bladder pain syndrome and feline idiopathic cystitis;

5. Based on the literature review and systematic review, conduct a laboratory evaluation of biomarkers tight junction protein 1 and E-cadherin in canine and feline urinary bladder tissues, using immunohistochemistry and polymerase chain reaction.

In summary, this thesis aims to improve inter-pathologist agreement with a novel application of logistic regression modelling, as well as expand the current knowledge on urinary bladder disease demographics, histological features and biomarker analysis in an Australian canine and feline population.

Chapter 2 Relationship of signalment data to disease outcome in canine and feline bladder pathology records in South East Queensland, Australia

Introduction

Disorders of the urinary bladder are common in dogs and cats. Bladder diseases account for 7% of cats hospitalised in veterinary clinics in the United States (317) while up to 4.6% of cats are affected by feline lower urinary tract signs (FLUTS) (171, 198, 258, 259). An estimated 14% of North American dogs are afflicted with a urinary tract infection at some point during their lifetime (240) and 1.5-3% of dogs admitted into veterinary care are diagnosed with urolithiasis (318). Urinary bladder neoplasms account for 1.5-2% of all canine neoplasms, while the proportionate morbidity (or prevalence) of neoplasia cases in cats is much lower (76). Despite the importance of bladder disease in dogs and cats, there is a paucity of prevalence data in an Australian context. Existing literature is primarily North American and European in origin, particularly for neoplastic and non- infectious conditions.

Many conditions affecting feline and canine urinary bladders are paralleled in humans, including urothelial carcinoma (UC) (201, 204, 320, 330), urinary tract infection/bacterial cystitis (244, 397), and non-infectious cystitis (interstitial cystitis/bladder pain syndrome) (216, 224). Urinary bladder disorders can be grouped into four broad categories – neoplastic, infectious, and inflammatory bladder conditions with or without uroliths (bladder stones) (74). Some diseases bridge multiple categories, for example, bacterial cystitis has both infectious and inflammatory components (242).

Typical animal signalment information such as breed, age, sex and neuter status has been suggested for the major categories of bladder disease in dogs and cats. Female dogs greater than seven to eight years, and female cats older than ten years are at higher risk for bacterial urinary tract infections (244, 397). In cats, non-infectious cystitis (feline interstitial cystitis; FIC) has been associated with young to middle age (98, 110, 231), and male cats have been overrepresented (49). Bladder neoplasia is more common in older male cats (446) and in older female neutered dogs with Scottish terriers overrepresented (204).

Importantly, signalment associations with bladder wall disease in an Australian population of cats and dogs have not been reported. This information would enhance awareness of veterinarians and pet owners to potential risk factors for bladder wall diseases in pets and will likely support the use of Australian dogs and cats as comparative models for human bladder diseases along with their international counterparts. The primary aim of this study was to determine the proportion of bladder 53

wall diseases in cats and dogs that had bladder tissue submitted to the University of Queensland Veterinary Laboratory Service (UQVLS) between January 1994 and March 2016. A secondary aim was to investigate the associations of species, breed, age, sex and neuter status as putative risk factors for bladder wall diseases in dogs and cats. Measures of risk in this study include both proportionate morbidity ratios obtained via proportionate morbidity calculations, and odds ratios obtained from multivariable logistic regression modelling.

Hypothesis and objectives

We hypothesise that the demographics of dogs and cats in this dataset will be comparable to those from the international published literature.

The objective of this chapter is to describe the demographics of animals within this dataset and explore the utility of logistic regression modelling in identifying associations between animal factors and the bladder disease diagnosis.

Material and Methods

2.3.1 Study Population

In a retrospective cross-sectional study, data was extracted from the UQVLS pathology archive for all dogs and cats with bladder tissue submitted between January 1994 to March 2016 (Figure 2-1). Bladder wall tissue was collected at The University of Queensland veterinary teaching hospitals (St Lucia and Gatton, Queensland Australia), other specialist and primary accession veterinary clinics in South East Queensland, and via the UQVLS anatomical pathology department. Tissue collection was undertaken for a variety of reasons, as outlined in Table 2-1. Twenty biopsy and 72 post-mortem records had no clinical history recorded in the UQVLS pathology database, so the reason for collection was unable to be ascertained.

Table 2-1: Reasons for bladder tissue collection.

Sampling reason Count Clinical biopsy Suspected neoplasia (bladder mass present) 31 Acute or chronic LUTS of unknown cause. 17 For culture, obtained during cystotomy for urolith 15 removal Suspected chronic bladder wall infection 9 Bladder rupture and biopsy taken during repair 5 Involved in a research trial 1

Post-mortem Targeted sampling due to a history of urinary tract signs 12 sampling of unknown cause Targeted sampling due to a history of upper urinary tract 11 disease Targeted sampling due to a history of known or 7 suspected urethral obstruction Targeted sampling due to a history of known or 3 suspected bladder neoplasia Euthanised as part of a research project 3 Post-mortem performed for non-bladder reasons, but 114 bladder tissue was collected as part of routine sampling Unable to ascertain 92 Total 320 LUTS – lower urinary tract signs.

2.3.2 Data collection and case definitions

Records of dogs and cats with bladder histology (submitted January 1994 to March 2016) were obtained via a search enquiry for canine and feline pathology reports containing at least one of the following terms: bladder, cystitis, transitional cell, urinary, or urothelial. The terms ‘bladder’ and ‘urinary’ alone were deemed insufficient as urinary tract infection is often termed ‘cystitis’ and bladder neoplasia may be recorded as transitional cell carcinoma or urothelial carcinoma without mention of the words ‘bladder’ or ‘urinary.’ The database search process is outlined in

Figure 2-1. From the search results, pathology records were excluded if they only mentioned gallbladder (no urinary bladder) or if the animal was less than six months of age in order to have only mature bladder wall tissue included. Depending on age, immature bladder wall can have umbilical artery or urachal remnants (283). These cases were excluded in order to standardise the dataset to mature bladder wall samples only. Records were also excluded when urinary bladder had been discussed in the gross section of the pathology report but no histology had been taken, and when no diagnosis could be made due to poor sample size or quality.

The following information was exported from the UQVLS database: case reference code; species (canine or feline); pathology report including history (if any submitted); date of pathology report; animal date of birth; animal breed; and animal sex and neuter status (e.g. female neutered = FN, male entire = ME).

The following parameters were determined from the extracted information: year of report; ‘has history’ (whether or not there was clinical history submitted with tissue sample); ‘urinary sign history’ (documentation of the presence or absence of clinical signs of lower urinary tract disease including frequent urination, straining to urinate or gross haematuria, important in determining routine or targeted sampling during post-mortem); ‘sampling method’ (whether the bladder sample was submitted as a clinical biopsy or obtained during post-mortem examination); ‘diagnosis’; ‘primary disease’ (the primary organ system causing the death or illness of the animal diagnosed at post- mortem or in clinic); animal age at diagnosis; sex and neuter status (if not recorded in database but ascertainable from pathology report description), and breed. Breeds were categorised based on the United Kingdom Kennel Club breed categories (395) and the Governing Council of the Cat Fancy (394). Dogs were grouped into gundog, hound, pastoral, terrier, toy dog, utility and working dog categories, while cats were grouped as Burmese, domestic, foreign cat, Persian cat, and semi-longhair. If the pathology report did not specify a patient’s breed it was recorded as unknown.

Diagnosis was classified as the primary diagnosis assigned to bladder tissue ranked in this order – neoplasia, urolithiasis (bladder stones), cystitis (inflammation in the bladder due to infectious or non- infectious causes), normal bladder wall or other diagnoses (with reported diagnoses including ‘haemorrhage only’, ‘oedema only’, autolysis, or peritonitis – inflammation of the bladder serosa due to other abdominal disease). Due to the histological nature of this study, many cases were lacking in urine culture results, making it impossible to categories cases as having or not having bacterial cystitis. Data was managed using Microsoft Excel (287).

Cases were defined based on the final diagnosis as assigned by the original pathologist in the pathology reports. ‘Cystitis’ technically means inflammation of the bladder wall, which may occur, with or without infectious organisms. There are no publications on normal numbers of resident lymphocytes in the canine or feline bladder wall. In humans, one paper reports normal urothelium and submucosa containing up to 42 lymphocytes per field (size unspecified but probably 150x magnification)(78). Another early paper reported small numbers of lymphocytes in the lamina propria of normal human bladder, particularly CD4+ lymphocytes (140).

When bladder stones were present, the histological diagnosis was often ‘cystitis’ as bladder wall inflammation can occur in these cases, however these records were separated into a different group, ‘urolithiasis’, as the inflammation was attributed to the physical trauma of the stone, in addition to potential bacterial infection. Clinical biopsy sample records were assigned to the urolithiasis category if bladder stones/uroliths were mentioned in the submitted history regarding the current clinical presentation or cause of death/reason for euthanasia.

Where neoplasia of any type had been identified in the bladder wall, records were placed into the ‘neoplasia’ category (Table 2-2), even if there was concurrent inflammation or urolithiasis, as neoplasia was considered to be the predominant pathologic process. Records were categorised as ‘normal’ if the bladder was reported as normal or was recorded as being sampled for histology but not mentioned in the diagnosis section (we assumed that absent comments on urinary bladder, in the presence of comments on other diseased tissues, meant that no abnormalities were found in the bladder tissue). All other diagnoses were grouped together in the ‘other’ category, including peritonitis, traumatic bladder rupture, and healed scar tissue (Table 2-3).

Table 2-2: Types of bladder neoplasia in the dataset

Species Type of neoplasia Count Dogs n = 66 Urothelial carcinoma 42 Soft tissue sarcoma* 7 Poorly differentiated carcinoma 6 Metastatic carcinoma 2 Metastatic adenocarcinoma 2 Fibropapilloma/fibroma 2 Not definitive 5 Cats n = 7 Urothelial carcinoma 4 Lymphosarcoma 1 Metastatic carcinoma suspected mesothelial 1 Not definitive 1 *including 3 leiomyoma and 2 leiomyosarcoma.

Table 2-3: Breakdown of pathology records where the diagnosis for the bladder tissue was assigned to the ‘other’ category (n=46).

Species Category Pathology report diagnosis Count (n) Canine No definitive Haemorrhage and/or oedema 8 n = 35 diagnosis Congestion 3 Necrosis, no identified cause 1 Not reported 1 Systemic disease Peritonitis 4 Vasculitis 2 Coagulopathy 2 Neospora caninum smooth muscle myositis 1 Systemic lupus erythematosus (fibrosis and 1 vasculitis) Trauma Trauma and bladder rupture 3 Preservation artefact Autolysis only 3 Iatrogenic Healing surgical wound 1 Trauma from catheterisation 1 Other Bladder polyp 1 Papillary epithelial hyperplasia 1 Congenital malformation 1 Smooth muscle hypertrophy 1 Feline No definitive Focal haemorrhage 2 n = 11 diagnosis Haemorrhage and oedema 1 Ulceration and fibrosis 1 Systemic disease Feline infectious peritonitis 2 Focal haemorrhage, suspected coagulopathy 2 Iatrogenic Changes secondary to previous cystostomy 1 Trauma Trauma and bladder rupture 2

2.3.3 Statistical analysis

2.3.3.1 Descriptive statistics and proportionate morbidities

Descriptive statistics (median, range and frequency as appropriate) were determined for bladder case records stratified by species. Associations between species and each of age, diagnosis, calendar year, sex, neuter status, breed, and sampling method were assessed using a Chi-squared test for categorical variables, Fisher’s exact test for normally distributed variables or Wilcoxon rank sum test to evaluate the equality of the medians for continuous variables.

A proportionate ‘bladder disease’ morbidity was used as the measure of risk for this study, and Poisson distribution was assumed. Proportionate morbidity (PM, expressed as a percentage) was deemed an appropriate measure of risk because the denominator in this study is a group of pathology records that had bladder tissue processed through the UQVLS, which do not necessarily represent the 59

general population of pet dogs and cats from which bladder diseases could potentially arise. Proportionate morbidity for each bladder disease outcome in dogs or cats (cystitis, neoplasia and urolithiasis) was calculated by dividing the total number of cases for that outcome by the total number of canine or feline cases in the dataset (cystitis, urolithiasis, neoplasia and normal combined). ‘Other’ diagnoses were not included in the PM calculation as they were often diagnoses of sample collection artefact (haemorrhage or oedema only), preservation artefact (autolysis), or diseases not directly involving the urinary bladder such as peritonitis or haemorrhage due to a systemic coagulopathy. The 95% confidence interval for proportionate morbidity was estimated using exact binomial distribution. For calculation of proportionate morbidity and age, animal age was separated into four categories – less than 4 years, 4 ≤ 8 years, 8 ≤ 11 and >11 years. These age groups were chosen based on the distribution of age throughout the dataset, based on 8 years being the mean age in both dogs and cats, as well as being the median age for dogs who make up the majority of the dataset. Proportionate morbidity ratio (PMR) was used to determine the presence and measure of associations between disease outcomes and animal factors or calendar year or sampling method, based on the proportionate morbidities. PMRs and their 95% confidence intervals were calculated based on the Poisson distribution.

2.3.3.2 Biopsy as a diagnostic tool

The Mantel-Haenszel procedure was used to quantify the association between the use of bladder biopsy as a diagnostic tool (as opposed to post-mortem sampling) and cystitis, urolithiasis or neoplasia diagnostic outcomes, while taking confounders into account. The Mantel-Haenszel procedure tests whether it is reasonable to assume that the odds ratios are equal (at the population level) across species, with the null hypothesis that all strata have the same odds ratio, which need not be equal to one. This test was implemented using the EpiR package (65) implemented within R studio (340). Cystitis, neoplasia and urolithiasis records were pooled and stratified by species and whether the bladder sample was obtained via a) clinical case investigation and biopsy or b) routine post- mortem examination and sampling. Crude odds ratios were calculated for each stratum and overall crude and Mantel-Haenszel odds ratios were calculated and compared using the procedures by Mantel and Haenszel as well as Bailar (21, 276). The Chi-squared homogeneity test was used to evaluate the significance of odds ratios between strata.

2.3.3.3 Multivariable simple logistic regression model

A multivariable logistic regression model with binomial link function was built to explore the

association between potential explanatory variables and each of the disease outcomes (cystitis, neoplasia or urolithiasis compared to normal), using methods similar to previous publications by JA (8). ‘Other’ diagnoses were not included in the logistic regression models as there was deemed to be too much variation in this category to make a meaningful contribution to the analysis. The dataset used for the logistic regression models comprised the following potential explanatory variables: a) species, b) age, c) sex, d) neuter status, e) breed, f) calendar year, and g) biopsy (‘yes’ for biopsy or ‘no’ for post-mortem sampling).

First, the association between each potential explanatory variable and each outcome was examined using univariable logistic regression models. The ‘best fit’ of the explanatory variable (continuous or categorical) was determined by graphical assessment of the relationship between the log odds of the outcome by categories of an explanatory variable (108). A likelihood ratio test P value ≤0.25 was used as a criterion for entry of an explanatory variable into a multivariable logistic regression model. Variables to be included in the final multivariable model were selected by a forward stepwise elimination procedure. All variables eliminated in the multivariable model were reintroduced in the final model one at a time in search for statistical significance at alpha level ≤0.05. Variables were retained in the multivariable model if the likelihood ratio test P values were ≤0.05 when establishing the model involving main effects (167) or when an excluded variable (likelihood ratio test P values were >0.05) included in the final multivariable model changed covariate estimates by 20% (and was declared as a confounder). Species and biopsy variables were forced into the final model once it was stable. Over-dispersion (or variance inflation factor; the ratio of the Pearson Chi-square goodness of fit statistic to its degrees of freedom) was estimated. A Chi-square test was then used to test for over- dispersion in the data at P ≤ 0.05. Over-dispersion was declared, and adjustment was required if Chi- square P-value was ≤0.05. The overall fit of the final model was evaluated by Hosmer–Lemeshow goodness-of-fit statistic (167). Further diagnostics, including the calculation of leverage and delta- betas, were used to identify any outliers or highly influential observations using the influence.ME package (302) in R (340).

Results

2.4.1 Data summary

The UQVLS database search yielded 320 records for dogs and cats that had bladder wall histology fitting the inclusion criteria (

Figure 2-1), summarised in Table 2-4. For canine records, there were 257 total submissions, with the 61

bladder diagnoses comprised of 85 cystitis cases (33%), 66 neoplasia cases (25%), 51 normal bladders (20%), 20 urolithiasis cases (8%) and 35 other diagnoses (14%). There were 63 feline submissions comprised of 24 cystitis cases (38%), 17 normal bladders (27%), 7 neoplasia cases (11%), 4 urolithiasis cases (6%) and 11 other diagnoses (18%) related to the urinary bladder.

There were 98 biopsy (86 dogs and 12 cats) and 222 post-mortem records (171 dogs and 51 cats). Dogs accounted for 88% of the biopsy records and 77% of the post-mortem records (while making up 80% of the dataset).

Table 2-4: Count data for canine and feline records with bladder tissue submitted to UQVLS January 1994 to March 2016.

Variable Canine (n = 257) Feline (n = 63) P value Age (months) 0.20 Mean (SD) 93.90 (51) 86.78 (70) Median (Q1-Q3) 96.0 (60.0-132.8) 60.0 (25-141.2)

Diagnosis 0.14 Cystitis 85 (33) 24 (38) Neoplasia 66 (25) 7 (11) Urolithiasis 20 (8) 4 (6) Other 35 (14) 11 (18) Normal 51 (20) 17 (27)

Sex 0.17 Male 115 (45) 36 (57) Female 124 (48) 25 (40) Unknown 18 (7) 2 (3)

Neuter status 0.05 Neutered 120 43 Entire 17 1 Unknown 120 19

Calendar year 0.75 1994-1997 33 (13) 8 (13) 1998-2001 34 (13) 10 (16) 2002-2005 35 (14) 5 (8) 2006-2009 85 (33) 19 (30) 2010-2013 38 (15) 10 (16) 2014-2016 32 (12) 11 (17)

Breed N/A Domestic 38 (60) Foreign 6 (10) Burmese 3 (5) Persian 3 (5) Semi-Longhair 2 (3) Pastoral 49 (19) Toy dog 29 (11) Utility dog 28 (11) Terrier 27 (11) Hound 26 (10) 63

Variable Canine (n = 257) Feline (n = 63) P value Gundog 24 (9) Working dog 20 (8) Unknown 54 (21) 11 (17)

Biopsy 0.04 Yes 86 (33) 12 (19) No 171 (67) 51 (81) Q1-Q3 - quartile one to quartile three; SD standard deviation. N/A not applicable, breeds had to be evaluated separately for dogs and for cats due to different categories.

2.4.2 Signalment descriptive data

For 50 (16%) records the animal age could be ascertained; 37 dogs (14%) and 13 cats (21%). The mean and quartiles for age were not significantly different between dogs and cats. Dogs and cats had a mean age of just under 8 years (93.9 months 86.78 months respectively). The mean age for dogs was 8 years, while it was 5 years for cats (Table 2-4). Breed was recorded for 79% (203/257) of dogs and 83% (52/63) of cats. The most common breed groups were pastoral breed dogs (19%) and domestic cats (60%).

Of the 257 dogs in the dataset, 115 (45%) were male, 124 (48%) were female and 18 (7%) did not have sex recorded. For cats, there were 36 (57%) male, 25 (40%) female and two (3%) were not recorded. The proportions of each sex were not statistically different between dogs and cats (P = 0.17). Of the 181 animals (57%) with neuter status recorded, there were 137 dogs (53%) and 44 (70%) cats, with canine cases consisting of 49 male neutered (36% of all dogs that had sex and neuter status recorded), 9 male entire (6%), 71 female neutered (52%) and 8 female entire animals (6%). Feline cases consisted of 25 male neutered (57% of all cats that had sex and neuter status recorded), 1 male entire (2%), 18 female neutered (41%) and no female entire animals.

Seventy-one percent (n = 228) of all records had some form of clinical history submitted with the tissue or body; 185 (72%) dogs and 43 (68%) cats. Complete signalment information comprising animal age, sex, neuter status and breed was recorded in only 130 (41%) animals - 99 (39%) dogs and 31 (49%) cats. All these 130 had clinical historical information recorded as well. Of the 130 records with full signalment information, 94 (72%) of these were submitted from 2008-2016. Forty- three percent (139/320) of all records came from this period.

Eighty records (25%, 4 biopsy and 77 post-mortem) had a non-bladder primary disease recorded in the pathology report belonging to one of the following categories: cardiac (n = 4), endocrine (n = 3),

gastrointestinal (n = 11), haematopoietic (n = 1), hepatic (n = 7), infectious (n = 2), neoplastic (n = 6), neurological (n = 14), renal (n = 6), reproductive (n = 3), respiratory (n = 4), skeletal (n = 1), systemic (n = 3), traumatic (n = 4) and undetermined (found dead/collapsed, n = 11).

The proportion of cases submitted in each four-year period remained constant in both cats (8-17%) and dogs (13-15%), apart from a spike in case numbers in 2006-2009. The cases submitted during the 2006-2009 period were primarily normal bladders (34% versus 15% normal throughout all other year groups), and the ratio of biopsy to post-mortem samples was similar to the entire dataset (65% post- mortem in this period compared to 68% in the overall population). One third of all cases were submitted in the 2006-2009 period (33% of dogs and 30% of cats).

2.4.3 Biopsy as a diagnostic tool

One third of canine samples (33%) and 19% of feline samples were obtained via clinical biopsy (P = 0.04, Table 2-4). Of dogs that had bladder tissue submitted to the UQVLS during the study period, the odds of undergoing a biopsy compared to necropsy, irrespective of the diagnostic outcome was 11.64 (95% CI 3.45-38.83). For cats, the odds of having had a biopsy compared to necropsy were 2.22 (95% CI 0.42-11.84). For canine and feline cases combined, the Mantel-Haenszel adjusted odds of using a biopsy to yield a diagnostic result was 7.52 (95% CI 2.92-19.40). See Appendix 2: Crude and adjusted odds ratio of biopsy sampling method stratified by disease cases and non-cases.

2.4.4 Proportionate morbidity and proportionate morbidity ratios

Dogs and cats showed a similar pattern of proportionate morbidity for the three diagnostic outcomes and the normal category. Comparing feline to canine records, the only statistically significant differences were a decreased proportion of cystitis cases in the 2006-2009 period (P = 0.02, Table 2-6), and many significant results in the neoplasia outcome (Table 2-5). Compared to dogs, cats had a proportionate morbidity ratio of 0.45 (0.21-0.99, P = 0.05). For dogs and cats combined, there was a marked association between age and neoplasia, with increasing proportionate morbidity ratios (PMRs) for neoplasia in each progressive age category (Table 2-5). The reference group was less than 4 years of age. The 4 ≥ 8-year-old group had a PMR of 8.59 (95% CI 1.97-37.35, P < 0.01), the 8 ≥ 11 year group had a PMR of 9.25 (95% CI 2.15-39.88, P < 0.01) and animals greater than 11 years had a PMR of 17.62 (95% CI 4.20-73.97, P < 0.01).

Cases of bladder neoplasia were almost three times more likely (PMR 2.7, 95% CI 1.70-4.27) to have had the tissue sample obtained via biopsy than via post-mortem examination (P < 0.01). Sex was

documented in 61 neoplasia records (92%), and in the PMR calculations was not statistically significantly associated with neoplasia (P = 0.25). Of dogs with neoplasia outcome and sex recorded (n=61), the ratio of males compared to females was the same for all neoplasms (57% male) as well as when evaluating UC alone (56% male). Cats with neoplasia and sex recorded (n=7) had a proportion of 28% male across all neoplasms and 25% male for UC alone.

Table 2-5: Total number of records to number of cases, proportionate morbidity (PM) and proportionate morbidity ratios (PMR) of canine and feline neoplasia records as confirmed by histology findings.

Variable Total Cases PM (95%CI) Coef (SE) PMR (95% CI) P value (Neoplasia) Species Canine 222 66 0.30 (0.24-0.36) -1.21 (0.12) Reference Feline 52 7 0.13 (0.06-0.26) -0.79 (0.40) 0.45 (0.21-0.99) 0.05

Age < 4 years 73 2 0.04 (0.00-0.13) -3.60 (0.71) Reference 4≤ 8 years 71 16 0.28 (0.17-0.42) 2.15 (0.75) 8.59 (1.97-37.35) <0.01 8≤11 years 58 18 0.35 (0.22-0.49) 2.22 (0.74) 9.25 (2.15-39.88) <0.01 > 11 years 68 28 0.49 (0.36-0.63) 2.87 (0.73) 17.62 (4.20-73.97) <0.01 Unknown 50 9 0.21 (0.10-0.39) 1.88 (0.78) 6.57 (1.42-30.41) 0.02

Year 1994-1997 31 7 0.20 (0.8-0.37) -1.49 (0.38) Reference 1998-2001 37 9 0.24 (0.12-0.41) 0.07 (0.50) 1.08 (0.40-2.89) 0.88 2002-2005 38 16 0.42 (0.26-0.59) 0.62 (0.45) 1.86 (0.77-4.53) 0.17 2006-2009 92 27 0.29 (0.2-0.40) 0.26 (0.42) 1.30 (0.57-2.98) 0.54 2010-2013 40 10 0.25 (0.13-0.41) 0.10 (0.49) 1.11 (0.42-2.91) 0.84 2014-2016 36 4 0.11 (0.03-0.26) -0.71 (0.63) 0.49 (0.14-1.68) 0.26

Sex Female 127 29 0.23 (0.16-0.31) -1.47 (0.19) Reference Male 129 39 0.3 (0.22-0.39) 0.28 (0.25) 1.32 (0.82-2.14) 0.25 Unknown 18 5 0.28 (0.10-0.53) 0.20 (0.48) 1.22 (0.47-3.14) 0.69

Biopsy No 189 33 0.17 (0.12-0.24) -1.75 (0.17) Reference Yes 85 40 0.47 (0.36-0.58) 0.99 (0.24) 2.70 (1.70-4.27) <0.01 * Total = cystitis, neoplasia, urolithiasis and normal (other diagnoses excluded).

Table 2-6: Total number of records to number of cases, proportionate morbidity (PM) and proportionate morbidity ratios (PMR) of canine and feline cystitis records as confirmed by histology findings.

Variable Total* Cases PM (95% CI) Coef (SE) PMR (95% CI) P value (cystitis) Species Canine 222 85 0.38 (0.32-0.45) -0.98 (0.1) Reference Feline 52 24 0.46 (0.32-0.61) 0.19 (0.23) 1.21 (0.77-1.90) 0.42

Age < 4 years 73 34 0.56 (0.42-0.70) -0.86 (0.18) Reference 4≤ 8 years 71 16 0.33 (0.21-0.47) -0.46 (0.29) 0.63 (0.36-1.12) 0.11 8≤ 11 years 58 15 0.29 (0.17-0.43) -0.50 (0.31) 0.61 (0.33-1.13) 0.12 > 11 years 68 23 0.40 (0.28-0.54) -0.23 (0.27) 0.80 (0.46-1.37) 0.41 Unknown 50 21 0.46 (0.33-0.65) -0.01 (0.28) 0.99 (0.57-1.72) 0.97

Year 1994-1997 31 18 0.51 (0.34-0.69) -0.54 (0.24) Reference 1998-2001 37 16 0.43 (0.27-0.60) -0.29 (0.34) 0.74 (0.38-0.92) 0.39 2002-2005 38 18 0.47 (0.31-0.64) -0.2 (0.33) 0.82 (0.42-1.57) 0.54 2006-2009 92 26 0.28 (0.19-0.39) -0.72 (0.31) 0.49 (0.27-0.89) 0.02 2010-2013 40 17 0.43 (0.27-0.59) -0.31 (0.34) 0.73 (0.38-1.42) 0.36 2014-2016 36 14 0.39 (0.23-0.57) -0.40 (0.36) 0.67 (0.33-1.35) 0.26

Sex Female 127 53 0.42 (0.33-0.51) -0.88 (0.14) Reference Male 129 49 0.38 (0.30-0.47) -0.09 (0.20) 0.91 (0.62-1.34) 0.64 Unknown 18 7 0.39 (0.17-0.64) -0.07 (0.40) 0.93 (0.42-2.05) 0.86

Biopsy No 189 77 0.41 (0.34-0.48) -0.90 (0.11) Reference Yes 85 32 0.38 (0.27-0.49) -0.08 (0.21) 0.92 (0.61-1.40) 0.71 * Total = cystitis, neoplasia, urolithiasis and normal (‘other’ diagnoses excluded).

Variable Total* Cases PM (95%CI) Coef (SE) PMR (95% CI) P value (Urolithiasis) Species Canine 222 20 0.09 (0.06-0.14) -2.41 (0.22) Reference Feline 52 4 0.08 (0.02-0.19) -0.15 (0.55) 0.85 (0.29-2.50) 0.77

Age < 4 years 73 6 0.09 (0.03-0.20) -2.68 (0.45) Reference 4≤ 8 years 71 6 0.12 (0.05-0.24) 0.02 (0.58) 1.03 (0.33-3.19) 0.96 >8 years 126 6 0.05 (0.02-0.11) 0.54 (0.58) 0.58 (0.19-1.80) 0.34 Unknown 50 6 0.14 (0.05-0.28) 0.38 (0.58) 1.46 (0.47-4.53) 0.51

Year 1994-1997 31 4 0.23 (0.10-0.40) -2.05 (0.5) Reference 1998-2001 37 6 0.16 (0.06-0.32) 0.23 (0.65) 1.26 (0.35-4.45) 0.72 2002-2005 38 3 0.08 (0.02-0.21) -0.49 (0.76) 0.61 (0.14-2.73) 0.52 2006-2009 92 4 0.04 (0.01-0.11) -1.09 (0.71) 0.34 (0.08-1.35) 0.12 2010-2013 40 3 0.08 (0.02-0.2) -0.54 (0.76) 0.58 (0.13-2.60) 0.48 2014-2016 36 4 0.11 (0.03-0.26) -0.15 (0.71) 0.86 (0.22-3.44) 0.83

Sex Female 127 9 0.07 (0.03-0.13) -2.65 (0.33) Reference Male 129 13 0.10 (0.05-0.17) 0.35 (0.43) 1.42 (0.61-3.33) 0.41 Unknown 18 2 0.11 (0.01-0.35) 0.45 (0.78) 1.57 (0.34-7.26) 0.57

Biopsy No 189 16 0.08 (0.05-0.13) -2.47 (0.25) Reference Yes 85 8 0.09 (0.04-0.18) 0.11 (0.43) 1.11 (0.48-2.60) 0.81 * Total = cystitis, neoplasia, urolithiasis and normal (‘other’ diagnoses excluded).

2.4.5 Multivariable models

Age, sex and undergoing bladder wall biopsy influenced the odds of being diagnosed with neoplasia. After controlling for the effect of other variables in the model, there was a highly significant trend for animals to have an increased risk of neoplasia as age increased, compared to the less than four years age group (P < 0.01, Table 2-8). Dogs and cats combined had a 6.8 times higher risk of neoplasia at 4-8 years of age (OR 6.84, 95% CI 1.76-45.38, P = 0.01), 13 times the risk at 8-11 years (OR 13.04, 95% CI 3.38-86.51, P < 0.01), and 20 times the risk of neoplasia when over 11 years of age (OR 20.63, 95% CI 5.58-134.36, P < 0.01). Male animals were almost twice as likely to be diagnosed with neoplasia compared to females (OR 1.84, 95% CI 1.00-3.46, P = 0.05). In addition, dogs or cats having a biopsy procedure were three times more likely to be diagnosed with neoplasia than those undergoing post-mortem examination (OR 3.29, 95% CI 1.76-6.28, P < 0.01).

Neuter status, age and calendar year all influenced the odds of cystitis diagnosis. After controlling for the effect of other variables in the model (Table 2-8), neutered canine and feline cases were 78% less likely (OR 0.22 95% CI 0.07-1.09, P = 0.01) to be diagnosed with cystitis compared with entire animals. Animals between the ages of 4 years and 8 years had a reduced risk of cystitis (OR 0.25, 95% CI 0.08-0.71, P < 0.01) compared to all other age categories combined. There was an important temporal association as well; the odds of cystitis in the 2006-2009 period were reduced by 50% (OR 0.51, 95% CI 0.23-1.09, P = 0.09) compared with all other time periods combined. No significant associations were found for cases with a diagnosis of urolithiasis in the multivariable models. There was no significant association between disease diagnosis and any of the breed groups.

Table 2-8: Coefficients (standard errors) and odds ratios (95% CI) of final multivariable logistic regression model fitted on records diagnosed with cystitis (n = 109), neoplasia (n = 73) and urolithiasis (n = 24) for canine and feline pathology records containing bladder histology. Cystitis Neoplasia Urolithiasis Variable Coef (SE) OR (95% CI) P Coef (SE) OR (95% CI) P Coef (SE) OR (95% CI) P Intercept 0.07 < 0.01 0.07 Species 0.47€ <0.01€ 0.69€ Canine Reference - -- Reference - -- Reference - -- Feline 0.34 (0.39) 1.40 (0.64-3.01) 0.39 -0.77 (0.46) 0.46 (0.17-1.10) 0.10 -0.20 (0.58) 0.82 (0.23-2.28) 0.72 Biopsy 0.28€ 0.08€ 0.77€ No Reference - -- Reference - -- Reference - -- Yes 0.50 (0.42) 1.64 (0.71-3.78) 0.24 1.19 (0.32) 3.29 (1.76-6.25) <0.01 0.11 (0.45) 1.12 (0.44-2.67) 0.80 Sex NA 0.05 NA Female NA NA NA Reference - -- NA NA NA Male NA NA 0.61 (0.31) 1.84 (1.00-3.46) 0.05 NA NA NA Neutered 0.01€ NA NA No Reference - -- NA NA NA NA NA NA Yes -1.51 (0.57) 0.22 (0.07-1.09) 0.01 NA NA NA NA NA NA Age <0.01 <0.01€ NA (years, y) € <4y Reference - -- Reference - -- NA NA NA 4 ≤ 8y -1.37 (0.54) 0.25 (0.08-0.71) 0.01 1.9 (0.79) 6.84 (1.76-45.38) 0.01 NA NA NA 8 ≤ 11 y -0.44 (0.53) 0.64 (0.22-1.80) 0.41 2.57 (0.79) 13.04 (3.38-86.51) <0.01 NA NA NA >11y 0.08 (0.45) 1.09 (0.45-2.65) 0.85 3.03 (0.77) 20.63 (5.58-134.36) <0.01 NA NA NA Calendar 0.07€ NA NA year period Other Reference - -- NA NA NA NA NA NA periods 2006- -0.67 (0.40) 0.51 (0.23-1.09) 0.09 NA NA NA NA NA NA 2009 Coef = coefficient; CI = confidence interval; OR = odds ratio; SE = standard error. € P values derived from likelihood ratio test comparing the reduced model with saturated model for each variable. The P values shows the statistical significance of the explanatory variable to overall model fit.

Discussion

This dataset was a difficult one to define, hence the statistical techniques applied were selected with great care. Many animal disease studies in the veterinary field evaluate the proportions of diseases in large or university veterinary hospital submissions (98, 202, 244, 397), whereas the population in this dataset comprised animals that were presented to veterinarians (UQ teaching hospital and general practices in South-East Queensland, Australia) that then underwent bladder biopsy, or were sent for post-mortem at UQVLS and had bladder tissue sampled during that procedure. As mentioned above, the histological nature of this study and relative lack of (positive or negative) urine culture results made it impossible to categorise cases as having or not having bacterial cystitis. When comparing ‘cystitis’ cases to existing literature, we are forced to look to published work on canine bacterial cystitis in dogs, and feline bacterial cystitis as well as the non-infectious feline idiopathic cystitis in cats, as both these diseases are likely to be represented in our ‘cystitis’ disease category. Multivariable modelling was performed to show the effect of each variable on the disease outcome alone and in combination with other significant variables. This process highlighted some associations that were not evident when evaluating the PMs and PMRs.

Neoplasia was significantly more common in dogs than cats in this dataset, with dogs having more than double the risk of the overall dataset population for bladder neoplasia. This is consistent with published reports of bladder neoplasia being much more common in dogs than in cats (203). It is important to note that the neoplasia category in our analysis dataset includes all types of bladder neoplasia, even though the majority are UC (Table 2-2). Population level literature on bladder neoplasia typically includes only urothelial carcinoma (UC) as this is the most common neoplasm of the urinary bladder in dogs and cats (203). We found an increase in the risk of neoplasia with increasing age, with up to 20 times higher odds in dogs and cats greater than 11 years of age. These results are consistent with the literature reports of older animals being at increased risk for bladder neoplasia (446).

In this dataset, bladder neoplasia was more common in males than in females. This finding was only present in dogs, which skewed the overall population, however sex was not significantly associated with neoplasia (P = 0.25). This finding was unique to the logistic regression model, highlighting the utility of this method in taking into account any interacting or confounding variables. Finding more males than females in the bladder neoplasia group is in contrast to the literature on UC in dogs, where females have up to three times higher risk than males (94, 202), however the female to male ratio does level out in the high risk breeds (Shetland sheepdogs, West Highland white terriers and beagles) 71

(203). The canine male to female ratio was comparable when evaluating all neoplasms together compared to evaluating UC alone. Our finding of higher risk for neoplasia in male dogs could be attributed to a different sex distribution of urinary bladder neoplasms in Australian dogs, or perhaps differences in the disease process or socioeconomic factors leading to male dogs getting sampled more than females and skewing our dataset. For cats, there were more females than males in the neoplasia group. Bladder cancer is rare in cats thus literature is sparse, and there have been conflicting reports of sex association for this disease in cats (64). The small sample size for cats impedes evaluation of sex association for feline bladder neoplasia in this study.

The literature reports neutered dogs being at higher risk of bladder neoplasia (203), however our PMRs and multivariable modelling showed no association between neoplasia and neuter status. Scottish terriers (terrier category), Shetland sheepdogs (pastoral category) and beagles (hound category) are known to have an increased risk for bladder neoplasia (203). We saw a trend towards increased risk for neoplasia in pastoral and terrier breeds, however these changes were not statistically significant.

For the cystitis outcome, it is important to note that our definition of cystitis (histological evidence of inflammation in the bladder wall in the form of inflammatory cell infiltrate in addition to haemorrhage and/or oedema) is different from the commonly used clinical definition of cystitis (used in cases with lower urinary tract clinical signs and often with positive bacterial culture indicating urinary tract infection). The cystitis disease category in this study could not distinguish between bladder inflammation due to bacterial or other infection, or idiopathic inflammation (feline idiopathic cystitis), due to absence of bacterial culture results for most cases. Another important point is that positive bacterial culture (bacteriuria) does not necessarily mean the patient has colonisation or infection of the urinary tract, which further complicates the issue (339, 428).

Our analysis found no difference between cystitis and species or sex, whilst the literature reports female dogs to be at increased risk for bacteriuria (155, 244, 397). Our results may be attributed to the unavoidable combining of infectious and non-infectious cystitis, however, our multivariable model showed that even when species was accounted for there was no difference in the risk of cystitis between males and females. Neutered animals were found to have a significantly decreased risk for cystitis which is consistent with some published work on canine bacteriuria (155), however there is scant published evidence on neuter status as a risk factor for canine urinary tract infections.

For the cystitis outcome, there was a decreased risk for cystitis in animals between four and eight

years of age, which is consistent with published material. Increased risk for younger and older age groups makes sense biologically as idiopathic cystitis is more common in younger cats (50, 110), and urinary tract infection risk is higher in dogs and cats greater than seven to ten years of age (231, 241). There was a decrease in the number of cystitis cases submitted during 2006-2009 (P = 0.02) which likely attributed to the increased proportion of normal bladder tissue sampled during this period. There was a trend towards a higher risk for cystitis in pastoral, hound and toy breeds, however this was not statistically significant.

We found no significant associations with the urolithiasis outcome in the PMs, PMRs or the multivariable model, most likely due to the limited case numbers.

The Mantel-Haenszel adjusted odds ratio indicates that the odds of having a bladder wall disease are 7.5 times higher if a biopsy is taken compared to a necropsy sample. This makes sense from a clinical standpoint as a biopsy procedure is typically performed when a disease is suspected. Biopsy was very strongly associated with neoplasia, as patients suspected to have a bladder neoplasm are more likely to have a biopsy taken to confirm the diagnosis.

We observed an increase in the total number of bladder tissue samples submitted during 2006-2009, with one third of the entire dataset obtained during this four-year period. The total number of histopathology submissions was investigated subsequent to this finding. Between and including 1994- 2013 there were approximately 30 000 histopathology submissions (including diagnostics, teaching and research). 2006-2009 had 25% of these, compared to 21–22% in each four-year period between 1994 and 2005, and only 11% during 2010–2013 (likely due to the vet school campus moving in 2010). The total increase in all pathology submissions during 2006-2009 could explain the finding of more bladder submissions during this period, however a lower number of total submissions was not identified during the 2010-2013 period where we saw less bladder submissions. Hypotheses for the 2006-2009 total pathology submission increase include increased marketing and promotion to local clinics for post-mortems, and changes in staff members leading to more complete routing sampling.

For animals with concurrent disease, the bladder findings were either incidental or not the main cause for the animal’s death. No significant associations with bladder disease were detected for any breed group. Our sample size was limited for cats, preventing adequate analysis of breed associations.

Substandard information on pathology submission forms is a frequent problem in both human and veterinary pathology and was observed in this study, with complete signalment information comprising animal age, sex, neuter status and breed recorded in less than half of all records. The 73

majority of these were from 2008-2016, suggesting an improvement in submission and/or data recording procedures in more recent years. In human medicine, higher quality clinical information on pathology requisition forms is associated with decreased turnaround time (P < 0.001) and improved outcomes for skin biopsies (298, 351). In veterinary pathology, up to 88% of biopsy submission forms (n = 510) were found to be deficient in at least one key area (43). The absence of clinical history in almost a third of records means that cases of urolithiasis (bladder inflammation on histology) may have been inadvertently categorised as cystitis as the crucial clinical history and diagnostic imaging results were missing. The clinical history may not have been submitted with the case or may have not been entered into the electronic database during processing.

The uniqueness of this dataset means that we cannot make direct comparisons to other population level studies or the general Queensland or Australian dog and cat population, however the information presented may be useful for client and veterinarian awareness. The retrospective nature of this study does have some inherent elements of recall bias and information bias that could affect external validity of this study. It is vital that patient signalment information is recorded and entered into data storage systems as it should be available for both pathological interpretation and for future research projects involving animal diseases. The absence of this information likely impaired the adjustment of records by breed, and therefore evaluation of breed as a risk factor for cystitis, neoplasia or urolithiasis. Missing data in our dataset was likely to have had a bidirectional impact on the proportionate morbidities – some may be overestimated while others underestimated. A further limitation is the lack of standardisation of pathology reporting and the potential for missed records due to misspelling, or the disease diagnosis containing slightly different terms to the ones used in our search, although we endeavoured to capture these variations.

Conclusion

In conclusion, this study analyses a unique dataset of canine and feline pathology records containing bladder histology. The main findings were a higher risk of bladder neoplasia in dogs compared to cats, increasing risk for bladder neoplasia with age, increased associations between biopsy procedure and being diagnosed with bladder disease, and decreased risk for cystitis in neutered animals. Whilst neoplasia and urolithiasis are consistently applied diagnoses, there is some ambiguity in the use of the term cystitis. The case definition for cystitis in this chapter being an infiltration of inflammatory cells along with haemorrhage and/or oedema, the absence of urolithiasis, with or without positive bacterial culture. This study highlights the utility of linear regression modelling for analysing animal disease data and presents the first summary of proportionate morbidity of bladder diseases in dogs 74

and cats in an Australian context. Further studies are needed to determine prevalence and risk factors for cystitis, neoplasia and urolithiasis in the Australian dog and cat population.

Chapter 3 A novel role for logistic regression modelling – predicting diagnosis of canine and feline urinary bladder disease based on histological features

Introduction

Anatomic pathologists are highly trained in the use of light microscopy for analysis of tissue specimens at 20 to 600 times magnification to evaluate the cellular components and characteristics of a disease or physiological process. Pathologists are able to make a diagnosis through the recognition of one or more histological findings, ideally combined with the patient’s clinical history (341). However, it is well known that pathologists are influenced by cognitive bias (6, 128) – a facet of the discipline that has both benefits and limitations (156). Examples of cognitive bias include (a) confirmation bias – looking for evidence of a favoured hypothesis; (b) diagnostic drift – slight variations in scoring values over the course of a study; as well as (c) tunnel vision - also known as anchoring, the tendency to rely too heavily on the first information presented; (d) avoidance of extreme scoring ranges; as well as (e) availability bias – referring to what most easily comes to mind, which means less familiar diseases are forgotten (6, 272). When a pathologist looks at a slide microscopically, they incorporate the observed changes with their knowledge obtained via extensive training and experience to formulate an ordered differential list (24). The diagnosis with the highest probability is then reported as the final diagnosis. In summary, pathologists incorporate a wide range of knowledge and experience to provide the most likely diagnosis based on the microscopic changes they observe.

Manual histopathology is generally qualitative or semi-quantitative as opposed to quantitative, and at times has low agreement between individual pathologists (93, 418) which has prompted recent advances in computer learning and computer-based image analysis (6). Mathematical models for the pathologist diagnostic processes do exist, however very few have been validated (91). Mathematical models are the foundation of machine learning and artificial intelligence, an interesting area of medical predictive modelling (63, 266). Artificial intelligence and deep learning software is steadily being introduced to the human pathology field as a tool to help automate mundane tasks, streamline sample accession, and score basic neoplastic lesions, with any questionable slides then forwarded to the pathologists for evaluation (63, 266). Logistic regression is a statistical method used to explain the relationship between data categories and outcomes (108). Logistic regression models were found to perform similarly to machine learning models in a systematic review of studies comparing the two methods for clinical prediction modelling (79). The output of logistic regression is an odds ratio which reflects the likelihood that an event/disease occurs as depending on dichotomous explanatory 76

variables (for example, what is the odds of a patient developing diabetes if they are overweight or not overweight) (379). In a stepwise procedure, logistic regression is used to build the best possible model that explains the relationship between the outcome/s of interest and the explanatory variables in the dataset (108, 379). To do this, the model uses logit (log-odds) transformation to predict the probability of the outcome/s occurring in every possible combination of explanatory variables, thus is a sort of simulation (108, 379). Logistic regression modelling was chosen for this study because it allows prediction of outcome probability from a combination of continuous and discrete independent variables (35).

Logistic regression multivariate models have potential to predict disease risk in both the human and veterinary medical fields by providing an objective probability of the disease occurring given the combination of variables (158). Predictive models have been used to estimate disease probability in some human medical fields such as cardiac exercise testing, and to estimate the risk of cardiac disease given multiple test results and patient factors (35, 293, 451). Lung cancer is another area of human medicine that has utilised predictive models based on logistic regression (100). In the veterinary field, predictive logistic regression models have been used in genetics (27), ultrasonography (296, 297), surgery and surgical prognosis development (150, 344) as well as predicting disease outbreaks (369), however their use in veterinary histopathology has been limited thus far to occasional studies in wildlife (144). Predictive logistic regression models have potential to assist decision making in veterinary histopathological diagnoses and improve pathologist and veterinarian awareness of the risk of a disease occurring in an individual patient.

Disorders of the urinary bladder are common in dogs and cats; therefore, bladder tissue is relatively easy to acquire. Bladder diseases account for 7% of cats hospitalised in veterinary clinics in the United States (317). An estimated 14% of North American dogs are affected by a urinary tract infection at some point during their lifetime (240) and 1.5-3% of dogs admitted into veterinary care are diagnosed with urolithiasis (bladder stones) (318). Urinary bladder neoplasms account for 1.5-2% of all canine neoplasms, while the prevalence of bladder neoplasia in cats is much lower (76). Urinary bladder disorders can be grouped into four broad categories – neoplastic, infectious, and inflammatory bladder conditions with or without uroliths (bladder stones) (74). Some diseases bridge multiple categories, for example, bacterial cystitis has both infectious and inflammatory components (242). The overall aim of this study was to explore the role of logistic regression modelling in creating a predictive model for histological diagnosis, using Australian canine and feline urinary bladder specimens as a pilot study tissue.

Hypothesis and objectives

We hypothesise that there will be some histological findings that are unique to certain urinary bladder diseases, and that logistic regression modelling will enable the formation of a predictive model for histological diagnosis of urinary bladder disease in cats and dogs.

The objectives of this chapter are to:

1. Evaluate the microscopic changes of urinary bladder diseases in a retrospective collection of canine and feline urinary bladder tissue.

2. Use logistic regression modelling to identify any associations between the microscopic changes and disease diagnosis.

3. Use any significant associations from Aim 2 to create a predictive model to assist decision making in the diagnosis of canine and feline urinary bladder conditions.

Methods

3.3.1 Study population and data collection

Records of dogs and cats with bladder histology submitted to the University of Queensland Veterinary Laboratory Service (UQVLS) between January 1994 and March 2016 were obtained via a search enquiry on the histopathology database for canine and feline pathology reports containing at least one of the following terms: bladder, cystitis, transitional cell, urinary, urothelial. The terms ‘bladder’ and ‘urinary’ alone were deemed insufficient as urinary tract infection is often termed ‘cystitis’ and bladder neoplasia may be recorded as transitional cell carcinoma or urothelial carcinoma without mention of the words ‘bladder’ or ‘urinary.’ The database search process is outlined in Figure 3-1, with the initial search yielding 1210 results. From the search results, pathology records were excluded if they only mentioned gallbladder (not urinary bladder, n = 747), if the record contained only cytology and not histology (n = 24) or if the animal was less than six months of age (n = 46), as immature bladder wall can have umbilical artery or urachal remnants (283). These cases were excluded in order to standardise the dataset to mature urinary bladder samples only. Records were also excluded when urinary bladder had been discussed in the gross section of the pathology report, but no histology had been taken (n = 68), and when no diagnosis could be made due to poor sample size or quality (n = 5). For the included records, tissue blocks were retrieved (where available, 83 unable to be found or sample not of diagnostic quality), and slides recut at 4 µm thickness and stained 78

with haematoxylin and eosin (H&E) (Leica ST5020 autostainer). Samples were collected under the University of Queensland animal ethics ANRFA/SVS/259/16.

In addition to the cases obtained from UQVLS, a trip to Murdoch University School of Veterinary and Life Sciences (MUSVLS) was undertaken in late 2017, with the intent to perform the same type of search on their pathology database as the one performed at UQVLS. The search function on the MUSVLS database prevented a similar search being performed so instead, a primary disease category had to be selected, followed by a search for keywords within each year of archived records. We searched cases with primary tags of LUR, LUT and URO, all used for ‘lower urinary tract’ disease over various time periods, then within those cases searched for records containing at least one of the following: cystitis, urin*, transit*, uroth*, urolith*, TCC and UC to capture as many variations as possible related to the urinary bladder, transitional epithelium, urothelium and urolithiasis. Pathology reports were found and cross referenced to identify cases that had archived blocks containing bladder tissue (n = 71). Finally, tissue blocks were identified, and slides recut and stained with H&E using a manual staining method.

A selection of Brisbane veterinary clinics were recruited to be involved in this project based on their geographical location, and a small number of prospective bladder biopsies were obtained from dogs and cats undergoing cystotomy at those clinics as part of disease diagnosis or treatment (n = 19, ANRFA/SVS/259/16). In addition, a small number of prospective samples were obtained from dogs and cats submitted to the UQVLS necropsy service, as part of routine paid post-mortem examination, or as part of the veterinary undergraduate post-mortem classes. Finally, eleven slides containing canine or feline urinary bladder specimens were donated by a local veterinary diagnostic pathology company.

3.3.2 Histological analysis

The dataset used for the histological analysis included the UQ dataset outlined above and in Figure 3-1 (UQVLS pathology database) as well as cases collected from the MUSLVS diagnostic pathology database, and prospective samples obtained during the project from Brisbane veterinary clinics, the UQVLS diagnostic pathology service and a local veterinary pathology laboratory. A wide range of histological features were scored for each slide, outlined in Appendix 3: All animal, sampling and diagnostic histological variables measured on each bladder slide and used in the logistic regression modelling process, with the raw data in Appendix 4: Raw data from analysis of histological features of all bladder specimens in the dataset.. These histological features were selected based on a literature review of microscopic changes in urinary bladder disease in people (71, 264, 380, 402, 417). Retrieved cases were interrogated by authors EJ (pathology trainee) and RA (Diplomate of the American College of Veterinary Pathologists), and after microscopic slide review cases were 81

categorised ranked in this order – neoplasia, urolithiasis (bladder stones), cystitis (inflammation in the bladder wall which may be due to infectious or non-infectious causes), normal bladder wall, or other diagnoses (anything that did not fit in the other categories). As part of the slide review process, some diagnoses were changed from the original report based on our interpretation of histological findings (Table 3-2).

There are no publications on the normal number of resident lymphocytes in the bladder wall of dogs or cats, and very few in human medicine. In humans, one paper reports normal urothelium and submucosa containing up to 42 lymphocytes per field (size unspecified but probably 150x magnification) (78). Another early paper reported ‘small numbers’ of lymphocytes in the lamina propria of normal human bladder, particularly CD4+ lymphocytes (140). There are no equivalent reports for canine and feline urinary bladders. Based on the human literature and a review of 56 normal and diseased bladder specimens at the beginning of this project, we decided that up to 20 scattered lymphocytes per 100x magnification field of the mucosa and submucosa were allowed for normal bladders, with greater than 20 leading to classification as cystitis. Normal bladders could have occasional lymphocytes scattered throughout the urothelium, but no neutrophils.

Urolithiasis is not a histological diagnosis per se, however these cases were identified and separated based on clinical imaging or gross post-mortem findings. Records were assigned to the urolithiasis category if uroliths were mentioned in the submitted history regarding the current clinical presentation for biopsy cases, or if they were reported in the necropsy report.

Where neoplasia of any type had been identified in the bladder wall, records were placed into the ‘neoplasia’ category, even if there was concurrent inflammation or urolithiasis, as neoplasia was considered to be the predominant pathologic process. Tissue was categorised as normal if the bladder wall had no abnormalities or had up to 20 scattered submucosal lymphocytes without haemorrhage and oedema.

It is important to note that urothelial loss may also occur as a post-mortem change or as part of tissue handing during biopsy, however, we endeavoured to prevent overlap here by using separate variables for urothelial loss (denudation) and ulceration. The features of pathological ulceration include one or more of the following cellular changes in addition to the urothelial loss: urothelial cell flattening and sliding, submucosal or urothelial inflammation, submucosal haemorrhage or submucosal oedema in the remaining urothelial layer and/or submucosa (457).

All other diagnoses were grouped together in the ‘other’ category, including haemorrhage only, 82

oedema only, autolysis only, peritonitis (inflammation of the bladder serosa), traumatic bladder rupture, and healed scar tissue (Table 3-3). Cases with normal bladder tissue and other diagnoses combined were considered as the base category to which the other three categories of diagnostic outcomes were compared in the statistical modelling. The ‘other’ diagnosis group was combined with normal bladders for two reasons. First, to simulate the array of histological features that may be observed in absence of one of our main three diagnoses; second, to increase the power of this group as the normal bladder cases alone were a relatively small group. Data was managed using Microsoft Excel (287).

3.3.3 Logistic regression

Multinomial logistic regression analysis was utilised to analyse univariable associations between risk factor variables - histological variables and species; and the dependent variable, the diagnoses - cystitis, neoplasia and urolithiasis, compared to normal and other diagnoses combined. As there were only two records with granulomatous submucosal inflammation, these cases were combined with lymphocytic inflammation into one category as these types of inflammation both occur with chronic disease processes. This created a dichotomous variable for submucosal inflammation type (neutrophilic versus non-neutrophilic inflammation). For urothelial inflammation (Error! Reference s ource not found.), there were no category 4 cases (>50% of the urothelium containing inflammatory cells) and only three cases in category 3, therefore categories 2 and 3 were combined. Eventually this was reduced to the dichotomous presence or absence of urothelial inflammation for the multivariate model. A multiple Wald test was computed to evaluate the statistical significance of all categories together for any histological variable (108). Variables for which Wald’s P < 0.10 were considered for multivariate analysis.

Forward model selection was performed and the overall P-values and odds ratios (OR) with 95% confidence intervals (CI) for explanatory variables were recorded. The stepwise selection process was stopped once all covariates were significantly (P < 0.05) contributing to the model in any of the four outcome categories (cystitis, neoplasia, urolithiasis, baseline of normal/other). First order interactions between explanatory variables were also explored. Variables not selected for the original multivariate model were added back one at a time, with significant variables retained. Using this approach, we were able to identify variables that by themselves were not significantly related to the outcome, but make an important significant contribution in the presence of other variables (56). The Hausman– McFadden Test was used to test the assumption that the model odds ratios for each level of the histological variable were independent of the other levels (160). 83

Several approaches were used to conduct regression diagnostics, as previously described (161). At first, three binary logistic regression models were created from the final multivariate model (baseline/cystitis, baseline/neoplasia, baseline/urolithiasis) and regression diagnostics were carried out on each model as described in Dohoo and colleagues’ Veterinary Epidemiological Research (108). Thus, covariate patterns with outlying standardised Pearson residual and delta–beta values were identified, and the models were then rerun excluding cases from within these patterns and the changes in the resulting coefficients were examined. Model fit was then assessed using the Hosmer– Lemeshow goodness-of-fit test. Secondly, a recently developed overall goodness-of-fit-test for multinominal logistic regression models (126) was applied using the —logitgof— command in Stata (382).

Predicted probabilities were calculated from the multinomial model using the —margins—command in Stata, where the predicted probability of each diagnostic outcome was calculated at each level of the individual variable and variable combinations, while all other variables in the model were at their means. Stata 15.0 was used for all statistical analyses (382).

Results

The dataset consisted of 338 cases (Table 3-1) including 102 cases of cystitis (30% off all samples) 84 neoplasia (25%) and 42 urolithiasis (12%). There were 63 normal bladders (19%) and 47 other diagnoses (14%). During the slide review process, 23 cases had their diagnosis changed due to different interpretation of the cellular changes observed by the author (Table 3-2). The most common reason for a changed diagnosis was reclassifying a cystitis diagnosis as normal due to paucity of inflammatory infiltrates (n = 7).

There were no records in the level 4 category for urothelial inflammation or for submucosal haemorrhage amount. Stepwise construction of the multinomial logistic regression model revealed six variables that were significantly associated with disease diagnosis, compared to the normal/other disease baseline category (herein referred to as ‘baseline’). These significant animal and histological risk factors are shown in Table 3-5 (for more detail on the variables see Appendix 3: All animal, sampling and diagnostic histological variables measured on each bladder slide and used in the logistic regression modelling process).

Table 3-1: Count data of all reviewed histology slides by species, following slide review.

Diagnosis Canine (%) Feline (%) Total (%) Cystitis 81 (30) 21 (30) 102 (30) Neoplasia 74 (28) 10 (14) 84 (25) Other 34 (13) 13 (18) 47 (14) Normal 40 (15) 23 (32) 63 (19) Urolithiasis 38 (14) 4 (6) 42 (12) 267 71 338

Table 3-2: Original and revised diagnoses of cases that had the diagnosis changed during the slide review process

New diagnosis Original diagnosis Count Reason for new diagnosis Too many inflammatory cells +/- other signs Normal 2 of inflammation (haemorrhage and oedema) Cystitis Too many inflammatory cells +/- other signs Other 1 of inflammation (haemorrhage and oedema) No diagnosis* 2 Presence of neoplastic cells Neoplasia Was serositis, identified epitheliotropic Other 1 neoplastic cells Too few inflammatory cells or no other signs Cystitis 7 of inflammation (haemorrhage and oedema)

Normal No diagnosis* 2 Normal bladder tissue Was ‘oedema’ only, deemed normal amount Other 1 of tissue separation Normal 1 Haemorrhage was the only finding Haemorrhage only (2), localised peritonitis, Other Cystitis 4 haemorrhage and oedema No diagnosis* 2 Haemorrhage was the only finding

Total 23 *No diagnosis – it was unclear whether there was an error in entering information, or the initial pathologist did not record an original diagnosis.

Table 3-3: Breakdown of pathology records where the diagnosis for the bladder tissue was assigned to the ‘other’ category (n = 47)

Species Category Pathology report diagnosis n Canine No definitive diagnosis Haemorrhage and/or oedema 9 n = 34 Congestion 4 Not reported 1 Systemic disease Peritonitis 3 Coagulopathy 2 Vasculitis 1 Neospora caninum smooth muscle myositis 1 Trauma Trauma and bladder rupture 5 Preservation artefact Autolysis only 4 Iatrogenic Healing surgical wound 1 Trauma from catheterisation 1 Other Bladder polyp 1 Papillary epithelial hyperplasia 1 Feline No definitive diagnosis* Focal haemorrhage 3 n = 13 Haemorrhage and oedema 2 Congestion only 2 Smooth muscle hypertrophy 1 Systemic disease Focal haemorrhage, suspected coagulopathy 2 Feline infectious peritonitis 1 Trauma Trauma and bladder rupture 1 Anatomic Cyst 1 *Potential feline idiopathic cystitis cases

Table 3-4: Neoplasia types

Species Type of neoplasia Count Dogs n = 74 Urothelial carcinoma 53 Soft tissue sarcoma* 6 Poorly differentiated carcinoma 4 Lymphoma 2 Fibropapilloma/fibroma 3 Haemangiosarcoma 2 Not definitive 4 Cats n = 10 Urothelial carcinoma 7 Lymphosarcoma 3 *Bladder wall leiomyoma, leiomyosarcoma.

mlogit dx3 i.species i.uroth_ulcer i.lymph_aggregates i.sm_inflam_type_2 i.sm_haem i.uroth_inflamm , baseoutcome(34)

Figure 3-2: The final logistic regression model developed in Stata (382).

The final logistic regression model is shown in Figure 3-2. The six significant variables in the model when other variables were accounted for were species, urothelial ulceration, urothelial inflammation, submucosal lymphoid aggregates, neutrophilic submucosal inflammation, and having a moderate amount of submucosal haemorrhage. Species was significant for the urolithiasis outcome, with canine cases having five times higher odds of having urolithiasis than cats (OR 5.41, CI 1.07-27.46, P=0.04). Slides with urothelial ulceration were 13 times more likely to be diagnosed with urolithiasis (OR 13.45, CI 3.73-48.56, P < 0.01) compared to slides without urothelial ulceration. The presence of inflammatory cells infiltrating the urothelial cell layer was significantly associated with all disease groups, but particularly urolithiasis (OR 10.23, CI 2.49-42.06, P < 0.01). Urothelial inflammation was also strongly associated with neoplasia (OR 5.45, CI1.7-17.52, P < 0.01) and cystitis (OR 4.56, CI 1.35-15.42, P = 0.02) compared to cases that did not have urothelial inflammation Table 3-5.

The presence of submucosal lymphoid aggregates was significantly associated with cystitis (OR 6.16, CI 1.59-23.87, P = 0.01) and neoplasia (OR 4.59, CI 1.20-17.61, P = 0.03). Neutrophilic submucosal

inflammation (allocated ‘type 2’ during data collection) was significantly associated with all disease groups compared to the baseline category, particularly urolithiasis where slides with neutrophilic submucosal inflammation were 17 times more likely to belong to the urolithiasis group (OR 17.05, CI 5.26-55.23, P = 0.01). Neutrophilic submucosal inflammation was also strongly associated with cystitis (OR 6.47, CI 2.81-14.88, P < 0.01), and moderately associated with belonging to the neoplasia category (OR 2.82, CI 1.23-6.49, P = 0.02). Type 3 submucosal haemorrhage (26-50% of the submucosa contains extravasated erythrocytes, a ‘moderate’ amount of haemorrhage) was significantly associated with cystitis (OR 4.46, CI 1.43-13.93, P = 0.01). Predicted probabilities were calculated using the ‘—margins—' command in STATA, and significant probabilities (P ≤ 0.05) are displayed in Appendix 5: Significant predicted probabilities (P≤0.05) and 95% confidence intervals for diagnostic outcomes by histopathology finding patterns for canine cases submitted to UQVLS between 1994 and 2016 and collected from MUSVLS and southeast Queensland veterinary clinics from 2016-2019.

Diagnosis Variable Cystitis Neoplasia Urolithiasis OR (95%CI) P OR (95%CI) P OR (95%CI) P Species Feline Reference Reference Reference Canine 1.95 (0.71, 5.31) 0.19 2.03 (0.75, 5.52) 0.17 5.41 (1.07, 27.46) 0.04 Urothelial ulceration No Reference Reference Reference Yes 1.12 (0.43, 2.89) 0.82 0.90 (0.33, 2.43) 0.83 13.45 (3.73, 48.56) <0.01 Lymphoid aggregates No Reference Reference Reference Yes 6.16 (1.59, 23.87) 0.01 4.59 (1.20, 17.61) 0.03 1.71 (0.32, 9.12) 0.53 Neutrophilic submucosal inflammation No Reference Reference Reference Yes 6.47 (2.81, 14.88) <0.01 2.82 (1.23, 6.49) 0.02 17.05 (5.26, 55.23) 0.01 Amount of submucosal haemorrhage 1 Reference Reference Reference 2 1.98 (0.81, 4.82) 0.13 1.12 (0.48, 2.57) 0.80 1.01 (0.3, 3.39) 0.99 3 4.46 (1.43, 13.93) 0.01 0.39 (0.09, 1.76) 0.22 1.7 (0.40, 7.30) 0.47 Urothelial inflammation No Reference Reference Reference Yes 4.56 (1.35, 15.42) 0.02 5.45 (1.7, 17.52) <0.01 10.23 (2.49, 42.06) <0.01

3.4.1 Graphic representation of the six significant variables

Each graph below depicts one of the six variables that were significantly associated with disease diagnosis. The graphs have been designed so that each entry depicts a disease diagnosis (cystitis, neoplasia, urolithiasis and normal/other) without and with the significant variable. For example, in Figure 3-6, disease probability by submucosal inflammation type, the probability of having cystitis if there is mononuclear inflammation (MONO) as the primary inflammation type in the submucosa is 26% (first column). The fourth entry depicts the probability of being diagnosed with neoplasia (NEO) if the sample has neutrophilic (NEU) inflammation as the primary inflammation type in the submucosa.

Figure 3-3 depicts disease probability by species. Each entry on the x-axis is a disease diagnosis, either feline or canine. The first entry shows the probability of having cystitis if the patient is a feline (F, 0.33, 33%), compared to canine (C, 37%). The biggest difference is in the normal/other diagnosis. Figure 3-4 depicts the probability for each diagnosis group, stratified by the absence (N) or presence (Y) or urothelial inflammation of any type.

Figure 3-5 depicts disease probability depending on the absence (N) or presence (Y) of any amount of urothelial ulceration. The most obvious finding is a markedly decreased probability of diagnosing urolithiasis if there is an absence of urothelial ulceration.

Figure 3-6 depicts the probability of each disease group, stratified by the main type of submucosal inflammation (mononuclear/non-neutrophilic versus neutrophilic). If the main type of submucosal inflammation is neutrophilic, then there is increased probability of having cystitis or urolithiasis, slightly decreased probability of having neoplasia, and a markedly decreased probability of being diagnosed as normal or with other diagnoses.

Figure 3-7 depicts disease probability depending on the amount of haemorrhage in the submucosa. Data is shown for each disease, and the options of having no, mild or moderate amounts of submucosal haemorrhage. For example, there is a markedly increased probability of being diagnosed with cystitis if the sample has a moderate amount of submucosal haemorrhage compared to a mild amount or no haemorrhage.

Figure 3-8 depicts disease probability by the absence (N) or presence (Y) of submucosal lymphoid aggregates.

DISEASE PROBABILITY BY SPECIES

0.6

0.5

0.4

0.3

PROBABILITY 0.2

0.1

-0.1 CYS_F CYS_C NEO_F NEO_C URO_F URO_C N/O_F N/O_C Probability Low High

Figure 3-3: Probability for bladder outcomes by species (95% CI). CYS = cystitis; NEO = neoplasia, URO = urolithiasis; N/O = normal/other category; _F = feline; _C = canine.

DISEASE PROBABILITY BY UROTHELIAL INFLAMMATION 0.6

0.5

0.4

0.3

PROBABILITY 0.2

0.1

-0.1 CYS_N CYS_Y NEO_N NEO_Y URO_N URO_Y N/O_N N/O_Y Probability Low High

DISEASE PROBABILITY BY UROTHELIAL ULCERATION 0.5 0.45 0.4 0.35 0.3 0.25

PROBABILITY 0.2 0.15 0.1 0.05 0 CYS_N CYS_Y NEO_N NEO_Y URO_N URO_Y N/O_N N/O_Y Probability Low High

DISEASE PROBABILITY BY SUBMUCOSAL INFLAMMATION TYPE 0.6

0.5

0.4

0.3 PROBABILITY

0.2

0.1

0 CYS_MON CYS_NEU NEO_MON NEO_NEU URO_MON URO_NEU N/O_MON N/O_NEU Probability Low High

DISEASE PROBABILITY BY AMOUNT OF SUBMUCOSAL HAEMORRHAGE 0.9 0.8 0.7 0.6 0.5 0.4

0.3 PROBABILITY 0.2 0.1 0 -0.1

Probability Low High

DISEASE PROBABILITY BY SUBMUCOSAL LYMPHOID AGGREGATES 0.7

0.6

0.5

0.4

0.3

PROBABILITY 0.2

0.1

-0.1 CYS_N CYS_Y NEO_N NEO_Y URO_N URO_Y N/O_N N/O_Y Probability Low High

Discussion

Five histological variables and one animal variable were found to be significantly associated with one or more disease group using logistic regression modelling – species, urothelial inflammation, urothelial ulceration, type of submucosal inflammation, amount of submucosal haemorrhage and presence of submucosal lymphoid aggregates. Species was particularly significant for the diagnosis of urolithiasis. Our finding of dogs having five times higher odds of being diagnosed with urolithiasis compared to cats is not surprising as urolithiasis is reported more commonly in dogs than cats (398). Urolithiasis is thought to be present in up to 1.5% of cats seen at veterinary practices and up to 3% of dogs (318), which is reflected in our findings. It is generally uncommon for cystoscopic bladder biopsies to be obtained from dogs with urinary tract symptoms, and even less common for cats due to their small urethral size (317), however, bladder biopsy is most likely to be performed when a patient is already undergoing a cystotomy procedure for something like urolith removal, or investigation of a mass. Bladder wall biopsy may also be appropriate for cases of chronic refractory urinary tract infection (386) and can be used to investigate for other underlying disease states.

Urothelial ulceration was significantly associated with urolithiasis, which makes sense biologically as this disease is characterised by the presence of a urolith in the bladder lumen. The bladder stone/s causes physical trauma to urothelial cells, sometimes to the point of complete urothelial loss (180, 318). Urothelial loss may also occur as a post-mortem change or as part of tissue handing during biopsy, however we endeavoured to prevent misclassification here by using separate variables for urothelial loss (denudation) and ulceration. Post-mortem or tissue handling-induced urothelial loss showed minimal tissue reaction suggesting it was artefactual, as opposed to pathological urothelial ulceration that occurred alongside inflammatory infiltrates and/or haemorrhage, fibrin and oedema.

The presence of urothelial inflammation was strongly associated with all disease groups but was not able to differentiate between them. It is normal to have low numbers of resident lymphocytes within the urothelium (140), however it makes biological sense that numbers would increase in response to physical trauma such as uroliths, the presence of microorganisms, or in response to ‘foreign’ neoplastic cells (most neoplasms in this dataset were urothelial carcinomas which arise from the urothelial cells). In addition, neoplasms weaken tissue architecture and can make the tissue more sensitive to trauma or infection, further increasing leukocyte levels.

Neutrophilic submucosal inflammation was increased in association with all disease outcomes while non-neutrophilic inflammation (mononuclear; primarily lymphocytic or lymphoplasmacytic

infiltrates) was not significantly associated with any disease group compared to baseline. As mentioned, the presence of scattered lymphocytes in the urothelium and submucosa can be normal (78, 140), however neutrophils are not part of the resident leukocyte population and are only present when stimulated by inflammation, which is consistent with our results. Neutrophils will marginate within capillaries secondary to tissue handling during sampling, however neutrophils were only counted towards submucosal inflammation if they had transmigrated out of the blood vessels and into the submucosa.

The presence of submucosal lymphoid aggregates was associated with cystitis and neoplasia, but not with urolithiasis. Aggregates of lymphocytes (often with germinal centres) are a part of normal immune surveillance in many tissues (208, 349), but in the human urinary bladder are thought to represent chronic inflammation (69). There are no published standards for the role of lymphoid aggregates in canine or feline bladders, however the chronic inflammation theory is plausible for these species as well. Chronic inflammation is frequent in cases of recurrent cystitis (305, 368), while tumour infiltrating leukocytes are commonly observed alongside neoplasia (204). Animals with recurrent bacterial cystitis may have predisposing breed, anatomical or metabolic factors that have caused them to have bladder inflammation more frequently throughout their life prior to histological sampling (280, 292, 305, 368), thus resulting in the presence of lymphoid aggregates. One potential limitation of this work is that cases of chronic cystitis are more likely to undergo clinical biopsy than acute cystitis patients, so there is likely to be some skewing of our dataset to contain a higher proportion of chronic cystitis cases than would occur in the general pet population. Uroliths may take a more acute clinical course, with animals showing clinical signs sooner than those with low grade cystitis or with neoplasia. This may lead to earlier detection of disease and therefore tissue samples being obtained earlier in the disease process before the formation of lymphoid aggregates. The constant presence of the urolith/s may also lead to repeated traumatic insult, therefore repeatedly stimulating acute inflammatory cells. Lastly, there is likely to be little antigenic drive in urolithiasis to provoke an acquired immune response typically occurring in lymphoid tissue. Conversely, infection and neoplasia are associated with abnormal self or non-self antigens, therefore would be more likely to develop an acquired, lymphocytic immune response (5, 182) and therefore develop lymphoid follicles.

Finally, the most severe category of haemorrhage observed in the histological analysis (a moderate amount of haemorrhage – extravasated erythrocytes occupying 26-50% of the submucosal area, label ‘3’) was significantly associated with the cystitis outcome when compared to baseline. Submucosal

haemorrhage was observed in all disease groups, but it was highest in the cystitis group. This is likely due to a number of factors, primarily the vasoactive cytokines released during acute inflammation causing increased vascular permeability and leakage (151). Secondly, we observed a higher level of neutrophilic inflammation in the cystitis group (six times the odds of the baseline category). Submucosal neutrophils are most likely to be associated with haemorrhage because one of the enzymes released, elastase, can contribute to the breakdown of arteriolar walls (218). In addition, bacteria can induce endothelial damage, which may result in submucosal haemorrhage in some cystitis cases (457). Submucosal haemorrhage may also occur as a sequela to tissue handing during biopsy, however tissue handling does not result in extravascular inflammatory cell infiltrates.

When comparing ‘cystitis’ cases to existing literature, we are required to look to published work on bacterial cystitis in dogs, and bacterial cystitis as well as non-infectious idiopathic cystitis in cats, as all these diseases are likely to be represented in our ‘cystitis’ disease category. For the cystitis outcome, it is important to note that our definition of cystitis for this analysis was histological evidence of inflammation in the bladder wall in the form of inflammatory cell infiltrate in addition to haemorrhage and/or oedema, in the absence of uroliths and with or without positive bacterial urine culture, which is different from the commonly used clinical definition of cystitis (patients with lower urinary tract clinical signs and often with positive bacterial culture indicating urinary tract infection). Due to the histological and primarily retrospective nature of this study, many cases were lacking urine culture results, making it impossible to definitively categorise those cases as having or not having bacterial cystitis. In addition, microorganisms are rarely visible on light microscopy slides due to their very small size, so presence or absence of bacterial cystitis could not be assessed with any consistency with the samples in this study.

When bladder stones were present, the histological diagnosis made by the initial pathologist was often ‘cystitis’, because bladder wall inflammation frequently occurs in these cases. Urolithiasis is not a histological diagnosis, however we chose to separate these records into a different group, ‘urolithiasis’ and not include them in the ‘cystitis’ category, as the inflammation can be attributed to the physical trauma of the stone in addition to potential bacterial infection. It was recognised that bacterial cystitis may lead to urolithiasis, namely struvite stone formation secondary to urease- producing bacteria (314, 366). Conversely, uroliths can impair bladder defences, facilitating the development of bacterial cystitis (314). Up to 50% of canine bladder stones are thought to be infection-related (387), a phenomenon that is much more common in dogs than in cats (169). The small number of cases in this study with confirmed concurrent bacterial cystitis and urolithiasis were

categorised with other urolithiasis cases instead of other bacterial cystitis cases for several reasons. Firstly, urine culture for bacterial detection may not be performed routinely on every urolith case, therefore the urolithiasis category in this dataset is likely to already contain some cases with both urolithiasis and bacterial urinary tract infection. Secondly, the cystitis disease group is already a heterogenous combination of a) cases with bacterial or fungal infection confirmed by urine culture, b) cases with suspected bacterial infection without bacterial culture, and c) cases of suspected feline idiopathic cystitis (no infectious organism involved). Thirdly, the mechanical damage caused by uroliths may lead to a different inflammation profile to bacterial cystitis alone, therefore cases with uroliths and bacterial infection were deemed more similar histologically to the urolithiasis cases than cases of cystitis without bladder stones. Therefore, the decision was made to add urolithiasis/bacterial cystitis cases to the urolithiasis group. The author realises that there is likely to be overlap between the cystitis and urolithiasis groups and it would be ideal to have culture results for each of these cases to enable more accurate stratification of these diseases, however this was not possible due to its retrospective nature of this study.

Almost 7% of records had the diagnosis changed after slide review, highlighting the subjectiveness of diagnostic pathology. This issue was most overt when differentiating mild cystitis from normal bladder tissue. ‘Cystitis’ technically means inflammation of the bladder wall, which may occur with or without infectious organisms. There are no publications on normal numbers of resident lymphocytes in the canine or feline bladder wall. In humans, one paper reports normal urothelium and submucosa containing up to 42 lymphocytes per field (size unspecified but probably 150x magnification) (78). Another early paper reported small numbers of lymphocytes in the lamina propria of normal human bladder, particularly CD4+ lymphocytes (140), however there are no equivalent reports for canine and feline urinary bladders. Based on the human literature, we made the decision to allow up to 20 scattered lymphocytes per 100x magnification field in a normal bladder sample, with greater than 20 lymphocytes per 100x magnification field leading to classification as cystitis. There are currently no standards for interpreting these kinds of lesions in veterinary pathology. The pathologist must make a diagnosis based on the clinical history and their experience; however, we aim to improve this area by providing a more objective tool for diagnosis of urinary bladder disease based on the logistic regression analysis performed in this study.

The primary limitation of this study is its retrospective nature. A retrospective study design was chosen for two reasons. Firstly, despite bladder disease occurring commonly in dogs and cats, bladder biopsy is not frequently performed and therefore a suitable sample size of bladder histological

specimens would be very difficult to attain via a prospective study. This particularly applies to the rarely biopsied but highly clinically significant waxing and waning syndrome of feline idiopathic cystitis. Secondly, accessing 20 years of archived pathology records provides a detailed look at the historical nature of bladder disease samples in an Australian tertiary institution – an analysis that has not previously been performed. A limitation of this approach is that ideally, mathematical models should be developed from prospective data to minimise missing data and tighten the definition of subject groups (158).

There is potential for unavoidable crossover between pathological and sampling artefact for some variables such as submucosal haemorrhage, however this was accounted for when possible (for example, differentiating pathological and sampling related urothelial loss). Further, the combination of normal and other diagnoses is likely to have had some impact on the results. The two groups were combined to increase case numbers in the baseline category and to replicate the range of other diseases that may be encountered. In future it would be prudent to recruit a larger sample size to enable the comparison of disease categories to a group of normal bladders without any other disease combined with it. Bladder wall samples obtained at necropsy were included in this analysis. It is unknown how many of our records had bladder disease found incidentally, that is, bladder disease was not the primary cause of death or euthanasia, due to absent or incomplete clinical history in many cases. However, the histological changes of each disease are expected to be consistent regardless of any other organ dysfunction, therefore these cases should not skew this analysis.

Some limitations for using logistic regression modelling are that odds ratios are the only output, the sequence of entering variables into the model is based purely on the statistical criteria, and it can be difficult to interpret interactions between variables (206). Secondly, the use of logistic regression means that our probabilities for disease outcome based on histological variables were calculated by keeping all other variables in the model at their mean value. As with any statistical modelling, the data groupings are going to have an impact on the results. For future research into this area, it would be prudent to 1) Perform the modelling on prospectively collected data, or if retrospective data is used then; 2) utilise a large enough dataset where more data stratifications can be made, for example, the separation of bacterial cystitis and non-infectious cystitis, as this would likely provide more meaningful information on those two categories of cystitis. Finally, an additional limitation of this work is the lack of a gold standard to confirm the diagnosis - in our case, histopathology is in fact the gold standard, however there are still elements of subjectivity and bias to this process (6, 272). Histology is simply part of the diagnostic process, and in practice is best incorporated with the clinical

findings and animal signalment - a feat that a logistic regression model would find difficult.

Conclusion

The overall aim of this chapter was to evaluate the utility of logistic regression in formulating a predictive probability model for veterinary histopathology. In summary, we used logistic regression modelling on a histology dataset of canine and feline urinary bladders from Eastern and Western Australia. This modelling procedure identified six significant variables that were associated with disease outcome compared to baseline – urothelial ulceration, urothelial inflammation, neutrophilic submucosal inflammation, submucosal lymphoid aggregates, amount of submucosal haemorrhage, and species. These six variables will be used to create a predictive probability tool for bladder disease diagnosis. The next step in this work will then be the validation of the tool in a controlled setting, to evaluate its potential use for real world diagnostic veterinary pathology applications.

Chapter 4 Concordance of pathologist bladder biopsy assessment with and without the use of a predictive tool.

Introduction

Inter-pathologist variation has been widely recognised across the human and veterinary medical fields (10, 83, 345, 445), with a Fleiss’ kappa statistic of greater than 40% generally deemed to be a fair to good level of agreement, and greater than 75% an excellent level of agreement (134). The kappa statistic measures the level of agreement between observers, and the likelihood that agreement would occur by chance alone (90). There are many causes for inter-pathologist variation including level of experience (429), sample quality and processing (444), and the organ system being examined. For example, variation between pathologists is high when assessing gastrointestinal (GIT) biopsies in cats and dogs due to the difficulty in interpreting the degree of resident leukocytes compared to pathological inflammation, the focal nature of some gastrointestinal diseases and GIT tissue sampling, and the numerous systems or grading schemes for evaluation (444). Inter-observer variability is a potential hazard to human and animal health in the form of diagnostic discrepancies, potentially resulting in incorrect diagnoses and therefore inappropriate subsequent therapeutics and disease management (445, 454).

A further limitation to disease diagnosis in veterinary pathology may be a lack of standardization of pathology reporting, where pathologists may agree that an abnormality is present, but disagree on the specific disease diagnosed (445). The current literature suggests that clearly defined histological criteria improves pathologist agreement regarding the assessment and interpretation of histological changes (141, 445). Further, standardised nomenclature for diagnoses is beneficial, particularly in sub-specialties like toxicologic pathology where clear communication to regulatory agencies and between pathologists during primary and peer-review is critical (294, 333). There exists standardised terminology for use on urinary bladder tissues in rodent toxicologic pathology (367), however to date, there is no standardised histological criteria for urinary bladder assessment in companion animals.

As part of a large retrospective analysis of urinary bladder histology of 236 canine and feline bladder cases from the University of Queensland (UQ) pathology archives, a predictive tool was developed from the results of logistic regression modelling on associations between histological features and the disease diagnosed. The primary bladder wall diagnosis was assigned to each tissue section ranked in this order – neoplasia, urolithiasis (bladder stones), cystitis (inflammation in the bladder due to infectious or non-infectious causes), normal bladder wall or other diagnoses (including sampling

100

artefacts such as haemorrhage and oedema, autolysis, and peritonitis - inflammation of the outer bladder wall along with other organs due to other abdominal disease).

The previous work presented in Chapter 3 suggested a problematic discrepancy in diagnoses of bladder disease. As part of the slide review process conducted in Chapter 3, recuts of all 236 cases were evaluated by two readers, and 23 cases had their diagnosis changed. Seven of these were simply ‘changed’ because no diagnosis had been recorded in the pathology report database to begin with. This left 16 cases (7%) that had the reported diagnosis altered based on the histological changes observed during slide review by two veterinary pathologists. Urinary bladder tissue would be deemed an ‘easy’ tissue to evaluate by pathologists due to its limited range of morphological responses to injury, however this result suggests that there may still be inconsistency in diagnosis, particularly regarding distinguishing inflamed from normal bladder tissue. Discrepancy was commonly related to the number of leukocytes present, and the concurrent presence of haemorrhage and oedema. The author has been unable to find published information on the normal numbers of submucosal resident lymphocytes in the bladder wall of dogs and cats. There are few publications on urothelial lymphocyte infiltration, with normal human urothelium and submucosa containing up to 42 lymphocytes per field (size unspecified but probably 150x magnification) (78). Another early paper reported small numbers of lymphocytes in the lamina propria of normal human bladder, particularly CD4+ lymphocytes (140).

In the UQ canine and feline bladder histology study, a tissue was assigned a diagnosis of ‘normal’ if the submucosa contained up to 20 lymphocytes per low power (100x) field, without haemorrhage and oedema. These leukocytes were deemed to be normal resident lymphocytes. If a section contained greater than 20 lymphocytes per low power field or had up to 20 but had concurrent vascular reaction (haemorrhage and/or oedema), or had any neutrophil infiltration, then the tissue was deemed to be inflamed and assigned to the cystitis group. It is important for both animal and human health that the correct diagnosis is made, as incorrectly diagnosing normal samples as cystitis may result in unnecessary treatment.

In addition, substandard information on pathology submission forms is a frequent problem in both human and veterinary pathology. In one veterinary study of 510 biopsy submissions, up to 88% of forms were found to be deficient in at least one key area (43). In human medicine, higher quality clinical information on pathology requisition forms is associated with decreased turnaround time (p < 0.001) and improved outcomes (351). Further, absence of clinical history resulted in lower diagnostic accuracy of bronchial cytology specimens (341). Regarding clinical history in toxicologic 101

pathology studies, Crissman and colleagues state that “if this information is not available initially, selected tissues may need to be re-evaluated to ensure accurate diagnoses and interpretations.” (88).

The Society of Toxicologic Pathology provides a list of guidelines for pathologist interpretation of research study material, including the following;

The study pathologist should have access to the study protocol and all protocol amendments, all study data including the intended pharmacologic target and mechanism of action, in-life study data, clinical pathology data, organ weight data, necropsy findings, toxicokinetic information, and (when possible) data from previous studies with the same test article. (294)p808

In summary, pathologist bias does exist based on the patient history, however a high quality patient history is vital in arriving at the correct diagnosis (376). It is acknowledged that a strong positive bias provided by the historical and ancillary information is critical for a pathologist to make the best diagnosis (294).

Hypothesis and objectives

We hypothesise that there will be higher agreement (measured using the kappa statistic) when pathologists diagnose bladder diseases with access to the clinical history, compared to without history. Secondly, we hypothesise that there will be higher agreement between veterinary pathologists when using our predictive tool for bladder disease diagnosis compared to making diagnoses without using predictive probabilities. Thirdly, we hypothesise there will be good agreement between the test pathologists’ diagnoses and the reference diagnosis.

The objectives for this chapter are to:

1. Develop a predictive tool for bladder disease diagnoses formulated from logistic regression modelling of histological changes of archived bladder tissue samples from the UQSVS and MUSVLS.

2. Test and evaluate inter-pathologist agreement on bladder tissue diagnosis with and without clinical history, and with and without predictive probabilities for each diagnosis group (cystitis, neoplasia, urolithiasis, normal).

3. To evaluate how well the test pathologists agreed with the reference diagnosis.

102

Materials and Methods

4.3.1 Sample size

The sample size for this project was calculated using the kappa sample size calculation. We assume baseline agreement (k0) = 0.5 (445), and assume the lower limit of agreement (kL) to be >= 0.2. For the recruitment of four pathologists to assess the slides, the number of slides needs to be at least 23 (n = 23).

The cases used for the concordance study were selected from a bladder disease case material pool by the author (EJ) until the desired sample size was reached, using the below criteria.

All case records of canine and feline patients with bladder tissue collected via biopsy or at necropsy between 1994-2016, for animals greater than six months of age, were obtained from the University of Queensland veterinary pathology database (See Figure 2-1). Bladder sections were recut from formalin fixed paraffin embedded blocks at 4-micron thickness for slide review by two veterinary pathologists. The slides were all reviewed and scored according to criteria determined by the author as part of the thesis literature review (Appendix 3: All animal, sampling and diagnostic histological variables measured on each bladder slide and used in the logistic regression modelling process), with the resultant diagnosis forming the reference diagnosis for this study.

Slide selection criteria for this study were as follows: the sample was full thickness (contains outer muscularis with or without serosa), the tissue section on the microscope slide was at least 0.5 cm x 0.5 cm in size and oriented in cross section, and samples represent varying degrees of severity of each diagnosis group (cystitis, neoplasia, urolithiasis or normal) for both dogs and cats. These rules provided an eventual 25 slides consisting of seven cystitis (two canine, five feline), six neoplasia (three canine and three feline), six urolithiasis including one with concurrent urinary tract infection (four canine and two feline), and six normal bladders (four canine and two feline) (See figures in Appendix 6: Bladder histology photomicrographs of cases used in the pathologist agreement study, Chapter 4.)

Slides were digitally scanned by the author on a Leica Aperio CS2 slide scanner (Serial number 50019) and ScanScope software and were available for each study participant to access electronically via guest access to the university’s Aperio eSlide Manager (version 12.4.0.5043, 2018). All slides were scanned using a 40x objective which produces an image with a resolution of 0.25μM/pixel. Snapshots were taken prior to scanning and the scanning region cropped as close to the tissue as

103

possible to minimise the file size. The default focus point setting was used, with additional focus points added to a total of 6-8 focus points per square centimetre of tissue evenly distributed over the tissue.

4.3.2 Building the predictive tool

In Chapter 3, multinomial logistic regression modelling using R studio identified six significant variables that were associated with disease diagnosis - species, urothelial ulceration, urothelial inflammation, submucosal inflammation type, presence of lymphoid aggregates and amount of submucosal haemorrhage. Predicted probabilities were calculated from the multinomial model using the —margins—command in Stata 15.0 (382), where the predicted probability of each diagnostic outcome was calculated at each level of the individual variable and variable combinations, while all other variables in the model were at their means.

A spreadsheet was built to record the findings, diagnoses and comments from each pathologist evaluating the slide set under the three sequential conditions - with no information (‘First’ read), then with animal signalment and clinical history (abbreviated to shorthand in the table headings below, ‘Hx’) and finally with the adjunct use of the predictive tool (‘Tool’) (Figure 4-1). The test spreadsheet was built in Microsoft Excel using four different worksheets (the third and fourth identical but with separate sheets for canine and feline cases) to be completed in order, with each worksheet having different conditions, with instructions both emailed to the pathologists, and also present within the Microsoft Excel document. For the first (no signalment or history) and second (signalment and history available) worksheets, most cells contained a drop-down list with yes or no options, or a list of inflammatory cell types, except for the morphological diagnosis and comment columns which allowed free text responses. For etiological diagnosis, pathologists could select one response from a drop-down list containing cystitis, neoplasia, urolithiasis, normal, or other. To reduce complexity of the task and the subsequent statistical analysis, pathologists were only asked to provide drop-down answers to a few histological features - urothelial ulceration, submucosal oedema, submucosal haemorrhage, presence of submucosal inflammation, primary type of submucosal inflammation, presence of muscularis inflammation, primary type of muscularis inflammation, and the presence of microorganisms (Appendix 7: A sample of the Microsoft Excel worksheet from the pathologist agreement study.) These features are ones that we deemed would be representative of the disease process. It was asked in the accompanying instructions that any other histological changes or lesions, or other comments on the slide, be recorded in the free text comment boxes.

104

Animal Use of No case signalment + predictive information clinical tool history

Figure 4-1: Sequence of assessment of the 25-slide set by each pathologist.

The third worksheet was designed so that the probabilities for each disease for every possible combination of variables (derived from the logistic regression model) could be stored in a hidden worksheet, then the table visible to the pathologists would populate with the probability for each disease based on the combination of histological variables observed by the pathologist, before the pathologist was prompted to make a diagnosis. In addition to displaying the probabilities and confidence intervals in a table, a chart was also set up to graphically represent the probabilities for each disease for every slide evaluated.

To summarise, our four study pathologists were not given a criteria for diagnosis but were asked to make the diagnosis based on their own individual criteria for the first two reads (without then with clinical history). Then, for the third read using the predictive tool, some probabilities would be displayed based on their responses to certain histologic features (derived from Chapter 3), with the goal to evaluate the influence of those probabilities, that is, whether there was higher or lower agreement between the participants then they were using the predictive tool. The instructions provided to the pathologists are shown in Appendix 7: A sample of the Microsoft Excel worksheet from the pathologist agreement study.

4.3.3 Pathologists

Four veterinary pathologists were selected based on the following criteria: specialist certification with the American College of Veterinary Pathologists (ACVP), and a minimum of 5 years working in a diagnostic environment that involved the review of tissues from dogs and cats, and from a variety of geographical locations (United States of America, Canada, New Zealand and Australia) to reduce bias based on country of work. Pathologists viewed the slides on the computer they use routinely for their work using Aperio’s eSlide Manager (version 12.4.0.5043, 2018).

105

Table 4-1: Histological criteria to be assessed by the pathologists in worksheets one and two.

Column heading Potential answers Slide code Provided Ulceration Yes, No SM_oedema Yes, No SM_haem Yes, No SM_inflamm Yes, No SM_inflamm_type Lymphocytic Lymphoplasmacytic Neutrophilic Granulomatous No inflammation Det_inflamm Yes, No Det_inflamm_type Lymphocytic Lymphoplasmacytic Neutrophilic Granulomatous No inflammation Organisms Yes, No Morphological diagnosis Free form box Etiological diagnosis Normal Other Cystitis Neoplasia Urolithiasis Comments Free form box Det detrusor muscle/muscularis; haem haemorrhage; inflamm inflammation; SM submucosal.

106

For worksheets 3 and 4 (third read of the slides, with canine and feline slides on separate worksheets), histological variables were based on the significant variables from the logistic regression modelling, thus pathologists were required to answer the following (Table 4-2).

Table 4-2: Column headings and potential answers for the pathologists participating in the agreement study.

Column heading Potential answers Slide code Provided Urothelial ulceration Yes, No Submucosal lymphoid aggregates Yes, No Neutrophilic submucosal inflammation Yes, No Urothelial inflammation Yes, No Amount of submucosal haemorrhage Mild Moderate Severe Your diagnosis Normal Other Cystitis Neoplasia Urolithiasis Comments Free form box

Pathologists emailed back their completed spreadsheets, which were then deidentified and randomly assigned a number from one to four (P1-P4). Data was managed via Microsoft Office Excel (287).

4.3.4 Statistical analysis

The final data from the four pathologists was combined to form a single file with unique slide identifier (ID), species, conditions (no animal information, signalment & history and with predictive tool) and the etiological diagnosis of the bladder by each pathologist. The Fleiss kappa coefficient was used to compute the inter-rater reliability measures as an agreement measure for multiple categorical variables (133, 134). According to Fleiss et al. (2003), the kappa value can range from -1 (no agreement) to +1 (perfect agreement) with k = 0 indicating that the agreement is no better than what would be obtained by chance (134). The interpretation of the magnitude of Fleiss kappa is that values greater than 0.75 represent excellent agreement beyond chance, values between 0.40 and 0.75 represent fair to good agreement beyond chance and values below 0.40 or so may be taken to represent poor agreement beyond chance (134).

The statistical significance test for the Fleiss kappa coefficient is based on the following hypothesis:

107

Null hypothesis (H0): kappa = 0. The agreement is the same as chance agreement.

Alternative hypothesis (Ha): kappa ≠ 0. The agreement is different from chance agreement.

All analyses were conducted in R (R: A language and environment for statistical computing, 2018) (340). P4 had some technical issues that prevented their assessment of all the slides. The MICE R package was used for imputing the missing diagnoses from this pathologist. As the outcome here is a multinomial variable, the polytomous logistic regression imputation model for unordered categorical data was implemented by the MICE algorithm as described by van Buuren and Groothuis-Oudshoorn in 2011 (416).

For the evaluation of pathologist accuracy compared to the reference diagnosis, cases for which any pathologist had made a diagnosis of ‘other’ could not be compared against the reference diagnosis and were removed.

108

Results

The diagnoses from all pathologists and slide reading conditions were collated in Microsoft Excel (Table 4-3). The ‘Reference diagnosis’ is the diagnosis given by the original pathologist on the case and confirmed by the author of this project. Cases were chosen to be representative of the range of lesions observed in our dataset.

Table 4-3: The count data from all study pathologists, P1-P4.

No animal information Signalment & history With predictive tool Diagnosis Reference P1 P2 P3 P4 P1 P2 P3 P4 P1 P2 P3 P4 Cystitis 7 7 14 17 6 6 14 11 5 7 14 11 4 Neoplasia 6 4 4 3 3 4 4 3 3 4 4 3 3 Urolithiasis 6 9 0 1 6 9 0 8 5 7 0 8 3 Normal 6 5 2 2 1 6 2 2 5 5 2 2 5 Other 0 0 5 2 3 0 5 1 3 2 5 1 4 Total 25 25 25 25 19 25 25 25 21 25 25 25 17 Reference = reference diagnosis, the diagnosis given by the original pathologist on the case and confirmed by the author of this project.

From the count data, it is evident that there is a large amount of variation between the four pathologists throughout the intervention steps. The diagnosis for each case by each pathologist is shown in Table 4-4. For some cases there is good agreement between pathologists, and each pathologist stays consistent with their diagnosis, particularly for slides with neoplasia. However, it can be seen in Table 4-4 that there are many cases where the pathologists have diagnosed the slides differently, or their diagnosis has changed throughout the sequential slide readings. This is particularly the case for slides with urolithiasis or cystitis, and for normal bladder slides.

109

Table 4-4: Diagnoses for every slide with the reference diagnosis, grouped by pathologist (P1-P4), under three different conditions (First, Hx and Tool).

Case Species P1 P2 P3 P4 Reference diagnosis number First Hx Tool First Hx Tool First Hx Tool First Hx Tool 02/0691A Canine Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neoplasia - UC 03/0244B Feline Neo Neo Neo Neo Neo Neo Neo Neo Neo . . . Neoplasia - UC 04/1068A Canine Neo Neo Neo Cys Cys Cys Uro Uro Uro . . . Neoplasia - UC 05/1429 Feline Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neo Neoplasia - UC 07/2196 Canine Cys Cys Cys Cys Cys Cys Cys Cys Cys . . . Cystitis, mild 08/1399D Feline Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cystitis secondary to lower motor neuron bladder 08/1885J Canine Cys Cys Cys Cys Cys Cys Cys Uro Uro Uro Uro Uro Urolithiasis - Cystitis + uroliths at necropsy 10/0259B Canine Cys Cys Othb Oth Oth Oth Cys Cys Cysd Neo Neo Neo Neoplasia - Leiomyosarcoma 11/0220A Canine Uro Uro Uro Cys Cys Cys Cys Uro Uro Uro Uro Uro Urolithiasis - Cystitis + stones removed surgically 12/0516 Feline Uro Uro Uroa Cys Cys Cys Cys Uro Uro Oth Oth Othl Cystitis, suspect FIC - urethral obstruction, no uroliths at necropsy 13/0197A Feline Uro Uro Uroa Cys Cys Cys Cys Cys Cys Oth Oth Othl Cystitis, suspect FIC - clinical signs of stranguria. No stones or obstruction at necropsy 14/0707A Feline Uro Uroa Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cystitis, suspect FIC - clinical urethral obstruction, no uroliths at necropsy 14/0993B Canine Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cys Cystitis 15/0893E Canine Norm Norm Norm Norm Norm Norm Cys Cys Cys . . . Normal 15-017E feline Norm Norm Norm Oth Oth Othf Cys Cys Cysd . . . Neoplasia - Epitheliotropic lymphoma (metastasis) 16/0034I Canine Norm Norm Norm Norm Norm Norm Cys Cys Cysd Norm Norm Norm Normal 110

17/0311 Canine Norm Norm Norm Oth Oth Othg Oth Oth Othk Oth Norm Norm Normal 17/0398 Canine Cys Norm Cysc Oth Oth Othg Cys Cys Cys . . . Normal 17/0503A feline Uro Uro Uroa Neo Neo Neo Cys Uro Uro Cys Uro Uro Urolithiasis - Urethral obstruction and bladder rupture, uroliths at necropsy. 17/1335A Feline Cys Cys Cysd Cys Cys Cys Norm Norm Norm Cys Norm Norm Normal 17/1336A Feline Norm Norm Norm Oth Oth Othh Norm Norm Norm Uro Norm Norm Normal 17/1337A Feline Uro Uro Othe Cys Cys Cysi Cys Cys Cysd Uro Uro Uro Urolithiasis (cystotomy) + UTI (positive culture) 18/0073A Canine Uro Uro Uro Cys Cys Cysi Othj Uro Uro Uro Uro Uro Urolithiasis 19/0528A Feline Uro Uro Uro Cys Cys Cys Cys Uro Uro Uro . . Cystitis, suspect FIC - clinical urethral obstruction, no uroliths at necropsy 19/0729A Canine Uro Uro Uro Cys Cys Cys Cys Uro Uro Cys . . Urolithiasis + UTI Key: First = First read of the slide, with no animal information; Hx = Second read of the slide, with animal signalment and clinical history; Tool = Third read of the slide, with the predictive tool. Cys = Cystitis; FIC = Feline idiopathic cystitis; Neo = Neoplasia; Norm = Normal; Oth = Other diagnosis; UC = Urothelial carcinoma; Uro = Urolithiasis. Full stop indicates slides that were unable to be assessed due to technical issues. Comments from the pathologists: a Suspected FIC cases, called these urolithiasis as many have plugs that act similarly to stones; b Polypoid cystitis, mild inflammation, secondary to stones; c Follicular cystitis; d Normal/minimal clinically insignificant cystitis; e Marked haemorrhage; f Urothelial hyperplasia with glandular metaplasia; g Urothelial hyperplasia; h Focal mucosal ulceration, intraluminal haemorrhage; i Obstruction/uroliths?; j Poorly scanned, had trouble evaluating; k Myositis; l Mucosal haemorrhage and necrosis.

111

4.4.1 Statistical analysis of inter-pathologist agreement of test pathologists.

In the first analysis, the agreement solely between the four pathologists undergoing the test was examined, excluding the relevance of the reference diagnosis. Fleiss kappa was computed to assess the agreement between the four pathologists in diagnosing canine and feline bladder diseases in 25 patients via bladder tissue assessment with and without the use of a predictive probability tool (Table 4-5, Figure 4-2). The Fleiss kappa (k) measures of overall agreement were found to be 0.07 (95% CI -0.03-0.17, P = 0.16), 0.27 (95% CI 0.18-0.37, P < 0.01) and 0.381 (95% CI 0.29-0.47, P < 0.001), respectively for no animal information, signalment and history, and predictive tool conditions. According to Fleiss classification (134), the 0.071 kappa for no animal information, represents a statistically non-significant (P < 0.170) poor agreement between the four pathologists.

There was a fair to good agreement between the four pathologists in terms of rating patients as having neoplasia or cystitis (for the clinical history reading); but there was poor agreement in rating patients as having normal bladder tissue, urolithiasis and ‘other’ diagnoses. Notably, when cases were selected by the reference pathologists, all cases were either normal, urolithiasis, cystitis or neoplasia categories with no true ‘other’ diagnoses. A category for ‘other’ diagnoses was included in the logistic regression modelling (combined with normal to make the baseline category), therefore the predictive tool provided probabilities for ‘normal/other’ combined. This is why the ‘other’ diagnosis category was included in this experiment, even though there were no cases with this diagnosis based on the reference diagnosis.

For the first read with no animal information, there was good agreement for making a diagnosis of neoplasia (kappa = 0.50, P < 0.001), but poor agreement for all other diagnosis groups. For the second read where clinical history and signalment was made available, there was a fair to good agreement between the four pathologists in terms of rating patients as having cystitis (kappa = 0.50, P < 0.001) and normal bladder tissue (kappa = 0.40, P < 0.001), and excellent agreement (kappa = 0.77, P < 0.001) in rating patients as having neoplasia. There was poor agreement in rating patients as having urolithiasis or ‘other’ diagnoses.

Borderline fair to good agreement (kappa = 0.38) between the four pathologists in the diagnosis of bladder disease was found when using the predictive probabilities diagnosis tool, meaning this third read had the highest overall agreement of the three slide reading conditions. In using this tool, there was excellent agreement (kappa = 0.75, P < 0.001) in diagnosing cases as having neoplasia and a fair to good agreement (kappa = 0.45, P < 0.001) between the four pathologists rating cases as having

112

normal bladder tissue; but there was poor agreement in rating patients as having cystitis, urolithiasis, or ‘other’ diagnoses.

113

Table 4-5: Inter-pathologist agreement kappa statistics for the three conditions of diagnosing bladder syndrome in canine and feline patients.

Inter-pathologist agreement: Fleiss Kappa Statistics Overall Kappa 95% CI Detailed Kappa Kappa Z P value LCL UCL Diagnosis Kappa Z P value

0.071 1.402 0.161 -0.028 0.17 cystitis 0.01 0.118 0.906 neoplasia 0.502 6.144 <0.001 No animal normal 0.259 3.175 0.001 information other -0.02 -0.25 0.803 urolithiasis -0.157 -1.928 0.054

0.274 5.587 <0.001 0.178 0.37 cystitis 0.502 6.144 <0.001 neoplasia 0.765 9.366 <0.001 Signalment and normal 0.399 4.883 <0.001 history other -0.01 -0.124 0.902 urolithiasis 0.132 1.616 0.106

0.381 8.377 <0.001 0.292 0.47 cystitis 0.259 3.175 0.001 neoplasia 0.752 9.21 <0.001 Predictive tool normal 0.446 5.466 <0.001 probabilities other 0.054 0.666 0.505 urolithiasis 0.306 3.742 <0.001 Z = standard normal score, CL = confidence limit, LCL = lower confidence limit, UCL = upper confidence limit.

114

Figure 4-2: Bar plot showing the inter-pathologist agreement kappa statistics with 95% CI for the three conditions of diagnosing bladder syndrome in canine and feline patients

115

4.4.2 Statistical analysis of agreement between the test pathologists and the reference diagnosis.

The accuracy of a test refers to the ability of that test to give a true measure of the substrate/item being tested, with a highly accurate test having a high sensitivity and specificity (108). In general, accuracy is measured similarly to concordance or agreement, with >75% being the highest and therefore acceptable level of accuracy (134). Low accuracy (< 0.75) was observed in this study across all three slide reading conditions.

The second analysis compared the agreement of the four test pathologists to the ‘reference’ diagnosis generated by the pathologists designing the study, including the author. The agreement between pathologists’ diagnosis of the bladder diseases and the reference diagnosis was poor when no animal information was provided (kappa = 0.33, P = 0.002). A fair to good agreement was observed between the four pathologists and the reference diagnosis when signalment and clinical history was available (kappa = 0.47, P = 0 .006) and this was improved further by the predictive probability tool (kappa = 0.55, P = 0.037) (Table 4-6 and Figure 4-3).

Table 4-6: Classification accuracy and kappa statistics for pathologist agreement with the reference diagnosis for the three conditions of diagnosing bladder syndrome in canine and feline patients.

Accuracy and Kappa Statistics Accuracy (95% CI) Agreement Accuracy LCL UCL P-value Kappa P value

No animal information 0.5 0.397 0.603 <0.001 0.328 0.002

Signalment and history 0.606 0.503 0.703 <0.001 0.472 0.006

Predictive tool probabilities 0.67 0.566 0.764 <0.001 0.554 0.037 LCL = lower confidence limit, UCL = upper confidence limit.

116

Figure 4-3: Bar plot showing classification accuracy for the four pathologists’ assessment of bladder tissue in canine and feline patients against the reference diagnosis.

117

Discussion

Broadly speaking, the current roles of the veterinary pathologist are many include recognising and interpreting gross, histological and ultrastructural lesions, interpreting histochemical and immunohistochemical tests, understanding the underlying disease process and pathogenesis, correlating these results with clinical findings, and communicating these findings to the submitting veterinarian or agency (295). Histopathology forms a large part of a veterinary pathologist’s workload, however it has been widely demonstrated that pathologists fall victim to various processing factors and individual cognitive biases that can impact their diagnosis, including, just to name a few, their level of experience (429), the sample quality and processing parameters (444), and the organ system being examined and its potential multiple grading schemes (444). The goal of this study was to investigate inter-pathologist agreement when evaluating canine and feline bladder histology, and to evaluate the effect of a predictive tool on that inter-pathologist agreement. A secondary goal was to evaluate how well the test pathologists agreed with the author-generated reference diagnosis. In the first analysis, the four test pathologists were compared for their agreement with each other.

The preliminary analysis revealed minimal effect of the predictive tool on inter-pathologist agreement (how was each pathologist agreed with the other participants). Interestingly, there was a negative effect of the predictive tool on the diagnosis of cystitis, with agreement between pathologists lower for cystitis diagnosis with the predictive tool compared to the clinical history read. This could be attributed to the subjective and variable interpretation of subtle histological features, such as leukocytic infiltrates, by individual pathologists causing the predictive tool to display higher or lower probabilities for cystitis compared to another test pathologist, as well as the fact that there is frequently overlap in histological features in urinary bladder diseases, particularly between cystitis and urolithiasis. In addition, the limited number of categories for selection by the pathologists may have resulted in the study participants being forced to place a case in a category that may not have quite fit, although, some recent work in human pathology suggests that interpathologist agreement decreases with an increasing number of potential categories (375). The second analysis evaluated each pathologist compared to the reference diagnosis. Our finding of increasing agreement with the reference diagnosis when clinical history was made available, then further increasing agreement when the predictive tool was applied suggests that there could be some value in the use of such a tool, particularly once confounding variables have been resolved as discussed below. It is important to note that the overall accuracy of a test is dependent on disease prevalence, therefore further analysis of sensitivity and specificity are required to interpret the overall accuracy of histological diagnosis of

118

bladder disease, and it would be wise to repeat this study on a cross section of bladder disease cases from the general population to accurately reflect the prevalence of these diseases in the general population (9).

There was some variation within each pathologist too, for example, P3 increased the number of diagnoses of urolithiasis in conjunction with access to clinical history and signalment. It makes sense that urolithiasis should be diagnosed considering the clinical history, as urolithiasis is not typically a histological diagnosis but a clinical or post-mortem diagnosis.

The marked variation between pathologists could be explained by several factors. Firstly, the free text comments from all pathologists described a difficulty in differentiating inflammatory cell types on the digital images. This is known to be an issue with digitised slides, and can be improved by a number of factors including thinner sections (three micron sections were ideal in one study) and further optimisation of scanner settings for the tissue and staining characteristics of each individual laboratory (452).

Digitally scanned microscope slides are becoming more commonplace in veterinary pathology (13, 34, 434), however it is vital that the imaging is validated to ensure the same, if not better, diagnostic performance as glass slides (327). The College of American Pathologists Pathology and Laboratory Quality Center recommends validation on whole slide images using at least 60 routine cases, and comparing intra-observer agreement between the digital and glass slides viewed at least two weeks a part (327), which was not conducted in this study. It can be challenging to digitize slide images with enough detail for interpretation, particularly for pathologists who perform their daily duties on a light microscope. In future studies it would be useful to validate the scanned images, or at least repeat the experiment using a glass slide set as a comparison. In addition, future work could include a training period for pathologists using the software and reading the digital slides or selecting only pathologists whose day to day work is performed on digital slide images. Secondly, there was some variation in the interpretation of the instructions by some pathologists. For example, P2 never diagnosed a case with urolithiasis, even with the clinical history available which at times stated there was a urolith in the bladder, so perhaps they misunderstood or misinterpreted the instructions to make a diagnosis in conjunction with clinical history for that set of slides.

There were several slides where the pathologist altered their diagnosis once or twice throughout the experiment. This occurred most commonly with cases of cystitis or urolithiasis, and particularly in cases suspected to have feline idiopathic cystitis. It is not surprising that the histological features of

119

these lesions created some confusion as FIC is uncommonly biopsied, and typically involves a mild to moderate inflammatory response (which technically qualifies as cystitis) as well as urothelial compromise which could be interpreted as physical trauma from urolithiasis (216). There was also some discrepancy between diagnosis of cystitis or normal, with borderline poor to fair agreement for the diagnosis of normal bladder tissue. Some pathologists commented that they called a case ‘cystitis’ even though the inflammation was minimal and/or likely clinically insignificant. This could be explained by the concept of context bias - the tendency to call a sample abnormal when viewed alongside other samples with a high disease prevalence (6, 12). This finding also raises the question of what encompasses normal leukocyte numbers for urinary bladder, and there is unfortunately no clear answer at this stage. To date, there is no standard criteria for resident leukocyte numbers in veterinary histology of the bladder. In Chapter 3 we decided on normal bladder being allowed to contain up to 20 lymphocytes per low power field (100x magnification) in the absence of haemorrhage and/or oedema based on the scant studies of leukocytes in normal human bladder tissue (78, 140). Another cognitive factor that could explain the under diagnosis of normal tissues in this study is that pathologists tend to avoid extreme scoring ranges including the lowest severity grading ranged for lesions i.e. normal tissue (6, 186).

No cases with ‘other’ diagnoses were included in this study, so it was interesting to observe the frequency with which our study pathologists selected this option. It is known that when provided with an ‘intermediate’ option for tumour grading, pathologists tend to most frequently diagnose the intermediate category and avoid the extreme ranges (89), so what we see here may be a similar phenomenon. It is also possible that the diagnosis of ‘other’ by our study pathologists was made as a last resort when they didn’t think their diagnosis fitted exactly into one of the other categories, as was found in a consensus review of human breast pathology cases (11). Most of the ‘other’ diagnoses in our dataset were on cases of cystitis or normal bladder and could be attributed to the reasons described above; an unclear definition of normal bladder leukocyte numbers, an unclear definition of histological features of FIC, and avoidance of extreme scoring ranges.

One pathologist commented in the free text that they classified suspected cases of feline lower urinary tract obstruction as urolithiasis, as they deemed the mucous plugs to have a similar physiological effect, whereas other pathologists classified this type of change as cystitis. Future work using this data will include an in-depth analysis of the free text morphological descriptions and comments.

One limitation of this study was that P4 had some technical issues that unfortunately prevented their assessment of all the slides, so the MICE R package was used for imputing the missing diagnoses 120

from the 4th pathologist. There is a risk that the imputed data may not have been the same as the real data, however this is unlikely to have changed the overall agreement results.

The literature suggests that clearly defined histological criteria improves pathologist agreement regarding the assessment and interpretation of histological changes (141, 445). Our application of a predictive probability tool was the first step towards providing a standardised criterion for evaluation of bladder disease in dogs and cats, however our preliminary analysis suggests that the tool had no effect on pathologist agreement. This finding indicates that further research is required in this area, particularly to explore other histological contexts and diagnoses, as well as to explore methodological confounders in this study including the impact of reading digital scans, and potential misinterpretation of instructions.

Conclusion

In conclusion, we found good levels of agreement between four veterinary pathologists evaluating canine and feline bladder sections to diagnose bladder neoplasia, however there was poor to fair agreement for the diagnosis of cystitis, urolithiasis and normal bladder tissue. Agreement between pathologists improved when signalment and clinical history was provided, with mixed results when a predictive probability tool was added. There was a fair to good agreement for the diagnosis of neoplastic lesions. The predictive tool did prove valuable in increasing the accuracy of the pathologists’ diagnosis compared to the reference diagnosis, however further exploration of this area is warranted using a dataset more representative of the prevalence of these diseases in the general cat and dog population. Future work by the author of this study plans to repeat this experiment using the glass slides to remove the possible confounder of suboptimal digital slide quality.

121

Chapter 5 Systematic review Biomarkers of bladder pain syndrome and feline interstitial cystitis

Introduction

Feline idiopathic cystitis (FIC) is a spontaneously occurring, non-infectious bladder disease of cats that is believed to have many comparable properties to interstitial cystitis/bladder pain syndrome (IC/BPS, herein referred to as BPS) in people (174). Similar pathogenic mechanisms are proposed for FIC and BPS (174, 223, 354). There are two subtypes of BPS that have different histological features and are hypothesised to be separate diseases (195, 267). FIC is widely supported as a model for BPS, particularly non-Hunner type disease (159, 174), as naturally occurring FIC reproduces a more comparable bladder disease to BPS than any other inflammatory bladder disease model (440).

The definition of feline idiopathic cystitis (FIC) is inconsistent throughout the literature, but the most common definition of FIC is the presence of chronic, waxing and waning clinical signs of irritative voiding (pollakiuria, stranguria, dysuria) in the absence of neoplasia or bladder infection, with or without the presence of uroliths or urethral plugs (50, 214, 440). Cats with FIC have a 12.5% mortality rate that increases to 26% in cats with obstructive FIC, primarily attributable to owner-elected euthanasia following repeated obstructive or non-obstructive episodes (98, 142). In cats with acute nonobstructive episodes, clinical signs resolve within 7 days in 92% of cases (215). Recurrence has been reported in 39-65% of cats within two years of the first episode (152, 213, 278). It has been proposed that like BPS, FIC is part of a broader systemic condition involving the brain, peripheral nerves, and other organ systems (51, 52). It is thought that FIC is similar to the non-Hunner BPS (50, 271).

In people, the umbrella term interstitial cystitis/bladder pain syndrome encompass any idiopathic condition involving chronic, debilitating disease of the urinary bladder of more than six weeks duration in the absence of any other bladder pathology (99, 151, 255, 304, 417). Pain upon bladder filling is a defining feature of BPS (92, 417), along with nocturia, urgency and increased frequency (92). The prevalence of BPS as diagnosed in women is approximately 0.5%, although prevalence of symptoms may be up to 12% in women (353) and up to 0.04% in men (18, 92, 151). BPS may be associated with Hunner’s lesions - erythematous areas of the bladder mucosa lacking normal capillary structure, identified on cystoscopy (7). Many studies have suggested that the two main types of interstitial cystitis/bladder pain syndrome (BPS, with and without Hunner’s lesions) are actually separate entities (127, 151, 230, 267, 335), with Hunner-BPS more common in patients over 50 years

122

of age (252).

BPS is diagnosed based on the patient’s history and symptoms, and the use of questionnaires such as the O’Leary–Sant Interstitial Cystitis Symptom Index (ICSI), the Interstitial Cystitis Problem Index (ICPI) and the Pain Urgency and Frequency score (PUF) (87). In addition to these, other diagnostic tools for BPS include the following; physical examination, use of a frequency volume chart, urinalysis and urine culture (to exclude bacterial infection), use of urinary markers such as antiproliferative factor (APF) and the potassium sensitivity test (both controversial (183)), as well as Tamm–Horsfall protein (THP) or monocyte chemotactic protein 1 (MCP-1) and the intravesical anaesthetic challenge, urodynamic studies (controversial), and cystoscopy to rule out bladder neoplasia, with or without bladder biopsy with staining for mast cells (87, 129, 191, 334, 417). Diagnosis of FIC typically involves objective findings such as physical examination, urinalysis and urine culture, and presence of FLUTS in the absence of obvious neoplasia or bacteriuria (136). Cystoscopy is technically difficult to perform in cats due to their small size, therefore FIC is a diagnosis of exclusion utilising urinalysis and abdominal ultrasound (136).

Biomarkers are proteins or quantifications of gene expression that can provide information on the pathogenesis of a disease, and can be used as an objective marker of disease presence (443). Use of biomarkers for disease diagnosis is a popular area of research as they have the potential to enable disease diagnosis using minimally invasive techniques, such as urine or serum analysis. It is important for disease diagnosis that the desired marker is highly specific for the disease of interest (122).

Inflammation in the urinary bladder, as in any organ, involves a complex interaction of cytokines, chemokines, growth factors, neurotrophins and other compounds, forming a communication network between the immune system and the nervous system (433). Chemokines are chemotactic cytokines that attract leukocytes to sites of injury, and comprise four subfamilies – CC, CXC, CX3C and C (441). In academic writing, the chemokine and its receptor are often recorded together, as changes in one will result in the same directional change in the other, for example, C chemokine ligand 2 and its receptor would be recorded as CCL2/CCR2. Chemokines have a variety of complex roles including pro-inflammatory and anti-inflammatory actions, peripheral and central nerve sensitisation, and nociceptive/pain perception functions (146). As an example, induced cystitis in rat models increases the expression of CXCL12/CXCR4, CX3CL1/CX3CR1, CCL2/CCR2 and CXCL1 in bladder tissue, and CCL2 and CXCL1 in urine (16, 17, 422), with various chemokine elevations being found in BPS (86, 414). Cytokine receptors have also been implicated in hyperalgesia syndromes (82), while various cytokines have been found to increase during disease in rat cystitis models and BPS including 123

interleukin (IL)-6, IL-1a and IL-4 (124, 310, 377).

5.1.1 Biomarkers in BPS

There are many biomarkers that have been evaluated in BPS covering genetic and protein markers in urine, serum, tissue and faeces (44, 177). Mast cells (MC) and their products have been proposed as biomarkers of BPS (357, 417), although their role has been controversial (115, 262). Methylhistamine and histamine, markers of mast cell activation, were found to be increased in BPS urine in one study (118). Methylhistamine was again examined in urine more recently, however, and was not found to be significantly increased in the BPS group (121). Increased presence of mast cells in tissue is closely associated with decreased E-cadherin expression in urothelial cells, both of which have been observed in BPS (247). E-cadherin is a tight junction protein, as is zona occludens-1, and both are thought to be decreased in BPS (247). Significantly decreased urothelial expression of E-cadherin in patients with IC/BPS (25.1 ± 16.3), ketamine related cystitis (11.0 ± 11.3), and recurrent urinary tract infection (26.2 ± 5.0) was found compared to healthy controls (42.4 ± 16.7), patients with spinal cord injury (44.4 ± 18.8) or patients with bladder outlet obstruction (42.8 ± 14.3) (all P < 0.05) (246). Nerve growth factor (NGF) is another marker that has been investigated in BPS. NGF is elevated in tissue from BPS patients as well as other bladder diseases compared to controls (255), and also increased in BPS urine compared to controls (403).

Two urine biomarkers of interest in BPS include glycoprotein-51 (GP-51) (58) and antiproliferative factor (APF) (124, 189), as these two markers showed little overlap between disease and control groups (122). Other potential urine markers include epidermal growth factor (EGF, (124, 187)), heparin-binding epidermal growth factor-like growth factor (HB-EGF, (124, 187)), and interleukin- 6 (IL-6) (124, 443). HB-EGF plays a role in the replication of epithelial cells, and in BPS is downregulated by APF (187). EGF stimulates epithelial proliferation, and this is upregulated in BPS (187). Erickson and colleagues evaluated the presence of 14 markers in the urine of BPS patients undergoing various treatments compared to healthy controls and found that IL-6 had the highest correlation with symptom score, and APF had the least overlap between the two study groups (124). Differences were not found in the expression of glycosaminoglycans (GAGs), epitectin, hyaluronic acid, IL-8, IL-1 and nitrates and nitrites (124). A more recent study suggested that IL-8 may be involved in the pathogenesis of BPS (409). The findings of Erickson et al. 2002 need to be interpreted with caution, as BPS patients in this study were undergoing various treatments at the time of marker analysis.

124

Neurological abnormalities are one of the major hypothesised feature of BPS and FIC. Increased immunoreactivity for substance P, a neurokinin neurotransmitter and increased density of neurokinin- 1 receptors has been associated with BPS (173, 291). In conclusion, there are a number of biomarkers that could be promising for diagnosis and/or treatment of BPS, however there is likely to be a combination of biomarkers, rather than one single biomarker, that is relevant for disease diagnosis or treatment (178).

5.1.2 Biomarkers in FIC

The University of Sydney’s Online Mendelian Inheritance in Animals (OMIA) website places the feline species third with regard to the total number of inherited diseases that occur similarly in humans, behind dogs and cattle, and second in usefulness of potential human disease models (396). Various markers have been explored to provide information on the pathogenesis of feline idiopathic cystitis (FIC), including urinary fibronectin (232, 406), urinary thioredoxin (406, 407) and urinary trefoil factor 2 (TFF2) (233).

Trefoil factor 2 (TFF2) may be associated with compromised urothelial immune response and repair and has potential for use as a diagnostic biomarker for FIC (233). TFF2 was found in the urine of control cats, but not in the urine of cats with FIC (233). In the normal bladder biopsies, TFF2 was expressed in the mucosa and muscle layers, as well as immediately subjacent to the urothelium (233). In biopsies of FIC cats, the expression was markedly decreased compared to controls (233).

Other potential urine FIC markers include increased galectin-7 and decreased I-FABP (407). Tissue colocalization of thioredoxin, NF-kB p65 and p38 MAPK (all signal transduction molecules) have been evaluated in FIC. Thioredoxin and NF-kB p5 were co-expressed while co-expression of thioredoxin and p38 MAPK was observed in healthy bladder tissue but not in FIC (407). These findings suggest altered cell signalling pathways in FIC (407), however other disease groups were not included therefore the findings may represent general inflammatory changes.

Recently, serum markers for FIC were investigated; increased levels of serum interleukin 12 (IL-12), interleukin 18 (IL-18), C-X-C motif chemokine ligand 12 (CXCL12, also known as SDF-1) and fms- related tyrosine kinase 3 ligand (Flt3L) were detected in FIC cats compared to healthy controls (329). Interleukin-12 is released mainly from antigen presenting cells, plays a major role in T-helper 1 (Th1) cell responses, and was found to be increased in the urine of overactive bladder patients in one study, however it has not been studied in BPS (413). Intravesicular IL-12 has been trialled as a treatment for superficial urothelial carcinoma, and one study found that BPS-like symptoms were induced in some 125

patients receiving the treatment (437). Interleukin-18 is also a pro-inflammatory Th1 cytokine, and works in synergy with IL-12 to induce IFN-y expression in many immune cells (329, 437). CXCL12 is a chemokine that stimulates migration of immune cells expressing the CXCR4 receptor, and has also been shown to induce pain via stimulation of nociceptive neurons (311, 329). Flt3L is produced mainly by natural killer cells but is produced by many immune and non-immune cell types, and is a proinflammatory factor in rheumatoid arthritis and other immune disease (329, 342).

Another proposed pathogenic mechanism for FIC is neurogenic inflammation with increased C-fibre neuron sensitivity, and upregulation of the stress response system (SRS), indicating increased sympathetic activation (72) due to environmental stressors (439, 440). Increased immunoreactivity for substance P has been associated with FIC (54, 173). Finally, altered acetylcholine synthesis and release in the oesophageal mucosa has been identified in FIC cats, suggesting that systemic cholinergic changes may be involved in altering systemic sensory function and visceral hyperalgesia (38). A more recent found altered serotonin release and therefore acetylcholine release from the mucosa of FIC cat bladder tissue compared to healthy controls (172).

In summary, FIC and BPS patients share some similarities in biomarker findings, however a much larger number of biomarkers have been evaluated in BPS than FIC, therefore there is room for further study into this area.

5.1.3 The importance of this review

The identification of a sensitive and specific biomarker for diagnosing and differentiating Hunner’s and non-Hunner’s BPS in people would be of great use in these patients. A specific BPS biomarker would enable faster, less invasive and more definitive diagnosis and enable more accurate targeting of treatments, as well as allow more effective monitoring of response to treatment. For feline patients, a diagnostic marker for FIC would enable definitive diagnosis of this disease, leading to better prepared and more knowledgeable clinicians and pet owners, and thus better therapeutic interventions for these patients.

This review of the current proposed biomarkers for BPS and FIC aims to identify trends in current research and suggest avenues for further investigation towards the identification of a diagnostic biomarker that would improve disease BPS and FIC diagnosis and treatment, therefore improving patient welfare and reducing suffering in afflicted people and cats. Further, identification of common biomarkers between BPS and FIC would strengthen the case for FIC as a comparative model for BPS.

126

Hypothesis and objectives

It is hypothesised that at least one biomarker investigated in BPS will be specific enough for one of the subtypes (HIC or NHIC) to allow non-invasive differentiation of these conditions, and diagnosis compared to healthy controls and other urinary bladder diseases.

It is also hypothesised that the evaluation of biomarkers in both subtypes of bladder pain syndrome and feline idiopathic cystitis will consolidate FIC as a good disease model for BPS.

The objectives for this chapter are to:

1. Identify all urine and tissue protein biomarkers that are expressed differentially in BPS (HIC and NHIC) from healthy controls,

2. Identify all urine and tissue protein biomarkers that have been found to be associated with FIC compared to healthy controls or other disease states, and

3. Compare biomarker expression of HIC, NHIC and FIC.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology was used as a guideline during the research process and while writing this manuscript (238). The electronic database searches are outlined in Appendix 8: Database search strings for systematic review. The author (EJ) was involved in the database search, initial screening and secondary screening, with input from JA and RA. After data extraction and secondary screening, 10 studies were included for meta-analysis (Figure 5-1 and Figure 5-2).

5.3.1 Criteria for considering studies for this review and outcome measures

For the initial database search, the decision was made to include only peer-reviewed original research that was published, as well as review papers to include in a narrative review on the same topic. Case control, cross-sectional and prospective studies were allowed.

The selected studies were to include participants with Hunner and/or non-Hunner BPS subtypes compared to healthy controls or other disease groups, or feline idiopathic cystitis compared to a healthy cat population or cats with other bladder diseases. Studies were excluded if they did not differentiate between the two subtypes of BPS, as these two subtypes have quite a different

127

histological profile so are likely to have different biomarker profiles as well. From the initial database search, there were a relatively small number of studies evaluating BPS compared to other disease groups. There was a lot of variation among the other disease groups, hence the decision was made to include only studies that compared the disease to a normal control group population. This decision went against one of our initial study objectives, however the decision was made to strengthen the results comparing the two subtypes of BPS to each other, and to FIC.

The intervention here was the diagnostic test in the form of biomarker analysis. Intervention methods included enzyme-linked immunosorbent assay (ELISA) including multiplex immunoassay kits, as well as immunohistochemistry (IHC), PCR, light and infrared spectrophotometry, Western blot, colorimetric assays and cell culture with subsequent exposure of the cells to various urine biomarkers.

The literature search was conducted in February 2019. In order to identify relevant studies, an electronic search was conducted using databases PubMed® (with and without MESH terms to encompass a wider range of studies), Web of Science®, Scopus®, and Congress Library®. A search using Google Scholar was also undertaken, to capture any studies that were not present in the databases, particularly grey literature and conference proceedings. The selected databases were chosen for their reputation for presenting high-level scientific literature. For feline studies, two additional sources were used - CABI (Centre for Agriculture and Bioscience International) VetMed Resource, and the Veterinary Information Network (VIN). VIN was utilised to include conference proceedings describing original work as well as published research. Keywords were selected to encompass the largest possible collection of desired studies and are outlined in Appendix 8: Database search strings for systematic review.

For BPS studies, the term ‘bladder pain syndrome’ is relatively recent and is not universally used, therefore ‘interstitial cystitis’, the previous name of the disease, was also included. Likewise, for ‘feline idiopathic cystitis’ and ‘feline interstitial cystitis.’ Where possible, a NOT string was used to exclude studies whose title included rodent-related words, to exclude studies that were about rodent models. Grey literature was included in the Google Scholar and Veterinary Information Network searches to encompass some potentially unpublished findings. The literature on FIC is sparse compared to that on BPS, therefore the decision was made to include all publication years for FIC (Google Scholar yielded too many search results thus only the last ten years were included), and only the most recent ten years (2009-2018) for BPS studies.

Initial database search - bladder pain syndrome

128

Studies were included if: they pertained to biomarkers in BPS (urine, serum, stool, and/or tissue markers); the keyword ‘biomarker’ was present in the methods or results sections of the abstract, and not just in the study introduction or ‘future research into biomarkers required’; they appeared to contain original research on biomarkers; the search terms were present it the title or abstract or keywords list and there was an abstract in English available; they were published within the last ten years (included from 2009 onwards); they described original work comparing BPS patients to healthy controls or other disease groups; they were a review paper on BPS biomarkers (for the narrative part of this review).

Studies were excluded if they were pertaining to animal or inflammatory models of BPS (typically mouse or rat models), if they described a case report or case series, or if they were pertaining to overactive bladder disease (OAB), a similar but distinctly different disease to BPS.

Initial database search - feline idiopathic cystitis

Studies were included if: they were pertaining to biomarkers in FIC (urine, serum, stool, and/or tissue markers); the keyword ‘biomarker’ was present in the methods or results sections of the abstract, and not just in the study introduction or ‘future research into biomarkers required’; they appeared to contain original research on biomarkers; the search terms were present it the title or abstract or keywords list and there was an abstract in English available; they were published in any year except for Google Scholar, where the ten-year limitation was used to reduce the results to a reasonable number to evaluate); they described original work comparing FIC patients to healthy controls or other disease groups; they were a review paper on FIC biomarkers (for the narrative part of this review).

Studies were excluded if they were pertaining to animal or inflammatory models of BPS (typically mouse or rat models), if they were a case report or case series, or if the only mention of FIC was in that study’s reference list.

5.3.2 Screening of studies

Screening of eligible studies was performed in two stages: primary and secondary screening. Primary screening: Titles and abstracts were screened after the initial search to eliminate duplicates and ‘abstract only’ studies. Narrative reviews were moved into a separate folder for use in the narrative component of this review.

Secondary screening: Eligible studies were read in full to evaluate the quality of results, relevance

129

and originality of the work, and the grouping of participants. This was performed by EJ.

For inclusion in the meta-analysis, it was decided that studies had to have reported Hunner-type BPS and non-Hunner type disease separately. These two subtypes are increasingly being considered as different disease entities (157, 210); therefore it was deemed prudent for these two subtypes to be evaluated separately from one another. Studies were progressed to the meta-analyses component of this review if they: described original work about a measurable biomarker (i.e. not gene expression) and examined BPS or FIC (not rodent models or cell culture/in vitro results) and the two BPS subtypes were separated and analysed separately (not combined), and there were exact values and units reported for the results. Studies that met these criteria underwent detailed data extraction by the author.

Data extraction and management

Within each study, comparisons were recorded separately as ‘reports,’ for example, a separate row was used for NHIC compared to controls, and HIC compared to controls, as the population information was often different. The following information was extracted from each of the 48 studies: Title and author details, year of publication, type of study, substrate/sample evaluated, test method, disease group name, disease group inclusion criteria (how the study authors defined their disease group), control group population, control group inclusion criteria, source population (adult patients being seen in hospitals or clinics with various clinical signs, newly or previously diagnosed with BPS etc), method of sampling, justification of sample size, sample size, age (median, and standard deviation where reported) and sex of the disease patients, laboratory sample type, and disease duration.

For studies that did not report enough information on disease group inclusion criteria, whether or not only one or both BPS subtypes were included, or had insufficient results reporting, corresponding authors were contacted in an attempt to obtain this information. Data extraction was managed using Microsoft Excel (287) and Review Manager version 5.3 (393). In Review Manager, one comparison was used so the human and feline studies could be compared to each other, as the program does not allow direct analysis between two comparison groups.

5.3.3 Assessment of risk of bias in included studies

The evaluated biases were selected based on the review manager software bias types and included random sequence generation, selection bias, blinding of participants, blinding of outcome assessment,

130

incomplete outcome data, selective outcome reporting, and other bias (393). Other biases found included lack of age and sex matching of controls, one study in which patients had all undergone a procedure prior to sample collection that can be used as a form of treatment for BPS, and potential confounding factors in FIC studies in which most cases had urethral obstruction, a factor that may influence the disease state of the bladder.

Allocation/selection bias: Bias relating to the selection of patients and control study participants. All studies included in the meta-analysis were case control studies. These studies were deemed to have a low level of bias regarding the patient selection, so long as patient inclusion criteria were discussed.

Blinding (performance and detection bias): None of the studies specified whether the authors and personnel involved were blinded to the study groups, likely due to the nature of case control studies. All study participants underwent measurement of biomarkers. Outcomes (results of the biomarker measurements) were recorded in absolute values, hence there was no potential bias between patients and controls here (detection bias).

Attrition bias: Attrition bias was not applicable as all studies selected for meta-analysis were case- control studies.

Selective reporting (reporting bias): Subsets of data were not reported in some studies. For 16 studies it was unclear whether the diseased patients had Hunner or non-Hunner type disease. Seven studies did not report adequate results (i.e. mean/median as well as standard deviation/interquartile range). In addition, the eventual inclusion of only peer-reviewed original research in the meta-analysis brings some bias, as important results may not have been published, particularly negative results.

5.3.4 Unit of measurement issues

Unit of measurement was an issue in this meta-analysis, as there was inconsistency across studies regarding the substance being measured (urine, serum, tissue), and the units used, mainly whether the results were absolute or normalised to tissue or urine protein levels. There are conflicting theories on whether normalising to protein levels is necessary (391), with the emerging thought that biomarker standardisation to protein levels is not necessary and in fact may underestimate biomarkers in urine (346).

In some studies, different outcomes were measured in different body parts (tissue versus serum versus urine), and these measurements were deemed too different to be compared to each other. Studies were

131

only compared via meta-analysis if their measurements were on the same substrate (urine, tissue or serum). The decision was made to combine measurement units (absolute or normalised to protein levels) within a substrate, for example, measurements of urine pg/ml and urine pg/mg creatinine could be compared to each other.

5.3.5 Dealing with missing data

Where data was missing in the studies the authors were contacted. If no reply was received, the study was excluded from the analysis.

5.3.6 Data synthesis and meta-analysis

Biomarker (outcome) results were input into Review Manager 5.3 (393) as continuous data. Where studies provided only interquartile range and median, the mean and standard deviation were calculated using the Review Manager calculator. Outcomes were analysed using the inverse variance statistical method, a random effects model, and the standardised mean difference effect measure. A random effects model was chosen for its assumption that there is a distribution of effect, not just a single value at which an effect occurs (154). The standardised mean difference measure of effect was selected to account for the studies not all using identical units. Study and total confidence intervals were set to 95%. The disease and control groups were compared using the inverse variance-weighted average method (IVW), which summarizes the size of effect from different studies by calculating the weighted mean of the effect sizes using the inverse variance of the individual studies as weights (226).

The meta-analysis was performed in Review Manager 5.3, using the inverse variance weighted average test. The Z value (overall effect) summarises effect size from multiple studies. This value provides a summary of effect of the outcome (biomarker) between the disease group and the control group and was deemed significant if P ≤0.05. Complete results are in Appendix 12: Meta-analysis for all biomarkers for which a meta-analysis was possible (more than one study for that biomarker, n=5). and Appendix 13: Summary of results from the Review Manager analysis of all 45 outcomes (biomarkers). For consistency, the decision was made to perform meta-analysis only on urine samples. It was deemed inappropriate by the author of this thesis to combine tissue, serum and urine results, as the meta-analysis results could be misleading.

The eventual key objective of the meta-analysis was to identify biomarkers that are significantly associated with disease (HIC/NHIC/FIC) compared to healthy controls. A second aim was to compare FIC to BPS (HIC and NHIC), however this function was not available in the Review

132

Manager software. Heterogeneity among studies was evaluated using the Inconsistency index (I2 - statistic) (164). The I2 value describes the percentage of variation between studies that is due to heterogeneity rather than chance (164). Typically, I2 values of 25%, 50% and 75% were considered as low, medium and high heterogeneity respectively. Potential sources of heterogeneity were investigated using subgroup analyses. Heterogeneity was assessed using the I2 value across all subgroups.

The following steps were taken for each outcome. First, overall heterogeneity was evaluated. If there was significant subgroup heterogeneity (if the overall Chi-squared statistic had a P value of ≤0.05), the following steps were employed: removing the FIC subgroups (if present) to compare only human subgroups. Next, evaluate NHIC alone and HIC alone to see if this solves the heterogeneity. Next, try the different measurement groups alone (such as analysing results in urine pg/ml separately from urine pg/mg Cr).

Secondly, if significant heterogeneity could be resolved, then the overall effect measure (Z value) could be interpreted. This value was accepted as the overall measure of effect (overall significant difference between disease and control groups) if the P value was ≤0.05.

Thirdly, for each biomarker that had significant overall effect between the disease and control groups (Z value not zero, and a P value ≤0.05), a forest plot was used to graphically illustrate the standardised mean difference of each study as well as the risk of bias assessment for each study.

Results

5.4.1 Description of studies

The included studies were all case-control studies. Eight out of ten included studies used some form of ELISA. Ten studies were included for meta-analysis, seven BPS studies and three FIC studies (Table 5-1, listed in Appendix 9: List of included studies). In total, authors of 22 studies were contacted to clarify results, or clarify classification of BPS subtypes. Five authors replied and so the study was included. Two authors initially replied regarding patient information; however, they did not reply to a second email regarding incomplete results reporting, thus a total of 17 studies could not be included without clarification from the authors. Results of the database search are outlined in Figure 5-1 (BPS) and Figure 5-2 (FIC). For database search details see Appendix 8: Database search strings for systematic review.

133

Thirty-five studies were excluded from the systematic review, see Appendix 10: List of excluded studies. and Appendix 11: Characteristics of excluded studies. The main reasons for exclusion were not differentiating between the two BPS subtypes (n = 16), and inadequate reporting of results (n = 7, Table 5-2). Some studies had multiple reasons for exclusion, however the primary reason is reported in the table.

134

Figure 5-1: Flow diagram for inclusion of FIC studies. 135

Figure 5-2: Flow diagram for inclusion of BPS studies.

136

Table 5-1: Characteristics of included studies

Disease Control Study Year Substrate Test method Biomarkers group group (303) 2018 Urine ELISA kits HIC and Healthy CXCL9, CXCL10, CXCL11, TNFSF14, PGE2, IL-6, NHIC control MCP1, MIP1a, NGF, HB-EGF (423) 2018 Urine ELISA HIC and Healthy MIF NHIC control (177) 2013 Serum Immunoassay NHIC Healthy CRP, NGF, IL-1B, IL-6, TNF-a, IL-8 controls (86) 2013 Tissue Immunoassay, NHIC Healthy IL-1a, IL-3, IL-12, IL-18, IFN-a2, LIF, M-CSF, MIF, and urine multiplex analysis controls SCF, TNF-B, TRAIL, VCAM-1, ICAM-1, IL-2Ra, CTACK/CCL27, MCP-3/CCL7, SDF-1/CXCL12, CXCL1, MIG/CXCL9, B-NGF, SCGFB, HGF (414) 2012 Urine Immunoassay HIC and Healthy CXCL-1, CXCL-10, NGF, IL-6, MCP-1, RANTES, IL- NHIC controls 1ra, VEGF (245) 2009 Urine Immunoassay NHIC Healthy NGF, PGE2 controls (217) 2010 Urine ELISA NHIC Healthy NGF controls (323) 2017 Urine Other FIC Healthy Urine protein, UPC, N-acetyl-B-D-glucosaminidase index controls (329) 2018 Serum Immunoassay FIC Healthy CCL2, CCL5, CXCL1, CXCL12, CXCL8, Flt3L, GM- controls CSF, IFN-y, IL-12 (p40), IL-13, IL-18, IL-1B, IL-2, IL-4, IL-6, PDGF-BB, SCF, sFas, TNF-a (232) 2011 Urine Other FIC Healthy Fibronectin controls

137

Table 5-2: Reasons for study exclusion

Primary reason for exclusion for BPS studies Count BPS types not differentiated 16 Inadequate reporting of results (no units, only partial 7 reporting) Not actually about biomarkers (testing or genes) 3 In vitro only, not patients 2 Controls not healthy 1 Inadequate selection of control cases 1 Categorical results 1 Duplicate (was not detected in initial duplicate search) 1 Abstract only 1 Total 33

The main types of outcome measures were urine measurements in pg/ml or pg/mg of creatinine (Cr) as well as tissue protein levels at pg/mg protein, and serum pg/ml. The feline studies had more variability, with one study measuring an enzyme index, N-acetyl-β-D-glucosaminidase index, (323), and another using spectrophotometry to report pixel intensity of an extracellular matrix protein, fibronectin (407). There were 45 outcomes across five broad categories – CC chemokines, CXC chemokines, cytokines, growth factors and other miscellaneous biomarkers. Twenty-eight biomarkers had study data from humans only, 15 biomarkers had data including both cats and people, while two biomarkers only had data for cats.

Forty-one biomarkers were examined in only one or two studies, leaving only four biomarkers (9%) investigated in greater than two studies in the meta-analysis. The NGF outcome was the most frequently studied biomarker with 5 studies and 436 participants. IL-6 had 4 studies with 249 participants, CCL2 had 3 studies with 117 participants, and CXCL1 had 3 studies with 105 participants.

It was noted that results from the Corcoran 2013 study often appeared in a similar pattern. This study was the only included study to report both urine and tissue levels of the same biomarkers. For 19/23 biomarkers (83%), the urine levels were higher in control patients compared to diseased (lower urine levels in the BPS disease state), while in tissue samples the biomarker levels were higher in diseased patients compared to controls.

5.4.2 Risk of bias in included studies

Several bias types were observed in the included studies. Selected outcome reporting and incomplete

138

outcome data were the most commonly observed biases in this analysis, noted in 3/10 and 1/10 studies respectively. The bias of selective outcome reporting was the most common, for example, one study (Corcoran et. al 2013) did not report all results from the urine cytokine analysis. In addition, Furuta and colleagues’ study (2018) included BPS patients that had already undergone a hydrodistension procedure. Hydrodistension used to be a mainstay of BPS diagnosis. It has been used less commonly in the last few years for diagnosis, however it may also be used as a treatment option (453).

139

Table 5-3: Human and feline data availability by outcome (biomarker)

Outcome Outcome name Human Feline acronym data data CC chemokines CCL2/MCP-1 CC chemokine ligand 2/ Monocyte chemotactic protein 1 Yes Yes CCL27/CTACK CC chemokine ligand 27/cutaneous T-cell-attracting Yes No chemokine CCL3/MIP1a CC chemokine ligand 3/macrophage inflammatory Yes No protein 1-alpha CCL5 CC chemokine ligand 5/ Regulated on Activation, Yes Yes (RANTES) Normal T Cell Expressed and Secreted CCL7/MCP-3 CC chemokine ligand 7/Monocyte chemotactic protein 3 Yes No

CXC chemokines CXCL1 (GRO1) Chemokine (C-X-C motif) ligand 1 (previously called Yes Yes growth-regulated oncogene 1) CXCL10 Chemokine (C-X-C motif) ligand 10 Yes No IL-8/CXCL8 Interleukin 8/chemokine (C-X-C motif) ligand 8 Yes Yes MIG/CXCL9 Monokine induced by gamma interferon/chemokine (C- Yes No X-C motif) ligand 9 SDF1a/CXCL12 Stromal cell-derived factor 1/chemokine (C-X-C motif) Yes Yes ligand 12

Cytokines Flt-3L FMS-like tyrosine kinase 3 ligand Yes Yes ICAM-1 Intercellular Adhesion Molecule 1 Yes No IFN-a Interferon alpha Yes No IFN-y Interferon gamma Yes Yes IL-12 Interleukin 12 Yes No IL-12p40 Interleukin 12p40 Yes Yes IL-13 Interleukin 13 Yes Yes IL-16 Interleukin 16 Yes No IL-18 Interleukin 18 Yes Yes IL-1a Interleukin 1 Yes No IL-1B Interleukin 1 beta Yes No IL-1R Interleukin 1 receptor antagonist Yes No IL-2Ra Interleukin 2 receptor alpha Yes No IL-3 Interleukin 3 Yes No IL-4 Interleukin 4 Yes Yes IL-6 Interleukin 6 Yes Yes LIF Leukemia inhibitory factor Yes No MCSF Macrophage colony-stimulating factor Yes No 140

Outcome Outcome name Human Feline acronym data data MIF Macrophage migration-inhibitory factor Yes No SCF Stem cell factor Yes Yes TNF-a Tumour necrosis factor alpha Yes Yes TNF-B Tumour necrosis factor beta Yes No TNFSF14 Tumour necrosis factor superfamily member 14 Yes No TRAIL Tumour necrosis factor-related apoptosis-inducing ligand Yes No VCAM-1 Vascular cell adhesion protein 1 Yes No

Growth factors HB-EGF Heparin-binding epidermal growth factor-like growth Yes No factor HGF Hepatocyte growth factor Yes No NGF Nerve growth factor Yes No SCGFB Stem cell growth factor beta Yes No VEGF Vascular endothelial growth factor Yes No

Others CRP C reactive protein Yes No PGE2 Prostaglandin E2 Yes No sFas Soluble cell death receptor Yes Yes NAG N-acetyl-beta-D-glucosaminidase No Yes Fibronectin Fibronectin No Yes

5.4.3 Meta-Analysis

Only five biomarkers were measured in urine in more than one study: NGF, MIF, CCL2 (MCP-1), CXCL10, and IL-6. Of the nine feline studies included for data extraction, only four quantified biomarkers in urine. Two of these studies were included for meta-analysis, however these studies evaluated urine protein, urine protein-creatinine ratio and N-acetyl-B-D-glucosaminidase index, and urine levels of fibronectin. These biomarkers were not measured in any of the human studies; thus, it was not possible to compare feline and human results. Therefore, no feline studies could be included in the meta-analysis. The meta-analysis results are summarised in Appendix 12: Meta-analysis for all biomarkers for which a meta-analysis was possible (more than one study for that biomarker, n=5).

The meta-analysis results revealed only NGF to have a significant overall difference between disease and control patients. Of the remaining biomarkers with urine measurements from more than one study, MIF, CCL2 and CXCL10 had no significant differences overall, or within sensitivity analysis

141

of each individual BPS subtype. IL-6 did not have a significant overall effect, however when the HIC subtype was evaluated alone, there was a significant difference in expression between BPS patients and healthy controls (Figure 5-8), suggesting that there is potential for IL-6 to be a useful biomarker for Hunner-type BPS. Meta-analysis forest plots from Review Manager for these 5 biomarkers are shown below in Figure 5-3 to Figure 5-8. The risk of bias section to the right of the forest plot indicates the risk of bias as interpreted by the thesis author (EJ). Green signifies low risk, while yellow signifies potential moderate risk. None of these five biomarkers contained any studies with a bias deemed high risk.

For NGF (Figure 5-3), there was no difference between the HIC and NHIC groups (Tau = 0.01; χ2 = 6.57; P = 0.36; I2 = 9%). The overall effect for NGF between studies was significant (Z = 3.48; P < 0.001). The test for overall effect evaluates whether the effect size is zero, and it applies a value to this. The Z result here of 3.48 means the effect size is NOT zero, and the significant P value indicates that this is a statistically significant finding. This biomarker only includes urine data from BPS patients, thus its use in investigating the pathogenesis or diagnosis of FIC cannot be evaluated. It seems that NGF could be a valid biomarker to use in distinguishing BPS from healthy controls, however further work needs to be done in evaluation of NGF between the two BPS subtypes, and in differentiating the BPS subtypes from other disease states.

The meta-analysis for CCL-2 (Figure 5-4) showed a high and statistically significant level of heterogeneity between the HIC and NHIC groups (Tau = 0.32; χ2 = 11.47; P = 0.009), as the I2 value is 74% and the P value for this is ≤0.05. In addition, the overall effect (Z value) is not significant (Z = 0.62; P = 0.54). From this analysis, CCL-2 does not appear to be a useful biomarker for bladder pain syndrome.

The meta-analysis for MIF (Figure 5-5) showed a very high (I2 = 99%) and significant level of heterogeneity between the HIC and NHIC groups (Tau = 11.72; χ2 = 187.87; P ≤ 0.01). In addition, the overall effect (Z value) is not significant (Z = 1.25; P = 0.21). The forest plot shows that one study (Vera 2018) found a much higher urine level of MIF in HIC patients. However, this effect cannot be interpreted alone as this category only contains one study. It is possible that this could be a significant biomarker for HIC, but more research is needed.

The meta-analysis for CXCL10 (Figure 5-6) showed a very high (I2 = 90%) and significant level of heterogeneity between the HIC and NHIC groups (Tau = 1.13; χ2 = 28.88; P ≤ 0.01). The overall effect (Z value) is trending towards significance (Z = 1.70; P = 0.09), however cannot be accepted

142

considering the subgroup differences (heterogeneity). The high level of subgroup differences may be alleviated if more studies could be included, therefore further research is required into CXCL10 as a biomarker for BPS.

The meta-analysis for IL-6 (Figure 5-6) showed a high (I2 = 83%) and significant level of heterogeneity between the HIC and NHIC groups (Tau = 0.55; χ2 = 17.45; P ≤ 0.01). The overall effect (Z value) is not statistically significant (Z = 0.80; P = 0.42). This analysis suggests that IL-6 is not a good biomarker for HIC/NHIC when these two subtypes are combined.

Sensitivity analysis was performed as part of the meta-analysis for every biomarker, as discussed in the methods section. For IL-6, the analysis of HIC patients alone, without including NHIC patients, did result in a significant overall effect (Figure 5-8, Z = 3.76, P < 0.01). This suggests that there could be a difference between HIC patients and controls for this marker, but not between NHIC and controls. It is important to note that there is only each subgroup (HIC and NHIC) only contained one study, making this a very small sample size of two studies.

In summary, we were able to achieve only a fraction of our initial study objectives. We were able to identify only five biomarkers (out of 45) that had urine quantification data from more than one study. Due to the very low study number for each biomarker, we were not able to compare biomarker expression between HIC and NHIC. For the same reasons, we were not able to compare biomarker expression between HIC and/or NHIC, and FIC. The low number of included studies pertaining to FIC severely limited our evaluation of FIC biomarkers. In addition, we chose to exclude studies involving other disease groups to try and reduce variability between an already very heterogenous group of studies. This decision prevented us fulfilling our objective of comparing BPS and FIC disease groups to other diseases as well as healthy controls.

143

Figure 5-3: Review Manager meta-analysis of NGF urine measurements

144

Figure 5-4: CCL2 meta-analysis.

Figure 5-5: MIF meta-analysis.

145

Figure 5-6: CXCL10 meta-analysis.

Figure 5-7: IL-6 meta-analysis.

146

Figure 5-8: IL-6 HIC only (row 1.8.1 and 1.8.3)

147

Discussion

This systematic review revealed one urine biomarker, nerve growth factor, that was significantly associated with bladder pain syndrome compared to healthy controls. Additionally, IL-6 was found to be increased in Hunner-type BPS compared to controls. IL-6 is one of the more frequently studies potential biomarkers for BPS and was examined in the second highest number of studies in this analysis. Subgroup differences could be resolved in this analysis by evaluating the Hunner-type BPS subtype alone, suggesting that IL-6 could be useful in differentiating the two BPS subtypes. It is important to note that the IL-6 outcome contained only one study per subgroup, therefore this result must be interpreted considering the small sample size.

Nerve growth factor is produced by urothelial cells and smooth muscle cells (384) and can stimulate production of cytokine-like stem cell factor, a potent activator of mast cells (255). Neurotrophins such as nerve growth factor (NGF) are involved in peripheral sensitisation of nociceptors, leading to hypersensitivity of these nerve pathways (22, 114). NGF is secreted by the target organ and is then absorbed by sympathetic and sensory nerve fibres, promoting survival of the nerves and also contributing to expression of neuropeptides (253). It is important to note that statistical significance does not equate with clinical importance (33), however nerve growth factor is known to be involved in the pathogenesis of BPS as well as other inflammatory bladder diseases, thus it is likely that nerve growth factor truly is associated with the disease compared to healthy controls. This brings up one of the study objectives which we were unable to fulfil, namely including other disease groups in the meta-analysis. Of the studies that fulfilled the initial inclusion criteria (differentiated BPS subtypes, adequate results reporting), only two studies evaluated BPS compared to other diseases. The decision was made to remove the ‘other disease’ comparison as a feature of this study due to the very small case number that could be evaluated. Unfortunately, this means that this meta-analysis lost the ability to detect biomarkers that were likely specific for BPS. As such, the association of NGF with BPS compared to healthy controls does not mean that this biomarker is specific to BPS. More work is required in this area, comparing BPS to other urinary bladder diseases, as NGF may simply be elevated with bladder inflammation, and not uniquely elevated in BPS.

Another initial study objective that could not be fulfilled was the comparison of FIC biomarkers and BPS biomarkers. There were a relatively small number of studies on biomarkers in FIC, and despite many human studies evaluating urine biomarkers in BPS, very few feline studies measured biomarkers in urine. This may be due to logistical limitations in the collection of urine samples from cats. There were two included studies that did measure biomarkers in urine of FIC cats, however these 148

biomarkers were not measured in any of the human studies so comparison between BPS and FIC was impossible. If FIC is to continue as an animal model for BPS, further study into the expression of important BPS biomarkers identified by this review, NGF and IL-6, in normal and FIC affected cats is a priority.

Early on in the systematic review planning, it was decided that studies would only be included for meta-analysis if they differentiated between the two BPS subtypes, due to their differences in gross and histological lesions (195, 267), and the hypothesis that the Hunner and non-Hunner subtypes of BPS could in fact be separate diseases (230, 335). A higher than expected proportion of studies were found to have not differentiated between the subtypes, thus this inclusion parameter is highly likely to have affected the results of our meta-analysis. However, of the five biomarkers for which a meta- analysis was possible, four of these biomarkers showed significant heterogeneity between the HIC and NHIC subtypes. This finding supports the contention that the two subtypes of BPS occur via different pathophysiologic mechanisms and should not be combined in any further studies.

Going forward, it would be prudent to use a combination of cystoscopy and histology to differentiate the two BPS subtypes in order to obtain a more accurate representation of these conditions. The inclusion in this systematic review of only studies that differentiated the two BPS subtypes will have introduced a level of bias, however the benefit of separating the two subtypes was deemed by the author to be greater than the increased sample size but potential diagnostic bias of including non- differentiated studies.

It was interesting to note the inverse pattern of results displayed in the Corcoran 2013 study. For most of the biomarkers in that study, BPS patients had lower levels of the biomarker in the urine (compared to healthy controls) while simultaneously having higher tissue levels of many biomarkers in BPS compared to healthy controls. Potential reasons for BPS patients having lower urine levels but higher tissue levels than controls may be that those biomarkers are being more actively produced in the tissue but not being released into the urine. However, this conflicts with the hypothesis that the pathogenesis of BPS involves increased bladder wall permeability (92, 116, 177). Urine output was not reported in this study; however, biomarker measurements were normalised to urine protein which should have accounted for the reduced urine retention time in BPS patients with irritative voiding symptoms. Further, the biomarkers with this expression pattern may have a protective effect, leading to higher urine levels in healthy patients that may protect the bladder from BPS in some way. It is important to note that this study sample size was quite small with only ten patients in each disease group and in the control group, therefore a larger study would be required to accept these findings. 149

There are two main biases that may have affected the results of this review. Firstly, only one person (EJ) reviewed studies for inclusion in the meta-analysis. Ideally two people would have individually selected studies for inclusion. Secondly, only published literature and conference proceedings were included, resulting in a potential publication bias. Publication bias is important because positive results more likely to be published (154). For biomarker analysis, the absence of a biomarker (particularly in non-BPS bladder disease states) is very useful in evaluating biomarker specificity.

Conclusion

Nerve growth factor was revealed as being a promising biomarker for bladder pain syndrome. Ongoing research should continue to explore the pathogenesis of FIC and BPS, particularly the investigation of urine and tissue biomarkers that may elucidate disease pathogenesis as well as new methods for disease diagnosis and treatment. Discussion between lab groups will be valuable in terms of evaluating negative results and producing uniformity across biomarker measurement techniques and results reporting. Increased uniformity in these two areas would greatly enhance the ability of the systematic review process to determine suitable biomarkers for the diagnosis of BPS and FIC. Feline idiopathic cystitis appears to be a good disease model for non-Hunner BPS based on clinical presentation and histopathology, however the results obtained from this meta-analysis are insufficient to allow biomarker comparison between FIC and either BPS subtype. Thus, a priority for future research would be the investigation of potential BPS biomarkers, including nerve growth factor, in cats.

150

Chapter 6 Biomarker investigation of canine and feline bladder disease using a combined approach of polymerase chain reaction and immunohistochemistry.

Introduction/background

Biomarkers are proteins or genes that can provide information on the pathogenesis of a disease, and have potential to be used as an objective marker of disease presence (443). The biomarker of interest may increase or decrease in disease states compared to normal (19). It is important that the desired marker is specific for the disease being diagnosed (122). Biomarkers are a popular area of research as they have the potential to enable disease diagnosis using minimally invasive techniques, such as urine or serum analysis. Diagnostic biomarkers can be measured in a variety of substrates so when making comparisons it is important to compare the same substrates, for example, expression levels of a protein may be different in urine compared to bladder tissue compared to serum.

Many bladder conditions affecting cats and dogs also occur in people, including urothelial carcinomas (UC), urinary tract infection (UTI), and non-infectious cystitis. Urinary bladder neoplasms account for 1.5-2% of all canine neoplasms, while the proportionate morbidity (or prevalence) of neoplasia cases in cats is much lower (76). Escherichia coli (E. coli) is an important bladder pathogen in people, dogs and cats as there is a uropathogenic strain (UPEC) that targets the lower urinary tract. UPEC in cats and dogs shows a high rate of antimicrobial resistance (163, 299), while the canine strains have zoonotic potential, posing a human health risk (299, 456). Feline idiopathic cystitis (FIC) is a spontaneously occurring, non-infectious bladder disease of cats that is believed to have many comparable properties to bladder pain syndrome (BPS) in people, however the pathogenesis of these two diseases has not been fully elucidated. There are two subtypes of BPS that have different histological features, non-Hunner and Hunner type BPS (195, 267). Feline idiopathic cystitis is

151

widely supported as a model for BPS, particularly non-Hunner type disease (159, 174). Naturally occurring FIC reproduces the most comparable bladder disease to BPS than any other inflammatory bladder disease model (440), however many biomarkers that have been evaluated in BPS have not been evaluated in FIC or other feline bladder diseases.

Various biomarkers have been proposed as being important in the pathogenesis of bladder pain syndrome (BPS) and potentially useful for diagnosis. Many of these biomarkers are related to urothelial dysfunction and/or neurological dysfunction of the urinary bladder. It is important to note that the two BPS subtypes are frequently combined in studies, which presents a problem in terms of biomarker specificity. The literature on BPS to date has been fraught with confusion over the best way to diagnose BPS, with up to 60% of patients with clinical BPS not meeting all of the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) criteria (122). There has also been inconsistency surrounding the two subtypes of disease (122, 247). For this reason, going forward, it would be prudent to diagnose all BPS patients using cystoscopy and histopathology as per the European Society for Study of Interstitial Cystitis (ESSIC) guidelines, and stratify patients by subtype when investigating disease biomarkers. As suggested by Jhang and colleagues in 2016, a combination of several different biomarkers may be diagnostic for BPS, rather than a single biomarker alone (177).

6.1.1 Biomarkers in human bladder disease

Bladder cancer is relatively straightforward to diagnose in people with the use of ultrasound, cytology and cystoscopic biopsy, however biomarker analysis of urothelial carcinoma (UC) in people has included a plethora of genetic markers in urine, tissue and blood (not discussed here) and many protein markers for potential screening but these often have low sensitivity for detection of low grade neoplasms (28, 229, 390). Biomarkers for diagnosis or prognosticating of UC in people include the bladder tumour antigen test (270), the recently identified protein AHNAK2 (447), decreased amount or function of junction proteins such as tight junction protein 1 (408) or cadherins (279, 338), as well as uroplakin II (236) and uroplakin IIIa (388), and other cell cycle, cell signalling, DNA damage repair, and tumour cell invasion markers (45). As with any neoplasm, bladder cancer can provoke an inflammatory response (308), therefore it is important that biomarkers are specific for the neoplasm itself and not simply altered due to the inflammatory process. Diagnostic biomarker investigation is less important for urinary tract infections, as the inciting organisms can typically be identified with urinalysis and urine culture.

There has been an abundance of research into potential biomarkers for bladder pain syndrome (BPS)

152

for diagnosis and to improve the current understanding of the pathogenesis of this disease (177). This is because the disease is ‘idiopathic’ with no defined cause and a poorly defined clinical syndrome, and thus a specific biomarker might improve diagnosis or point to therapeutic strategies. Mast cells and their products have been proposed as biomarkers of BPS (357, 417), although their role has been controversial (115, 118, 121, 262). E-cadherin is a tight junction protein that appears to be decreased in BPS compared to healthy controls, and whose decrease has been associated with an increased presence of mast cells (247). In addition, Shie and Kuo found an association between decreased E- cadherin and increased BPS patient pain scores (visual analogue scale), as well as increased urothelial apoptosis and decreased cell proliferation in BPS patients (373). However, E-cadherin has been found to be decreased in other urinary bladder diseases such as urinary tract infections, putting into question its specificity for BPS (246).

Tight junction protein 1 (TJP-1, also known as zona occludens-1) is another tight junction protein, located on the cytoplasmic side of the cell membrane and belonging to the family of membrane- associated guanylate kinase like proteins (29, 247). TJP-1 is thought to be involved both in maintaining intercellular tight junctions, and in the regulation of cell growth and proliferation. TJP-1 was found to be decreased in BPS patient biopsies compared to healthy controls and patients with overactive bladder disease (247), however alterations have also been detected in UC so this protein may not be specific to BPS (408).

Nerve growth factor (NGF) is another marker that has been investigated for BPS. NGF is produced by urothelial cells and smooth muscle cells (384) and can stimulate production of cytokine-like stem cell factor, a potent activator of mast cells (255). NGF was found to be increased in tissue from BPS patients as well as other bladder diseases compared to controls (255). Neurotrophins such as nerve growth factor (NGF) are involved in peripheral sensitisation of nociceptors, leading to hypersensitivity of these nerve pathways (22, 114). NGF is secreted by the target organ and is then absorbed by sympathetic and sensory nerve fibres, promoting survival of the nerves and also contributing to expression of neuropeptides (253). NGF is elevated in tissue from BPS patients as well as other bladder diseases compared to controls (255), and also increased in BPS urine compared to controls (403). However, NGF is also elevated in overactive bladder disease (370).

Some potential urine markers for the diagnosis of BPS include epidermal growth factor (EGF, (124, 187)), heparin-binding epidermal growth factor-like growth factor (HB-EGF, (124, 187)), and interleukins 6 and 8 (IL-6 and IL-8) (124, 443). EGF is primarily produced by cells in the thick ascending limb of Henle and the distal convoluted tubule and stimulates bladder epithelial cell growth 153

(181, 190). Epidermal growth factor levels differed between the two BPS subtypes in one study (459) and were elevated in BPS compared to controls in another (187). In one study evaluating urine HB- EGF, levels were comparable in both subtypes, however the combining of both subtypes has an unknown effect (459) and HB-EGF is decreased in the serum of BPS patients (187). Interleukin-6 is a cytokine involved in innate immunity and is mainly produced by mast cells and macrophages, however its expression can also be induced in other cell types in disease states (254). IL-6 was found to be more easily detectable in patients with severe bladder inflammation compared to mild inflammation, however this sample size was small (123). IL-6 has also been associated with patient pain and nocturia scores (124, 254), and has been found to be increased in patients with BPS (124). Another interleukin, IL-8, has been reported as altered in BPS patients (409), but has also been reported as significantly different in BPS patients compared to controls (124). We hypothesise that because that an array of interleukins and other inflammatory markers become increased or decreased when there is inflammation secondary to any disease process, alterations in these targets may simply be due to an inflammatory response which may differ between individual patients. It is important that any biomarker under investigation for a particular disease is evaluated in other disease states as well as normal patients and patients with the disease of interest.

Two further urine biomarkers of particular interest in BPS are glycoprotein-51 (GP-51) (58) and antiproliferative factor (APF) (124, 189), as they seem to be more specific to BPS (122). Erickson and colleagues evaluated the presence of 14 biomarkers in the urine of BPS patients undergoing various treatments compared to healthy controls and found that APF was significantly increased in BPS patients compared to controls (124). The findings of Erickson et al. 2002 need to be interpreted with caution, however, as BPS patients in this study were undergoing various treatments at the time of marker analysis. Urine levels of APF appear to be increased in bladder pain syndrome compared to control patients with a variety of other bladder diseases (189, 459). Further, BPS urothelial cells secreted more APF than controls in vitro, and urine APF activity was decreased by HB-EGF (187). This suggests that decreased or downregulated HB-EGF may be the inciting cause of increased APF activity in the urothelial cells and urine of BPS patients. In summary, HB-EGF plays a role in the replication of epithelial cells, and in BPS is downregulated by APF (232). EGF stimulates epithelial proliferation, and this is upregulated in BPS (232).

Finally, uroplakin III (UPK3) genes were significantly upregulated in non-Hunner BPS compared to healthy control (458) as well as being associated with UC (388).

154

6.1.2 Canine and feline bladder disease biomarkers

Human bladder cancer is increasingly being evaluated in terms of genetic biomarkers and transcriptome analysis, with canine UC sharing a very similar transcriptome with patterns similar to those observed in human UC (268, 343). Six cancer metabolite biomarkers were useful to differentiate between UC and normal cases in one study - urea, choline, methyl guanidine, citrate, acetone and beta-hydroxybutyrate (460), however these biomarkers were not evaluated in other canine bladder diseases so they may not be specific for UC. The bladder tumour antigen test was a commonly used test to screen for human bladder cancer, and a canine equivalent was found to be useful in differentiating between urine from dogs with UC (of the bladder or prostate) and healthy or unhealthy dogs without bladder disease, but had low specificity (41%) for UC compared to other urinary tract disease (162). The bladder tumour antigen seemed to be elevated in inflammation which occurs with neoplasms as well as non-neoplastic disease. A more recent and more promising diagnostic biomarker for canine bladder cancer is the BRAF gene mutation (96, 289). Finally, 96 proteins in one study were differentially expressed in UC compared to urine from dogs with UTI or normal bladders (41). It appears to be difficult to find a diagnostic biomarker for canine UC that is specific to that disease, however the 96 proteins identified by Bracha and colleagues warrant further investigation in a larger population.

Feline UC occurs very rarely, thus they are not good candidates for a disease model for human UC and there have been few studies on biomarkers for this disease. In a small study of 11 cats, cyclooxygenases were associated with mean survival time in cats treated with COX inhibitors, however these enzymes are associated with general inflammation so may not be specific to UC in terms of diagnosis (39).

In cats, the primary comparative disease of interest is FIC for bladder pain syndrome (440). Various biomarkers have been explored to provide information on the pathogenesis of feline idiopathic cystitis (FIC), including urinary fibronectin (232, 406), urinary thioredoxin (406, 407) and urinary trefoil factor 2 (TFF2) (233). Fibronectin is a high molecular weight glycoprotein found in plasma, basement membranes and extracellular matrix and plays a role in cell adhesion, migration, growth and differentiation, and is present in the matrix coating the urothelium (232, 326). Lemberger and colleagues studied cats with obstructive FIC and found a high level of fibrosis in FIC tissue as well as altered fibronectin levels, thus it was not clear whether the fibronectin or the fibrosis came first in the FIC disease process (232). There were some cats with urinary tract infection and urolithiasis that also had obstruction, however these cats did not have fibronectin levels significantly different to the 155

control group, suggesting that the obstructive component is not related to the increased urine levels of fibronectin in FIC (232). The findings of this study suggest that fibronectin could be a specific marker for FIC. Fibronectin has not been published in any BPS studies to this author’s knowledge.

Trefoil factor 2 (TFF2) is a low molecular weight, protease-resistant protein that is expressed primarily in gastric mucous neck cells and other tissues, however little is known about its role in the urinary tract (130). It may be associated with impaired urothelial immune response and repair and has potential for use as a diagnostic biomarker for FIC (233). TFF2 was found in the urine of control cats, but not in FIC cat urine (233), suggesting that it may be decreased in disease. In the normal bladder biopsies TFF2 was expressed in the mucosa and muscle layers, as well as immediately subjacent to the urothelium, whilst in biopsies of FIC cats, the expression was markedly decreased compared to controls (233). This protein has not been evaluated in other disease states, however, so may not be specific to FIC.

Tissue colocalization of the signal transduction molecules thioredoxin, NF-kB p65 and p38 MAPK was examined, finding that in FIC thioredoxin and NF-kB p65 were co-expressed while co-expression of thioredoxin and p38 MAPK was observed in healthy feline bladder tissue (407). These findings suggest altered cell signalling pathways in FIC (407), however, other disease groups were not included in this analysis, thus the findings may represent general inflammatory changes. Some potential urine FIC biomarkers include galectin-7 and I-FABP (fatty acid-binding protein 2), which are associated with bladder tissue fibronectin (407).

Substance P (SP, also called neurokinin 1, NK1) is a neurokinin neurotransmitter associated with sensory nerve fibres and has been found some studies of BPS bladder tissue (61, 325). Substance P is believed to play a role in pain, anxiety, nausea, vasodilation, leucocyte infiltration and mast cell degranulation, as well as being involved in the neurogenic inflammation cascade (291, 336). The presence of substance P binding sites (tachykinin/neurokinin receptor 1) in feline bladder tissue with or without evidence of FIC has been investigated, with autoradiography revealing an increase in the expression of high affinity NK1 binding sites in FIC tissue (54). Glycosaminoglycans appear to be decreased in the urine of FIC cats compared to controls, similar to findings in BPS (53).

One of the major biomarkers that has been evaluated in tissue of patients with bladder pain syndrome but not FIC to date is nerve growth factor. Biomarkers such as E-cadherin and zona occludens-1 (tight junction protein) have been evaluated in FIC patients compared to healthy controls, however the sample size was very small with less than 10 patients, an no other disease states were evaluated (159).

156

In summary, FIC and BPS patients share some similarities in biomarker findings; however, many more biomarkers have been evaluated in BPS than FIC, therefore there is room for further study into this area. For the laboratory-based chapter in this thesis, tight junction protein-1 was chosen to evaluate in cats and dogs due to its relative specificity for bladder pain syndrome in people. E- cadherin has been evaluated in bladder tissue of cats with FIC, so this marker was also chosen to evaluate in other feline urinary bladder diseases, to evaluate its specificity for FIC.

157

Table 6-1: Potential biomarkers for canine and feline bladder disease.

Disease Biomarker Main findings Sample Population* Method Reference Canine Urea, choline, All differentiated between Urine 40 dogs with UC NMR spectroscopy (460) UC Methyl guanidine, urine from diseased and 42 healthy controls on urine citrate, acetone and normal dogs supernatant -hydroxybutyrate Bladder tumour BTA was useful at screening Urine 88 UC (162) antigen test for urinary tract disease but 82 healthy controls had low specificity for UC 71 other bladder disease compared to other urinary 28 sick dogs with non- tract disease. bladder disease BRAF gene PCR assay identified the Urine and 48 UC tissue (96, 289) mutation mutation in 83% of UC and tissue 27 PC tissue PC patients 23 UC urine 3 PC urine 37 cystitis or healthy controls FIC Fibronectin Fibronectin increased in FIC Frozen 27 obstructive FIC (7 Western blot (406) Total protein on day 0 urine provided follow up Bradford assay Thioredoxin Total protein increased in FIC samples) (total protein) (day 0) then decreased in FIC 18 healthy controls (3 months) Thioredoxin increased in FIC (day 0) GAGs Urine GAG to creatinine ratio Urine 24 FIC (active and Spectrophotometry (53) decreased in FIC inactive, male and female) 27 controls TFF2 11 protein spots different Frozen 18 FIC Electrophoresis (233) between FIC and normal. 10 urine 18 healthy controls Western blot were albumin, 1 TFF2 supernatant IHC (decreased in FIC) 158

Fibronectin Urine fibronectin increased in Urine 18 FIC with and without Electrophoresis (232) FIC Tissue obstruction Western blot Similar in 12 UTI IHC (Bouin’s) controls/UTI/urolithiasis 12 urolithiasis IHC (2 FIC, 4 controls) – less 18 healthy controls fluorescence for fibronectin in FIC than in controls Galectin-7 Galectin-7 increased in FIC Frozen Urine: 26 FIC with Western blot (407) I-FABP urine, increased signal urine urethral obstruction, IHC intensity in urothelium of FIC supernatant 20 healthy controls cases I-FABP decreased to half the Tissue: 3 obstructive FIC, physiological level. 4 controls *FIC cats were all male unless otherwise stated. BRAF the gene encoding the protein B-Raf; CK7 cytokeratin 7; COX-2 cyclooxygenase-2; CXCL12 C-X-C motif chemokine 12; FFPI formalin-fixed, paraffin-embedded; FIC feline idiopathic cystitis; Flt3L FMS-like tyrosine kinase 3 ligand; GAGs glycosaminoglycans; I-FABP fatty acid-binding protein 2; IHC immunohistochemistry; IL-12 interleukin-12; IL-18 interleukin-18; NK1R neurokinin 1 receptor; PC prostatic carcinoma; TFF2 trefoil factor 2; UC urothelial carcinoma; UPK3 uroplakin III; UTI urinary tract infection.

159

The presence of diagnostic biomarkers can be evaluated using several laboratory techniques, including polymerase chain reaction (PCR), immunohistochemistry (IHC), enzyme-linked immunosorbent assays (ELISA) and Western blotting. These techniques all have benefits and limitations; PCR is more sensitive for detection of genetic material, while immunohistochemistry is less specific but has the benefit of showing the precise cellular location of the target protein. This project was primarily histology based; therefore, immunohistochemistry was chosen as a primary tool for evaluating the presence and cellular location of potential biomarker proteins. In addition to IHC, PCR was chosen to validate the IHC findings. That is, to use primers for the genes encoding the biomarker proteins to confirm any changes in biomarkers that were observed in the IHC experiments. It was expected that PCR would also be more sensitive at detecting lower magnitude changes in the target proteins between disease groups. In veterinary research, PCR is commonly performed on formalin-fixed, paraffin-embedded tissues (436), however it is known that the formalin fixation process can reduce the quality of genetic material in the tissue due to formation of protein crosslinks during the fixation process (112, 381). Fixation in formalin particularly affects RNA, due to incomplete lysis of tissue and high levels of RNases (131, 149). Some studies found minimal effect of fixation time on RNA quality (149), however less than 48 hours in formalin is generally used as a rough guideline (30, 411), while three to six hours in formalin has been described as optimal for human biopsies (381).

6.1.3 Internal controls

The use of an internal standard is of great importance in both IHC and PCR. In IHC, control tissues are used to confirm that the antibody is attaching to the protein of interest, which should be highly or specifically expressed in that chosen control. For PCR, an internal control is used to determine gene expression levels relative to a known gene. A common internal standard used in PCR is the use of reference (or housekeeping) genes (332). The inclusion of a known amount of these reference genes enables standardisation of the PCR process and helps control for variable RNA integrity and purity, variable cDNA concentration, as well as tissue type differences (332). A literature review of feline reference genes was performed, revealing several genes that had reported reliability in feline tissue as well as cross reactivity with canine genes.

One paper evaluated ten feline reference genes and found that the optimal set of reference genes depended on the tissue being used (332). Expression of reference genes varies in different tissues (194, 332), however no literature was found on reference gene expression in urinary bladder. Thus, the ten reference genes used in Penning’s 2007 paper were selected for inclusion in this project, to 160

evaluate their expression in feline (and canine) urinary bladder and to standardise the biomarker test results (Table 6-4)(332).

6.1.4 Conclusions

In conclusion, there has been marked variation in case number, study design, samples analysed, and biomarkers measured in studies evaluating bladder diseases in humans as well as cats and dogs, therefore it is not surprising that there is marked variation in the biomarker research to date for these diseases. Many biomarker studies look at healthy controls compared to the disease of interest but do not include other bladder diseases, hence it is difficult to confirm whether results are due specifically to the disease in question, or simply due to general bladder inflammation. For this reason, we have chosen to evaluate normal feline and canine bladder tissue as well as bacterial cystitis, urolithiasis and urothelial carcinoma (UC) in both species, and suspected FIC cases.

This PhD project had access to tissue samples only, therefore urine and serum biomarkers could not be evaluated. The biomarkers chosen for investigation were tight junction protein-1, E-cadherin, fibronectin-1, uroplakin IIIa, interleukin 8, and nerve growth factor. Tight junction protein-1 was chosen because it has been evaluated in BPS, but not in FIC or other urinary bladder diseases. If this protein is specific for FIC, we would expect that it would not be associated with any canine bladder disease, or with any other feline bladder diseases. E-cadherin has been evaluated in BPS and UC in people, therefore we aim to investigate its expression in canine and feline UC, and FIC compared to normal bladders. Fibronectin-1 has been identified in FIC compared to normal bladders and other bladder diseases, so we aim to be the second group to confirm these findings in cats and compare the expression of fibronectin-1 in canine bladder diseases. Uroplakin IIIa has been associated with BPS and UC in people but has not been evaluated in the same diseases in cats and dogs. IL-8 has been evaluated in BPS but has not been investigated in feline or canine bladder diseases to date. Nerve growth factor has been identified as a promising biomarker for bladder pain syndrome but has not been investigated in FIC or other bladder diseases in cats or dogs. Due to the comparative potential of dogs and cats for UC and BPS respectively, it will be useful to investigate the above biomarkers from the human diseases in their companion animal counterparts.

In summary, the biomarkers chosen for this study on canine and feline bladder tissue samples are tight junction protein 1 (TJP-1), E-cadherin, fibronectin-1 (FN-1), uroplakin IIIa (UPK3A), interleukin 8 (IL-8) and nerve growth factor (NGF). Due to technical and optimisation issues, paired PCR and IHC experiments were attempted only for TJP-1.

161

Hypothesis and objectives

We hypothesise that archived formalin-fixed paraffin-embedded tissue blocks will be suitable for biomarker investigation using polymerase chain reaction and immunohistochemistry; and that there will be differences in biomarker expression between different canine and feline bladder diseases and normal bladder tissue. We also hypothesise that dogs and cats will show similar biomarker alterations to those found in the human bladder diseases for which these animals are used as comparative models (UC and BPS).

The objectives for this chapter are to:

1. Perform RNA extraction on formalin-fixed, paraffin-embedded tissue blocks and compare DNA quality to that from samples optimally collected into RNA later and compare PCR results to immunohistochemistry results for the same biomarkers on the same samples.

2. Investigate the expression of biomarkers that have been differentially expressed in human bladder disease across a variety of canine and feline urinary bladder diseases and normal bladder tissues.

Methods

6.3.1 Sample size

To estimate the desired sample size to have statistical power, we used an open access web page, OpenEpi sample size calculator for frequency in a population, which uses the hypothesized (or estimated) percentage frequency of the outcome factor in the population (95). The sample size calculations were done using a population size of 10000 (representing the general population), a confidence limit of 5% (0.05), and a design effect of 1.0 (for random sampling).

Tight junction protein-1 and E-cadherin were the first antibodies to be optimised for IHC, therefore these markers were used as the basis for the sample size calculation. For normal samples, the proportion expected to have immunolabelling for E-cadherin and TJP-1 is all samples, or 99%. When 99% is entered as the expected proportion, the sample size is 16 for normal bladders. For cases of bladder neoplasia there is expected to be a lower level of expression of tight junction markers, perhaps around 90% based on a study of feline mammary neoplasia (103). If the proportion of gene expression is expected to be 90%, then the required sample size would be 137 to accurately measure a true difference in gene expression. 162

The expected gene expression for these biomarkers is unknown for urolithiasis and cystitis, however the study will be underpowered as we did not have access to 137 cases of bladder neoplasia for dogs or for cats. The eventual sample size selected for this trial (50 samples across all disease groups and cats and dogs) is underpowered, and therefore this will act as a pilot study.

6.3.2 Samples

For this study, sample size was dictated by availability of samples, as well as cost of processing. Forty-six formalin-fixed, paraffin-embedded (FFPE) blocks containing feline and canine bladder tissue encompassing all bladder disease groups, and four control tissues (canine kidney and small intestine, and feline kidney and stomach) were available for the study. In addition, 10 cases were supplied by the Murdoch School of Veterinary and Life Sciences (slides only for IHC, blocks not available for PCR) and five bladder biopsies from dogs and cats were collected into RNAlater (Cat. No. AM7022, Invitrogen) and formalin for histology to provide paired samples for PCR analysis and immunohistochemistry.

Only tissue blocks that contained full thickness pieces of bladder tissue (i.e. included external muscularis) were selected for RNA extraction. After the initial IHC experiments, it was found that 14 samples had >50% of the urothelium denuded as part of the disease process/sample collection, therefore these were removed from the IHC analysis. A further 3 IHC samples were removed due to too young an age, not bladder (accidental sampling of incorrect block) and no tissue being left in the block after PCR. Following initial PCR experiments, 8 samples did not undergo reverse transcription due to reagent limitations, thus these samples were removed from the IHC experiments as they would no longer represent paired samples. The final number of samples was 55 that were intended for PCR with 47 that ended up being processed for PCR, as well as 39 IHC samples including some additional FFPE samples obtained from Murdoch University (Table 6-2).

163

Table 6-2: Samples for PCR and IHC

Species Preservation Tissue Disease Initial Final Final type sample number number number for PCR for IHC Canine FFPE Bladder Cystitis 6 4 3 UC 6 4 4 Urolithiasis 6 5 5 Normal 6 4 7 Urolithiasis and 1 1 0 UTI Kidney Normal 1 1 NA Small Normal 1 1 NA intestine RNAlater Bladder Normal 1 1 NA Urolithiasis and 1 0 NA UTI Total 29 21 19 Feline FFPE Bladder Confirmed UTI 5 5 4 Malakoplakia 0 0 1 UC 4 4 4 Urolithiasis 1 1 1 Normal 5 5 7 Suspected FIC 5 5 1 FIC/urolithiasis 1 1 1 Kidney Normal 1 1 NA Stomach Normal 1 1 NA RNAlater Bladder Normal 2 2 NA UTI + urethral 1 1 NA obstruction Total feline 26 26 20 Total canine and feline samples 55 47 39 FFPE formalin-fixed, paraffin-embedded; FIC feline idiopathic cystitis; IHC immunohistochemistry; NA not applicable; PCR polymerase chain reaction; UC urothelial carcinoma; UTI bacterial urinary tract infection.

164

6.3.3 Polymerase Chain Reaction

6.3.3.1 RNA isolation and extraction from biopsies in RNAlater

Sample preparation for RNA isolation was performed using a protocol developed by Wright (449) employing the TRIzol reagent protocol (Invitrogen, Cat. No. 15596026). Tissue biopsies were weighed and wrapped in heavy duty aluminium foil or placed in a sterile measuring cup and pulverised with a sterile metal hammer or rod. Every effort was made to prevent thawing of the tissue, by repeatedly applying liquid nitrogen to the measuring cup or foil packet. Once biopsies were pulverised, the powdered tissue was placed in a tube and stored at minus 20 degrees Celsius.

TRIzol was added to each pulverised sample. Samples were further homogenised using a 1000µL pipette tip and centrifuged at 12,000g for 10 minutes at 4oC. The supernatant was retained, and 0.2 ml chloroform was added. The tube was shaken vigorously for 15 seconds, then incubated at room temperature for 3 minutes. Samples were centrifuged at 12,000g for 15 minutes at 4oC. The supernatant was transferred to a new tube.

RNA extraction from the five RNA later samples was performed using a Qiagen RNeasy Mini Kit (Cat. No. 74104), following the manufacturer’s instructions and including the DNase step (Cat. No. 79254), as follows. The supernatant following pulverisation and homogenisation in TRIzol and chloroform was combined with one volume of 70% ethanol (as per the kit instructions, 2012, page 41), and mixed by pipetting. The sample was then centrifuged 700µL at a time on the provided RNeasy spin column placed in a 2ml collection tube for 15 seconds at 8,000g. The flow-through was discarded, and this step repeated until the entire sample had been centrifuged through the spin column. Next, the on-column DNase digestion step was performed by centrifuging 350µL RW1 buffer through the spin column for 15 seconds at 8,000g. Ten microlitres of the DNase I stock solution was mixed with 70µL buffer RDD and mixed. The DNase I incubation mix was added to the RNeasy spin column membrane and incubated on the benchtop for 15 minutes. The spin column was then washed with 350µL RW1 buffer then twice with 500µL buffer RPE, centrifuging for 15 seconds at 8,000g then the RNA was eluted by adding 50µL of RNase-free water to the spin column membrane and centrifuging for 1 minute at 14,000g. This final step yielded a total of 48µL RNA eluate.

6.3.3.2 RNA extraction from FFPE samples

Formalin-fixed, paraffin-embedded (FFPE) tissue samples ranged in size from 5x5mm to 10x20mm. This large variation in size was difficult to accommodate, as the samples with smaller area were often

165

also small in depth, rendering the cutting of many sections impossible. For the larger samples, it was deemed important to include the entire width of the section to encompass all layers of the bladder wall, and all pathological changes that were occurring within the tissue. Each tissue block was cut into 10µm sections using a microtome, and as much paraffin wax removed from around the tissue as possible. The tissue pieces were placed into a sterile 1.5ml Eppendorf tube, and 10-micron sections were cut until there was enough tissue to reach a 3mm depth of tissue in the tube when gently compacted. The microtome was cleaned with alcohol between each block to minimise cross contamination. Each blade was used for three blocks, with each block being cut with a new section of the blade.

RNA was extracted from FFPE samples using the Qiagen RNeasy FFPE extraction kit (Cat. No. 73504) and Qiagen deparaffinisation solution (Cat. No. 19093) according to kit instructions as described below. The larger volumes recommended by the manufacturer when processing >2 tissue sections per sample were used in all steps unless otherwise stated.

Deparaffinisation: 320µL deparaffinisation solution was added to each sample tube, vortexed vigorously for 20 seconds, centrifuged briefly, incubated at 56oC for 3 minutes, and allowed to cool at room temperature. 240µL of Buffer PKD (Proteinase K Digestion) was added, vortexed and centrifuged at 11,000g for 1 minute.

Tissue digestion: Proteinase K (10µL) was added to the lower colourless phase following deparaffinisation and mixed gently by pipetting. Samples were incubated at 56oC for 15 minutes followed by incubation at 80oC for 15 minutes. The lower colourless phase was transferred to a new 2ml microcentrifuge tube. Samples were incubated on ice for 3 minutes, then centrifuged at 14,100g for 15 minutes. The supernatant was retained.

DNA elimination: DNase Booster Buffer was added to each sample, equivalent to one tenth of the total sample volume (25µL) and 10µL DNase I added. Tubes were mixed by inverting, centrifuged briefly and incubated at room temperature for 15 minutes.

RNA extraction: 500µL Buffer RBC (red blood cell lysis buffer) was added to each sample, and briefly vortexed. 1000µL of 100% ethanol was added to each sample and mixed by pipetting. The sample was added in 700µL aliquots to the RNeasy MinElute spin column placed in a 2ml collection tube, and centrifuged at 8,100g for 15 seconds, with the resulting flow through discarded. Once all the sample had passed through the spin column, 500µL Buffer RPE was added and centrifuged at 8,100g for 15 seconds, and the flow through discarded. Another 500µL Buffer RPE was added and 166

tubes centrifuged at 8,100g for 2 minutes. The spin column was placed into a new 2ml collection tube and centrifuged at 14,100g for 5 minutes with the tube lids open to dry the spin column membrane. Finally, the spin column was placed into a new 1.5ml tube. 50µL of RNase-free water was added directly to the spin column membrane and the tubes centrifuged at 14,100g for 1 minute to elute the RNA, resulting in 48µL of RNA eluate. Following all RNA extractions, the RNA purity and quantity was measured using a NanodropTM ND-1000 spectrophotometer (Thermo Fisher Scientific Australia Pty Ltd).

6.3.3.3 Genomic DNA elimination and Reverse transcription

Reverse transcription (RT) was performed using the Qiagen Quantitect® Reverse Transcription Kit (Cat. No. 205311). Each genomic DNA elimination reaction contained 2µL gDNA wipeout buffer 7x, 5µL of RNA template, and 7µL RNase-free water, to make a total volume of 14µL. Each sample was mixed, centrifuged briefly and incubated for 5 minutes at 42oC, then placed on ice. Each reverse transcription reaction contained 1µL Quantiscript Reverse Transcriptase enzyme, 4µL 5x Quantiscript RT Buffer, and 1µL RT primer mix per reaction. Six microlitres of the reverse transcription reaction was added to the 14µL genomic DNA elimination reaction. Samples were mixed, centrifuged briefly, incubated at 42oC for 30 minutes. The reactions were incubated at 95oC for 3 minutes to stop the RT reaction and cDNA stored at 4oC, then at -20oC. A NanodropTM 1000 spectrophotometer (Thermo Fisher Scientific Australia Pty Ltd) was used to evaluate cDNA quantity and quality. See Appendix 16: Nanodrop results for extracted RNA and reverse transcribed cDNA.

6.3.3.4 Primer design

Primers were designed for the following sequences: nerve growth factor beta subunit (NGF), tight junction protein 1 (TJP-1), uroplakin IIIa (UPK3A), interleukin-8 (IL-8) and fibronectin-1 (FN-1) (Table 6-3). A primer for E-cadherin was unable to be formulated as no feline gene sequence was able to be found. The beta subunit for NGF was chosen to match the immunohistochemistry antibody for this biomarker. For uroplakin 3A, the gene sequence also matched unintended templates in cats, therefore there was a reasonable likelihood that these primers would cross-react with other proteins, thus it would be important to correlate any findings with the immunohistochemistry results.

For each biomarker, two primer sets were designed from known feline sequences available on the NCBI (National Center for Biotechnology Information) nucleotide database: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?CMD=search&DB=nucleotide. This resulted in the ‘original’ primers. In order to increase specificity, a ‘modified’ version of each primer was created 167

by adding at additional 10 nucleotides to the 3’ end of both the forward primer and the reverse primer. For all primer sets, the reverse primer was entered into the reverse complement bioinformatics web page https://www.bioinformatics.org/sms/rev_comp.html to obtain the complement nucleotide sequence. Primer sets were numbered based on the section of gene sequence to which the primer was designed, therefore the numbers in the primer set names are not associated with the protein (as an example, TJP2 and TJP4 are both primers for the protein TJP-1).

Even though the target proteins are canine and feline, most antibodies were raised against the human protein, therefore protein BLAST (Basic Local Alignment Search Tool) was performed on all biomarkers to determine the amount of cross reactivity in feline and canine samples. This procedure involved using the NCBI protein BLAST website, entering the human gene reference number and running a BLAST search to compare the sequence to any available feline and canine sequences (See section Canine and feline protein homology).The UPK3A antibody was raised against a bovine immunogen, however this PCR or IHC was not performed leaving areas for potential future exploration.

168

Table 6-3: Original and modified biomarker primers, designed from the National Center for Biotechnology Information nucleotide database.

Code Gene Genbank Sequence Forward primer (5’ to 3’) Reverse primer (5’ to 3’) Amp Supplied Supplied target reference number (bp) Tm fwd Tm rev NGF1_orig NGF AJ639860.1 CACCTGTGGCTGCAGAT TGTGAGTCCTGTTGAAG 72 59.8 58.3 ACTC GAGG NGF2_orig NGF XM_004001117.4 TAAAAAGCGGCGACTGC CCAAGTCCAGACCCTGA 86 60.0 57.1 GT GTA FN5_orig FN-1 XM_003991118.4 TCCACACCCCAATCTTCA CAGAAGTGCCAGGAAGC 119 59.0 59.0 CG TGA TJP2_orig TJP-1 XM_019831950.2 TCTTGGAAATGCCGCTG TCTGCAGTGGATGCCTCT 301 59.0 59.1 AGT TG TJP4_orig TJP-1 XM_019831950.2 AGAAAGGGAGCAGAGG TTCAGGGAGAACACGCC 126 59.2 59.1 ACCA ATC UPK3A5_orig UPK3A XM_003989368.4 GGTTCAAGGCCATGCTT CTGTCTGCTGGAAGGTG 285 58.1 59.1 CC GAG INTERLEU1_orig IL-8 AF158598.1 GTGGCCCACACTGTGAA CTGCACCCACTTTTGCTT 89 58.9 59.3 AAC GG INTERLEU5_orig IL-8 NM_001009281.1 GACCCCAAGCAAAAGTG ACTGCATGAAGTGCTGA 172 59.8 58.2 GGT AGTG NGF1 (mod) NGF Genbank CACCTGTGGCTGCAGAT TGTGAGTCCTGTTGAAG 92 70.8 71.6 AJ639860.1 ACTCAGGGTCTGGA GAGGCAGCACCACC NGF2 (mod) NGF XM_004001117.4 TAAAAAGCGGCGACTGC CCAAGTCCAGACCCTGA 106 73.6 69.5 GTTCACCCCGCG GTATCTGCAGCCA FN5 (mod) FN-1 XM_003991118.4 TCCACACCCCAATCTTCA CAGAAGTGCCAGGAAGC 139 69.1 66.5 CGGACCAGAGAT TGAATACCGTTTC TJP2 (mod) TJP-1 XM_019831950.2 TCTTGGAAATGCCGCTG TCTGCAGTGGATGCCTCT 321 66.7 65.6 AGTTACAAGGAGA TGTAAGATCCTA TJP4 (mod) TJP-1 XM_019831950.2 AGAAAGGGAGCAGAGG TTCAGGGAGAACACGCC 146 69.5 63.8 ACCAAGCAGCTCTT ATCATATACTAGA

169

UPK3A5 (mod) UPK3A XM_003989368.4 GGTTCAAGGCCATGCTTCCGC CTGTCTGCTGGAAGGTG 304 73.4 73.5 TCTGGGCC GAGCTCAGCGGGG INTERLEU1 IL-8 GenBank GTGGCCCACACTGTGAA CTGCACCCACTTTTGCTT 109 66.4 72.9 (mod) AF158598.1 AACTCAGAAATCA GGGGTCCAGGCA INTERLEU5 IL-8 NM_001009281.1 GACCCCAAGCAAAAGTG ACTGCATGAAGTGCTGA 192 69.7 65.8 (mod) GGTGCAGAAGGTT AGTGAATTTGCTTG Tm melting temperature.

170

6.3.3.5 Synthetic positive control gene sequences

The gene sequences from which the biomarker primers were designed were synthesised by Bioneer Pacific, based on the protein BLAST procedure as described above. These gene sequences were used as synthetic positive controls to optimise PCR conditions for the primers (Appendix 14: Biomarker control genes synthesised by Bioneer Pacific.)

6.3.3.6 Primer optimisation

6.3.3.6.1 Biomarkers

Synthetic biomarker positive control oligonucleotides were reconstituted according to manufacturer instructions and quantified using the NanodropTM 1000 spectrophotometer (Thermo Fisher Scientific Australia Pty Ltd). The gene length and concentration were entered into the on line calculation tool, https://cels.uri.edu/gsc/cndna.html to calculate the number of copies of the DNA template in the solution. Synthetic controls were serially diluted 10-fold from a copy number of 1010 to 100, where 100 represents one DNA copy. Primers were prepared as a 100µM solution (100picomoles/µL) and working 10µM and 1µM solutions prepared.

Biomarker primer optimisation was performed for both original and modified design primers for TJP- 1 number 2, and NGF beta number 1 primer sets. Bioneer’s AccuPower GreenStar qPCR premix (Cat. No. K-6200) and Bioline MyTaq DNA polymerase (Cat. No. BIO-21105) were trialled. Twenty microlitre reactions were prepared according to manufacturer instructions for each Taq master mix. Synthetic positive controls were serially diluted from 103, 102, 101, to 100 copy number and one no template reaction included in each optimisation experiment with a volume of 5µL nuclease-free water added to each reaction. Final primer concentrations were trialled between 0.1 and 1 µM. PCR tubes were then centrifuged prior to thermocycling in a Biorad T100 thermal cycler. Cycling conditions of optimisation experiments were initial denaturation at 95oC for 3 minutes; 30-45 cycles of denaturation at 95oC for 30 seconds, annealing for 30 seconds at 50-70oC, and extension for 1 minute at 72oC, before a final extension of 5 minutes at 72oC then cooling to 12oC. After 13 experiments, neither primer set could be optimised.

The final protocol for TJP-1, using the ‘TJP2_mod’ primer (Table 6-3), was as follows. Twenty microlitre reactions were prepared using the Bioline MyTaq kit, using primers at 0.5µM concentration with the addition of 5µL diluted cDNA (synthetic TJP-1 gene) at a final reaction concentration of 7,500, 750 and 75 gene copies. Cycling conditions were as follows: Denaturation at 95oC for 1

171

minute, followed by 40 cycles of denaturation at 95oC for 1 minute, annealing for 15 seconds at 60oC, and extension for 10 seconds at 72oC before cooling to 12oC.

Following the optimal protocol, PCR product (20uL) and the forward and reverse primers (10µL each) were sent to Australian Genome Research Facility St Lucia QLD for sequencing, which yielded 100% match for all variants of the TJP-1 gene, with a product length of 320bp (primer information states product size of 321bp for ‘TJP2 modified primer set’), therefore the PCR is correctly amplifying the target. Sequencing was not performed on the product from the NGF primers, as these were unable to be optimised to a point where they were working repeatably.

6.3.3.6.2 Reference genes

Reference gene primers were synthesised by Bioneer Pacific, based on the primers used in Penning’s 2007 paper (332). Primers were reconstituted to the recommended 100 pmoles/µL (100µM) concentration. Working 10 µM and 1 µM solutions were prepared.

Two genes were selected to trial first as they were the most stable in kidney samples in Penning’s 2007 paper (332) - ribosomal protein S7 (RPS7) and glyceraldehyde-3-phosphatedehydrogenase (GAPDH). After optimisation, GAPDH was working more reliably than RPS7, so was chosen to be tested on all the test samples using conventional PCR as described above for TJP-1 using 10µM concentration of GAPDH primers with the following cycling conditions: Denaturation at 95oC for 1 minute, followed by 50 cycles of denaturation at 95oC for 1 minute, annealing for 15 seconds at 60oC, and extension for 10 seconds at 72oC before cooling to 12oC.

Reference genes were first trialled on the RNAlater (optimally collected) samples. 25µL reactions were prepared using Bioline MyTaq 5x master mix and DNA polymerase (Cat. No. BIO-21105) reagent set as per the manufacturer’s instructions with a final concentration of 0.4µM of primers. The MyTaq protocol suggests using 100ng of eukaryotic template per 25µL reaction. cDNA was at a concentration of around 1500ng/µL, therefore each sample was diluted 1:50 to allow 5µL of template to deliver 150ng of template. Three subsequent 10-fold dilutions of this sample were prepared (delivering 150, 15, 1.5 and 0.15ng respectively to each reaction). PCR cycling was performed as per myTaq instructions, with annealing temperatures of 55.5oC (GAPDH) and 58oC (RPS7), 5 degrees below the melting temperatures stated on the primer manufacturer information sheets (Table 6-4). Denaturation was performed for one minute at 95oC, followed by 30 cycles of denaturation at 95oC for 15 seconds, annealing at the calculated temperature for 15 seconds and extension at 72oC for 10 seconds. 172

Table 6-4: Reference genes

o Gene Accession Forward (5’ to 3’) Reverse (5’ to 3’) Tm ( C) Amp size name* number (fwd, rev)† (bp)¥ B2M NM 001009876 TTTGTGGTCTTGGTCCTGCTCG TTCTCTGCTGGGTGACGGGA 62.0, 62.0 100 GAPDH NM 001009307 AGTATGATTCCACCCACGGCA GATCTCGCTCCTGGAAGATGGT 60.5, 60.6 101 GUSB NM 001009310 TGACATCACCATCAGCACCAGC GCCTTCCTCATCCAGAAGACGC 62.2, 62.3 114 HMBS EF453696 CAAACAGACAGTGTGGTGGCAG AGAATCTTGTCCCCTGTGGTGG 61.2, 61.0 92 HPRT EF453697 ACTGTAATGACCAGTCAACAGGGG TGTATCCAACACTTCGAGGAGTCC 61.3, 60.9 210 RPL17 AY738264 CTCTGGTCATTGAGCACATCC TCAATGTGGCAGGGAGAGC 57.8, 58.7 108 RPL30 AY700577 CCTCGGCAGATAAATTGGACTGTC TGATGGCCCTCTGGAATTTGAC 60.5, 59.6 111 RPS7 AY800278 GTCCCAGAAGCCGCACTTTGAC CTCTTGCCCACAATCTCGCTCG 63.2, 62.8 81bp RPS19 AY667161 TCATGCCCAGCCACTTTAGC GAGGTGTCAGTTTGCGTCCC 59.7, 60.3 116 YWHAZ EF458621 GAAGAGTCCTACAAAGACAGCACGC AATTTTCCCCTCCTTCTCCTGC 62.8, 59.3 115 *Abbreviations: B2M, beta-2-microglobulin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GUSB, beta-glucuronidase; HMBS, hydroxymethylbilane synthase; HPRT, hypoxanthine phosphoribosyltransferase; RPL17, ribosomal protein L17; RPL30, ribosomal protein L30; RPS7, ribosomal protein S7; RPS19, ribosomal protein S19; YWHAZ, tryptophan 5’-monooxygenase activation protein zeta isoform. o † Tm melting temperature of the gene ( C) from the Bioneer primer synthesis reports. fwd forward primer; rev reverse primer. ¥ Amplicon size derived from Penning’s paper (332).

173

6.3.3.7 Gel electrophoresis

Following conventional PCR thermocycling, PCR products were visualised using agarose gel electrophoresis. 1% agarose gels (Bioline, Cat. No. BIO-41025) were prepared in 1x sodium borate (SB) buffer, then 1µL of SYBR safe dye was added for each 20ml of SB buffer (to give SYBR concentration of 5x105). DNA ladder (100bp, Invitrogen, Cat. No. 15628019) was diluted with sterile water at 1:4 (25µL ladder + 75µL sterile water). 20µL of loading dye was added, to make a total volume of 120µL of working stock. The remaining DNA ladder was stored at -20oC. Five microlitres of sample (with loading dye) was placed into each well. Electrophoresis was performed at 90 volts (V) for one hour. Visualisation was performed using a Biorad Gel Documentation System with Bio rad Image Lab Version 3 software (36), and images saved electronically.

6.3.3.8 Real time PCR

Once the target and one reference gene were optimised using conventional PCR, real time quantitative PCR (qPCR) was performed. qPCR was firstly trialled using the GAPDH reference gene primers on the RNAlater (optimally collected) and then the FFPE samples to determine quality and quantity of RNA yield and cDNA synthesis. Those samples that demonstrated sufficient yields for detection, progressed to qPCR for the TJP-1 target primers.

Real time PCR was performed using Biorad Sso advanced SYBR green master mix (Cat. No. 1725270), made up to 20µL reactions of 10µL master mix, 0.5µL of 1µM GAPDH forward primer, 0.5µL of 1µM GAPDH reverse primer (final primer concentration of 0.025µM), 1µL template, and 8µL nuclease-free water. PCR was performed on a Biorad C1000 Touch thermocycler with CX96 real time PCR detection system, with cycling conditions as follows: Denaturation at 95oC for 30 seconds, followed by 50 cycles of denaturation at 95oC for 15 seconds and annealing at 60oC for 30 seconds. A melt curve was also performed between 65oC and 95oC, using 0.5oC increments and 5 seconds per step. Serial dilutions were performed on the RNAlater samples, with PCR being run on 1:10, 1:20 and 1:50 dilutions of the neat cDNA. 1:20 was initially chosen as this dilution would result in all reactions containing cDNA in the range of 50-100ng per reaction, as requested by the DNA polymerase instructions.

6.3.4 Immunohistochemistry

Many immunohistochemistry (IHC) antibodies are raised against the human version of the target protein, therefore it was important to run a NCBI protein BLAST in order to evaluate the homology

174

between humans and cats, and humans and dogs to see if antibodies raised against the human protein should have cross reactivity in our canine and feline samples. This process also allowed us to extrapolate how closely related the canine and feline proteins were to inform how much cross reactivity we should expect in the PCR portion of this experiment, as the primers were designed around feline gene sequences. However, PCR primers are likely to be more specific than IHC antibodies.

6.3.4.1 Samples

IHC was performed on 36 FFPE tissue sections. This 36 included the five RNAlater samples, for which tissue was also fixed in formalin but only for 24-48 hours, and also included five feline bladder wall samples obtained from Murdoch School of Veterinary and Life Sciences veterinary pathology archives during a previous experiment. Four-micron tissue sections were cut onto ionised slides, and heat fixed at 60oC for two hours, or 6 hours for repeat samples that did not stay on the slide properly during the initial IHC experiments.

6.3.4.2 Immunohistochemistry method

IHC for both TJP-1 and E-cadherin antibodies was optimised using healthy canine and feline bladder tissue. Attempts to optimise the FN-1 antibody for canine and feline samples were unsuccessful. Negative controls were omission of primary antibody (Dako antibody diluent only). Positive controls were normal canine and feline bladder tissue. The final protocol was routine deparaffinisation using a Leica ST5020 autostainer, then heat-induced epitope retrieval using the Dako PT Link (MY1903P055) at 95 degrees for 20 minutes (pH 9), followed by 5 minutes in buffer, then incubation for 1 hour at room temperature with the primary antibody (TJP-1 or E cadherin, 1:400). After one hour, slides were rinsed with buffer then underwent automatic staining with the Dako autostainer (AS5230D1704) then rehydration and cover slipping. The autostainer IHC protocol is as follows:

175

Table 6-5: Dako autostainer immunohistochemistry protocol after primary antibody incubation.

Code Reagent Volume (ml) Incubation (minutes) 1 Buffer 5 2 SM801 FLEX Peroxidase Block 200 5 3 Buffer 0 4 SM802 FLEX /HRP 200 20 5 Buffer 0 6 Buffer 5 7 SM803 FLEX DAB+ Sub-Chromo 400 10 8 Buffer 5 9 SM806 FLEX Hematoxylin 200 5 10 DI water 0 11 Buffer 5 12 DI water 0

As part of the thesis examination process, isotype negative controls were performed on a small number of slides. For TJP-1, the antibody concentration was 0.25mg/ml, therefore at a 1:400 dilution and 100µL of diluted antibody per slide, each slide was receiving 0.0625µg of antibody. The rabbit IgG was 5mg/ml, which was diluted 1:100 to obtain a 0.05µg/µL working solution. Then, 1.25µL of this working solution was added to each slide (made up to 100 µL with diluent) to obtain the same concentration of IgG. For E-cadherin, the antibody concentration was 0.5mg/ml, therefore at a 1:400 dilution and 100µL of diluted antibody per slide, each slide was receiving 0.125µg of antibody. The mouse IgG was 100µg/ml, therefore 0.125µL of this IgG was added to each slide (made up to 100 µL with diluent) to obtain the same concentration of IgG.

6.3.4.3 Immunohistochemistry scoring and statistical analysis

For immunohistochemistry, it is ideal to use as quantitative an approach as possible, which can be achieved using image analysis software. Unfortunately, current open access image analysis software can only evaluate cytoplasmic or nuclear staining and our proteins of interest are membranous proteins, therefore a manual method had to be used for the IHC scoring (421). The percentage of cells expressing positive immunolabelling was estimated and recorded. For both TJP-1 and E-cadherin, the presence or absence of membranous staining is the most valuable criterion in evaluating expression (120), therefore we focussed on this and not on the cytoplasmic staining which can be more variable. For TJP-1, high-grade tumours in people may show lower membranous staining higher cytoplasmic staining, however this has not been reported in animals thus we focussed only on

176

membranous staining.

Immunohistochemistry data was recorded, summarised, and graphically represented using Microsoft Excel (287). The statistical question to be answered from this part of the study was ‘Is the percentage of urothelial cells expressing TJP-1/E-cadherin immunolabelling associated with the disease diagnosed?’ As there were more than two disease groups to be compared and the data was not normally distributed, the Kruskal-Wallis statistical test was chosen to test whether the means were different between the different disease groups. The Kruskal-Wallis test was performed using RStudio (340). The Wilcoxon rank sum test was performed after the Kruskal-Wallis test in the case of any significant findings, to identify the disease groups responsible for the difference in means.

6.3.4.4 Immunohistochemistry validation

The goal of this study was to use PCR to validate the IHC, however due to reasons outside our control this was unfortunately not possible. Attempts were made to conduct Western blot analysis to validate the IHC antibodies, however this was interrupted by Covid-19 and was not able to be completed prior to thesis submission. The IHC does have internal controls in the form of endothelial labelling for TJP- 1, and the expected immunolabelling of urothelial cell junctions with both TJP-1 and E-cadherin.

Table 6-6: Antibodies used for immunohistochemistry

Target Type Clone Catalogue Dilution E-cadherin Mouse monoclonal 4A2C7 334000 1:400 TJP-1 Rabbit polyclonal NA 617300 1:400 NGF Rabbit polyclonal NA bs-0067R Not optimised UPK3A Mouse monoclonal AU1, batch 9272 10R-U103a Not optimised FN-1 Rabbit polyclonal NA ab2413 Not optimised IL-8 Mouse monoclonal NA* NA* NA* *An IL-8 antibody did not end up being purchased.

Results

6.4.1 Canine and feline protein homology

Protein BLASTs were performed for cats and dogs separately, to evaluate how similar the canine and feline proteins were to the human protein, as the immunohistochemistry antibodies were raised against the human proteins. The alignment boxes display all the accessions related to the protein of interest, and the top one or two have been displayed below. The E value depicts how likely the alignment is due to chance, with a typical cut-off being 1x10-6. The E values of zero indicate it is

177

extremely unlikely that the alignments are due to chance. Protein BLAST for TJP-1 (ZO-1) revealed a 92-93% alignment between human TJP-1 and feline or canine TJP-1. Protein BLAST revealed 82- 83% homology between human E-cadherin and canine or feline E-cadherin. See BLAST images in Appendix 15: Protein BLAST homology reports, displaying the top protein sequence for feline and canine TJP-1, and E-cadherin.

6.4.2 Optimising biomarker primers using synthetic controls

After all the optimisation steps outlined above, NGF was not able to be optimised.

For TJP-1, some bands were evident at the desired size of 321bp using the modified ‘TJP2’ primer (experiment 17). There is a strong band at the 103 dilution of the TJP-1 synthesised gene, then a weaker band at 102, and no band at 101. There is also a band present in the negative control (asterisk, this was a persistent problem for my initial experiments but one that was resolved with stricter pre- PCR protocols). This gel is the result of the optimal conditions outlined for TJP-1 in the methods section. The red colouration indicates a very high intensity of the PRC product in that location.

Figure 6-1: Experiment 17, optimised PCR for TJP-1, using TJP2_modified primer set, 10µM concentration.

178

6.4.3 Reference gene optimisation on sample extracts

Experiment 18 shows GAPDH (left half) and RPS7 (right half) primers, tested on 3 samples each - #51, 52 and 53. This experiment showed the detection of reference genes in the RNA later samples, in decreasing concentration with serial dilution of the sample. Thus, the samples collection and extraction are optimal. The expected 100 base pair amplicons for the GAPDH primers are visualised, however the RPS7 bands we expected to be around 80 base pairs look a little bigger. There are bands at around 40 base pairs which are suspected to be primer dimers.

Figure 6-2: Evaluation of GAPDH and RPS7 in some of the RNAlater samples (#51, 52 and 53).

Experiment 21 shows GAPDH (left half) and RPS7 (right half) primers, tested on 3 samples each - #51, 52 and 53. When testing FFPE samples, experiment 21 proved GAPDH to be more robust, with bands present in 4/7 samples, compared to 1/7 positive for RPS7 on the same samples. The last well in each set is a no template control which worked correctly (no band present).

179

Figure 6-3: Experiment 21, showing GAPDH on the left and RPS7 on the right, on the same set of samples.

Experiment 22 tested all FFPE samples for the presence of GAPDH. Small gels with 15 wells were run to produce a more detailed image than using a larger gel. The feline samples were initially run, which subsequently provided feline sample #1 to be used as a positive control in the canine PCR. Of the 50 FFPE samples tested, bands were only present in six samples – #1, 19, 31 and 39 (feline) and samples #6 and 50 (canine). At this point, the decision was made to move to real time PCR as this should be more sensitive in detecting reference gene cDNA if present.

180

Figure 6-4: Experiment 22, PCR for GAPDH on canine FFPE samples, with positive bands in the positive control well (sample #1) and sample #6 in well 8 (asterisk).

Figure 6-5: Experiment 22, PCR for GAPDH on canine FFPE samples, showing positive bands in the positive control well sample #1, and sample #50 in well 14 (asterisk).

181

* *

Figure 6-6: Experiment 22, PCR for GAPDH on feline FFPE samples, showing positive bands for samples #53 (positive control, well 2), and samples #1 and #19 (wells 3 and 14, asterisks).

* *

Figure 6-7: Experiment 22, PCR for GAPDH on feline FFPE samples, showing positive bands for samples #53 (positive control, well 2), and samples # 31 and #39 (wells 7 and 11, asterisks). 182

6.4.4 Real time PCR

Attempts to effectively purify RNA, and thus efficiently synthesise specific cDNA from the FFPE tissues, as expected was challenging and resulted in poor sensitivity in the quantitative real time PCRs. Real time PCR was performed on all samples in triplicate (experiments 28, 29 and 30). DNA was observed to be amplified only in the RNAlater samples. Positive results were present for samples 51, 53 and 54 in all three experiments. The fluorescence in all other samples up to 600 relative fluorescence units (RFU) was suggestive of primer dimer in all samples without PCR product. Any sample with fluorescence over 600 RFU was interpreted as real DNA product amplification. In addition, real DNA should begin to amplify before 30-35 cycles, therefore the samples that showed amplification at this time point were interpreted as real DNA product amplification.

Figure 6-8: Experiment 29, showing three clear amplification curves for samples 51, 53 and 54, while the remainder samples only have curves consistent with primer dimers.

183

Figure 6-9: Melt curve from experiment 28, GAPDH on all samples, showing melt peaks in samples 51, 53 and 54, with very slight but inadequate peaks in samples 52 (at 70 - d(RFU)/dT)and sample 1.

TJP-1 was evaluated on the RNAlater samples, using the ‘TJP2_modified’ primer that was optimised using conventional PCR and the synthetic gene sequence as a positive control. In experiments 25, 26 and 27, serial dilutions were run on samples 51, 53 and 54 which had previously yielded quantitative detection of the GAPDH reference gene. Samples did not show any evidence of amplifying TJP-1 (Figure 6-10). The melt curve shows a single peak corresponding to the specific amplification of positive control gene, indicating the success of the reaction.

184

Figure 6-10: Real time PCR amplification curve for experiment 27, showing a positive curve only for the well containing the synthetic control gene.

Figure 6-11: The melt peak for experiment 27, showing a single melt peak for the well containing the synthetic control gene.

In summary, the FFPE samples did not yield adequate levels of RNA to obtain readings for the reference genes, therefore these samples could not be used to evaluate the presence of our test gene, TJP-1. Three RNAlater samples did consistently yield acceptable results for the reference genes, however these samples tested negative for TJP-1.

185

6.4.5 Immunohistochemistry scoring

56 samples underwent IHC initially. Immediately, 3 samples were removed as described above (too young, not bladder, no tissue left after PCR) leaving 53. Fourteen samples were removed from the IHC due to inadequate amounts of urothelium remaining on the slide, and it was found that these 14 had >50% urothelial denudation from the initial histological analysis. Unfortunately, this included all 5 of the suspected FIC cases. This left the 39 cases in Table 6-2.

A further 3 samples were removed from analysis due to loss of urothelium during the IHC process, leaving 36 in total. Of these 36, four had loss of urothelium in the TJP-1 IHC, however the longer heat fixation time for E-cadherin resulted in intact urothelium, therefore these cases were retained for the E-cadherin results. Therefore, the final results include 32 samples of TJP-1, and 36 for E-cadherin.

Both TJP-1 and E-cadherin are located on the urothelial cell membrane, therefore the slides were scored according to the estimated percentage of urothelial cells that had positive membranous immunolabelling. ‘Positive’ immunolabelling was defined as equal or greater than the level of immunolabelling of endothelial cells for TJP-1 (internal control), and the presence of any immunolabelling for E-cadherin, above any background nonspecific staining Figure 6-12. Isotype negative controls, performed as part of the thesis examination process, showed identical staining to the omission of primary antibody controls for TJP-1 and E cadherin. The TJP-1 positive control did not show immunolabelling as strong as the original test slides, however this is hypothesised to be due to aging of both the antibodies and the tissue sections.

186

Figure 6-12: TJP-1 (left side) and E-cadherin (right side) immunohistochemistry. 1 and 2: TJP negative and positive controls, normal feline bladder. 3 and 4: E-cadherin negative and positive controls, normal feline bladder. 5 TJP-1 and 6 E-cadherin: Normal canine bladder showing faint staining (sample #41), with 30% of the cells exhibiting positive immunolabelling. 7 TJP-1 and 8 E-cadherin: Canine UC (sample #17), showing positive immunolabelling of the normal urothelium but variable immunolabelling within the neoplasm. 9: TJP-1, feline idiopathic cystitis (sample #28), showing scant urothelium remaining due to the disease process. 10: E- cadherin, feline urinary tract infection and urethral obstruction (sample #1), showing loss of urothelium during processing due to the submucosal haemorrhage. 187

Figure 6-13: TJP-1 (1-3) and E-cadherin (4-6) immunohistochemistry, normal feline bladder. 1: TJP-1 positive control. 2:TJP-1, negative control by omission of primary antibody. 3: TJP- 1, negative control using species-matched IgG. 4: E cadherin positive control. 2: E cadherin, negative control by omission of primary antibody. 3: E cadherin, negative control using species- matched IgG.

188

Table 6-7: Grading results showing percentage of urothelial cells with positive membranous staining for the tight junction protein-1 (TJP-1) and E-cadherin, by species and diagnosis.

Sample# Species Diagnosis TJP-1 E-cadherin 20 Canine Cystitis 30 5 32 Canine Cystitis 0 80 35 Canine Cystitis 30 80 6 Canine Normal 30 50 27 Canine Normal 70 80 29 Canine Normal 50 50 41 Canine Normal 60 80 52 Canine Normal 50 90 15 Canine Normal NU 100 2 Canine UC 0 100 8 Canine UC NU 80 17 Canine UC 0 30 37 Canine UC 10 10 21 Canine Urolithiasis 40 60 24 Canine urolithiasis NA 100 34 Canine Urolithiasis 30 80 45 Canine Urolithiasis 50 50 47 Canine Urolithiasis 0 20 59 Feline Neoplasia 0 30 4 Feline Normal NU 100 19 Feline Normal 60 90 31 Feline Normal 80 95 39 Feline Normal 80 80 51 Feline Normal 95 90 60 Feline Normal 0 10 7 Feline UC 0 80 12 Feline UC 0 30 14 Feline UC 0 5 22 Feline UC 0 10 61 Feline Urolithiasis 0 80 1 Feline UTI 80 100 18 Feline UTI 20 100 23 Feline UTI 10 50 62 Feline Cystitis 0 10 42 Feline FIC/Urolithiasis 0 90 58 Feline malakoplakia 0 10 *NU = no urothelium on slide post-IHC.

189

Percentage of urothelial cells with positive immunolabelling (error bars SD) 120.00

100.00

80.00

60.00

40.00 Percentage Percentage (%)

20.00

0.00 Dog Cat Dog Cat Dog Cat Dog Cat Dog Cat Dog Cat Dog Cat Dog Cat -20.00 Ecad Ecad TJP TJP Ecad Ecad TJP TJP Ecad Ecad TJP TJP Ecad Ecad TJP TJP Cys Cys Cys Cys Neo Neo Neo Neo Norm Norm Norm Norm Uro Uro Uro Uro Category by diagnosis, biomarker and species

6.4.6 Immunohistochemistry statistical analysis

The Kruskal-Wallis test was performed on the data for each biomarker, using the percentage of urothelial cells that had positive immunolabelling for that protein. For TJP-1, the mean percentage was significantly different across all disease groups (cystitis, neoplasia, urolithiasis and normal, P = 0.001). The mean percentage scores were not significantly different between dogs and cats (P = 0.33). For E cadherin, there was no significant difference in the mean scores between disease groups (P = 0.25) or species (P = 0.95). Overall, there was a statistically significant result in the means of TJP-1 across the disease groups. A post-hoc analysis (pairwise Wilcoxon rank sum test) showed that the significant overall difference could be attributed to the difference between neoplasia and normal (P < 0.05) and between cystitis and normal (P < 0.05).

190

Discussion

The primary result from this chapter is that formalin-fixed, paraffin-embedded archived tissue blocks are not a reliable source of genetic material for molecular work, however GAPDH is a reliable reference gene for feline bladder tissue in samples collected into RNAlater. Retrospective studies can be very useful for research purposes, particularly for rare or rarely sampled diseases such as FIC as they can provide a larger and more diverse collection of samples size than prospective studies. However, as we found here, the unknown and therefore assumed variability in sampling to fixation time, fixative type and time in fixative, and paraffin block processing and storage can significantly impact the quality and quantity of RNA that is able to be obtained from these specimens. Formalin is well known to be an excellent resource for preservation of tissue morphology, however its mechanism of fixation includes cross-linking of nucleic acids with proteins, modification of RNA by covalent addition of monomethylol moieties to the amino groups of bases, or methylene bridging between neighbouring bases and fragmentation of the RNA molecules (185, 392). In addition, RNA is highly prone to fragmentation, particularly if any sampling, fixation, storage or preanalytical conditions are less than ideal (234).

Literature on the use of FFPE tissues for RNA extraction has made many general recommendations in recent years, such as fixation in formalin for no more than 48 hours (67, 419), ideal nucleic acid yield within 1 year of paraffin embedding (347), and a postmortem interval of less than four hours to limit autolysis (239). Unfortunately, many of these parameters were not previously followed in veterinary pathology practice so our blocks did not undergo processing according to these recommendations, and to this day may not even be possible (such as sample collection within four hours). In addition, almost half of our blocks are ten years old or greater, therefore are likely to be more difficult in terms of extracting usable RNA of greater than 150bp (our GAPDH sequence was 101bp while the TJP2_mod primer amplicon was 321bp) (347). Even for more recently collected samples with known fixation times, the RNA extracted in our study was of inadequate quality compared to samples that were collected and stored in RNAlater. Therefore, it is recommended that archived tissue blocks with unknown sampling and fixative conditions not be used for RNA extraction and PCR unless multiple extraction attempts and optimisation can be performed. For future studies, it would be optimal to collect samples prospectively so all samples can be stored optimally, such as in RNAlater or frozen. For a rarely sampled disease such as FIC, prospective tissue sampling is unlikely to have yielded many samples due to the self-limiting nature of the disease and difficulty of obtaining cat bladder biopsies, therefore the only choice for us to explore the pathogenesis of this

191

disease with a reasonable sample size was to use archived material. A future study might consider the use of urine as the substrate of choice for feline biomarker interrogation, notwithstanding technical challenges in urine sampling in cats, however as this thesis focuses on tissue samples this was outside the scope of our study. There are a plethora of pre-PCR variables that can impact the yield of genetic material (381), and in this case further optimisation and trialling of different deparaffinisation and RNA extraction techniques may have resulted in better yields, including the use of an RNA heating step to increase RNA template length (419) or the addition of a saline washing step following sample rehydration (148). There have been reports of increased success of extracting genetic material of genes with shorter (100-200bp) amplicon lengths (3, 412), however our targets were around 100 base pairs and were still not detectable. In general, less than 48 hours fixation in formalin is recommended (30, 184, 411). Going forward, it would be worthwhile forming standard and recordable fixation procedures within institutions, where tissues are fixed in formalin for as little time as possible, or perhaps fixed both in formalin and also ethanol if practicable, and the sampling and fixation parameters can be easily made known to researchers in the case of a future retrospective study.

The PCR component of this study also showed that GAPDH is a good reference gene for urinary bladder tissue in cats and dogs as it was routinely expressed in the RNAlater samples in both conventional and real time (three out of four samples) PCR. The protein BLAST showed high gene homology between dogs and cats, and this was confirmed when our feline GAPDH gene primers worked similarly in our canine bladder samples (n = 2). It was unusual that both the GAPDH primers and the chosen TJP-1 primer set seemed to work on some samples in conventional PCR but did not seem to work in the same samples using real time PCR. These experiments were unfortunately interrupted by the Covid-19 pandemic, so it is unclear whether further optimisation experiments may have yielded positive TJP-1 results in the RNAlater samples. Only the ‘TJP2’ primer sets (both original and modified) were trialled, therefore it is possible that the ‘TJP4’ primer sets may yield better results, and it is hoped that these will be able to be trialled in the future.

The secondary goal of this study was to explore biomarker expression in canine and feline bladder diseases and compare this to the literature on biomarkers of human bladder diseases. Biomarkers are proteins (or quantifications of gene expression) that provide information on disease pathogenesis and can be used as an objective marker of disease presence; the biomarker of interest may increase (positive biomarker) or decrease (negative biomarker) in the disease of interest (443). As discussed, the PCR component of this study was unsuccessful, however we were able to obtain some results from the immunohistochemistry component. The IHC showed that TJP-1 was significantly associated

192

with the disease process. TJP-1 was decreased in neoplasia, cystitis and urolithiasis (though not statistically significant for urolithiasis) cases compared to normal bladders, suggesting that this marker is decreased in disease states. TJP-1 is known to be decreased in human bladder cancer and the loss of tight junctions contributes to the motility of neoplastic cells, causing increased cancer progression and metastasis (408). Our finding confirms that this protein is comparatively expressed in human as well as canine and feline urothelial carcinomas, further cementing the use of dogs particularly as a comparative model for human bladder cancer. The decreased immunolabelling of TJP-1 in both canine and feline cystitis cases is also comparable to the literature on TJP-1 in human bladder disease, where the protein was reported to be decreased in bladder pain syndrome compared to healthy controls and overactive bladder disease (247). Cystitis cases in this study included a combination of urinary tract infection and feline idiopathic cystitis cases, however the membranous immunolabelling was comparable between dogs and cats and between UTI and FIC, although due to loss of urothelium in the FIC cases there was only one FIC case in the final results. Based on our results, TJP-1 may simply be decreased in inflammation and may not be specific for a particular disease. Further research is needed in this area, particularly in BPS patients if this protein is to be explored as a diagnostic biomarker that would need to be specific for BPS.

E-cadherin showed a moderate but not statistically significant decrease in cases of neoplasia compared to normal bladders. This is consistent with the literature on cadherins in human cancers, where the cadherin switch phenomenon occurs where E-cadherin is replaced by abnormal expression of N-cadherin, causing dysfunction of tight junctions (46). This switching process has been associated with tumour progression and metastasis (46).

The major limitation for this study was that the FFPE samples did not yield adequate quality and/or quantity of RNA for PCR analysis. It is unusual that some samples showed slight positive results using GAPDH on conventional PCR, but none of the FFPE samples showed positive GAPDH results on the real time PCR which should be a more sensitive modality. It is known that RNA degradation levels impact reference gene expression in real time PCR (228). It is possible that this result could be due to human error, as the RNA extraction process needs to be very precise, particularly the tissue homogenisation step (228). Unknown problems may have occurred at any step in the process from initial tissue collection and fixation, to sample collection for RNA extraction, the extraction process itself, storage of samples, and reverse transcription, despite every effort being made to control all variables in the RNA extraction steps and beyond. One well researched example is the postmortem interval, for example, a postmortem interval of greater than four hours is likely to result in suboptimal

193

preservation of various urothelial cellular components (125), with ultrastructural changes detectable on scanning electron microscopy as soon as 60 seconds after the death of the animal (84). In addition, urine composition is quite dynamic and sensitive to altered homeostasis for example, a fasted patient is likely to have an altered urine pH therefore any sick animal with reduced appetite may have alterations in their urine composition, the role of which on the urothelial cells is unknown (84).

In an ideal world, further optimisation would have been performed on these samples. Future potential work includes replicating the reverse transcription step but using GAPDH and TJP-1 specific primers, to try and further amplify these genes in the test samples. Of the seven FFPE samples that showed a positive band for GAPDH in at least one conventional PCR experiment, five of these were collected by EJ during this PhD project so had a relatively short sampling to fixation interval, as well as only 24 hours in formalin, which are likely to have increased the chance of obtaining usable RNA. The other two samples were from 2008 and 2014 and the fixation time is not known, however it is possible that these samples could have had a lower fixation time as well. In addition, the post-mortem interval has some impact on RNA quality, as RNA starts degrading immediately after death and in fact the level of RNA degradation has been used to estimate post-mortem interval (119). Secondly, optimisation was not successful for the NGF primers, which unfortunately prevented assessment of this promising BPS biomarker in our samples. Future work on this protein could include further optimisation experiments, and repeating reverse transcription on the extracted RNA using the NGF primers to further amplify this gene if present, however in light of the FFPE samples not amplifying the reference genes, prospective sampling into RNAlater may be preferred for ongoing work into these biomarkers.

Some limitations for the IHC component of this study include possible recall bias during scoring as the diagnosis was hidden but the case numbers for all bladder cases were the same throughout the PhD project which involved multiple assessments of the same slides, thus the author became familiar with some of the case numbers. In addition, the selection of urothelial target proteins for investigation in bladder diseases that often involve denudation of the urothelium was found to be a flaw. Unfortunately, in some bladder diseases such as FIC and urolithiasis, the urothelium is lost as part of the disease process which severely limits the ability to evaluate target proteins that are located in urothelial cells. Future work would best be performed only on optimally collected samples so there is the best chance of an intact urothelial layer, or investigating potential non-urothelial biomarkers of bladder disease such as nerve growth factor beta in optimally collected tissue samples, urine or serum (209, 217). Further, the storage of patient and clinical data was limited for some cases, meaning that

194

urine culture results were sometimes absent. This resulted in cases with UTI confirmed by culture as well as suspected (not cultured) urinary tract infection cases both being included into the ‘cystitis’ category. In addition, it is important to note that despite using archived bladder samples from a 20- year period, this study was underpowered, and these results should be interpreted with caution. This highlights the difficulty in conducting large scale studies in veterinary medicine.

We found it difficult to achieve our goal of confirming the comparative potential of urinary bladder diseases in dogs and cats. The wide variety of diagnostic biomarkers evaluated in people and the variability of their results made it difficult to select markers that could be of use in dogs and cats. Furthermore, as most laboratory reagents are formulated for use on human samples, it was challenging to optimise these for use in companion animals.

Conclusion

In conclusion, this study was a pilot investigation into the utility of formalin-fixed, paraffin-embedded archived tissue samples for PCR, and the expression of E-cadherin and tight junction protein-1 in canine and feline bladder disease. The main finding was that FFPE tissue blocks were not a reliable source of genetic material for molecular work, likely due to variability in sample fixation and storage. Secondly, we found a significant decrease of tight junction protein-1 immunolabelling in neoplasia and cystitis cases, suggesting that this protein is decreased in disease states and may be useful in differentiating between some canine and feline urinary bladder diseases. We recommend the introduction of institutional protocols for sample fixation and processing, or at the very least, recording of these variables so they can be known for future retrospective studies.

195

Chapter 7 Discussion

The primary reason for beginning this PhD project was a noticed gap in knowledge on canine and feline urinary bladder diseases in Australia. Globally, dogs and cats commonly present to veterinary hospitals with urinary bladder disease, with almost one quarter of feline presentations to emergency veterinary clinics being attributable to urologic disorders (227). Despite their clinical importance and immense comparative potential to human diseases, there are few papers published on bladder diseases in the Australian dog and cat population. This project aimed to fill this gap, using statistical investigations into the demographics of Australian dogs and cats with bladder disease, complemented with an in-depth histological analysis of bladder disease cases, a statistical exploration of the relationship between histological features and disease diagnosis and finally the formation of a predictive tool for urinary bladder disease diagnosis which we validated with diagnostic exercises on veterinary pathologists. Finally, a laboratory-based study was undertaken to evaluate the utility of archived formalin fixed paraffin embedded tissue blocks for polymerase chain reaction (PCR), and the use of a combined approach of immunohistochemistry (IHC) and PCR to evaluate biomarker expression in canine and feline normal and diseased bladder samples. We have covered many facets of canine and feline bladder disease pathology, using an innovative approach that could also be applied to other organ systems.

Many conditions affecting feline and canine urinary bladders are paralleled in humans, including urothelial carcinoma (UC) (113, 204, 330), bacterial cystitis (244, 397), and non-infectious cystitis (interstitial cystitis/bladder pain syndrome, BPS) (216, 224). Dogs and cats provide great comparative models for many human diseases, particularly the less common diseases for which it is more difficult to obtain human tissue for research purposes such as brain cancers (174, 207). Our first primary thesis goal was to interrogate the comparative potential of canine and feline bladder diseases. We fulfilled this goal by conducting a systematic review to compare biomarker expression between human bladder pain syndrome and feline idiopathic cystitis, and a lab-based chapter to evaluate the potential use of archived formalin fixed, paraffin embedded tissue samples for biomarker investigation of canine and feline urinary bladder diseases using PCR and IHC. The systematic review was initially undertaken to evaluate studies on potential diagnostic biomarkers in urine of human bladder pain syndrome patients (BPS, encompassing both Hunner and non-Hunner subtypes), compared to studies on any potential urine biomarkers of feline idiopathic cystitis. Due to a low number of feline studies and vast variability between samples analysed in the human studies, the review was stripped back to include only BPS studies evaluating urine biomarkers, compared to healthy controls.

196

The systematic review identified several problems within the scientific literature in this area - namely a lack of consistency surrounding the diagnostic criteria for bladder pain syndrome. This is not unexpected with an idiopathic disease, but regardless presents an issue for comparison of studies, both within human medicine and comparatively to veterinary species, if the studies are not evaluating the exact same disease. This leads to the question of FIC being an appropriate model - if we do not have a clear disease definition for BPS then how can we say FIC is a good model for this disease? Which disease? The Hunner or non-Hunner type? Are there additional subtypes that have been classified under ‘interstitial cystitis/bladder pain syndrome’? As far as we know, FIC appears to be the most similar to the non-Hunner subtype of BPS (50, 174, 271), however it is clear that further research is needed in this field. It is important that research continues into the pathogenesis and diagnostic criteria of BPS as well as FIC, so we can continue to improve our knowledge of these diseases and decide whether they are in fact the same disease in different species. One of the major initial intentions of this work was to interrogate the pathogenesis of FIC for comparative purposes, however the rarity of diagnostic specimens from this condition was a barrier. This suggests that it would be valuable to pursue a collaborative multicentre approach to collect high quality prospective samples from cats fitting the FIC case definition. This would be another necessary step to facilitate the development of cats with FIC as a natural model for BPS.

Another hurdle we faced on the comparative medicine front is the lack of reagents formulated for use in canine and feline tissue experiments. It proved time consuming and only variably successful to optimise immunohistochemistry antibodies for our canine and feline samples as the antibodies are typically raised against the human antigen which may or may not be homologous with the canine or feline target. It would be helpful to continue research and development into this type of reagent, particularly for those diseases which are regarded as comparative models for human diseases.

The second overarching thesis goal was to investigate a noted lack of agreement between pathologists when formulating a diagnosis on the same tissue. One of the problems that lead to this goal was the inconsistency and poor agreement of veterinary pathologists when analysing some tissues, particularly gastrointestinal biopsies where levels of inter-observer agreement sit at around 40% (445). Logistic regression is a statistical technique which could improve pathologist concordance that has not been utilised much in veterinary histopathology to date, and is slowly becoming more widely used (144, 275). This project aimed to utilise logistic regression modelling to evaluate associations between disease diagnosis and histological features on a tissue that is straightforward to evaluate, urinary bladder, insomuch as it has a limited repertoire of morphologic responses to injury, to see if

197

the novel use of this method in veterinary histopathology would improve diagnostic concordance. Our preliminary results suggest that the predictive tool did not substantially improve the inter- pathologist agreement, however the agreement between the study pathologists was under 40% across the entire dataset, suggesting that techniques to improve pathologist concordance would indeed be highly beneficial for the speciality, and in improving diagnostic outcomes. Notably however, there were several confounding factors which might have contributed to the low concordance we observed, such as a lack of familiarity with the evaluations method workflow, and the loss of visual discrimination when reading slides off a digital scan versus a light microscope, which is still the most commonly employed method in veterinary pathology. Further exploration of these confounders is warranted, as well as exploration of other diagnostic contexts, before logistical regression is abandoned as a tool to improve concordance.

When a pathologist makes a diagnosis, they do so via incorporation of the visible microscopic changes with the clinical information about the case combined with their specialist level and experience to formulate an ordered differential list (24). Differential diagnoses are then progressively eliminated based on the probability of it occurring based on the microscopic observations and the clinical circumstances, then the diagnosis with the highest probability is then reported as the final diagnosis (24). This process occurs in a similar way to a computer or statistical program providing an answer with the highest probability based on the machine learning algorithm (374). There has been much discussion over the last decade surrounding the incorporation of artificial intelligence into the medical field, particularly in the disciplines of radiology and pathology - disciplines that extract medical information from images. There was initially (and remains) some concern about the loss of jobs in these disciplines due to increased automation and artificial intelligence. Can computers diagnose pathology by identifying the relevant histological features? Maybe. Can they combine multiple histological features and subtleties to tie in with the clinical presentation of the animal, explain the pathological process to the clinician and provide a prognosis? Not in the same way that a pathologist could. Therefore, pathologists are likely to retain their jobs, but as discussed by Jha and Topol, clinicians in pathology and radiology need to adapt to become information specialists as well as diagnosticians (176). In fact, the automation of tasks like cell counts, screening of neoplasms and evaluating mitotic rates may in fact increase the job satisfaction of veterinary pathologists. The combined ‘skills’ of pathologists and computers can outperform either one alone (1).

The human pathology field has begun to utilise artificial intelligence and deep learning software as a tool to help automate mundane tasks, streamline the accession process, and score neoplastic lesions,

198

with any questionable slides then evaluated by the pathologists (63, 266). Statistical analysis is the first step towards a more integrative, similar approach in veterinary histopathology, where the pathologist is not replaced by computers or artificial intelligence, but where the pathologist is able to use statistical and computerised methods as an adjunct to their own skills and experience in order to help standardise pathology diagnoses across the globe. In human pathology, a recent study applying deep learning computers to prostate cancer histology found new associations with prostate cancer and prognosis that humans had not found yet (455), suggesting there could be a whole new world out there for veterinary pathologists if we could only harness this technology.

Logistic regression has been widely used in human pathology to examine associations between histological features and particular prognostic factors (132, 205, 273, 277, 322), however this technique has not been widely used in veterinary histopathology to date. The results from our logistic regression modelling of histological changes showed that six variables can help to differentiate between different bladder diseases, and between diseased and normal bladders, providing a more objective way to interpret histological findings. This has exciting potential for use in other disease processes that are more difficult to diagnose. One example is inflammatory bowel disease - it can be difficult to differentiate IBD in dogs and cats from normal intestines, which are home to a variable population of leukocytic hence potentially inflammatory cells in a normal state (93). Problems arise when pathologists are faced with small biopsies that may only contain mucosa and small amounts of submucosa, and it can be virtually impossible to differentiate normal from inflamed tissue. As an example, logistic regression modelling may be able to be applied to intestinal disease in the same fashion - use a large dataset to encompass a wide variety of histological changes then use logistic regression modelling to look for associations between histological findings and the disease diagnosed. However, because inflammatory intestinal diseases are more difficult to diagnose on histology, it is important to have a thorough clinical picture and multiple specialist opinions on the histological diagnosis to make sure the logistic regression model is accurate. Once validated, this method could save pathologists a lot of time and stress by helping them to decide between diagnoses when the histological features may be subtle or equivocal. One benefit of statistical modelling is that the results can feed into the development of artificial intelligence programs that could do things like count the inflammatory cells by area and produce the likelihood of the tissue being diseased or normal. Logistic regression comes into its own in this field compared to other modelling techniques, as it allows prediction of outcome probability from a combination of continuous and discrete independent variables (424).

199

In our study, two histological variables were strongly associated with all disease groups compared to normal but were not able to differentiate between them - the presence of urothelial inflammation and having neutrophilic submucosal inflammation as opposed to mononuclear inflammation. These variables can therefore not be used to help differentiate between bladder disease states, but this type of categorisation could be a useful first step in a process-streamlining artificial intelligence program. Our results can be used to formulate a very basic decision tree as shown below. This is by no means exhaustive and other variables would need to be accounted for, and of course the binary ‘yes or no’ approach is a vast simplification of the complex biological processes that make up pathology. However, many classification solutions in pathology begin with this kind of system, then improve on it from there (206, 401).

Figure 7-1: A simple example of the type of decision tree that can be formulated by using logistic regression modelling.

In conclusion, this thesis has performed a detailed analysis of the comparative potential of canine and feline bladder diseases and has investigated the role of logistic regression modelling in improving agreement in veterinary pathology. Logistic regression modelling is a promising tool for veterinary pathology, and future work in this area includes reducing confounding variables such as whole slide image quality. Dogs and cats continue to be regarded as good comparative models for human bladder diseases, however this work highlights the importance of pinning down a universal case definition for the two subtypes of bladder pain syndrome and evaluating the two subtypes separately to inform

200

pathogenesis on these two conditions. Further, a collaborative multicentre approach would be invaluable to collect high quality prospective samples of feline idiopathic cystitis cases to allow further investigation into this disease.

201

Chapter 8 References

1. AI Achieves Near-Human Performance in Diagnosing Breast Cancer. ECN. 2016. 2. Nomina Histologica Veterinaria. International Committee on Veterinary Histological Nomenclature; 2017. 3. Abed FM, Dark MJ. Determining the utility of veterinary tissue archives for retrospective DNA analysis. PeerJ. 2016;4:e1996. 4. Abernethy MG, Rosenfeld A, White JR, Mueller MG, Lewicky-Gaupp C, Kenton K. Urinary Microbiome and Cytokine Levels in Women With Interstitial Cystitis. Obstet Gynecol. 2017;129(3):500-6. 5. Abraham SN, Miao Y. The nature of immune responses to urinary tract infections. Nature reviews Immunology. 2015;15(10):655-63. 6. Aeffner F, Wilson K, Martin N, Black J, Hendriks C, Bolon B, Rudmann D, Gianani R, Koegler S, Krueger J, Young G. The Gold Standard Paradox in Digital Image Analysis: Manual Versus Automated Scoring as Ground Truth. Archives of Pathology & Laboratory Medicine. 2017;141(9):1267-75. 7. Akiyama Y, Homma Y, Maeda D. Pathology and terminology of interstitial cystitis/bladder pain syndrome: A review. Histology and Histopathology. 2019:18028. 8. Alawneh JI, Stevenson MA, Williamson NB, Lopez-Villalobos N, Otley T. The effects of liveweight loss and milk production on the risk of lameness in a seasonally calving, pasture fed dairy herd in New Zealand. Prev Vet Med. 2014;113(1):72-9. 9. Alberg AJ, Park JW, Hager BW, Brock MV, Diener-West M. The use of "overall accuracy" to evaluate the validity of screening or diagnostic tests. Journal of general internal medicine. 2004;19(5 Pt 1):460-5. 10. Allen TC. Second opinions: pathologists' preventive medicine. Arch Pathol Lab Med. 2013;137(3):310-1. 11. Allison KH, Reisch LM, Carney PA, Weaver DL, Schnitt SJ, O'Malley FP, Geller BM, Elmore JG. Understanding diagnostic variability in breast pathology: lessons learned from an expert consensus review panel. Histopathology. 2014;65(2):240-51. 12. Allred DC, Harvey JM, Berardo M, Clark GM. Prognostic and predictive factors in breast cancer by immunohistochemical analysis. Mod Pathol. 1998;11(2):155-68. 13. American College of Veterinary Pathologists. Report of the future practice of pathology task force. 2019 September 5 2019. 14. Andersson KE, McCloskey KD. Lamina propria: the functional center of the bladder? Neurourol Urodyn. 2014;33(1):9-16. 15. Apostolidis A, Jacques TS, Freeman A, Kalsi V, Popat R, Gonzales G, Datta SN, Ghazi-Noori S, Elneil S, Dasgupta P, Fowler CJ. Histological changes in the urothelium and suburothelium of human overactive bladder following intradetrusor injections of botulinum neurotoxin type A for the treatment of neurogenic or idiopathic detrusor overactivity. Eur Urol. 2008;53(6):1245-53. 16. Arms L, Girard BM, Malley SE, Vizzard MA. Expression and function of CCL2/CCR2 in rat micturition reflexes and somatic sensitivity with urinary bladder inflammation. Am J Physiol Renal Physiol. 2013;305(1):F111-22. 17. Arms L, Girard BM, Vizzard MA. Expression and function of CXCL12/CXCR4 in rat urinary bladder with cyclophosphamide-induced cystitis. American journal of physiology Renal physiology. 2010;298(3):F589-F600. 18. Arora HC, Shoskes DA. The enigma of men with interstitial cystitis/bladder pain syndrome. Translational Andrology and Urology. 2015;4(6):668-76. 19. Arshad H, Alfonso JCL, Franke R, Michaelis K, Araujo L, Habib A, Zboromyrska Y, Lücke E, Strungaru E, Akmatov MK, Hatzikirou H, Meyer-Hermann M, Petersmann A, Nauck M, Brönstrup M, Bilitewski U, Abel L, Sievers J, Vila J, Illig T, Schreiber J, Pessler F. Decreased plasma 202

phospholipid concentrations and increased acid sphingomyelinase activity are accurate biomarkers for community-acquired pneumonia. Journal of Translational Medicine. 2019;17(1):365. 20. Bacha WJJ, Bacha LM. Urinary System. Color Atlas of Veterinary Histology. 3rd ed: John Wiley & Sons, Incorporated; 2012. p. 183-94. 21. Bailar JCI, Ederer F. Significance factors for the ratio of Poisson variable to its expectation. The Biometric Society. 1964;20:639-43. 22. Barde YA. The nerve growth factor family. Prog Growth Factor Res. 1990;2(4):237-48. 23. Barsanti J. Genitourinary Infectious. In: Greene CE, editor. Infectious Diseases of the Dog and Cat. 4th ed. St Louis, MO, USA: Elsevier; 2012. p. 1013-31. 24. Bartels PH. The diagnostic pattern in histopathology. American Journal of Clinical Pathology. 1989;91(4 Suppl 1):S7-13. 25. Bartges J, Barsanti J. Bacterial urinary tract infection in cats. In: Bonagura J, editor. Current Veterinary Therapy XIII. Philadelphia, PA: Saunders; 2000 p. 880-2. 26. Bartges JW, Finco DR, Polzin DJ, Osborne CA, Barsanti JA, Brown SA. Pathophysiology of Urethral Obstruction. Veterinary Clinics of North America: Small Animal Practice. 1996;26(2):255- 64. 27. Bartolomé N, Segarra S, Artieda M, Francino O, Sánchez E, Szczypiorska M, Casellas J, Tejedor D, Cerdeira J, Martínez A, Velasco A, Sánchez A. A Genetic Predictive Model for Canine Hip Dysplasia: Integration of Genome Wide Association Study (GWAS) and Candidate Gene Approaches. PLoS One. 2015;10(4):e0122558. 28. Batista R, Vinagre N, Meireles S, Vinagre J, Prazeres H, Leao R, Maximo V, Soares P. Biomarkers for Bladder Cancer Diagnosis and Surveillance: A Comprehensive Review. Diagnostics (Basel). 2020;10(1). 29. Bauer H, Zweimueller-Mayer J, Steinbacher P, Lametschwandtner A, Bauer HC. The dual role of zonula occludens (ZO) proteins. J Biomed Biotechnol. 2010;2010:402593. 30. Ben-Ezra J, Johnson DA, Rossi J, Cook N, Wu A. Effect of fixation on the amplification of nucleic acids from paraffin-embedded material by the polymerase chain reaction. J Histochem Cytochem. 1991;39(3):351-4. 31. Benzimra C, Job C, Pascal Q, Bureau S, Combes A, Bongrand Y, M RF. Malakoplakia of the Bladder in a 4-Month-Old Puppy. J Am Anim Hosp Assoc. 2019;55(5):261-5. 32. Berent A. Cystourethroscopy in the cat: What do you need? When do you need it? How do you do it? J Feline Med Surg. 2014;16(1):34-41. 33. Berman NG, Parker RA. Meta-analysis: neither quick nor easy. BMC medical research methodology. 2002;2. 34. Bertram CA, Gurtner C, Dettwiler M, Kershaw O, Dietert K, Pieper L, Pischon H, Gruber AD, Klopfleisch R. Validation of Digital Microscopy Compared With Light Microscopy for the Diagnosis of Canine Cutaneous Tumors. Veterinary Pathology. 2018;55(4):490-500. 35. Bhatti IP, Lohano HD, Pirzado ZA, Jafri IA. A Logistic Regression Analysis of the Ischemic Heart Disease Risk. Journal of Applied Sciences. 2006;6:785-8. 36. Biorad Laboratories Inc. Biorad Image Lab. 3 ed2010. 37. Birder LA. Urinary bladder, cystitis and nerve/urothelial interactions. Autonomic Neuroscience. 2014;182:89-94. 38. Birder LA, Hanna-Mitchell AT, Mayer E, Buffington CA. Cystitis, co-morbid disorders and associated epithelial dysfunction. Neurourology and Urodynamics. 2011;30(5):668-72. 39. Bommer NX, Hayes AM, Scase TJ, Gunn-Moore DA. Clinical features, survival times and COX-1 and COX-2 expression in cats with transitional cell carcinoma of the urinary bladder treated with meloxicam. J Feline Med Surg. 2012;14(8):527-33. 40. Bowles M. Stalking stones: An overview of canine and feline urolithiasis. Veterinary Medicine. 2008;103(10).

203

41. Bracha S, McNamara M, Hilgart I, Milovancev M, Medlock J, Goodall C, Wickramasekara S, Maier CS. A multiplex biomarker approach for the diagnosis of transitional cell carcinoma from canine urine. Analytical Biochemistry. 2014;455:41-7. 42. Bradley MS, Burke EE, Grenier C, Amundsen CL, Murphy SK, Siddiqui NY. A genome-scale DNA methylation study in women with interstitial cystitis/bladder pain syndrome. Neurourology and Urodynamics. 2018;37(4):1485-93. 43. Brannick EM, Zhang J, Zhang X, Stromberg PC. Influence of submission form characteristics on clinical information received in biopsy accession. Journal of Veterinary Diagnostic Investigation. 2012;24(6):1073-82. 44. Braundmeier-Fleming A, Russell NT, Yang W, Nas MY, Yaggie RE, Berry M, Bachrach L, Flury SC, Marko DS, Bushell CB, Welge ME, White BA, Schaeffer AJ, Klumpp DJ. Stool-based biomarkers of interstitial cystitis/bladder pain syndrome. Sci Rep. 2016;6:26083. 45. Bruchbacher A, Soria F, Hassler M, Shariat SF, D’Andrea D. Tissue biomarkers in nonmuscle- invasive bladder cancer: any role in clinical practice? Curr Opin Urol. 2018;28(6):584-90. 46. Bryan RT. Cell adhesion and urothelial bladder cancer: the role of cadherin switching and related phenomena. Philosophical transactions of the Royal Society of London Series B, Biological sciences. 2015;370(1661):20140042. 47. Budreckis DM, Byrne BA, Pollard RE, Rebhun RB, Rodriguez CO, Jr., Skorupski KA. Bacterial urinary tract infections associated with transitional cell carcinoma in dogs. J Vet Intern Med. 2015;29(3):828-33. 48. Buffington CA. Developmental influences on medically unexplained symptoms. Psychotherapy and Psychosomatics. 2009;78(3):139-44. 49. Buffington CA, Chew DJ, Kendall MS, Scrivani PV, Thompson SB, Blaisdell JL, Woodworth BE. Clinical evaluation of cats with nonobstructive urinary tract diseases. Journal of the American Veterinary Medical Association. 1997;210(1):46-50. 50. Buffington CA, Chew DJ, Woodworth BE. Feline interstitial cystitis. Journal of the American Veterinary Medical Assocation. 1999;215(5):682-7. 51. Buffington CAT. Comorbidity of Interstitial Cystitis with Other Unexplained Clinical Conditions. The Journal of Urology. 2004;172(4):1242-8. 52. Buffington CAT. Pandora syndrome: Rethinking our approach to idiopathic cystitis in cats: Veterinary Medicine dvm360.com; 2011. 53. Buffington CAT, Blaisdell JL, Binns SP, Woodworth BE. Decreased urine glycosaminoglycan excretion in cats with interstitial cystitis. J Urol. 1996;155:1801-4. 54. Buffington CAT, Wolfe SA. High affinity binding sites for [3H] substance P in urinary bladders of cats with interstitial cystitis. Journal of Urology. 1998;160(2):605-11. 55. Buffington CT. Idiopathic cystitis in domestic cats - beyond the urinary tract. Journal of Veterinary Internal Medicine. 2011;25:784-96. 56. Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source Code Biol Med. 2008;3:17. 57. Burton EN, Cohn LA, Reinero CN, Rindt H, Moore SG, Ericsson AC. Characterization of the urinary microbiome in healthy dogs. PLoS One. 2017;12(5):e0177783. 58. Byrne DS, Sedor JF, Estojak J, Fitzpatrick KJ, Chiura AN, Mulholland SG. The urinary glycoprotein GP51 as a clinical marker for interstitial cystitis. J Urol. 1999;161:1786-90. 59. Byron JK. Urinary Tract Infection. Veterinary Clinics of North America: Small Animal Practice. 2019;49(2):211-21. 60. Cai T, Mazzoli S, Mondaini N, Meacci F, Nesi G, D'Elia C, Malossini G, Boddi V, Bartolettie R. The role of asymptomatic bacteriuria in young women with recurrent urinary tract infections: To treat or not to treat? Clinical Infectious Diseases. 2012;55(6):771-7. 61. Caito M. Neuron-specific enolase and Substance P immunoreactivity in urinary bladder of cats with idiopathic cystitis: The Ohio State University; 1995. 204

62. Cameron ME, Casey RA, Bradshaw JWS, Waran NK, Gunn-Moore DA. A study of environmental and behavioural factors that may be associated with feline idiopathic cystitis. Journal of Small Animal Practice. 2004;45:144-7. 63. Campanella G, Hanna MG, Geneslaw L, Miraflor A, Werneck Krauss Silva V, Busam KJ, Brogi E, Reuter VE, Klimstra DS, Fuchs TJ. Clinical-grade computational pathology using weakly supervised deep learning on whole slide images. Nat Med. 2019;25(8):1301-9. 64. Cannon CM, Allstadt SD. Lower urinary tract cancer. Veterinary Clinical North American Small Animal Practice. 2015;45(4):807-24. 65. Carstensen B, Plummer M, Laara E, Hills M. Epi: A Package for Statistical Analysis in Epidemiology. R package version 2.32 ed2018. 66. Casey HW, Ayers KM, Robinson FR. The Urinary System. In: Benirschke K, Garner FM, Jones TC, editors. Pathology of Laboratory Animals. New York: Springer-Verlag; 1978. 67. Castiglione F, Rossi Degl'Innocenti D, Taddei A, Garbini F, Buccoliero AM, Raspollini MR, Pepi M, Paglierani M, Asirelli G, Freschi G, Bechi P, Taddei GL. Real-time PCR analysis of RNA extracted from formalin-fixed and paraffin-embeded tissues: effects of the fixation on outcome reliability. Appl Immunohistochem Mol Morphol. 2007;15(3):338-42. 68. Cheng L, Lopez-Beltran A, Bostwick DG. Bladder Cancer: General Features. In: Cheng L, Lopez-Beltran A, Bostwick DG, editors. Bladder Pathology: John Wiley & Sons, Inc.; 2012. p. 137- 59. 69. Cheng L, Lopez-Beltran A, Bostwick DG. Inflammatory and infectious conditions. In: Cheng L, Lopez-Beltran A, Bostwick DG, editors. Bladder Pathology. 1st ed: John Wiley & Sons, Inc.; 2012. 70. Cheng L, Lopez-Beltran A, Bostwick DG. Normal anatomy and histology. Bladder Pathology. 1st ed. New Jersey: John Wiley & Sons Inc.; 2012. 71. Cheng L, MacLennan GT, Lopez-Beltran A. Histologic grading of urothelial carcinoma: a reappraisal. Human Pathology. 2012;43(12):2097-108. 72. Chew D, Buffington C.A.T., editor Pandora Syndrome: It's more than just the bladder. American Association of Feline Practitioners; 2013; Dallas, Texas. 73. Chew DJ, Buffington T, Kendall MS, Osborn SD, Woodsworth BE. Urethroscopy, Cystoscopy, and Biopsy of the Feline Lower Urinary Tract. Veterinary Clinics of North America: Small Animal Practice. 1996;26(3):441-62. 74. Chew DJ, DiBartola SP, Schenck PA. Canine and Feline Nephrology and Urology. 2nd ed. Missouri: Saunders; 2011. 75. Chew DJ, DiBartola SP, Schenck PA. Nonobstructive Idiopathic or Interstitial Cystitis in Cats. In: Chew DJ, DiBartola SP, Schenck PA, editors. Canine and feline nephrology and urology. Second ed. Missouri: Saunders Elsevier; 2011. p. 306-40. 76. Chew DJ, DiBartola SP, Schenck PA. Tumors of the urinary system. Canine and feline nephrology and urology. 2 ed. Missouri: Saunders; 2011. 77. Choi D, Han JY, Shin JH, Ryu CM, Yu HY, Kim A, Lee S, Lim J, Shin DM, Choo MS. Downregulation of WNT11 is associated with bladder tissue fibrosis in patients with interstitial cystitis/bladder pain syndrome without Hunner lesion. Sci Rep. 2018;8(1):9782. 78. Christmas T. Lymphocyte sub‐populations in the bladder wall in normal bladder, bacterial cystitis and interstitial cystitis. British Journal of Urology. 1994;73(5):508-15. 79. Christodoulou E, Ma J, Collins GS, Steyerberg EW, Verbakel JY, Van Calster B. A systematic review shows no performance benefit of machine learning over logistic regression for clinical prediction models. J Clin Epidemiol. 2019;110:12-22. 80. Cianciolo RE, Mohr FC. Urinary system. In: Maxie MG, editor. Jubb, Kennedy and Palmer's Pathology of Domestic Animals. 2. 6th ed: Elsevier; 2015. 81. Cianciolo RE, Mohr FC, Aresu L, Brown CA, James C, Jansen JH, Spangler WL, van der Lugt JJ, Kass PH, Brovida C, Cowgill LD, Heiene R, Polzin DJ, Syme H, Vaden SL, van Dongen AM,

205

Lees GE. World Small Animal Veterinary Association Renal Pathology Initiative: Classification of Glomerular Diseases in Dogs. Vet Pathol. 2016;53(1):113-35. 82. Clark AK, Old EA, Malcangio M. Neuropathic pain and cytokines: current perspectives. J Pain Res. 2013;6:803-14. 83. Coblentz TR, Mills SE, Theodorescu D. Impact of second opinion pathology in the definitive management of patients with bladder carcinoma. Cancer. 2001;91(7):1284-90. 84. Cohen SM, Ohnishi T, Clark NM, He J, Arnold LL. Investigations of Rodent Urinary Bladder Carcinogens: Collection, Processing, and Evaluation of Urine and Bladders. Toxicologic Pathology. 2007;35(3):337-47. 85. Cohrs RJ, Martin T, Ghahramani P, Bidaut L, Higgins PJ, Shahzad A. Translational Medicine definition by the European Society for Translational Medicine. New Horizons in Translational Medicine. 2015;2(3):86-8. 86. Corcoran A, Yoshimura N, Tyagi V, Jacobs B, Leng W, Tyagi P. Mapping the cytokine profile of painful bladder syndrome/interstitial cystitis in human bladder and urine specimens. World Journal of Urology. 2013;31(1):241-6. 87. Cox A, Golda N, Nadeau G, Curtis Nickel J, Carr L, Corcos J, Teichman J. CUA guideline: Diagnosis and treatment of interstitial cystitis/bladder pain syndrome. Can Urol Assoc J. 2016;10(5- 6):E136-E55. 88. Crissman JW, Goodman DG, Hildebrandt PK, Maronpot RR, Prater DA, Riley JH, Seaman WJ, Thake DC. Best practice guideline: Toxicologic histopathology. Toxicol Pathol. 2004;32:126-31. 89. Cross SS. Grading and scoring in histopathology. Histopathology. 1998;33(2):99-106. 90. Cross SS. Kappa statistics as indicators of quality assurance in histopathology and cytopathology. J Clin Pathol. 1996;49:597-9. 91. Crowley RS, Naus GJ, Stewart J, Friedman CP. Development of visual diagnostic expertise in pathology -- an information-processing study. Journal of the American Medical Informatics Association. 2003;10(1):39-51. 92. Davis NF, Brady CM, Creagh T. Interstitial cystitis/painful bladder syndrome: epidemiology, pathophysiology and evidence-based treatment options. European Journal of Obstetrics & Gynecology and Reproductive Biology. 2014;175:30-7. 93. Day MJ, Bilzer T, Mansell J, Wilcock B, Hall EJ, Jergens A, Minami T, Willard M, Washabau R, World Small Animal Veterinary Association Gastrointestinal Standardization G. Histopathological standards for the diagnosis of gastrointestinal inflammation in endoscopic biopsy samples from the dog and cat: a report from the World Small Animal Veterinary Association Gastrointestinal Standardization Group. Journal of Comparative Pathology. 2008;138 Suppl 1:S1-43. 94. de Brot S, Robinson BD, Scase T, Grau‑Roma L, Wilkinson E, Boorjian SA, Gardner D, Mongan NP. The dog as an animal model for bladder and urethral urothelial carcinoma: Comparative epidemiology and histology. Oncol Lett. 2018;16(2):1641-9. 95. Dean AG, Sullivan KM, Soe MM. OpenEpi: Open Source Epidemiologic Statistics for Public Health, Version 3.01 2013 [Available from: www.OpenEpi.com. 96. Decker B, Parker HG, Dhawan D, Kwon EM, Karlins E, Davis BW, Ramos-Vara JA, Bonney PL, McNiel EA, Knapp DW, Ostrander EA. Homologous Mutation to Human BRAF V600E Is Common in Naturally Occurring Canine Bladder Cancer--Evidence for a Relevant Model System and Urine-Based Diagnostic Test. Mol Cancer Res. 2015;13(6):993-1002. 97. Defauw P, Van de Maele I, Duchateau L, Daminet S, editors. Evaluation of possible risk factors for feline idiopathic cystitis. 53rd BSAVA annual congress; 2010: British Small Animal Veterinary Association (BSAVA). 98. Defauw PA, Van de Maele I, Duchateau L, Polis IE, Saunders JH, Daminet S. Risk factors and clinical presentation of cats with feline idiopathic cystitis. Journal of Feline Medicine & Surgery. 2011;13(12):967-75.

206

99. Denson MA, Griebling, T.L., Cohen, M.B., Kreder, K.J. Comparison of cystoscopic and histological findings in patients with suspected interstitial cystitis. The Journal of Urology. 2000;164:1908-11. 100. Deppen SA, Blume JD, Aldrich MC, Fletcher SA, Massion PP, Walker RC, Chen HC, Speroff T, Degesys CA, Pinkerman R, Lambright ES, Nesbitt JC, Putnam JB, Jr., Grogan EL. Predicting lung cancer prior to surgical resection in patients with lung nodules. Journal of Thoracic Oncology. 2014;9(10):1477-84. 101. Di Capua-Sacoto C, Sanchez-Llopis A, O'Connor E, Martinez A, Ruiz-Cerda JL. Study of the apoptotic effect of urine as a diagnostic biomarker in patients with interstitial cystitis. Actas Urol Esp. 2016;40(9):570-6. 102. Di Capua-Sacoto C, Sanchez-Llopis A, Oconnor JE, Martinez-Romero A, Ruiz-Cerdá JL. Apoptotic effect as biomarker of disease, severity and follow-up in interstitial cystitis. Actas Urológicas Españolas (English Edition). 2018;42(4):262-6. 103. Dias Pereira P, Gartner F. Expression of E-cadherin in normal, hyperplastic and neoplastic feline mammary tissue. Vet Rec. 2003;153(10):297-302. 104. DiBartola SP. Clinical approach and laboratory evaluation of renal disease. In: Ettinger SJ, Feldman EC, editors. Textbook of veterinary internal medicine. 7th ed. Missouri: Elsevier Saunders; 2010. p. 1955-69. 105. Dinesh D, Behl SM, Singh P, Tayal R, Pal M, Chandolia RK. Diagnosis of urinary bladder diseases in dogs by using two-dimensional and three-dimensional ultrasonography. Vet World. 2015;8(7):819-22. 106. Dinis S, de Oliveira JT, Pinto R, Cruz F, Buffington CT, Dinis P. From bladder to systemic syndrome: concept and treatment evolution of interstitial cystitis. International Journal of Women's Health. 2015;7:735-44. 107. Dintzis R, McBride J. Anatomy, embryology, and histology. In: Hansel DE, McKenney JK, Stephensen AJ, Chang SS, editors. The Urinary Tract, A comprehensive guide to patient diagnosis and management. New York: Springer; 2012. p. 3-25. 108. Dohoo I, Martin W, Stryhn H. Veterinary Epidemiologic Research. 2nd ed. Charlottetown, PE, Canada: VER Inc.; 2009. 109. Donaldson JF, Ruhayel Y, Skolarikos A, MacLennan S, Yuan Y, Shepherd R, Thomas K, Seitz C, Petrik A, Türk C, Neisius A. Treatment of Bladder Stones in Adults and Children: A Systematic Review and Meta-analysis on Behalf of the European Association of Urology Urolithiasis Guideline Panel. European Urology. 2019;76(3):352-67. 110. Dorsch R, Remer C, Sauter-Louis C, Hartmann K. Feline lower urinary tract disease in a German cat population. Tierarztl Prax Ausg K Kleintiere Heimtiere. 2014;4:1-9. 111. Dorsch R, von Vopelius-Feldt C, Wolf G, Mueller RS, Straubinger RK, Hartmann K. Urinary tract infections in cats. Prevalence of comorbidities and bacterial species, and determination of antimicrobial susceptibility to commonly used antimicrobial agents. Tierarztl Prax Ausg K Kleintiere Heimtiere. 2016;44(4):227-36. 112. Douglas MP, Rogers SO. DNA damage caused by common cytological fixatives. Mutat Res. 1998;401(1-2):77-88. 113. Dow S. A Role for Dogs in Advancing Cancer Immunotherapy Research. Frontiers in immunology. 2020;10:2935-. 114. Dray A. Inflammatory mediators of pain. Br J Anaesth. 1995;75(2):125-31. 115. Dundore PA, Schwartz AM, Semerjian H. Mast cell counts are not useful in the diagnosis of nonulcerative interstitial cystitis. Journal of Urology. 1996;155:885-7. 116. Dyer AJ, Twiss CO. Painful bladder syndrome: An update and review of current management strategies. Current Urology Reports. 2014;15(2):384. 117. Eggertsdottir AV, Lund HS, Krontveit R, Sorum H. Bacteriuria in cats with feline lower urinary tract disease: a clinical study of 134 cases in Norway. J Feline Med Surg. 2007;9(6):458-65. 207

118. El-Mansoury M, Boucher W, Sant G, Theoharides T. Increased urine histamine and methylhistamine in interstitial cystitis. The Journal of Urology. 1994;152(2):350-3. 119. Elghamry HA, Hassan FM, Mohamed MI, Abdelfattah DS, Abdelaal AG. Estimation of the postmortem interval using GAPDH mRNA in skin and heart tissues of albino rats at different environmental conditions. Egyptian Journal of Forensic Sciences. 2018;8(1):69. 120. ElMoneim HMA, Zaghloul NM. Expression of E-cadherin, N-cadherin and snail and their correlation with clinicopathological variants: an immunohistochemical study of 132 invasive ductal breast carcinomas in Egypt. Clinics (Sao Paulo, Brazil). 2011;66(10):1765-71. 121. Erickson D, Kunselman A, Bentley C, Peters K, Rovner E, Demers L, Tomaszewski J. Is Urine Methylhistamine a Useful Marker for Interstitial Cystitis? The Journal of Urology. 2004;172(6):2256- 60. 122. Erickson DR. Urine markers of interstitial cystitis. Urology. 2001;57(Suppl 6A):15-21. 123. Erickson DR, Belchis DA, Dabbs DJ. Inflammatory cell types and clinical features of interstitial cystitis. J Urol. 1997;158(3 Pt 1):790-3. 124. Erickson DR, Xie SX, Bhavanandan VP, Wheeler MA, Hurst RE, Demers LM, Kushner L, Keay SK. A comparison of multiple urine markers for interstitial cystitis. J Urol. 2002;167:2461-9. 125. Erman A, Veranič P. Time- and temperature-dependent autolysis of urinary bladder epithelium during ex vivo preservation. Protoplasma. 2011;248(3):541-50. 126. Fagerland MW, Hosmer DW, Bofin AM. Multinomial goodness-of-fit tests for logistic regression models. Stat Med. 2008;27(21):4238-53. 127. Fall M, Johansson SL, Aldenborg F. Chronic interstitial cystitis: a heterogenous syndrome. Journal of Urology. 1987;137(1):35-8. 128. Fandel TM, Pfnur M, Schafer SC, Bacchetti P, Mast FW, Corinth C, Ansorge M, Melchior SW, Thuroff JW, Kirkpatrick CJ, Lehr HA. Do we truly see what we think we see? The role of cognitive bias in pathological interpretation. Journal of Pathology. 2008;216(2):193-200. 129. Fariello JY, Whitmore KE. Clinical Evaluation and Diagnosis of Bladder Pain Syndrome. In: Nordling J, Wyndaele JJ, Van de Merwe JP, Bouchelouche P, Cervigni M, Fall M, editors. Bladder pain syndrome, a book for clinicians. New York: Springer; 2013. p. 189-203. 130. Farrell JJ, Taupin D, Koh TJ, Chen D, Zhao CM, Podolsky DK, Wang TC. TFF2/SP-deficient mice show decreased gastric proliferation, increased acid secretion, and increased susceptibility to NSAID injury. J Clin Invest. 2002;109(2):193-204. 131. Finke J, Fritzen R, Ternes P, Lange W, Dolken G. An improved strategy and a useful housekeeping gene for RNA analysis from formalin-fixed, paraffin-embedded tissues by PCR. Biotechniques. 1993;14(3):448-53. 132. Fitzmaurice GM. Regression. Diagnostic Histopathology. 2016;22(7):271-8. 133. Fleiss JL. Measuring nominal scale agreement among many raters. Psychological Bulletin. 1971;76(5):378-82. 134. Fleiss JL, Levin B, Paik MC. The Measurement of Interrater Agreement. In: Shewart WA, Wilks SS, Fleiss JL, Levin B, Paik MC, editors. Statistical Methods for Rates and Proportions: John Wiley & Sons, Inc.; 2003. p. 598-626. 135. Formsa A, Saini NS. Canine Urolithiasis and Concurrent Urinary Bladder Abnormalities: Symptoms, Haematology, Urinalysis and Comparative Radiographic and Ultrasonographic Diagnosis. Veterinary Medicine. 2019;4(1):18-26. 136. Forrester SD, Towell TL. Feline idiopathic cystitis. Vet Clin North Am Small Anim Pract. 2015;45(4):783-806. 137. Foxman B. Epidemiology of urinary tract infections: Incidence, morbidity, and economic costs. Am J Med. 2002;113(1A):5S-13S. 138. Furuta A, Yamamoto T, Suzuki Y, Gotoh M, Egawa S, Yoshimura N. Comparison of inflammatory urine markers in patients with interstitial cystitis and overactive bladder. International Urogynecology Journal. 2018;29(7):961-6. 208

139. Gamper M, Regauer S, Welter J, Eberhard J, Viereck V. Are mast cells still good biomarkers for bladder pain syndrome/interstitial cystitis? J Urol. 2015;193(6):1994-2000. 140. Gardiner RA, Seymore GJ, Lavin MF, Strutton GM, Gemmell E, Hazan G. Immunohistochemical Analysis of the Human Bladder. British Journal of Urology. 1986;58(1):19- 25. 141. Geboes K, Riddell R, Ost A, Jensfelt B, Persson T, Lofberg R. A reproducible grading scale for histological assessment of inflammation in ulcerative colitis. Gut. 2000;47(3):404-9. 142. Gerber B, Eichenberger S, Reusch CE. Guarded long-term prognosis in male cats with urethral obstruction. Journal of Feline Medicine & Surgery. 2008;10(1):16-23. 143. Gillenwater JY, Wein AJ. Summary of the National Institute ofArthritis, Diabetes, Digestive and Kidney Diseases Workshop on Interstitial Cystitis. Journal of Urology. 1988;140:203-6. 144. Glueckert E, Clifford DL, Brenn-White M, Ochoa J, Gabriel M, Wengert G, Foley J. Endemic Skunk amdoparvovirus in free-ranging striped skunks (Mephitis mephitis) in California. Transboundary and Emerging Diseases.66(6):2252-63. 145. Goldschmidt M, Pena L, Rasotto R, Zappulli V. Classification and grading of canine mammary tumors. Vet Pathol. 2011;48(1):117-31. 146. Gonzalez EJ, Arms L, Vizzard MA. The role(s) of cytokines/chemokines in urinary bladder inflammation and dysfunction. Biomed Res Int. 2014;2014:120525. 147. Goo YA, Tsai YS, Liu AY, Goodlett DR, Yang CC. Urinary proteomics evaluation in interstitial cystitis/painful bladder syndrome: a pilot study. Int Braz J Urol. 2010;36(4):464-78; discussion 78-9, 79. 148. Gouveia GR, Ferreira SC, Ferreira JE, Siqueira SAC, Pereira J. Comparison of two methods of RNA extraction from formalin-fixed paraffin-embedded tissue specimens. Biomed Research International. 2014;2014:151724-. 149. Granato A, Giantin M, Ariani P, Carminato A, Baratto C, Zorzan E, Vascellari M, Bozzato E, Dacasto M, Mutinelli F. DNA and RNA isolation from canine oncologic formalin-fixed, paraffin- embedded tissues for downstream "-omic" analyses: possible or not? J Vet Diagn Invest. 2014;26(1):117-24. 150. Grohn YT, Fubini SL, Smith DF. Use of a multiple logistic regression model to determine prognosis of dairy cows with right displacement of the abomasum or abomasal volvulus. American Journal of Veterinary Research. 1990;51(12):1895-9. 151. Grover S, Srivastava A, Lee R, Tewari AK, Te AE. Role of inflammation in bladder function and interstitial cystitis. Therapeutic Advances in Urology. 2011;3(1):19-33. 152. Gunn-Moore DA, Shenoy CM. Oral glucosamine and the management of feline idiopathic cystitis. J Feline Med Surg. 2004;6(4):219-25. 153. Guo CC, Moore CK. Infection and inflammatory disorders. In: Hansel DE, McKenney JK, Stephensen AJ, Chang SS, editors. The Urinary Tract, A comprehensive guide to patient diagnosis and management. New York: Springer; 2012. 154. Haidich AB. Meta-analysis in medical research. Hippokratia. 2010;14(Suppl 1):29-37. 155. Hall JL, Holmes MA, Baines SJ. Prevalence and antimicrobial resistance of canine urinary tract pathogens. Veterinary Record. 2013;173(22):549. 156. Hamilton PW, van Diest PJ, Williams R, Gallagher AG. Do we see what we think we see? The complexities of morphological assessment. Journal of Pathology. 2009;218(3):285-91. 157. Han E, Nguyen L, Sirls L, Peters K. Current best practice management of interstitial cystitis/bladder pain syndrome. Ther Adv Urol. 2018;10(7):197-211. 158. Harrell FE. Regression modeling strategies. 2nd ed. New York City: Springer Science + Business Media; 2015. 159. Hauser PJ, VanGordon SB, Seavey J, Sofinowski TM, Ramadan M, Abdullah S, Buffington CA, Hurst RE. Abnormalities in Expression of Structural, Barrier and Differentiation Related

209

Proteins, and Chondroitin Sulfate in Feline and Human Interstitial Cystitis. Journal of Urology. 2015;194(2):571-7. 160. Hausman J, McFadden D. Specification Tests for the Multinomial Logit Model. Econometrica. 1984;52(5):1219-40. 161. Henning J, Hannon C, McKinnon A, Larkin R, Allavena R. The causes and prognoses of different types of fractures in wild koalas submitted to wildlife hospitals. Prev Vet Med. 2015;122(3):371-8. 162. Henry CJ, Tyler JW, McEntee MC, Stokol T, Rogers KS, Chun R, Garrett LD, McCaw DL, Higginbotham ML, Flessland KA, Stokes PK. Evaluation of a bladder tumor antigen test as a screening test for transitional cell carcinoma of the lower urinary tract in dogs. Am J Vet Res. 2003;64(8):1017-20. 163. Hernandez J, Bota D, Farbos M, Bernardin F, Ragetly G, Medaille C. Risk factors for urinary tract infection with multiple drug-resistant Escherichia coli in cats. J Feline Med Surg. 2014;16(2):75- 81. 164. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539-58. 165. Hohenfellner M, Nunes L, Schmidt R, Lampel A, Thüroff J, Tanagho E. Interstitial cystitis: increased sympathetic innervation and related neuropeptide synthesis. The Journal of Urology. 1992;147(3):587-91. 166. Hooton TM. Uncomplicated urinary tract infection. N Engl J Med. 2012;366:1028-37. 167. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. USA: John Wiley & Sons Inc; 2000. 168. Houbrechts M. Predisposing factors of Feline Idiopathic Cystitis in The Netherlands 2018. 169. Houston DM, Vanstone NP, Moore AEP, Weese HE, Weese JS. Evaluation of 21 426 feline bladder urolith submissions to the Canadian Veterinary Urolith Centre (1998-2014). Canadian Veterinary Journal. 2016;57:196-201. 170. Hunner GL. A rare type of bladder ulcer in women; report of cases. Boston Medical and Surgical Journal. 1915;172(18):660-4. 171. Ikeda Y, Birder L, Buffington C, Roppolo J, Kanai A. Mucosal muscarinic receptors enhance bladder activity in cats with feline interstitial cystitis. Journal of Urology. 2009;181(3):1415-22. 172. Ikeda Y, Wolf-Johnston A, Roppolo JR, Buffington CA, Birder L. Feline Interstitial Cystitis Enhances Mucosa-Dependent Contractile Responses to Serotonin. International Neurourology Journal. 2018;22(4):246. 173. Izgi K, Daneshgari F. Animal Models for Interstitial Cystitis/Painful Bladder Syndrome (IC/PBS). Erciyes Tıp Dergisi/Erciyes Medical Journal. 2013;35(3):103-7. 174. Jabr RI, Fry CH. Animal Models of Lower Urinary Tract Dysfunction. In: Conn M, editor. Animal Models for the Study of Human Disease: Elsevier; 2013. p. 461-81. 175. Janicki T. Chronic pelvic pain as a form of complex regional pain complex syndrome. Clinical Obstetrics and Gynecology. 2003;46(4):797-803. 176. Jha S, Topol EJ. Adapting to Artificial Intelligence: Radiologists and Pathologists as Information Specialists. JAMA. 2016;316(22):2353-4. 177. Jhang J-F, Kuo H-C. Pathomechanism of Interstitial Cystitis/Bladder Pain Syndrome and Mapping the Heterogeneity of Disease. International Neurourology Journal. 2016;20(Suppl 2):S95- 104. 178. Jhang JF, Kuo HC. Pathomechanism of Interstitial Cystitis/Bladder Pain Syndrome and Mapping the Heterogeneity of Disease. International Neurourology Journal. 2016;20:95-104. 179. Jin Y, Lin D. Fungal Urinary Tract Infections in the Dog and Cat: A Retrospective Study (2001- 2004). J Am Anim Hosp Assoc. 2005;41:373-81.

210

180. Jokinen MP, Seely JC. Urinary bladder, ureter and urethra. In: Suttie AW, Leininger JR, Bradley AE, editors. Boorman's Pathology of the Rat, Reference and Atlas. 2nd ed: Elsevier; 2018. p. 167-88. 181. Jorgensen PE, Hilchey SD, Nexo E, Poulsen SS, Quilley CP. Urinary epidermal growth factor is excreted from the rat isolated perfused kidney in the absence of plasma. Journal of Endocrinology. 1993;139:227. 182. Joseph M, Enting D. Immune Responses in Bladder Cancer-Role of Immune Cell Populations, Prognostic Factors and Therapeutic Implications. Frontiers in Oncology. 2019;9(1270). 183. Kabjorn-Gustafsson C, Peeker R. Biopsy Retrieval, Tissue Handling, Morphology, and Histopathological Characteristics. In: Nordling J, Wyndaele JJ, Van de Merwe JP, Bouchelouche P, Cervigni M, Fall M, editors. Bladder pain syndrome, a guide for clinicians. New York: Springer; 2013. p. 231-9. 184. Kaneko N, Tanimoto T, Morimoto M, Hayashi T, Shimokawa Miyama T, Hiraoka H, Itamoto K, Une S, Mizuno T, Okuda M. Use of formalin-fixed paraffin-embedded tissue and single-strand conformation polymorphism analysis for polymerase chain reaction of antigen receptor rearrangements in dogs. J Vet Med Sci. 2009;71(4):535-8. 185. Kashofer K, Viertler C, Pichler M, Zatloukal K. Quality control of RNA preservation and extraction from paraffin-embedded tissue: implications for RT-PCR and microarray analysis. PLoS One. 2013;8(7):e70714. 186. Kay EW, O'Dowd J, Thomas R, Alyusuf R, Sachithanandan S, Robinson R, Walsh CB, Fielding JF, Leader MB. Mild abnormalities in liver histology associated with chronic hepatitis: distinction from normal liver histology. J Clin Pathol. 1997;50(11):929-31. 187. Keay S, Kleinberg M, Zhang CO, Hise MK, Warren JW. Bladder epithelial cells from patients with interstitial cystitis produce an inhibitor of heparin-binding epidermal growth factor-like growth factor production. J Urol. 2000;164:2112-8. 188. Keay S, Nallar SC, Gade P, Zhang CO, Kalvakolanu DV. Oncosuppressor protein p53 and cyclin-dependent kinase inhibitor p21 regulate interstitial cystitis associated gene expression. Cytokine. 2018;110:110-5. 189. Keay S, Zhang CO, Hise MK, Hebel JR, Jacobs SC, Gordon D, Whitmore K, Bodison S, Gordon N, Warren JW. A diagnostic in vitro urine assay for interstitial cystitis. Urology. 1998;52:974-8. 190. Keay S, Zhang CO, Kagen DI, Hise MK, Jacobs SC, Hebel JR, Gordon D, Whitmore K, Bodison S, Warren JW. Concentrations of specific epithelial growth factors in the urine of interstitial cystitis patients and controls. J Urol. 1997;158(5):1983-8. 191. Keay S, Zhang CO, Marvel R, Chai T. Antiproliferative factor, heparin-binding epidermal growth factor-like growth factor, and epidermal growth factor: sensitive and specific urine markers for interstitial cystitis. Urology. 2001;57(6 Suppl 1):104. 192. Keller JJ, Chen YK, Lin HC. Comorbidities of bladder pain syndrome/interstitial cystitis: a population-based study. British Journal Of Urology International. 2012;110(11 Pt C):E903-9. 193. Kennedy SM, Lulich JP, Ritt MG, Furrow E. Comparison of body condition score and urinalysis variables between dogs with and without calcium oxalate uroliths. Journal of the American Veterinary Medical Association. 2016;249(11):1274-80. 194. Kessler Y, Helfer-Hungerbuehler AK, Cattori V, Meli ML, Zellweger B, Ossent P, Riond B, Reusch CE, Lutz H, Hofmann-Lehmann R. Quantitative TaqMan® real-time PCR assays for gene expression normalisation in feline tissues. BMC Molecular Biology. 2009;10(1):106. 195. Kim A, Han JY, Ryu CM, Yu HY, Lee S, Kim Y, Jeong SU, Cho YM, Shin DM, Choo MS. Histopathological characteristics of interstitial cystitis/bladder pain syndrome without Hunner lesion. Histopathology. 2017;71(3):415-24. 196. Kim S, Im Y, Choi H, Kang H, Kim J, Kim J. Urinary nerve growth factor correlates with the severity of urgency and pain. Including Pelvic Floor Dysfunction. 2014;25(11):1561-7. 211

197. Kind T, Cho E, Park TD, Deng N, Liu Z, Lee T, Fiehn O, Kim J. Interstitial Cystitis-Associated Urinary Metabolites Identified by Mass-Spectrometry Based Metabolomics Analysis. Sci Rep. 2016;6:39227. 198. Kirk CA, Lund E, Armstrong P. Prevalence of lower urinary tract disorders of dogs and cats in the United States. Waltham International Symposium Conference Proceedings. Vancouver, Canada2001. p. 61. 199. Kiuchi H, Tsujimura A, Takao T, Yamamoto K, Nakayama J, Miyagawa Y, Nonomura N, Takeyama M, Okuyama A. Increased vascular endothelial growth factor expression in patients with bladder pain syndrome/interstitial cystitis: its association with pain severity and glomerulations. British Journal Of Urology International. 2009;104(6):826-31; discussion 31. 200. Kiupel M, Webster JD, Bailey KL, Best S, DeLay J, Detrisac CJ, Fitzgerald SD, Gamble D, Ginn PE, Goldschmidt MH, Hendrick MJ, Howerth EW, Janovitz EB, Langohr I, Lenz SD, Lipscomb TP, Miller MA, Misdorp W, Moroff S, Mullaney TP, Neyens I, O'Toole D, Ramos-Vara J, Scase TJ, Schulman FY, Sledge D, Smedley RC, Smith K, P WS, Southorn E, Stedman NL, Steficek BA, Stromberg PC, Valli VE, Weisbrode SE, Yager J, Heller J, Miller R. Proposal of a 2-tier histologic grading system for canine cutaneous mast cell tumors to more accurately predict biological behavior. Vet Pathol. 2011;48(1):147-55. 201. Knapp DW, Adams LG, Degrand AM, Niles JD, Ramos-Vara JA, Weil AB, O'Donnell MA, Lucroy MD, Frangioni JV. Sentinel lymph node mapping of invasive urinary bladder cancer in animal models using invisible light. European Urology. 2007;52(6):1700-8. 202. Knapp DW, Glickman NW, DeNicola DB, Bonney PL, Lin TC, Glickman LT. Naturally- occurring canine transitional cell carcinoma of the urinary bladder, A relevant model of human invasive bladder cancer. Urologic Oncology. 2000;5:47-59. 203. Knapp DW, McMillan SK. Tumors of the Urinary System. In: Withrow SJ, Vail DM, Page RL, editors. Small Animal Clinical Oncology. 5th ed. Missouri: Elsevier; 2013. p. 572-82. 204. Knapp DW, Ramos-Vara JA, Moore GE, Dhawan D, Bonney PL, Young KE. Urinary bladder cancer in dogs, a naturally occurring model for cancer biology and drug development. Institute for Laboratory Animal Research Journal. 2014;55(1):100-18. 205. Knottenbelt DC, Patterson-Kane JC, Snalune KL. Vascular neoplasms. Clinical Equine Oncology2015. p. 332-41. 206. Knottnerus JA. Application of Logistic Regression to the Analysis of Diagnostic Data:Exact Modeling of a Probability Tree of Multiple Binary Variables. Medical Decision Making. 1992;12(2):93-108. 207. Koehler JW, Miller AD, Miller CR, Porter B, Aldape K, Beck J, Brat D, Cornax I, Corps K, Frank C, Giannini C, Horbinski C, Huse JT, O'Sullivan MG, Rissi DR, Mark Simpson R, Woolard K, Shih JH, Mazcko C, Gilbert MR, LeBlanc AK. A Revised Diagnostic Classification of Canine Glioma: Towards Validation of the Canine Glioma Patient as a Naturally Occurring Preclinical Model for Human Glioma. J Neuropathol Exp Neurol. 2018;77(11):1039-54. 208. Kolbjornsen O, Press CM, Moore PF, Landsverk T. Lymphoid follicles in the gastric mucosa of dogs. Distribution and lymphocyte phenotypes. Veterinary Immunology and Immunopathology. 1994;40:299-312. 209. Kordass U, Carlson R, Stein VM, Tipold A. Measurements of C-reactive protein (CRP) and nerve-growth-factor (NGF) concentrations in serum and urine samples of dogs with neurologic disorders. BMC Vet Res. 2016;12:7. 210. Koziol JA, Adams H-P, Frutos A. Discrimination Between the Ulcerous and the Nonulcerous Forms of Interstitial Cystitis by Noninvasive Findings. The Journal of Urology. 1996;155(1):87-90. 211. Kraijer M, Fink-Gremmels J, Nickel RF. The short-term clinical efficacy of amitriptyline in the management of idiopathic feline lower urinary tract disease: a controlled clinical study. Journal of Feline Medicine and Surgery. 2003;5:191-6. 212. Krieger JN. Urinary tract infections: what's new? J Urol. 2002;168(6):2351-8. 212

213. Kruger J, Conway T, Kaneene J, Perry R, Hagenlocker E, Golombek A, Stuhler J. Randomized controlled trial of the efficacy of short-term amitriptyline administration for treatment of acute, nonobstructive, idiopathic lower urinary tract disease in cats. Journal of American Veterinary Medical Association. 2003;222(6):749-58. 214. Kruger J, Osborne C, Goyal S, Wickstrom S, Johnston G, Fletcher T, Brown P. Clinical evaluation of cats with lower urinary tract disease. Journal of the American Veterinary Medical Assocation. 1991;199(2):211-6. 215. Kruger JM, Osborne CA, Lulich JP. Changing paradigms of feline idiopathic cystitis. Veterinary Clinical North American Small Animal Practice. 2009;39(1):15-40. 216. Kullmann FA, McDonnell BM, Wolf-Johnston AS, Lynn AM, Giglio D, Getchell SE, Ruiz WG, Zabbarova IV, Ikeda Y, Kanai AJ, Roppolo JR, Bastacky SI, Apodaca G, Buffington CAT, Birder LA. Inflammation and tissue remodeling in the bladder and urethra in feline interstitial cystitis. Frontiers in Systems Neuroscience. 2018;12(13):1-15. 217. Kuo HC, Liu HT, Tyagi P, Chancellor MB. Urinary Nerve Growth Factor Levels in Urinary Tract Diseases With or Without Frequency Urgency Symptoms. Luts-Lower Urinary Tract Symptoms. 2010;2(2):88-94. 218. Kuromitsu S, Yokota H, Hiramoto M, Morita S, Mita H, Yamada T. Increased concentration of neutrophil elastase in urine from patients with interstitial cystitis. Scandinavian Journal of Urology and Nephrology. 2008;42(5):455-61. 219. Kutlu O, Akkaya E, Koksal I, Bassorgun I, Ciftcioglu M, Sanlioglu S, Kukul E. Importance of TNF-related apoptosis-inducing ligand in pathogenesis of interstitial cystitis. International Urology and Nephrology. 2010;42(2):393-9. 220. Lamb LE, Janicki JJ, Bartolone SN, Peters KM, Chancellor MB. Development of an interstitial cystitis risk score for bladder permeability. PLoS One. 2017;12(10):e0185686. 221. Lane IF, Lennon EM, editors. Canine and feline urolithiasis updates and challenges. Georgia Veterinary Medical Association Fall Convention; 2018; Atlanta. 222. Larson J, Kruger JM, Wise AG, Kaneene JB, Miller R, Fitzgerald SD, Kiupel M, Maes RK. Nested case-control study of feline calicivirus viruria, oral carriage, and serum neutralizing antibodies in cats with idiopathic cystitis. J Vet Intern Med. 2011;25:199-205. 223. Lascelles BD, Knazovicky D, Case B, Freire M, Innes JF, Drew AC, Gearing DP. A canine- specific anti-nerve growth factor antibody alleviates pain and improves mobility and function in dogs with degenerative joint disease-associated pain. Bmc Veterinary Research. 2015;11:101. 224. Lavelle J, Meyers S, Ruiz W, Buffington C, Zeidel M, Apodaca G. Urothelial pathophysiological changes in feline interstitial cystitis: a human model. American Journal of Physiology-Renal Physiology. 2000;278(4):F540-53. 225. le Feber J, van Asselt E, van Mastrigt R. Afferent bladder nerve activity in the rat: a mechanism for starting and stopping voiding contractions. Urol Res. 2004;32(6):395-405. 226. Lee CH, Cook S, Lee JS, Han B. Comparison of Two Meta-Analysis Methods: Inverse- Variance-Weighted Average and Weighted Sum of Z-Scores. Genomics & informatics. 2016;14(4):173-80. 227. Lee H. Retrospective analysis of chief complaints and causes of dogs and cats during emergency visits. Seoul: Seoul National University; 2014. 228. Lee S, Trivedi U, Johnson C, Farquharson C, Bergkvist GT. Optimised isolation method for RNA extraction suitable for RNA sequencing from feline teeth collected in a clinical setting and at post mortem. Veterinary Research Communications. 2019;43(1):17-27. 229. Leiblich A. Recent Developments in the Search for Urinary Biomarkers in Bladder Cancer. Current Urology Reports. 2017;18(12):100-. 230. Leiby BE, Landis JR, Propert KJ, Tomaszewski JE, Interstitial Cystitis Data Base Study Group. Discovery of morphological subgroups that correlate with severity of symptoms in interstitial cystitis: a proposed biopsy classification system. Journal of Urology. 2007;177(1):142-8. 213

231. Lekcharoensuk C, Osborne CA, Lulich JP. Epidemiologic study of risk factors for lower urinary tract diseases in cats. Journal of the American Veterinary Medical Assocation. 2001;218(9):1429-35. 232. Lemberger SI, Deeg CA, Hauck SM, Amann B, Hirmer S, Hartmann K, Dorsch R. Comparison of urine protein profiles in cats without urinary tract disease and cats with idiopathic cystitis, bacterial urinary tract infection, or urolithiasis. American Journal of Veterinary Research. 2011;72:1407-15. 233. Lemberger SI, Dorsch R, Hauck SM, Amann B, Hirmer S, Hartmann K, Deeg CA. Decrease of Trefoil factor 2 in cats with feline idiopathic cystitis. BJU Int. 2011;107(4):670-7. 234. Lewis F, Maughan NJ, Smith V, Hillan K, Quirke P. Unlocking the archive--gene expression in paraffin-embedded tissue. J Pathol. 2001;195(1):66-71. 235. Li A, Lu H, Liu S, Zhang F, Qian X, Wang H. Clinical Investigation on the Relationship between Upper Urinary Tract Obstruction and the Accompanied Chronic Cystitis. Current Urology. 2008;2(3):142-5. 236. Li W, Liang Y, Deavers MT, Kamat AM, Matin SF, Dinney CP, Czerniak B, Guo CC. Uroplakin II Is a More Sensitive Immunohistochemical Marker Than Uroplakin III in Urothelial Carcinoma and Its Variants. American Journal of Clinical Pathology. 2014;142(6):864-71. 237. Li WM, Chou YH, Li CC, Liu CC, Huang SP, Wu WJ, Huang CH. Local factors compared with systemic factors in the formation of bladder uric acid stones. Urol Int. 2009;82(1):48-52. 238. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JPA, Clarke M, Devereaux PJ, Kleijnen J, Moher D. The PRISMA Statement for Reporting Systematic Reviews and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and Elaboration. PLOS Medicine. 2009;6(7):e1000100. 239. Lin J, Kawano H, Paparella MM, Ho SB. Improved RNA analysis for immediate autopsy of temporal bone soft tissues. Acta Otolaryngol. 1999;119(7):787-95. 240. Ling GV. Therapeutic strategies involving antimicrobial treatment of the canine urinary tract. Journal of the American Veterinary Medical Assocation. 1984;185(10):1162-4. 241. Ling GV, Norris CR, Franti CE, Eisele PH, Johnson DL, Ruby AL, Jang SS. Interrelations of organism prevalence, specimen collection method, and host age, sex, and breed among 8,354 canine urinary tract infections (1969-1995). Journal of Veterinary Internal Medicine. 2001;15(4):341-7. 242. Litster A, Moss S, Platell J, Trott DJ. Occult bacterial lower urinary tract infections in cats- urinalysis and culture findings. Veterinary Microbiology. 2009;136(1-2):130-4. 243. Litster A, Moss SM, Honnery M, Rees B, Trott DJ. Prevalence of bacterial species in cats with clinical signs of lower urinary tract disease: recognition of Staphylococcus felis as a possible feline urinary tract pathogen. Vet Microbiol. 2007;121(1-2):182-8. 244. Litster A, Thompson M, Moss S, Trott D. Feline bacterial urinary tract infections: An update on an evolving clinical problem. The Veterinary Journal. 2011;187(1):18-22. 245. Liu HT, Chancellor MB, Kuo HC. Urinary nerve growth factor levels are elevated in patients with detrusor overactivity and decreased in responders to detrusor botulinum toxin-A injection. Eur Urol. 2009;56(4):700-6. 246. Liu HT, Jiang YH, Kuo HC. Alteration of Urothelial Inflammation, Apoptosis, and Junction Protein in Patients with Various Bladder Conditions and Storage Bladder Symptoms Suggest Common Pathway Involved in Underlying Pathophysiology. Low Urin Tract Symptoms. 2015;7(2):102-7. 247. Liu HT, Shie JH, Chen SH, Wang YS, Kuo HC. Differences in mast cell infiltration, E-cadherin, and zonula occludens-1 expression between patients with overactive bladder and interstitial cystitis/bladder pain syndrome. Urology. 2012;80(1):225 e13-8. 248. Liu HT, Tyagi P, Chancellor MB, Kuo HC. Urinary nerve growth factor but not prostaglandin E2 increases in patients with interstitial cystitis/bladder pain syndrome and detrusor overactivity. Bju International. 2010;106(11):1681-5. 249. Liu S, Feng S, Luo D. Analysis of key genes and micro-RNA-mRNA regulatory networks in women with ulcerative interstitial cystitis/pain bladder syndrome. Int Urogynecol J. 2018. 214

250. Logadottir Y, Delbro D, Fall M, Gjertsson I, Jirholt P, Lindholm C, Peeker R. Cytokine expression in patients with bladder pain syndrome/interstitial cystitis ESSIC type 3C. J Urol. 2014;192(5):1564-8. 251. Logadottir Y, Delbro D, Lindholm C, Fall M, Peeker R. Inflammation characteristics in bladder pain syndrome ESSIC type 3C/classic interstitial cystitis. Int J Urol. 2014;21 Suppl 1:75-8. 252. Logadottir Y, Fall M, Kabjorn-Gustafsson C, Peeker R. Clinical characteristics differ considerably between phenotypes of bladder pain syndrome/interstitial cystitis. Scand J Urol. 2012;46(5):365-70. 253. Longo FM, Holtzman DM, Grimes ML, Mobley WC. NGF Actions in the peripheral and central nervous systems. In: Loughlin SE, Fallon JH, editors. Neurotrophic Factors. San Diego, USA: Academic Press; 1993. p. 209-56. 254. Lotz M, Villiger P, Hugli T, Koziol J, Zuraw BL. Interleukin-6 and interstitial cystitis. J Urol. 1994;152(3):869-73. 255. Lowe EM, Anand P, Terengi G, Williams-Chestnut RE, Sinicropi DV, Osborne JL. Increased nerve growth factor levels in the urinary badder of women with idiopathic sensory urgency and interstitial cystitis. British Journal of Urology. 1997;79:572-7. 256. Lulich J, Osborne C, Kruger J, editors. What constitutes a diagnosis of feline idiopathic cystitis. Proc ACVIM Forum; 2010. 257. Lulich JP, Berent AC, Adams LG, Westropp JL, Bartges JW, Osborne CA. ACVIM Small Animal Consensus Recommendations on the Treatment and Prevention of Uroliths in Dogs and Cats. Journal of Veterinary Internal Medicine. 2016;30(5):1564-74. 258. Lund EM, Armstrong PJ, Kirk CA. Health status and population characteristics of dogs and cats examined at private veterinary practices in the United States. Journal of the American Veterinary Medical Assocation. 1999;214(9):1336-41. 259. Lund HS, Rimstad E, Eggertsdottir AV. Prevalence of viral infections in Norwegian cats with and without feline lower urinary tract disease. Journal of Feline Medicine & Surgery. 2012;14(12):895-9. 260. Lund HS, Skogtun G, Sørum H, Eggertsdóttir AV. Antimicrobial susceptibility in bacterial isolates from Norwegian cats with lower urinary tract disease. Journal of Feline Medicine and Surgery. 2014;17(6):507-15. 261. Lusby AL, Kirk CA, Moyers TD, Toll PW, editors. Prevalence of asymptomatic bacterial urinary tract infections in morbidly obese dogs. Proceedings of the American College of Veterinary Internal Medicine Forum; 2011 15th-18th June 2011; Denver, CO. 262. Lynes W, Flynn S, Shortliffe L, Stamey T. The histology of interstitial cystitis. The American Journal of Surgical Pathology. 1990;14(10):969-76. 263. Lynes WL, Flynn SD, Shortliffe LD, Lemmers M, Zipser R, Roberts II LJ, Stamey TA. Mast cell involvement in interstitial cystitis. The Journal of Urology. 1987;138(4):746-52. 264. Lynes WL, Flynn SD, Shortliffe LD, Stamey TA. The histology of interstitial cystitis. The American Journal of Surgical Pathology. 1990;14(10):969-76. 265. Ma E, Vetter J, Bliss L, Lai HH, Mysorekar IU, Jain S. A multiplexed analysis approach identifies new association of inflammatory proteins in patients with overactive bladder. American Journal of Physiology - Renal Physiology. 2016;311(1):F28-F34. 266. Maddox TM, Rumsfeld JS, Payne PRO. Questions for Artificial Intelligence in Health Care. JAMA. 2019;321(1):31-2. 267. Maeda D, Akiyama Y, Morikawa T, Kunita A, Ota Y, Katoh H, Niimi A, Nomiya A, Ishikawa S, Goto A, Igawa Y, Fukayama M, Homma Y. Hunner-Type (Classic) Interstitial Cystitis: A Distinct Inflammatory Disorder Characterized by Pancystitis, with Frequent Expansion of Clonal B-Cells and Epithelial Denudation. PLoS One. 2015;10(11):e0143316.

215

268. Maeda S, Tomiyasu H, Tsuboi M, Inoue A, Ishihara G, Uchikai T, Chambers JK, Uchida K, Yonezawa T, Matsuki N. Comprehensive gene expression analysis of canine invasive urothelial bladder carcinoma by RNA-Seq. BMC Cancer. 2018;18(1):472. 269. Makino T, Kawashima H, Konishi H, Nakatani T, Kiyama H. Elevated Urinary Levels and Urothelial Expression of Hepatocarcinoma-intestine-pancreas/Pancreatitis-associated Protein in Patients With Interstitial Cystitis. Urology. 2010;75(4):933-7. 270. Malkowicz SB. The application of human complement factor H-related protein (BTA TRAK) in monitoring patients with bladder cancer. Urol Clin North Am. 2000;27(1):63-73, ix. 271. Malykhina A, Hanno P. How are we going to make progress treating bladder pain syndrome? Neurourology and Urodynamics. 2014;33(5):625-9. 272. Mamede S, van Gog T, van den Berge K, Rikers RM, van Saase JL, van Guldener C, Schmidt HG. Effect of availability bias and reflective reasoning on diagnostic accuracy among internal medicine residents. Journal of the American Medical Association. 2010;304(11):1198-203. 273. Mandala M, Imberti GL, Piazzalunga D, Belfiglio M, Labianca R, Barberis M, Marchesi L, Poletti P, Bonomi L, Novellino L, Di Biagio K, Milesi A, Guerra U, Tondini C. Clinical and histopathological risk factors to predict sentinel lymph node positivity, disease-free and overall survival in clinical stages I-II AJCC skin melanoma: outcome analysis from a single-institution prospectively collected database. Eur J Cancer. 2009;45(14):2537-45. 274. Mann PC, Vahle J, Keenan CM, Baker JF, Bradley AE, Goodman DG, Harada T, Herbert R, Kaufmann W, Kellner R, Nolte T, Rittinghausen S, Tanaka T. International Harmonization of Toxicologic Pathology Nomenclature: An Overview and Review of Basic Principles. Toxicologic Pathology. 2012;40(4_suppl):7S-13S. 275. Manning LK, Wünschmann A, Armién AG, Willette M, MacAulay K, Bender JB, Buchweitz JP, Redig P. Lead Intoxication in Free-Ranging Bald Eagles (Haliaeetus leucocephalus). Veterinary Pathology. 2018;56(2):289-99. 276. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. Journal of the National Cancer Institute. 1959;22:719-48. 277. Marchevsky AM, Shah S, Patel S. Reasoning with uncertainty in pathology: artificial neural networks and logistic regression as tools for prediction of lymph node status in breast cancer patients. Mod Pathol. 1999;12(5):505-13. 278. Markwell P, Buffington C, Chew D, Kendall M, Harte J, DiBartola S. Clinical evaluation of commercially available urinary acidification diets in the management of idiopathic cystitis in cats. Journal of American Veterinary Medical Association. 1999;214(3):361-5. 279. Martin TA, Jiang WG. Loss of tight junction barrier function and its role in cancer metastasis. Biochimica et Biophysica Acta (BBA) - Biomembranes. 2009;1788(4):872-91. 280. Martinez-Ruzafa I, Kruger JM, Miller R, Swenson CL, Bolin CA, Kaneene JB. Clinical features and risk factors for development of urinary tract infections in cats. Journal of Feline Medicine and Surgery. 2012;14(10):729-40. 281. Martinez I, Mattoon JS, Eaton KA, Chew DJ, DiBartola SP. Polypoid cystitis in 17 dogs (1978- 2001). J Vet Intern Med. 2003;17:499-509. 282. Mastracci L, Piol N, Molinaro L, Pitto F, Tinelli C, De Silvestri A, Fiocca R, Grillo F, Group AS. Interobserver reproducibility in pathologist interpretation of columnar-lined esophagus. Virchows Arch. 2016;468(2):159-67. 283. Mata JA, Livne PM, Gibbons MD. Urinary ascites: Complication of umbilical artery catheterization. Urology. 1987;30(4):375-7. 284. McGuire NC, Schulman R, Ridgway MD, Bollero G. Detection of Occult Urinary Tract Infections in Dogs With Diabetes Mellitus. Journal of the American Animal Hospital Association. 2002;38(6):541-4. 285. Mendoza-Lopez CI, Del-Angel-Caraza J, Quijano-Hernandez IA, Barbosa-Mireles MA. Analysis of lower urinary tract disease of dogs. Brazilian Veterinary Research. 2017;37(11):1275-80. 216

286. Michaud L. Bladder Diseases. In: Zhang Y, editor. Encyclopedia of Global Health. Thousand Oaks: SAGE Publications, Inc.; 2008. p. 260-3. 287. Microsoft Corporation. Microsoft Excel 365. 2010. 288. Moberg FS, Langhorn R, Bertelsen PV, Pilegaard LM, Sørensen TM, Bjørnvad CR, Damborg P, Kieler IN, Jessen LR. Subclinical bacteriuria in a mixed population of 179 middle-aged and elderly cats: a prospective cross-sectional study. Journal of Feline Medicine and Surgery. 2019:1098612X19874141. 289. Mochizuki H, Shapiro SG, Breen M. Detection of BRAF Mutation in Urine DNA as a Molecular Diagnostic for Canine Urothelial and Prostatic Carcinoma. PLoS One. 2015;10(12):e0144170. 290. Mohamaden WI, Hamad R, Bahr HI. Alterations of Pro-Inflammatory Cytokines and Tissue Protein Expressions in Cats with Interstitial Cystitis. Pakistan Veterinary Journal. 2019;39(2):151-6. 291. Moldwin RM. Similarities between interstitial cystitis and male chronic pelvic pain syndrome. Current Urology Reports. 2002;3:313-8. 292. Montalbetti N, Rued AC, Clayton DR, Ruiz WG, Bastacky SI, Prakasam HS, Eaton AF, Kullmann FA, Apodaca G, Carattino MD. Increased urothelial paracellular transport promotes cystitis. American Journal of Physiology-Renal Physiology. 2015;309(12):F1070-81. 293. Morise AP, Detrano R, Bobbio M, Diamond GA. Development and validation of a logistic regression-derived algorithm for estimating the incremental probability of coronary artery disease before and after exercise testing. Journal of the American College of Cardiology. 1992;20(5):1187- 96. 294. Morton D, Kemp RK, Francke-Carroll S, Jensen K, McCartney J, Monticello TM, Perry R, Pulido O, Roome N, Schafer K, Sellers R, Snyder PW. Best Practices for Reporting Pathology Interpretations within GLP Toxicology Studies. Toxicol Pathol. 2006;34(6):806-9. 295. Munson L, Craig LE, Miller MA, Kock ND, Simpson RM, Wellman ML, Sharkey LC, Birkebak TA. Elements of Good Training in Anatomic Pathology. Veterinary Pathology. 2010;47(5):995-1002. 296. Murakami T, Feeney DA, Bahr KL. Analysis of clinical and ultrasonographic data by use of logistic regression models for prediction of malignant versus benign causes of ultrasonographically detected focal liver lesions in dogs. American Journal of Veterinary Research. 2012;73(6):821-9. 297. Nakamura K, Kawamoto S, Osuga T, Morita T, Sasaki N, Morishita K, Ohta H, Takiguchi M. Left Atrial Strain at Different Stages of Myxomatous Mitral Valve Disease in Dogs. Journal of Veterinary Internal Medicine. 2017;31(2):316-25. 298. Nakhleh RE, Gephardt G, Zarbo RJ. Necessity of clinical information in surgical pathology. A college of American pathologists Q-Probes study of 771 475 surgical pathology cases from 341 institutions. Archives of Pathology and Laboratory Medicine. 1999;123:615-9. 299. Nam EH, Ko S, Chae JS, Hwang CY. Characterization and zoonotic potential of uropathogenic Escherichia coli isolated from dogs. J Microbiol Biotechnol. 2013;23(3):422-9. 300. Neugent ML, Gadhvi J, Palmer KL, Zimmern PE, De Nisco NJ. Detection of Tissue-resident Bacteria in Bladder Biopsies by 16S rRNA Fluorescence In Situ Hybridization. J Vis Exp. 2019(152). 301. Nicholson AG, Torkko K, Viola P, Duhig E, Geisinger K, Borczuk AC, Hiroshima K, Tsao MS, Warth A, Lantuejoul S, Russell PA, Thunnissen E, Marchevsky A, Mino-Kenudson M, Beasley MB, Botling J, Dacic S, Yatabe Y, Noguchi M, Travis WD, Kerr K, Hirsch FR, Chirieac LR, Wistuba II, Moreira A, Chung J-H, Chou TY, Bubendorf L, Chen G, Pelosi G, Poleri C, Detterbeck FC, Franklin WA. Interobserver Variation among Pathologists and Refinement of Criteria in Distinguishing Separate Primary Tumors from Intrapulmonary Metastases in Lung. Journal of Thoracic Oncology. 2018;13(2):205-17. 302. Nieuwenhuis R, te Grotenhuis M, Pelzer B. influence.ME: tools for detecting influential data in mixed effects models. R journal. 2012;4(2):38-47.

217

303. Niimi A, Igawa Y, Aizawa N, Honma T, Nomiya A, Akiyama Y, Kamei J, Fujimura T, Fukuhara H, Homma Y. Diagnostic value of urinary CXCL10 as a biomarker for predicting Hunner type interstitial cystitis. Neurourology and Urodynamics. 2018;37(3):1113-9. 304. Nordling J, Anjum FH, Bade JJ, Bouchelouche K, Bouchelouche P, Cervigni M, Elneil S, Fall M, Hald T, Hanus T, Hedlund H, Hohlbrugger G, Horn T, Larsen S, Leppilahti M, Mortensen S, Nagendra M, Oliveira PD, Osborne J, Riedl C, Sairanen J, Tinzl M, Wyndaele JJ. Primary evaluation of patients suspected of having interstitial cystitis (IC). Eur Urol. 2004;45(5):662-9. 305. Norris CR, Williams BJ, Ling GV, Franti CE, Johnson DL, Ruby AL. Recurrent and persistent urinary tract infections in dogs: 383 cases. Journal of the American Animal Hospital Association. 2000;36:484-92. 306. Northrup NC, Howerth EW, Harmon BG, Brown CA, Carmicheal KP, Garcia AP, Latimer KS, Munday JS, Rakich PM, Richey LJ, Stedman NL, Gieger TL. Variation among Pathologists in the Histologic Grading of Canine Cutaneous Mast Cell Tumors with Uniform Use of a Single Grading Reference. Journal of Veterinary Diagnostic Investigation. 2005;17(6):561-4. 307. O'Leary MP, Sant GR, Fowler FJJ, Whitmore KE, Spolarich-Kroll J. The interstitial cystitis symptom index and problem index. Urology. 1997;49(5A Suppl):58-63. 308. Offersen BV, Knap MM, Marcussen N, Horsman MR, Hamilton-Dutoit S, Overgaard J. Intense inflammation in bladder carcinoma is associated with angiogenesis and indicates good prognosis. British Journal of Cancer. 2002;87(12):1422-30. 309. Offiah I, Didangelos A, Dawes J, Cartwright R, Khullar V, Bradbury EJ, O'Sullivan S, Williams D, Chessell IP, Pallas K. The expression of inflammatory mediators in bladder pain syndrome. European Urology. 2016;70(2):283-90. 310. Ogawa T, Homma T, Igawa Y, Seki S, Ishizuka O, Imamura T, Akahane S, Homma Y, Nishizawa O. CXCR3 binding chemokine and TNFSF14 over expression in bladder urothelium of patients with ulcerative interstitial cystitis. J Urol. 2010;183(3):1206-12. 311. Oh SB, Tran PB, Gillard SE, Hurley RW, Hammond DL, Miller RJ. Chemokines and glycoprotein120 produce pain hypersensitivity by directly exciting primary nociceptive neurons. J Neurosci. 2001;21(14):5027-35. 312. Okafor CC, Pearl DL, Blois SL, Lefebvre SL, Yang M, Lund EM, Dewey CE. Factors associated with hematuric struvite crystalluria in cats. Journal of Feline Medicine & Surgery. 2019;21(10):922-30. 313. Olin SJ, Bartges JW. Urinary Tract Infections: Treatment/Comparative Therapeutics. Veterinary Clinics of North America: Small Animal Practice. 2015;45(4):721-46. 314. Osborne C, Lees G. Bacterial infections of the canine and feline urinary tract. In: Osborne C, Finco R, editors. Canine and feline nephrology and urology. First ed. Philadelphia: Williams & Wilkins.; 1995. p. 798-885. 315. Osborne CA, Caywood DD, Johnston GR, Polzin DJ, Lulich JP, Kruger JM, Ulrich LK. Feline Perineal Urethrostomy, A Potential Cause of Feline Lower Urinary Tract Disease. Veterinary Clinics of North America: Small Animal Practice. 1996;26(3):535-49. 316. Osborne CA, Kruger JM, Lulich JP. Feline Lower Urinary Tract Disorders Definition of Terms and Concepts. Veterinary Clinical North American Small Animal Practice. 1996;26(2):169-79. 317. Osborne CA, Kruger JM, Lulich JP, Polzin DJ. Disorders of the feline lower urinary tract. In: Osborne CA, Finco D, editors. Canine and feline nephrology and urology. 1st ed. Philadelphia Lippincott Williams and Wilkins; 1995. p. 625-9. 318. Osborne CA, Lulich JP, Bartges JW, Unger LK, Thumchai R, Koehler LA, Bird KA, Felice LJ. Canine and feline urolithiasis: relationship of etiopathogenesis to treatment and prevention. In: Osborne CA, Finco D, editors. Canine and feline nephrology and urology. Philadelphia: Lea & Febiger 1995. p. 798-888.

218

319. Osborne CA, Lulich JP, Kruger JM, Ulrich LK, Koehler LA. Analysis of 451,891 canine uroliths, feline uroliths, and feline urethral plugs from 1981 to 2007: perspectives from the Minnesota Urolith Center. Veterinary Clinical North American Small Anim Practice. 2009;39(1):183-97. 320. Otto W, Denzinger S, Fritsche HM, Burger M, Wieland WF, Hofstadter F, Hartmann A, Bertz S. The WHO classification of 1973 is more suitable than the WHO classification of 2004 for predicting survival in pT1 urothelial bladder cancer. Bju International. 2011;107(3):404-8. 321. Paepe D, Verjans G, Duchateau L, Piron K, Ghys L, Daminet S. Routine health screening: findings in apparently healthy middle-aged and old cats. Journal of Feline Medicine & Surgery. 2013;15(1):8-19. 322. Pai RK, Chen Y, Jakubowski MA, Shadrach BL, Plesec TP, Pai RK. Colorectal carcinomas with submucosal invasion (pT1): analysis of histopathological and molecular factors predicting lymph node metastasis. Modern Pathology. 2017;30(1):113-22. 323. Panboon I, Asawakarn S, Pusoonthornthum R. Urine protein, urine protein to creatinine ratio and N-acetyl-beta-D-glucosaminidase index in cats with idiopathic cystitis vs healthy control cats. Journal of Feline Medicine and Surgery. 2017;19(8):869-75. 324. Pang J, Borjeson TM, Parry NMA, Fox JG. Struvite Urolithiasis in Long-Evans Rats. Comparative medicine. 2015;65(6):486-91. 325. Pang X, Marchand J, Sant G, Kream R, Theoharides T, J. Increased number of substance P positive nerve fibres in interstitial cystitis. British Journal of Urology. 1995;75(6):744-50. 326. Pankov R, Yamada KM. Fibronectin at a glance. J Cell Sci. 2002;115(20):3861-3. 327. Pantanowitz L, Sinard JH, Henricks WH, Fatheree LA, Carter AB, Contis L, Beckwith BA, Evans AJ, Lal A, Parwani AV. Validating Whole Slide Imaging for Diagnostic Purposes in Pathology: Guideline from the College of American Pathologists Pathology and Laboratory Quality Center. Archives of Pathology & Laboratory Medicine. 2013;137(12):1710-22. 328. Parsons CL, Stein P, Zupkas P, Chenoweth M, Argade SP, Proctor JG, Datta A, Trotter RN. Defective Tamm-Horsfall protein in patients with interstitial cystitis. J Urol. 2007;178(6):2665-70. 329. Parys M, Yuzbasiyan-Gurkan V, Kruger JM. Serum cytokine profiling in cats with acute idiopathic cystitis. J Vet Intern Med. 2018;32(1):274-9. 330. Patrick DJ, Fitzgerald SD, Sesterhenn IA, Davis CJ, Kiupel M. Classification of canine urinary bladder urothelial tumours based on the World Health Organization/International Society of Urological Pathology consensus classification. Journal of Comparative Pathology. 2006;135(4):190- 9. 331. Peeker R, Enerback, L., Fall, M., Aldenborg, F. Recruitment, distribution and phenotypes of mast cells in interstitial cystiti. The Journal of Urology. 2000;163:1009-15. 332. Penning LC, Vrieling HE, Brinkhof B, Riemers FM, Rothuizen J, Rutteman GR, Hazewinkel HA. A validation of 10 feline reference genes for gene expression measurements in snap-frozen tissues. Vet Immunol Immunopathol. 2007;120(3-4):212-22. 333. Perry R, Farris G, Bienvenu JG, Dean C, Jr., Foley G, Mahrt C, Short B, Society of Toxicologic P. Society of Toxicologic Pathology position paper on best practices on recovery studies: the role of the anatomic pathologist. Toxicol Pathol. 2013;41(8):1159-69. 334. Peters KM, Jayabalan N, Bui D, Killinger K, Chancellor M, Tyagi P. Effect of Sacral Neuromodulation on Outcome Measures and Urine Chemokines in Interstitial Cystitis/Painful Bladder Syndrome Patients. Low Urin Tract Symptoms. 2015;7(2):77-83. 335. Peters KM, Killinger KA, Mounayer MH, Boura JA. Are ulcerative and nonulcerative interstitial cystitis/painful bladder syndrome 2 distinct diseases? A study of coexisting conditions. Urology. 2011;78(2):301-8. 336. Pinter E, Pozsgai G, Hajna Z, Helyes Z, Szolcsanyi J. Neuropeptide receptors as potential drug targets in the treatment of inflammatory conditions. Br J Clin Pharmacol. 2014;77(1):5-20. 337. Piyarungsri K, Tangtrongsup S, Thitaram N, Lekklar P, Kittinuntasilp A. Prevalence and risk factors of feline lower urinary tract disease in Chiang Mai, Thailand. Sci Rep. 2020;10(1):196. 219

338. Popov Z, Gil-Diez de Medina S, Lefrere-Belda M-A, Hoznek A, Bastufi-Garin S, Abbou CC, Thiery JP, Radvanyi F, Chopin DK. Low E-cadherin expression in bladder cancer at the transcriptional and protein level provides prognostic information. British Journal of Cancer. 2000;83(2):209-14. 339. Puchot ML, Cook AK, Pohlit C. Subclinical bacteriuria in cats: prevalence, findings on contemporaneous urinalyses and clinical risk factors. Journal of Feline Medicine & Surgery. 2017;19(12):1238-44. 340. R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2017. 341. Raab SS, Oweity T, Hughes JH, Salomao DR, Kelley CM, Flynn CM, D'Antonio JA, Cohen MB. Effect of clinical history on diagnostic accuracy in the cytologic interpretation of bronchial brush border specimens. American Journal of Clinical Pathology. 2000;114(1):78-83. 342. Ramos MI, Perez SG, Aarrass S, Helder B, Broekstra P, Gerlag DM, Reedquist KA, Tak PP, Lebre MC. FMS-related tyrosine kinase 3 ligand (Flt3L)/CD135 axis in rheumatoid arthritis. Arthritis Res Ther. 2013;15(6):R209. 343. Ramsey SA, Xu T, Goodall C, Rhodes AC, Kashyap A, He J, Bracha S. Cross-species analysis of the canine and human bladder cancer transcriptome and exome. Genes Chromosomes Cancer. 2017;56(4):328-43. 344. Reeves MJ, Curtis CR, Salman MD, Stashak TS, Reif JS. Validation of logistic regression models used in the assessment of prognosis and the need for surgery in equine colic patients. Preventive Veterinary Medicine. 1992;13(3):155-72. 345. Regan RC, Rassnick KM, Balkman CE, Bailey DB, McDonough SP. Comparison of first- opinion and second-opinion histopathology from dogs and cats with cancer: 430 cases (2001–2008). Veterinary and Comparative Oncology. 2010;8(1):1-10. 346. Reid CN, Stevenson M, Abogunrin F, Ruddock MW, Emmert-Streib F, Lamont JV, Williamson KE. Standardization of diagnostic biomarker concentrations in urine: the hematuria caveat. PLoS One. 2012;7(12):e53354-e. 347. Ribeiro-Silva A, Zhang H, Jeffrey SS. RNA extraction from ten year old formalin-fixed paraffin-embedded breast cancer samples: a comparison of column purification and magnetic bead- based technologies. BMC Molecular Biology. 2007;8:118-. 348. Richter B, Roslind A, Hesse U, Nordling J, Johansen JS, Horn T, Hansen AB. YKL-40 and mast cells are associated with detrusor fibrosis in patients diagnosed with bladder pain syndrome/interstitial cystitis according to the 2008 criteria of the European Society for the Study of Interstitial Cystitis. Histopathology. 2010;57(3):371-83. 349. Roccabianca P, Woo JC, Moore PF. Characterization of the diffuse mucosal associated lymphoid tissue of feline small intestine. Veterinary Immunology and Immunopathology. 2000;75:27-42. 350. Roe FJC. Toxicity Testing: Some Principles and Some Pitfalls in Histopathological Evaluation. Human Toxicology. 1988;7(5):405-10. 351. Romano RC, Novotny PJ, Sloan JA, Comfere NI. Measures of completeness and accuracy of clinical information in skin biopsy requisition forms, an analysis of 249 cases. American Journal of Clinical Pathology. 2016;146:727-35. 352. Rooney P, Srivastava A, Watson L, Quinlan LR, Pandit A. Hyaluronic acid decreases IL-6 and IL-8 secretion and permeability in an inflammatory model of interstitial cystitis. Acta Biomater. 2015;19:66-75. 353. Rosenberg MT, Hazzard M. Prevalence of interstitial cystitis symptoms in women: a population based study in the primary care office. Journal of Urology. 2005;174:2231-4. 354. Roth L, Walton AM, Leib MS, Burrows CF. A grading system for lymphocytic plasmacytic colitis in dogs. Journal of Veterinary Diagnostic Investigation. 1990;2(4):257-62.

220

355. Rubio-Diaz DE, Pozza ME, Dimitrakov J, Gilleran JP, Giusti MM, Stella JL, Rodriguez-Saona LE, Buffington CA. A candidate serum biomarker for bladder pain syndrome/interstitial cystitis. Analyst. 2009;134(6):1133-7. 356. Rutland BE, Nimmo J, Goldsworthy M, Simcock JO, Simpson KW, Kuntz CA. Successful treatment of malakoplakia of the bladder in a kitten. J Feline Med Surg. 2013;15(8):744-8. 357. Sant GR, Kempuraj D, Marchand JE, Theoharides TC. The mast cell in interstitial cystitis: role in pathophysiology and pathogenesis. Urology. 2007;69(4 Suppl):34-40. 358. Sant GR, Theoharides TC. The role of the mast cell in interstitial cystitis. Urologic Clinics of North America. 1994;21(1):41-53. 359. Sapin CdF, Silva-Mariano LC, Bassi JN, Grecco FB. Anatomo-pathological and epidemiological analysis of urinary tract lesions in dogs. Ciência Rural. 2016;46(8):1443-9. 360. Saposnik G, Redelmeier D, Ruff CC, Tobler PN. Cognitive biases associated with medical decisions: a systematic review. BMC medical informatics and decision making. 2016;16(1):138-. 361. Sarma KP. On the Nature of Cystitis Follicularis. The Journal of Urology. 1970;104(5):709-14. 362. Scholes D, Hawn TR, Roberts PL, Li SS, Stapleton AE, Zhao LP, Stamm WE, Hooton TM. Family history and risk of recurrent cystitis and pyelonephritis in women. J Urol. 2010;184:564-9. 363. Schrepf A, O’donnell M, Luo Y, Bradley CS, Kreder K, Lutgendorf S, Network MAttSoCPPR. Inflammation and inflammatory control in interstitial cystitis/bladder pain syndrome: Associations with painful symptoms. PAIN®. 2014;155(9):1755-61. 364. Schwalenberg T, Stolzenburg JU, Ho TP, Mallock T, Hartenstein S, Alexander H, Zimmermann G, Hohenfellner R, Denzinger S, Burger M, Horn LC, Neuhaus J. Enhanced urothelial expression of human chorionic gonadotropin beta (hCGbeta) in bladder pain syndrome/interstitial cystitis (BPS/IC). World J Urol. 2012;30(3):411-7. 365. Sculptoreanu A, de Groat WC, Buffington CA, Birder LA. Abnormal excitability in capsaicin- responsive DRG neurons from cats with feline interstitial cystitis. Experimental neurology. 2005;193(2):437-43. 366. Seaman R, Bartges J. Canine struvite urolithiasis. Compendium on Continuing Education for the Practicing Veterinarian. 2001;23(5):407-20. 367. Seely JC, Nyska A. National Toxicology Program Nonneoplastic Lesion Atlas: The National Toxicology Program of the United States Department of Health and Human Services; 2014 [A guide for standardizing terminology in toxicologic pathology for rodents.]. Available from: https://ntp.niehs.nih.gov/nnl/index.htm. 368. Seguin MA, Vaden SL, Altier C, Stone E, Levine JF. Persistent urinary tract infections and reinfections in 100 dogs (1989-1999). Journal of Veterinary Internal Medicine. 2003;17:622-31. 369. Selvaraju G, Balasubramaniam A, Rajendran D, Kannon D, Geetha M. Multiple linear regression model for forecasting Bluetongue disease outbreak in sheep of North-west agroclimatic zone of Tamil Nadu, India. Veterinary World. 2013;6(6):321-4. 370. Seth JH, Sahai A, Khan MS, van der Aa F, de Ridder D, Panicker JN, Dasgupta P, Fowler CJ. Nerve growth factor (NGF): a potential urinary biomarker for overactive bladder syndrome (OAB)? BJU Int. 2013;111(3):372-80. 371. Shahid M, Lee MY, Yeon A, Cho E, Sairam V, Valdiviez L, You S, Kim J. Menthol, a unique urinary volatile compound, is associated with chronic inflammation in interstitial cystitis. Scientific Reports. 2018;8(1). 372. Sheerin NS. Urinary tract infection. Medicine. 2011;39(7):384-9. 373. Shie JH, Kuo HC. Higher levels of cell apoptosis and abnormal E-cadherin expression in the urothelium are associated with inflammation in patients with interstitial cystitis/painful bladder syndrome. BJU Int. 2010;108:E136-E41. 374. Shipe ME, Deppen SA, Farjah F, Grogan EL. Developing prediction models for clinical use using logistic regression: an overview. Journal of Thoracic Disease. 2019:S574-S84.

221

375. Sim J, Wright CC. The kappa statistic in reliability studies: use, interpretation, and sample size requirements. Phys Ther. 2005;85(3):257-68. 376. Skov BG, Braendstrup O, Hirsch FR, Lauritzen AF, Nielsen W, Skov T. Are pathologists biased by clinical information?: a blinded cross-over study of the histopathological diagnosis of mesothelial tumours versus pulmonary adenocarcinoma. Lung Cancer. 1994;11:365-72. 377. Smaldone MC, Vodovotz Y, Tyagi V, Barclay D, Philips BJ, Yoshimura N, Chancellor MB, Tyagi P. Multiplex analysis of urinary cytokine levels in rat model of cyclophosphamide-induced cystitis. Urology. 2009;73(2):421-6. 378. Specht A, Kruger JM, Fitzgerald SD, Kiupel M, Hauptman J, editors. Histochemical and immunohistochemical light microscopic features of chronic feline idiopathic cystitis. 22nd annual American College of Veterinary Internal Medicine forum; 2004 June 9-12 Lakewood, Colo. : American College of Veterinary Internal Medicine. 379. Sperandei S. Understanding logistic regression analysis. Biochemia medica. 2014;24(1):12-8. 380. Srikousthubha, Sukesh, Raghuveer CV, Hingle S. Profile of lesions in cystoscopic bladder biopsies: a histopathological study. Journal of Clinical and Diagnostic Research. 2013;7(8):1609-12. 381. Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue processing on the content and integrity of nucleic acids. Am J Pathol. 2002;161(6):1961-71. 382. StataCorp. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC; 2017. 383. Stav K, Dwyer PL. Urinary bladder stones in women. Obstet Gynecol Surv. 2012;67(11):715- 25. 384. Steers WD, Kolbeck S, Creedon D, Tuttle JB. Nerve Growth Factor in the Urinary Bladder of the Adult Regulates Neuronal Form and Function. J Clin Invest. 1991;88:1709-15. 385. Stella J, Croney C, Buffington T. Effects of stressors on the behavior and physiology of domestic cats. Applied Animal Behaviour Science. 2013;143(2-4):157-63. 386. Sycamore KF, Poorbaugh VR, Pullin SS, Ward CR. Comparison of urine and bladder or urethral mucosal biopsy culture obtained by transurethral cystoscopy in dogs with chronic lower urinary tract disease: 41 cases (2002 to 2011). Journal of Small Animal Practice. 2014;55(7):364-8. 387. Syme HM. Stones in cats and dogs: What can be learnt from them? Arab journal of urology. 2012;10(3):230-9. 388. Szymanska B, Matuszewski MA, Dembowski J, Zdrojowy R, Dlugosz A. Uroplakin IIIa is a marker in bladder cancer but seems not to reflect chemical carcinogenesis. Biomed Res Int. 2018:1- 9. 389. Tammela T, Autio-Harmainen H, Lukkarinen O, Sormunen R. Effect of prolonged experimental distension on the function and ultrastructure of the canine urinary bladder. Ann Chir Gynaecol. 1991;80(3):301-6. 390. Tan WS, Tan WP, Tan M-Y, Khetrapal P, Dong L, deWinter P, Feber A, Kelly JD. Novel urinary biomarkers for the detection of bladder cancer: A systematic review. Cancer Treatment Reviews. 2018;69:39-52. 391. Tang KWA, Toh QC, Teo BW. Normalisation of urinary biomarkers to creatinine for clinical practice and research--when and why. Singapore medical journal. 2015;56(1):7-10. 392. Thavarajah R, Mudimbaimannar VK, Elizabeth J, Rao UK, Ranganathan K. Chemical and physical basics of routine formaldehyde fixation. Journal of oral and maxillofacial pathology : JOMFP. 2012;16(3):400-5. 393. The Cochrane Collaboration. Review Manager (RevMan). 5.3 ed. Copenhagen: The Nordic Cochrane Centre; 2014. p. [Computer Program]. 394. The Governing Council of the Cat Fancy. The Governing Council of the Cat Fancy 2017 [The UK's Premier Cat Registration body]. Available from: https://www.gccfcats.org/Cat-Breeds. 395. The Kennel Club. The Kennel Club Limited Breed Information Centre 2017 [Breed Information Centre for the Kennel Club, UK]. Available from: https://www.thekennelclub.org.uk/services/public/breed/Default.aspx. 222

396. The University of Sydney. OMIA - Online Mendelian Inheritance in Animals 2019 [Available from: https://omia.org/home/. 397. Thompson MF, Litster AL, Platell JL, Trott DJ. Canine bacterial urinary tract infections: new developments in old pathogens. The Veterinary Journal. 2011;190(1):22-7. 398. Tion M, Dvorska J, Saganuwan S. A review on urolithiasis in dogs and cats. Bulgarian Journal of Veterinary Medicine. 2015;18(1):1-18. 399. Tirlapur SA, Kuhrt K, Chaliha C, Ball E, Meads C, Khan KS. The 'evil twin syndrome' in chronic pelvic pain: a systematic review of prevalence studies of bladder pain syndrome and endometriosis. International Journal of Surgery. 2013;11(3):233-7. 400. Tirumuru S, Al-Kurdi D, Latthe P. Intravesical botulinum toxin A injections in the treatment of painful bladder syndrome/interstitial cystitis: a systematic review. International Urogynecology Journal. 2010;21(10):1285-300. 401. Tizhoosh HR, Pantanowitz L. Artificial Intelligence and Digital Pathology: Challenges and Opportunities. Journal of pathology informatics. 2018;9:38-. 402. Tomaszewski JE, Landis JR, Russack V, Williams TM, Wang L, Hardy C, Brensinger C, Matthews YL, Abele ST, Kusek JW, Nyberg LM. Biopsy features are associated with primary symptoms in interstitial cystitis: results from the interstitial cystitis database study. Urology. 2001;57(Suppl 6A):67-81. 403. Tonyali S, Ates D, Akbiyik F, Kankaya D, Baydar D, Ergen A. Urine nerve growth factor (NGF) level, bladder nerve staining and symptom/problem scores in patients with interstitial cystitis. Adv Clin Exp Med. 2018;27(2):159-63. 404. Torres NV, Santos G. The (Mathematical) Modeling Process in Biosciences. Frontiers in Genetics. 2015;6(354). 405. Torricelli FCM, Mazzucchi E, Danilovic A, Coelho RF, Srougi M. Surgical management of bladder stones: literature review. Revista do Colegio Brasileiro de Cirurgioes. 2013;40(3):227-33. 406. Treutlein G, Deeg CA, Hauck SM, Amann B, Hartmann K, Dorsch R. Follow-up protein profiles in urine samples during the course of obstructive feline idiopathic cystitis. Vet J. 2013;198(3):625-30. 407. Treutlein G, Dorsch R, Euler KN, Hauck SM, Amann B, Hartmann K, Deeg CA. Novel potential interacting partners of fibronectin in spontaneous animal model of interstitial cystitis. PLoS One. 2012;7(12):e51391. 408. Tsai KW, Kuo WT, Jeng SY. Tight Junction Protein 1 Dysfunction Contributes to Cell Motility in Bladder Cancer. Anticancer Res. 2018;38(8):4607-15. 409. Tseng-Rogenski S, Liebert M. Interleukin-8 is essential for normal urothelial cell survival. Am J Physiol Renal Physiol. 2009;297(3):F816-21. 410. Tsung JS. Institutional pathology consultation. Am J Surg Pathol. 2004;28(3):399-402. 411. Turashvili G, Leung S, Turbin D, Montgomery K, Gilks B, West R, Carrier M, Huntsman D, Aparicio S. Inter-observer reproducibility of HER2 immunohistochemical assessment and concordance with fluorescent in situ hybridization (FISH): pathologist assessment compared to quantitative image analysis. BMC Cancer. 2009;9:165. 412. Turashvili G, Yang W, McKinney S, Kalloger S, Gale N, Ng Y, Chow K, Bell L, Lorette J, Carrier M, Luk M, Aparicio S, Huntsman D, Yip S. Nucleic acid quantity and quality from paraffin blocks: defining optimal fixation, processing and DNA/RNA extraction techniques. Exp Mol Pathol. 2012;92(1):33-43. 413. Tyagi P, Barclay D, Zamora R, Yoshimura N, Peters K, Vodovotz Y, Chancellor M. Urine cytokines suggest an inflammatory response in the overactive bladder: a pilot study. Int Urol Nephrol. 2010;42(3):629-35. 414. Tyagi P, Killinger K, Tyagi V, Nirmal J, Chancellor M, Peters KM. Urinary chemokines as noninvasive predictors of ulcerative interstitial cystitis. J Urol. 2012;187(6):2243-8.

223

415. Uehling DT. Current concepts of the urinary bladder defenses against infection. Int Urogynecol J. 1991;2:32-5. 416. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software; Vol 1, Issue 3 (2011). 2011. 417. van de Merwe JP, Nordling J, Bouchelouche P, Bouchelouche K, Cervigni M, Daha LK, Elneil S, Fall M, Hohlbrugger G, Irwin P, Mortensen S, van Ophoven A, Osborne JL, Peeker R, Richter B, Riedl C, Sairanen J, Tinzl M, Wyndaele JJ. Diagnostic criteria, classification, and nomenclature for painful bladder syndrome/interstitial cystitis: an ESSIC proposal. European Urology. 2008;53(1):60- 7. 418. van Ginneken AM, van der Lei J, editors. Understanding differential diagnostic disagreement in pathology. Symposium on Computer Applications in Medical Care; 1991: American Medical Informatics Association. 419. van Maldegem F, de Wit M, Morsink F, Musler A, Weegenaar J, van Noesel CJ. Effects of processing delay, formalin fixation, and immunohistochemistry on RNA Recovery From Formalin- fixed Paraffin-embedded Tissue Sections. Diagn Mol Pathol. 2008;17(1):51-8. 420. Vapalahti K, Virtala A-M, Joensuu TA, Tiira K, Tähtinen J, Lohi H. Health and behavioral survey of over 8000 Finnish cats. Frontiers in veterinary science. 2016;3:70. 421. Varghese F, Bukhari AB, Malhotra R, De A. IHC Profiler: An Open Source Plugin for the Quantitative Evaluation and Automated Scoring of Immunohistochemistry Images of Human Tissue Samples. PLoS One. 2014;9(5):e96801. 422. Vera PL, Iczkowski KA, Wang X, Meyer-Siegler KL. Cyclophosphamide-induced cystitis increases bladder CXCR4 expression and CXCR4-macrophage migration inhibitory factor association. PLoS One. 2008;3(12):e3898. 423. Vera PL, Preston DM, Moldwin RM, Erickson DR, Mowlazadeh B, Ma F, Kouzoukas DE, Meyer-Siegler KL, Fall M. Elevated Urine Levels of Macrophage Migration Inhibitory Factor in Inflammatory Bladder Conditions: A Potential Biomarker for a Subgroup of Interstitial Cystitis/Bladder Pain Syndrome Patients. Urology. 2018;116:55-62. 424. Vollmer RT. Multivariate Statistical Analysis for Pathologists:Part I, The Logistic Model. American Journal of Clinical Pathology. 1996;105(1):115-26. 425. Waddell AH. Capillaria Feliscati in the bladder of cats in Australia. Aust Vet J. 1967;43:297. 426. Walker SJ, Colaco M, Koslov DS, Keys T, Evans RJ, Badlani GH, Andersson KE. Transcriptome analysis of bladder biopsy from interstitial cystitis/bladder pain syndrome patients. Genom Data. 2014;2:366-8. 427. Wallerström BI, Wågberg TI. Canine urolithiasis in Sweden and Norway: Retrospective survey of prevalence and epidemiology. Journal of Small Animal Practice. 1992;33(11):534-9. 428. Wan SY, Hartmann FA, Jooss MK, Viviano KR. Prevalence and clinical outcome of subclinical bacteriuria in female dogs. Journal of the American Veterinary Medical Assocation. 2014;245(1):106-12. 429. Warren AL, Donnon TL, Wagg CR, Priest H. Educational interventions to improve cytology visual diagnostic reasoning measured by eye tracking. J Vet Med Educ. 2015;42(1):69-78. 430. Warren JW. Bladder pain syndrome/interstitial cystitis as a functional somatic syndrome. Journal of Psychosomatic Research. 2014;77(6):510-5. 431. Warren JW, Jackson TL, Langenberg P, Meyers DJ, Xu J. Prevalence of interstitial cystitis in first-degree relatives of patients with interstitial cystitis. Urology. 2004;63(1):17-21. 432. Warren JW, van de Merwe JP, Nickel JC. Interstitial cystitis/bladder pain syndrome and nonbladder syndromes: facts and hypotheses. Urology. 2011;78(4):727-32. 433. Watkins LR, Maier SF. Implications of immune-to-brain communication for sickness and pain. Proc Natl Acad Sci U S A. 1999;96(14):7710-3. 434. Webster JD, Dunstan RW. Whole-slide imaging and automated image analysis: considerations and opportunities in the practice of pathology. Vet Pathol. 2014;51(1):211-23. 224

435. Weese JS, Blondeau J, Boothe D, Guardabassi LG, Gumley N, Papich M, Jessen LR, Lappin M, Rankin S, Westropp JL, Sykes J. International Society for Companion Animal Infectious Diseases (ISCAID) guidelines for the diagnosis and management of bacterial urinary tract infections in dogs and cats. The Veterinary Journal. 2019;247:8-25. 436. Weiss AT, Delcour NM, Meyer A, Klopfleisch R. Efficient and cost-effective extraction of genomic DNA from formalin-fixed and paraffin-embedded tissues. Vet Pathol. 2011;48(4):834-8. 437. Weiss GR, O'Donnell MA, Loughlin K, Zonno K, Laliberte RJ, Sherman ML. Phase 1 study of the intravesical administration of recombinant human interleukin-12 in patients with recurrent superficial transitional cell carcinoma of the bladder. J Immunother. 2003;26(4):343-8. 438. Wen H, Lee T, You S, Park S-H, Song H, Eilber KS, Anger JT, Freeman MR, Park S, Kim J. Urinary Metabolite Profiling Combined with Computational Analysis Predicts Interstitial Cystitis- Associated Candidate Biomarkers. J Proteome Res. 2014;14(1). 439. Westropp JL. Evaluation of the effects of stress on the sympathetic nervous system and hypothalamic-pituitary-adrenal axis in cats with feline interstitial cystitis. Missouri: Ohio State University; 2003. 440. Westropp JL, Buffington CAT. In vivo models of interstitial cystitis. Journal of Urology. 2002;167:694-702. 441. White FA, Bhangoo SK, Miller RJ. Chemokines: integrators of pain and inflammation. Nat Rev Drug Discov. 2005;4(10):834-44. 442. White JD, Cave NJ, Grinberg A, Thomas DG, Heuer C. Subclinical bacteriuria in older cats and its association with survival. J Vet Intern Med. 2016;30:1824-9. 443. Wilkonson DR, Erickson DR. Urinary and serologic markers for interstitial cystitis: An update. Current Urology Reports. 2006;7:414-22. 444. Willard M, Mansell J. Correlating clinical activity and histopathologic assessment of gastrointestinal lesion severity: current challenges. Vet Clin North Am Small Anim Pract. 2011;41(2):457-63. 445. Willard MD, Jergens AE, Duncan RB, Leib MS, McCracken MD, DeNovo RC, Helman RG, Slater MR, Harbison JL. Interobserver variation among histopathologic evaluations of intestinal tissues from dogs and cats. J Am Vet Med Assoc. 2002;220:1177-82. 446. Wilson HM, Chun R, Larson VS, Kurzman ID, Vail DM. Clinical signs, treatments, and outcome in cats with transitional cell carcinoma of the urinary bladder: 20 cases (1990-2004). Journal of the American Veterinary Medical Assocation. 2007;231(1):101-6. 447. Witzke KE, Großerueschkamp F, Jütte H, Horn M, Roghmann F, von Landenberg N, Bracht T, Kallenbach-Thieltges A, Käfferlein H, Brüning T, Schork K, Eisenacher M, Marcus K, Noldus J, Tannapfel A, Sitek B, Gerwert K. Integrated Fourier Transform Infrared Imaging and Proteomics for Identification of a Candidate Histochemical Biomarker in Bladder Cancer. Am J Pathol. 2019;189(3):619-31. 448. Wong C, Epstein SE, Westropp JL. Antimicrobial susceptibility patterins in urinary tract infections in dogs (2010-2013). J Vet Intern Med. 2015;29(4):1045-52. 449. Wright JD. Modulation of gap junction expression in injured equine superficial digital flexor tendon. The University of Queensland, School of Veterinary Science; 2013. 450. Wu Q, Li Y, Wang M, Pan XP, Tang YF. Fluorescence in situ hybridization rapidly detects three different pathogenic bacteria in urinary tract infection samples. J Microbiol Methods. 2010;83(2):175-8. 451. Xu W, Zhao Y, Nian S, Feng L, Bai X, Luo X, Luo F. Differential analysis of disease risk assessment using binary logistic regression with different analysis strategies. The Journal of international medical research. 2018;46(9):3656-64. 452. Yagi Y, Gilbertson JR. A relationship between slide quality and image quality in whole slide imaging (WSI). Diagnostic pathology. 2008;3(1):S12.

225

453. Yamada T, Murayama T, Andoh M. Adjuvant hydrodistension under epidural anesthesia for interstitial cystitis. International Journal of Urology. 2003;10(9):463-8. 454. Yamaguchi T, Mukai H, Akiyama F, Arihiro K, Masuda S, Kurosumi M, Kodama Y, Horii R, Tsuda H. Inter-observer agreement among pathologists in grading the pathological response to neoadjuvant chemotherapy in breast cancer. Breast Cancer. 2018;25(1):118-25. 455. Yamamoto Y, Tsuzuki T, Akatsuka J, Ueki M, Morikawa H, Numata Y, Takahara T, Tsuyuki T, Tsutsumi K, Nakazawa R, Shimizu A, Maeda I, Tsuchiya S, Kanno H, Kondo Y, Fukumoto M, Tamiya G, Ueda N, Kimura G. Automated acquisition of explainable knowledge from unannotated histopathology images. Nature Communications. 2019;10(1):5642. 456. Yuri K, Nakata K, Katae H, Yamamoto S, Hasegawa A. Distribution of Uropathogenic Virulence Factors among Escherichia coli Strains Isolated from Dogs and Cats. J Vet Med Sci. 1998;60(3):287-90. 457. Zachary J, McGavin M. Pathologic basis of veterinary disease. 6th ed. Missouri: Mosby; 2016. 458. Zeng Y, Wu XX, Homma Y, Yoshimura N, Iwaki H, Kageyama S, Yoshiki T, Kakehi Y. Uroplakin III-delta4 messenger RNA as a promising marker to identify nonulcerative interstitial cystitis. J Urol. 2007;178(4 Pt 1):1322-7; discussion 7. 459. Zhang CO, Li ZL, Kong CZ. APF, HB-EGF, and EGF biomarkers in patients with ulcerative vs. non-ulcerative interstitial cystitis. BMC Urol. 2005;5:7. 460. Zhang J, Wei S, Liu L, Nagana Gowda GA, Bonney P, Stewart J, Knapp DW, Raftery D. NMR- based metabolomics study of canine bladder cancer. Biochim Biophys Acta. 2012;1822(11):1807-14.

226

Appendix 1: Copy of Animal Ethics Approval Certificate ANRFA/SVS/259/16 by UQ Animal Ethics Committee.

227

228

259-16 #1, Approved.

Title of project/work

Pathological and clinicopathological features of feline and canine cystitis

Where and from whom the material will be obtained from and how it will be obtained (AEC numbers required if from other projects)

1. Urine collected during general anaesthetic, from dogs already undergoing diagnostic bladder endoscopy and/or biopsy for urinary tract disease, at the UQ veterinary medical centre or other

Brisbane specialist clinics.

2. Bladder wall biopsy tissue collected from dogs already undergoing diagnostic bladder endoscopy and/or biopsy for urinary tract disease, at the UQ veterinary medical centre or other

Brisbane specialist clinics.

3. Clinical and diagnostic records from the UQ veterinary medical centre for urinalysis results,

pathology results and clinical history of the abovementioned patients.

If native wildlife/endangered species samples, are permits required?

Confirm that you will be receiving tissue only; no animals will be euthanised solely for this work

Only tissue samples will be used. No animals will be euthanised solely for this work.

Why is the material required and what is hoped to be achieved from the work being conducted?

Cystitis and indications for clinicopathological and histopathological examination of the bladder in dogs and cats are poorly described. Through this work we are hoping to develop a grading scheme for cystitis in companion animals, and to explore the role of neurokinin-1, substance P, nerve growth factor (NGF) and mast cells in the pathogenesis of cystitis in companion animals to build a translational model for human interstitial cystitis. We will also explore the correlation between clinicopathology samples (urine, urinalysis, urine cytology) and biopsy/autopsy specimens of canine and feline bladder to establish the predictive reliability of various clinical diagnostic tests. This work 229

will be invaluable in guiding veterinary surgeons in the clinical management of cystitis, and in further exploring the role of these inflammatory markers in both veterinary and human interstitial cystitis.

How much material is required?

At least twenty (20) paired urine and biopsy samples

Any correlating archived urine, urinalysis, urine cytology and blood samples from these patients.

259-16 #1

Clinical and diagnostic patient records for the abovementioned animals. Relevant clinical data will be extracted from the patient records in a deidentified fashion (signalment, clinical history, treatment history, test results).

Who will be involved in this work?

MPhil student Emily Jones (a registered veterinarian with the Veterinary Surgeons Board of Queensland). Emily will be supervised by diagnostic pathologists Rachel Allavena and Chiara

Palmieri (anatomic pathology) and Karen Jackson (clinical pathology).

Where will the material be stored and how it will be stored?

UQ School of Veterinary Science Gatton Campus. Archival material will be returned to the pathology archives for indefinite storage after use.

Diagnostic records are on secure password protected databases within the School of Veterinary Science. De-identified data only will be extracted from the databases and stored on a password protected computer.

Where will the work be carried out?

UQ School of Veterinary Science Gatton Campus How long will the approval be required for? 3 years

230

Appendix 2: Crude and adjusted odds ratio of biopsy sampling method stratified by disease cases and non-cases. Biopsy Species Disease Status* OR (95% CI) Yes No Total Canine Cases 72 99 171 Non-cases 3 48 51 11.64 (3.48 –38.83) Feline Cases 8 27 35 Non-cases 2 15 17 2.22 (0.42–11.84) Total 85 189 274 Overall Crude 8.00 (3.09–20.74) Mantel-Haenszel adjusted 7.52 (2.92–19.40) *Cystitis, urolithiasis and neoplasia cases only. Non-cases are normal bladder samples. Chi-squared statistics for homogeneity of odds ratios across strata = (χ2 2.77, 1 d.f.), P = 0.10. 2 Chi-squared statistics homogeneity of overall Odds Ratio: (χ 16.11, 1 d.f.), P < 0.001.

231

Appendix 3: All animal, sampling and diagnostic histological variables measured on each bladder slide and used in the logistic regression modelling process

Variable label Variable description Level Value label Animal or source variables Species Species 1 Feline 2 Canine Source Source of sample 1 UQVLS 2 MUSLVS 3 Prospective sampling Sampling variables Full_thickness Full thickness sample 1 No (outer muscularis 2 Yes included) Uroth_denud Percent of urothelium 1 All intact (no denudation) that has been denuded 2 1-25% of urothelium has been denuded 3 26-50% 4 51-99% 5 100% Uroth_artefact Urothelial artefact 1 No causing denudation 2 Yes Oedema_artefact SM oedema artefact 1 No 2 Yes Diagnosis variables Uroth_ulcer Urothelial ulceration 1 No 2 Yes Uroth_hyperp Urothelial hyperplasia 1 No 2 Yes Uroth_react Reactive urothelium 1 No 2 Yes Neoplastic Neoplastic urothelium 1 No 2 Yes Neo_inv SM neoplastic invasion 1 No 2 Yes

232

Variable label Variable description Level Value label Uroth_inflamm Amount of urothelium 1 None that was being 2 1-25% infiltrated by inflammatory cells 3 26-50% 4 >50% SM_haem SM haemorrhage 1 No SM haemorrhage (none) amount 2 Haemorrhage present in up to 25% of the SM (mild) 3 Haemorrhage present in 26-50% of the SM (moderate) 4 Haemorrhage present in >50% of the SM (severe) SM_cong SM congestion 1 No 2 Yes SM_oedema SM oedema 1 No 2 Yes SM_inflamm SM inflammation 1 None amount 2 Occasional inflammatory cell per LPF (100x magnification) 3 <100 inflammatory cells per LPF 4 >100 inflammatory cells per LPF SM_inflamm_type SM inflammation type 1 Primarily lymphocytic 2 Neutrophilic 3 Lymphoplasmacytic 4 Granulomatous Gran_tiss SM granulation tissue 1 No 2 Yes Lymph_agg SM lymphoid 1 No aggregates 2 Yes Det_inflamm Detrusor/muscularis 1 No inflammation present 2 Yes Det_fib Detrusor/muscularis 1 No fibrosis 2 Yes Ser_inflamm Serosal inflammation 1 No

233

Variable label Variable description Level Value label 2 Yes Ser_haem Serosal haemorrhage 1 No 2 Yes Hyperemia Presence of 1 No hyperaemia 2 Yes Organisms Presence of 1 No microorganisms 2 Yes LPF, low power field (100x magnification). SM, submucosa.

234

Appendix 4: Raw data from analysis of histological features of all bladder specimens in the dataset.

235

tefact

ctive

Code Species Dx

denuded hyperp

thickness

oth_

full uroth_ uroth_ulcer uroth_artefact ur uroth_rea neoplastic uroth_haem uroth_inflamm gran_tiss sm_cong sm_haem sm_haem_dist sm_oedema oedema_ar sm_inflamm lymph_aggreg sm_inflam_type det_inflam det_inflam_type det_fib ser_inflam ser_hae organisms changed_ 07/1154 Canine Oth 2 3 1 1 1 1 1 1 1 1 1 2 2 2 1 2 1 2 1 N 1 1 1 1 1 00/0199 Canine Cys 2 1 1 1 1 1 1 1 1 1 1 2 2 2 1 3 1 3 1 N 1 1 1 1 1 00/0400 Canine Uro 2 3 1 1 1 1 1 1 1 1 1 2 1 2 1 3 2 3 1 N 1 1 1 1 1 11-355 Canine Oth 2 5 1 2 N N N N N 1 1 2 3 1 1 2 1 1 1 N 1 1 1 1 1 00/0871 Feline Cys 2 5 1 1 1 1 1 N N 1 1 4 2 1 1 4 1 2 2 2 1 1 2 1 1 00/1328 Canine Norm 2 4 1 1 1 1 1 N N 1 1 2 2 3 1 1 1 N 1 N 1 1 1 1 1 01/0243 Canine Neo 2 1 1 1 2 2 2 1 2 1 1 1 N 1 1 4 1 2 N N N N N 1 1 01/0609 Canine Neo 2 1 1 1 1 1 2 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 01/0289 Canine Cys 2 5 2 2 N N N N N 1 1 2 2 3 1 4 1 2 1 N 1 1 1 2 1 02/0691 Canine Neo 2 1 1 1 2 2 2 1 2 1 1 1 N 3 1 4 1 3 1 N 1 1 1 1 1 02/0300 Canine Cys 2 4 1 1 1 1 1 N N 1 2 3 3 3 1 3 2 3 1 N 1 1 2 1 1 02/0429 Canine Cys 2 4 1 2 1 1 1 N N 1 2 1 N 3 1 4 2 3 1 N 1 1 1 1 1 11-523 Canine Oth 2 3 1 2 1 1 1 1 1 1 2 1 N 2 1 3 1 1 1 N 1 1 1 1 1 02/0686 Canine Cys 2 4 2 1 1 1 1 2 2 1 1 1 N 2 1 4 2 3 1 N 1 1 1 1 1 03-433 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 2 1 1 1 2 1 1 1 N 1 1 1 1 1 04/0228 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 1 N 3 1 2 1 1 1 N 1 1 1 1 1 02/0975 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 1 N 2 2 3 1 3 1 N 1 1 1 1 1 02/1382 Feline Norm 2 5 1 2 1 1 1 N N 1 1 1 N 3 1 1 1 N 1 N 1 1 1 1 1 02-480 Feline Cys 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 3 1 3 2 3 2 2 1 1 1 02-543 Canine Cys 1 4 2 1 2 1 1 1 2 1 2 2 2 4 1 4 2 3 1 N 1 1 1 1 1 02-796 Canine Uro 2 3 2 1 1 1 1 2 2 1 1 2 2 2 1 4 1 2 2 2 1 2 1 1 2 03/0065 Feline Norm 2 3 2 1 1 1 1 2 1 1 2 2 1 2 1 2 1 1 1 N 1 1 1 1 2 04/0706 Canine Neo 2 1 1 1 2 2 2 2 2 1 1 2 2 3 2 4 2 3 1 N 1 1 1 1 1 03/0244 Feline Neo 2 1 1 1 2 2 2 1 2 1 1 1 N 1 1 2 1 1 2 1 1 1 1 1 1

236

03/0281 Canine Cys 2 1 1 1 2 1 1 2 2 1 1 2 2 1 1 4 1 3 1 N 1 1 1 1 1 03/0298 Canine Uro 2 1 1 1 1 1 1 1 2 1 1 1 N 3 1 3 1 3 1 N 1 1 1 1 2 03/1097 Canine Cys 1 2 1 2 1 1 1 1 2 1 1 3 3 3 1 4 1 3 N N N N N 1 1 03/1176 Feline Cys 1 5 2 1 N N N N N 1 2 4 2 4 1 3 1 2 N N N N N 2 1 03-343 Feline Cys 2 5 2 1 N N N N N 1 1 2 2 3 1 4 1 2 2 2 1 2 2 2 1 04/1068 Canine Neo 2 2 2 1 2 2 2 2 1 1 2 2 2 2 1 4 1 2 1 N 1 1 1 1 1 04/1500 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 2 2 1 1 3 1 1 1 N 1 1 1 1 1 05/1487 Canine Neo 1 2 2 1 2 2 2 1 1 1 1 1 N 1 1 4 1 3 N N N 1 1 1 1 04/0426 Canine Cys 1 2 2 1 1 1 1 2 2 1 1 3 2 4 2 3 2 3 1 N 1 N N 1 1 05/1742 Canine Neo 1 1 1 1 2 2 2 2 2 N N N N N N N N N N N N N N 2 1 04/0861 Canine Cys 2 4 1 1 1 1 1 1 1 1 1 3 2 5 1 4 1 2 2 2 1 1 1 2 1 06/0973 Canine Neo 1 1 1 1 2 2 2 2 2 N N N N N N N N N N N N N N 1 1 05/1437 Canine Oth 1 1 1 1 2 1 1 1 1 1 1 1 N 3 1 4 1 3 N N N N N 1 1 04/1386 Canine Uro 2 4 2 1 1 1 1 1 1 1 2 3 2 5 1 2 1 3 1 N 1 1 1 1 1 06-462 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 1 N 1 1 4 1 1 1 N 1 1 1 1 1 07/0906 Canine Oth 2 2 2 1 1 2 1 1 1 1 2 3 3 5 1 3 1 2 1 N 1 1 2 1 1 11/0331 Canine Oth 2 1 1 1 1 1 1 1 1 1 2 1 N 1 1 2 1 1 1 N 1 1 1 1 1 12/0587 Canine Oth 2 2 1 2 1 1 1 2 1 1 2 4 2 3 1 2 1 3 1 N 1 1 2 1 1 05/0006 Canine Cys 1 1 1 1 1 1 1 1 1 N N N N N N N N N N N N N N N 1 06-723 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 1 N 3 1 4 2 3 1 N 1 1 1 1 1 07/0472 Canine Neo 1 1 1 1 2 2 2 1 2 N N N N N N N N N N N N N N 1 1 05/1139 Canine Cys 2 5 N N N N N N N 1 1 1 N 3 1 4 1 2 1 N 1 1 1 1 1 15/0384 Canine Oth 2 1 1 1 1 1 1 1 2 1 2 1 N 2 2 2 1 1 1 N 1 1 2 1 1 05/1305 Canine Cys 1 5 N N N N N N N 2 1 2 2 3 1 4 1 2 1 N 1 1 1 2 1 05/1429 Feline Neo 2 1 1 1 2 2 2 2 1 1 1 2 1 3 1 4 1 3 2 2 1 2 2 1 1 04-395 Feline Oth 2 5 1 2 N N N N N N N N N N N N N N 1 N 1 1 1 1 1 07/0581 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 2 3 2 1 4 1 2 1 N 1 1 1 1 1 05/1587 Canine Uro 2 1 1 1 2 1 1 1 1 1 1 3 1 3 1 3 1 2 1 N 1 N N 1 1 07/0731 Canine Neo 2 1 1 1 2 2 2 2 1 1 1 1 N 1 1 3 1 3 1 N 1 1 1 1 1

237

04-0500 Feline Oth 2 1 1 1 1 1 1 1 1 1 1 2 3 4 1 2 1 1 1 N 1 1 1 1 1 06/0439 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 15/1023 Canine Oth 2 4 1 2 N N N N N 1 1 1 N 2 2 1 1 N 2 3 1 1 1 1 1 06/0546 Canine Norm 2 2 1 1 1 1 1 2 2 1 2 1 N 1 1 1 1 N 1 N 1 1 1 1 1 06/0731 Canine Cys 2 2 1 2 1 1 1 1 1 1 1 2 3 1 1 3 2 3 1 N 1 1 1 1 1 07/0907 Canine Neo 1 1 1 1 2 2 2 2 1 N N N N N N N N N N N N N N 1 1 07/1020 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 2 2 5 2 3 2 3 1 N 1 1 2 1 1 15/0393 Feline Oth 2 4 2 1 1 1 1 2 1 1 1 3 3 3 1 4 1 3 1 N 1 2 2 1 1 08/1541 Canine Oth 2 2 2 1 1 1 1 1 1 1 1 3 3 1 1 2 1 1 1 N 1 1 1 1 1 07/1101 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 1 N 3 1 2 1 3 1 N 1 1 1 1 1 07/1971 Canine Neo 2 1 2 1 2 2 2 1 1 1 1 2 2 1 1 3 1 1 1 N 1 1 1 1 1 07/0026 Canine Cys 2 5 2 1 N N N N N 1 2 4 2 3 1 4 1 2 1 N 1 2 2 1 1 07/0215 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 2 1 2 07/0239 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 2 2 3 1 4 1 3 1 N 1 1 1 1 1 07/0341 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 2 1 1 1 07/0361 Canine Cys 2 5 2 1 N N N N N 1 1 4 2 3 1 4 1 2 2 3 1 1 2 1 1 07-0114 Canine Neo 2 4 1 1 1 2 2 1 1 1 1 1 N 3 1 1 1 N 1 N 1 N N 1 1 07/0550 Canine Norm 2 2 2 1 1 1 1 2 1 1 1 2 2 2 1 2 1 1 1 N 1 1 1 1 2 07/0567 Canine Cys 2 2 1 2 1 1 1 1 2 1 1 1 N 3 1 4 1 3 1 N 1 1 1 1 1 07/0577 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 13/0428 Canine Oth 2 3 1 2 1 1 1 1 1 1 1 2 1 2 1 2 1 3 1 N 1 2 1 1 1 07-0377 Canine Neo 1 1 1 1 2 2 2 1 3 1 1 1 N 3 1 3 2 1 N N N N N 1 1 07/0628 Canine Norm 2 2 1 2 1 1 1 1 1 1 2 1 N 1 1 1 2 N 1 N 1 1 1 1 1 99/1417 Canine Oth 2 2 1 2 1 1 1 1 1 1 1 2 1 1 N 2 1 1 1 N 1 1 2 1 1 07/0660 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 07/0706 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 3 2 1 1 N 1 N 1 1 1 1 1 08/0256 Canine Neo 1 1 1 1 2 2 2 1 1 1 N N N N N N N N N N N N N 1 1 07/0781 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 07/0864 Feline Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1

238

07/0895 Feline Cys 2 5 1 1 1 1 1 N N 1 2 2 2 3 1 2 1 3 1 N 1 1 1 1 1 07/0904 Feline Norm 1 3 2 1 1 1 1 1 1 1 1 1 N 3 2 2 1 3 N N 1 1 1 1 2 96/0607 Feline Oth 2 3 2 1 1 1 1 1 1 1 1 4 3 5 1 3 1 2 2 2 1 2 2 1 1 08/0609 Canine Neo 1 1 1 1 2 2 2 2 2 1 N N N N N N 2 N N N N N N 1 1 07/0916 Feline Cys 2 5 1 1 1 1 1 2 N 1 1 3 2 4 1 3 1 2 2 2 1 1 2 1 1 08/0941 Feline Cys 2 4 2 2 1 1 1 2 1 1 1 4 2 4 1 3 1 3 2 3 1 2 2 1 1 96/1020 Canine Oth 2 5 1 1 1 1 1 1 1 1 2 4 2 5 1 1 1 N 1 N 1 1 2 1 1 08/0813 Canine Neo 1 1 1 1 2 2 2 2 2 1 1 2 2 1 1 4 1 1 N N N N N 1 1 07/1022 Canine Cys 1 1 1 1 2 1 1 1 1 1 1 3 3 3 1 4 1 1 N N N N 1 1 1 07/1031 Canine Cys 2 1 1 1 1 1 1 1 1 1 1 1 N 3 1 3 2 3 1 N 1 1 1 1 1 07/1059 Canine Norm 2 1 1 1 1 1 1 1 1 1 2 1 N 1 1 1 1 N 1 N 1 1 1 1 1 07/1089 Canine Cys 2 1 1 1 2 1 1 1 1 1 2 1 1 1 1 4 2 3 1 N 1 1 1 1 1 08/1436 Canine Neo 2 1 1 1 2 2 2 1 1 1 1 2 1 2 1 3 1 1 1 N 1 1 1 1 1 14/0586 Canine Oth 2 5 1 2 N N N N N 1 2 2 2 3 1 2 1 1 1 N 1 1 1 1 1 07/1819 Feline Neo 2 4 1 2 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 2 1 1 1 17-206 Feline Oth 2 5 1 1 N N N N N 1 2 2 1 3 1 4 1 2 2 2 1 2 1 1 1 07/1880 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 07/1968 Canine Cys 2 2 1 1 1 1 1 1 1 1 1 2 2 1 1 4 2 3 1 N 1 2 1 1 1 08/1694 Canine Neo 1 1 1 1 2 2 2 2 1 1 1 1 N 5 2 3 2 3 N N 1 1 N 1 1 07/2154 Canine Norm 2 1 1 2 1 1 1 1 1 1 1 1 N 1 2 1 1 N 1 N 1 1 1 2 1 07/2155 Canine Norm 2 3 1 2 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 07/2156 Canine Cys 2 1 1 1 1 1 1 1 2 1 2 1 N 1 1 3 1 3 1 N 1 1 1 1 1 07/2196 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 2 1 3 2 3 1 3 1 N 1 1 1 1 2 07-0069 Canine Cys 2 5 2 2 N N N N N 1 1 1 N 2 1 4 1 4 1 N 1 1 1 2 1 08-005 Canine Neo 1 2 2 1 2 2 2 1 1 1 1 2 1 3 1 3 1 1 N N N N N 1 1 08-791 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 1 1 2 1 3 N N N N N 1 1 07-1136 Canine Cys 1 2 1 1 1 1 1 1 1 1 1 1 N 3 2 2 1 1 1 N 1 N N 1 1 12/0445 Feline Cys 2 5 2 1 N N N N N 1 1 4 2 3 1 3 1 3 1 N 1 2 2 1 1 07-553 Canine Cys 1 1 1 1 1 1 1 1 1 1 1 1 N 1 1 3 1 1 N N N N N 1 1

239

07-894 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 2 2 3 1 3 1 1 1 N 1 1 1 1 1 07-914 Canine Cys 1 1 1 1 1 1 1 1 1 1 1 1 N 2 1 2 1 3 1 N 1 N N 1 1 09/0898 Canine Neo 2 1 1 1 2 2 2 2 1 1 1 2 2 3 2 1 1 N 1 N 1 N N 1 1 08/0239 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 1 09/1219 Canine Neo 2 1 1 1 2 2 2 2 2 1 1 2 1 3 1 2 2 1 1 N 1 1 1 1 1 09/1573 Canine Neo 1 1 1 1 2 2 2 2 2 1 1 1 N 3 2 3 1 2 N N N N N 1 1 09/1608 Canine Neo 2 1 1 1 2 2 2 2 2 1 1 1 N 3 2 4 2 2 1 N 1 1 1 1 1 08/0743 Canine Cys 2 2 2 1 1 1 1 2 1 1 1 2 1 3 1 4 1 3 1 N 1 1 1 1 1 09-0263 Canine Neo 2 1 1 1 2 2 2 2 2 1 1 1 N 2 1 4 2 3 1 N 1 1 2 1 1 08/0932 Canine Cys 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 3 1 3 1 N 1 1 1 1 1 12/0516 Feline Cys 2 4 2 1 1 1 1 2 1 1 1 4 2 5 1 3 1 3 1 N 2 1 2 1 1 08/1011 Canine Norm 2 3 1 2 1 1 1 1 1 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 1 08/1058 Canine Norm 2 1 1 1 1 1 1 1 1 1 2 1 N 1 1 2 1 1 1 N 1 1 1 1 1 08/1065 Feline Norm 2 3 1 2 1 1 1 1 1 1 2 1 N 2 2 2 1 1 1 N 1 1 1 1 1 08/1328 Canine Uro 1 2 2 1 1 1 1 1 1 1 2 2 2 3 2 1 1 N N N N N N 1 2 08/1399 Feline Cys 2 3 2 2 1 1 1 1 1 1 1 2 1 2 1 4 1 3 1 N 1 1 1 2 1 08/1426 Canine Cys 2 2 2 1 1 1 1 1 1 1 1 2 3 1 1 3 1 1 1 N 1 1 1 1 1 09-1008 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 2 1 3 1 1 N N N N N 1 1 08/1527 Feline Neo 2 3 1 2 2 2 2 1 1 1 1 2 1 4 1 2 1 3 1 N 1 2 1 1 1 15/0307 Feline Oth 2 2 2 1 1 1 1 1 1 1 1 3 3 3 1 2 1 3 1 N 1 2 2 1 1 08/1550 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 09-1027 Canine Neo 2 1 1 1 2 2 2 2 2 1 1 1 N 1 1 2 2 3 1 N 1 1 1 1 1 08/1695 Canine Norm 2 3 1 2 1 1 1 1 1 1 2 1 N 3 2 2 1 1 1 N 1 1 1 1 1 08/1713 Canine Norm 2 2 1 2 1 1 1 2 1 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 1 15-616 Feline Cys 2 4 1 1 N N 1 N N 1 2 3 2 3 1 4 1 2 2 2 1 2 1 1 1 08/1885 Canine Uro 2 3 2 1 1 1 1 1 1 1 1 1 N 3 1 4 2 3 1 N 1 1 1 1 1 09/0313 Feline Neo 2 2 1 2 2 2 2 1 1 1 1 1 N 2 1 2 1 3 1 N 1 1 1 1 1 08/1964 Canine Cys 2 4 2 1 1 1 1 1 1 1 1 4 2 3 1 4 1 2 1 N 1 1 2 1 1 09-756 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 2 1 1 1 N N N N N N 1 1

240

10/0033 Canine Neo 2 2 2 1 2 2 2 1 2 1 1 1 N 3 1 4 2 1 1 N 1 1 1 1 1 08-269 Canine Cys 2 3 1 2 1 1 1 1 1 1 1 2 1 2 1 3 2 1 1 N 1 1 1 1 2 10-0357 Canine Neo 2 2 2 1 2 2 2 1 1 1 1 1 N 1 1 4 2 3 1 N 1 1 1 1 1 10-461 Canine Neo 2 2 2 1 2 2 2 1 1 1 1 1 N 3 1 4 2 3 1 N 1 1 1 1 1 09/0066 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 2 2 3 2 2 1 1 1 N 1 1 1 1 1 08/1758 Feline Norm 2 3 1 2 1 1 1 1 1 1 2 1 N 2 2 2 1 1 1 N 1 1 1 1 1 11/0164 Canine Neo 2 2 2 1 2 2 2 1 2 1 2 1 N 3 1 2 1 1 1 N 1 1 1 1 1 09/0895 Canine Cys 1 2 1 2 1 1 1 N 1 1 2 1 N 1 1 3 1 3 N N N N N 1 1 11/0648 Canine Neo 1 1 1 1 2 2 2 1 2 1 1 1 N 3 2 1 1 N 1 N 1 N N 1 1 12/0514 Canine Neo 2 4 2 2 1 1 2 1 1 1 2 1 N 1 1 4 1 1 2 3 1 1 1 1 1 12/0634 Canine Neo 2 2 2 1 2 2 2 2 2 1 1 2 3 1 1 4 2 1 1 N 1 1 1 1 1 13/0479 Canine Neo 2 2 1 2 2 2 2 1 1 1 2 2 1 2 1 4 1 3 1 N 1 1 1 1 2 06/0088 Canine Oth 2 1 1 1 1 1 1 1 1 1 1 2 1 3 1 2 2 1 1 N 1 1 1 1 1 13-0082 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 5 1 2 1 1 N N N N N 1 1 09-0605 Feline Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 3 1 2 1 1 1 N 1 1 1 1 1 13-126 Canine Neo 2 1 1 1 1 2 2 2 1 1 2 2 3 3 1 3 1 1 1 N 1 1 1 1 1 14-81 Canine Neo 1 2 1 2 1 2 2 1 1 1 1 1 N 2 2 4 1 2 1 N 1 N N 1 1 09-1105 Canine Cys 2 2 2 1 1 1 1 1 1 1 1 4 2 3 1 3 1 2 2 2 1 2 2 1 1 09-1189 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 3 1 N 1 1 1 1 2 09-119 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 2 2 3 1 4 1 2 1 N 1 1 2 2 1 09-1231 Canine Unkn 2 1 1 1 1 1 1 1 1 1 1 2 2 1 1 4 2 3 1 N 1 1 1 1 1 15/0621 Canine Neo 2 1 1 1 2 2 2 1 2 1 1 2 2 1 1 4 1 3 1 N 1 1 1 1 2 09-773 Canine Cys 1 3 1 1 1 1 1 1 1 1 1 2 2 2 2 2 1 1 N N N N N 1 1 09-838 Canine Cys 1 3 1 1 1 1 1 1 1 1 1 1 N 2 1 2 1 3 N N N N N 1 1 15-107 Canine Neo 2 1 1 1 2 2 2 1 2 2 1 3 1 3 1 3 1 3 1 N 1 1 1 1 1 10/0125 Canine Cys 2 2 2 1 1 1 1 1 1 1 1 2 1 3 1 3 2 3 1 N 1 1 1 1 1 10/0187 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 1 10/0592 Canine Uro 2 1 1 1 1 2 1 1 1 1 1 1 N 4 1 3 1 3 2 3 1 2 2 1 1 94/0935 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 1 1 4 1 3 N N N N N 1 1

241

10-323 Canine Norm 2 4 1 2 1 1 1 1 1 1 1 1 N 2 1 2 1 1 1 N 1 1 2 1 1 96/0232 Canine Neo 2 1 1 1 1 1 2 1 1 1 1 1 N 1 1 3 1 2 1 N 1 1 1 1 2 96/1337 Canine Neo 2 1 1 1 2 2 2 1 2 1 1 2 1 4 1 2 1 3 1 N 1 1 1 1 1 06/1385 Canine Oth 2 5 1 2 N N N N N 1 1 3 3 1 1 3 1 3 2 3 1 1 1 1 1 11/0143 Feline Cys 2 2 1 2 1 1 1 2 1 1 1 2 2 3 1 2 1 1 1 N 1 1 1 2 1 06/1126 Canine Neo 1 1 1 1 2 2 2 2 1 1 1 1 N 1 1 3 1 3 N N N N N 1 1 11/0204 Canine Norm 2 5 1 2 N N N N N 1 1 1 N 3 2 2 1 1 1 N 1 1 1 1 1 11/0220 Canine Uro 2 3 2 1 1 1 1 2 1 1 1 3 3 2 1 1 1 N 1 N 1 1 1 1 2 07/0580 Canine Oth 2 4 2 2 1 1 1 1 1 1 1 3 2 3 1 1 1 N 1 N 1 N N 1 2 11/0595 Canine Cys 2 2 1 2 1 1 1 1 2 1 2 2 1 2 1 3 2 1 1 N 1 1 1 1 1 08/0179 Canine Neo 2 2 1 2 1 1 2 1 1 1 1 1 N 3 2 2 1 1 2 3 1 2 1 1 1 11/0771 Canine Cys 2 4 2 2 1 1 1 1 1 1 1 2 2 2 2 3 2 3 1 N 1 1 1 1 1 11/0814 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 2 2 5 2 2 1 1 1 N 1 1 1 1 1 11/0833 Canine Uro 2 2 2 1 1 1 1 1 1 1 1 3 3 4 1 3 1 3 1 N 1 1 1 1 1 11/0844 Canine Norm 2 4 1 2 1 1 1 1 1 1 1 2 1 3 2 2 1 1 1 N 1 1 1 1 1 11-084 Canine Uro 2 5 2 1 N N N N N 1 1 3 2 3 1 4 1 2 2 2 1 2 2 1 2 11-278 Canine Cys 2 4 1 1 1 1 1 1 1 1 1 3 2 5 1 4 2 2 2 2 1 N N 2 1 11-299 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 2 1 2 1 1 1 N 1 1 1 1 1 98/1775 Canine Oth 2 1 1 1 1 1 1 1 1 1 1 2 2 2 1 2 1 1 1 N 1 1 2 1 1 97/0108 Canine Oth 2 2 2 2 1 1 1 1 1 1 1 1 N 3 2 2 1 1 2 1 1 1 1 1 1 12/0031 Canine Cys 2 2 1 2 1 1 1 1 2 1 1 1 N 2 1 4 2 3 1 N 1 1 1 1 1 08/1944 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 3 2 1 1 N 1 N 1 1 1 1 1 12/0458 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 09/0352 Canine Neo 2 2 1 2 2 2 2 1 2 1 1 1 N 3 1 4 1 1 2 1 1 2 1 1 1 17/1336 Feline Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 07/0939 Canine Oth 1 2 1 2 1 1 1 1 1 1 1 2 2 3 1 1 1 N N N N N N 1 2 12/0628 Canine Cys 2 4 2 2 1 1 1 1 1 1 1 2 2 3 1 4 2 3 2 3 1 1 1 1 1 08/0305 Canine Neo 1 1 1 1 2 2 2 2 2 1 N N N N N N 1 N N N N N N 1 1 12/0792 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 3 2 2 1 1 1 N 1 1 1 1 1

242

96/0227 Canine Neo 2 1 1 1 2 1 2 1 2 1 1 1 N 2 1 3 1 3 1 N 1 N N 1 1 12-383 Canine Norm 2 5 1 2 N N N N N 1 1 1 N 2 2 1 1 N 1 N 1 1 1 1 1 13/0197 Feline Cys 2 4 1 2 1 1 1 1 1 1 1 4 2 3 1 3 1 3 1 N 1 1 2 1 1 07/1871 Feline Oth 2 1 1 1 1 1 1 1 1 1 1 3 3 2 2 1 1 N 1 N 1 2 2 1 1 15/0793 Canine Oth 2 1 1 1 1 1 1 2 2 1 1 2 3 1 1 4 1 3 1 N 1 1 2 1 2 08-0705 Canine Neo 1 1 1 1 2 1 1 1 1 1 1 1 N 2 1 3 1 3 N N N N N 1 1 13-0068 Canine Cys 1 4 1 1 N N N N N 1 1 2 2 3 2 4 1 2 N N N N N 1 1 12-006 Canine Neo 1 1 1 1 2 2 2 1 1 2 1 2 2 4 1 4 1 3 N N N N N 1 1 16-339 Canine Neo 2 2 1 2 1 1 1 1 1 1 1 1 N 4 1 1 1 N 1 N 1 2 1 1 1 13-352 Feline Neo 2 2 2 1 2 1 2 1 1 1 2 2 1 3 1 4 1 3 1 N 1 1 2 1 1 01/0023 Canine Neo 2 4 2 1 1 1 2 1 1 1 1 2 2 3 1 4 2 2 1 N 1 1 1 1 2 14/0009 Canine Cys 2 5 1 2 N N N N N 1 1 2 2 3 1 3 1 3 2 2 1 1 1 1 1 14/0034 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 00/0178 Canine Oth 2 3 1 1 1 1 1 1 1 1 2 2 2 5 1 2 1 2 1 N 1 1 2 1 1 14/0218 Canine Uro 2 3 1 1 1 1 1 2 1 1 1 2 3 3 2 1 1 N 1 N 1 1 1 1 2 11/0079 Canine Oth 1 1 1 1 1 2 1 1 2 1 1 2 2 3 2 2 1 3 N N 1 1 1 1 2 07/0647 Canine Oth 2 4 1 2 1 1 1 1 1 1 1 1 N 3 2 2 1 1 1 N 1 2 2 1 1 02/0532 Feline Oth 2 5 1 1 1 1 1 N N 1 1 4 2 5 1 1 1 N 1 N 1 1 1 1 2 14/0707 Feline Cys 2 4 2 1 1 1 1 1 1 1 1 3 3 3 1 4 1 2 2 2 1 1 1 1 1 14/0783 Canine Uro 2 4 2 1 1 1 1 1 1 1 1 2 2 4 1 4 1 2 2 2 1 1 1 1 1 14/0993 Canine Cys 2 4 1 2 1 1 1 1 1 1 1 1 N 2 1 4 2 3 1 N 1 1 1 1 1 14/1029 Feline Norm 2 4 1 2 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 14/1153 Feline Norm 2 4 1 2 1 1 1 1 1 1 1 1 N 2 1 1 1 N 1 N 1 1 1 1 1 14/1191 Canine Norm 2 3 1 2 1 1 1 2 1 1 1 1 N 2 2 2 1 1 1 N 1 1 2 1 1 14-553 Canine Cys 1 1 1 1 1 1 1 1 1 1 1 1 N 2 2 1 1 N N N N N N 1 1 03/0217 Canine Neo 2 1 1 1 1 1 2 1 2 1 2 3 3 2 2 4 1 3 1 N 1 1 1 1 1 15/0076 Canine Cys 2 4 2 1 1 1 1 1 1 1 1 2 3 2 1 2 1 3 1 N 1 2 2 1 1 15/0275 Canine Uro 2 2 2 1 1 1 1 1 1 1 1 2 3 3 1 1 1 N 1 N 1 1 1 1 1 15/0295 Canine Norm 2 4 1 2 1 1 1 1 1 1 2 1 N 1 1 1 1 N 1 N 1 1 1 1 1

243

13/0437 Feline Oth 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 2 2 1 1 97/0880 Canine Oth 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 14/0092 Feline Oth 2 2 1 2 1 1 1 1 1 1 1 2 1 3 1 1 1 N 1 N 1 1 1 1 1 15/0421 Canine Cys 2 4 1 2 1 1 1 1 1 1 1 1 N 2 2 4 1 3 1 N 1 1 1 1 1 15/0510 Canine Cys 2 4 2 1 1 1 1 1 1 1 1 1 N 1 1 3 2 3 1 N 1 1 1 1 1 03-453 Canine Neo 2 1 1 1 2 1 2 1 1 1 1 2 1 3 1 3 2 3 1 N 1 1 1 1 1 15/0626 Feline Norm 2 4 1 1 1 1 1 1 1 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 1 15/0731 Canine Cys 2 4 1 2 1 1 1 1 1 1 1 1 N 1 1 2 1 3 2 3 1 2 2 1 1 14/0584 Canine Oth 2 4 2 1 1 1 1 1 1 1 1 3 3 5 1 2 1 1 1 N 1 2 2 1 1 15/0869 Canine Cys 2 2 2 2 1 1 1 1 2 1 2 3 3 1 1 3 1 1 1 N 1 1 1 1 1 15/0893 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 2 1 3 1 N 1 1 1 1 1 06/1303 Canine Oth 2 2 2 1 1 1 1 2 2 2 1 3 2 3 1 4 1 2 1 N 1 1 2 1 1 15/1415 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 3 2 1 1 N 1 N 1 1 1 1 1 15-017 Feline Neo 2 4 1 2 1 1 2 2 2 1 2 1 N 1 N 2 1 1 1 N 1 2 2 1 2 05/0496 Canine Neo 2 1 1 1 1 1 2 1 2 1 1 1 N 3 2 2 1 1 1 N 1 N N 1 1 15-110 Canine Uro 1 1 1 1 1 1 1 1 2 1 1 1 N 3 1 3 1 3 N N N N N 1 1 05/1078 Canine Neo 2 1 1 1 1 1 2 1 1 1 2 1 N 3 1 2 2 3 1 N 1 1 1 1 1 15-595 Canine Norm 2 2 2 1 1 2 1 1 1 1 1 1 N 2 1 2 1 1 1 N 1 2 1 1 2 15-604 Canine Cys 1 N 1 2 1 1 1 1 1 1 1 3 3 2 2 4 1 2 N N N N N 2 1 07/0917 Feline Norm 2 3 1 1 1 1 1 1 N 1 1 3 2 3 2 2 1 1 1 N 1 1 1 1 1 15-627 Canine Cys 2 2 1 1 1 1 1 1 1 1 1 2 1 3 1 3 1 3 1 N 1 1 2 1 1 16/0006 Feline Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 1 1 2 2 1 1 N 1 1 2 1 1 16/0034 Canine Norm 2 2 1 2 1 1 1 1 1 1 1 1 N 2 1 2 1 1 1 N 1 1 1 1 1 16/0487 Canine Uro 2 5 2 1 N N N N N 1 2 3 3 2 1 4 1 2 2 2 1 2 2 1 2 16-0556 Canine Cys 1 N 1 2 1 1 1 1 1 1 1 1 N 3 1 3 1 1 1 N 1 N N 1 1 16-212 Canine Norm 2 5 1 2 N N 1 N N 1 2 1 N 5 1 1 1 N 1 N 1 1 1 1 1 10-634 Canine Neo 2 1 1 1 1 1 2 1 1 1 1 2 2 3 1 3 2 1 1 N 1 1 1 1 1 16-396 Feline Cys 2 4 1 1 1 1 1 1 1 1 1 3 2 5 1 3 1 3 1 N 1 1 1 1 1 16-732 Canine Cys 1 N 1 2 1 1 1 1 1 1 1 1 N 2 1 2 1 1 N N N N N 1 1

244

16-778 Canine Cys 1 N 1 2 1 1 1 1 1 1 1 1 N 3 2 1 1 N N N N N N 1 1 16V- Canine Uro 2 3 2 1 1 1 1 1 3 1 1 2 1 4 1 4 2 3 1 N 1 N N 1 1 00297 16V- Canine Cys 1 3 2 1 1 1 1 1 2 1 1 2 3 4 1 4 2 2 N N N N N 2 1 00539 16V- Canine Cys 1 1 1 1 2 1 1 1 2 1 1 4 2 4 2 4 1 3 N N N N N 1 1 00661 16V- Canine Uro 2 1 1 1 1 1 1 1 2 1 1 1 N 3 1 4 1 3 1 N 1 1 2 1 1 00720 16V- Canine Uro 1 1 1 1 2 1 1 2 2 1 2 3 2 3 1 4 1 3 N N N N N 1 1 01051 16V- Canine Uro 1 3 2 1 1 1 1 1 1 1 1 1 N 5 1 3 1 3 1 N 1 1 1 1 1 1112 17/1248 Canine Uro 2 2 2 1 1 2 1 1 2 1 1 3 3 4 1 2 1 3 1 N 1 1 2 1 1 17/1257 Feline Neo 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 1 1 N 2 2 1 1 1 1 1 17/1334 Feline Norm 2 4 1 2 1 1 1 2 1 1 1 1 N 1 1 1 1 N 1 N 1 2 1 1 1 17/1335 Feline Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 2 1 2 2 1 1 N 1 1 1 1 1 07-412 Feline Uro 2 2 2 2 1 1 1 1 1 1 1 3 2 2 1 3 1 3 2 4 1 1 1 1 1 17/1337 Feline Uro 2 5 2 1 1 1 1 N N 1 1 3 3 5 1 3 1 2 1 N 1 1 2 1 1 17-115 Canine Norm 2 3 1 2 1 1 1 1 1 1 1 1 N 2 1 1 1 N 1 N 1 1 1 1 1 05/1152 Canine Oth 2 1 1 1 1 1 1 1 1 1 1 1 N 3 1 2 1 2 1 N 1 2 2 1 1 17-362 Feline Cys 2 1 1 1 1 1 1 1 2 1 1 1 N 2 1 4 2 4 2 3 1 1 1 1 1 17V- Canine Cys 1 1 1 1 1 1 1 1 1 1 1 4 2 3 2 3 1 3 N N N N N 1 1 00067 18/0009 Canine Uro 2 1 1 1 2 1 1 1 1 1 1 2 2 3 1 2 1 2 1 N 1 1 1 1 1 18/0073 Canine Uro 2 2 2 1 1 1 1 1 1 1 1 2 2 4 1 2 1 3 1 N 1 1 1 1 1 18/0508 Canine Uro 2 1 2 1 1 2 1 1 2 1 1 2 1 4 1 4 2 3 1 N 1 1 1 1 1 18/0508 Canine Uro 2 2 2 1 1 1 1 2 2 1 1 2 1 3 1 4 2 3 1 N 1 1 2 1 1 18/0509 Canine Uro 1 N N N N N 1 N N N N N N N N N N N 1 N 1 1 1 1 1 18/0652 Canine Uro 2 4 2 1 1 1 1 1 1 1 1 1 N 2 1 3 2 3 1 N 1 1 1 1 1 18/0653 Canine Uro 2 5 2 2 N N 1 N N 1 1 2 2 3 2 2 1 1 1 N 1 1 1 1 1 245

18/0759 Canine Uro 2 4 2 1 1 1 1 1 1 1 1 2 1 4 1 3 1 3 1 N 1 1 1 1 1 18/0878 Canine Uro 2 5 2 1 N N 1 N N 1 1 2 1 2 1 3 1 2 1 N 1 1 2 1 1 18/1026 Feline Cys 2 5 2 1 N N 1 N N 1 2 4 3 2 1 3 1 2 1 N 1 1 2 1 1 18/1031 Canine Uro 2 4 2 2 1 1 1 2 1 1 1 4 2 2 1 4 1 3 1 N 1 1 1 1 1 18/1105 Feline Uro 2 4 2 1 1 1 1 2 1 1 2 3 2 4 1 2 1 2 1 N 1 1 2 1 1 18/1190 Canine Uro 2 3 2 2 1 1 1 1 2 1 1 3 3 4 1 4 1 3 1 N 1 1 2 1 1 18V- Canine Uro 1 2 2 1 1 1 1 1 1 1 1 2 2 3 2 3 1 3 N N N N N 1 1 00418 18V- Canine Cys 2 4 1 2 1 1 1 1 1 1 1 4 2 3 1 3 1 3 1 N 1 1 1 1 1 01308 18V- Canine Cys 1 2 2 1 1 1 1 1 2 1 1 2 2 4 2 4 2 2 N N N N N 1 1 01374 18V- Canine Cys 1 1 1 1 1 2 1 1 2 1 1 2 1 5 2 3 1 3 N N N N N 1 1 01388 19/0528 Feline Cys 2 4 2 1 1 1 1 1 1 1 1 4 2 5 1 4 1 2 2 2 1 2 2 2 1 15-368 Canine Neo 1 1 1 1 1 1 1 1 1 1 1 1 N 3 2 3 1 3 N N N N N 1 1 95/1270 Canine Uro 2 2 2 1 1 1 1 1 1 1 2 1 N 3 1 4 2 3 1 N 1 1 1 1 1 95/1363 Feline Cys 2 5 2 1 N N N N N 1 1 4 2 3 1 4 1 2 2 2 1 2 2 1 1 95/1377 Feline Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 3 2 2 1 1 1 N 1 1 1 1 1 95/1406 Canine Neo 2 4 2 1 1 1 2 1 1 1 1 1 N 3 1 2 2 1 1 N 1 1 1 1 2 96/0628 Canine Neo 2 2 1 2 1 1 2 2 1 1 1 1 N 2 1 3 1 3 1 N 1 1 1 1 1 02/0719 Canine Neo 2 1 1 1 2 2 2 2 1 1 1 1 N 1 1 3 1 1 1 N 1 1 2 1 2 06/0498 Canine Oth 2 3 2 2 1 1 1 1 1 1 2 3 2 4 1 2 1 3 2 3 1 1 2 1 1 08-508 Canine Neo 1 1 1 1 2 2 2 2 1 1 1 3 2 1 1 3 1 1 N N N N N 1 1 96/0644 Canine Uro 2 4 2 1 1 1 1 1 1 2 1 1 N 2 1 3 1 1 1 N 1 2 2 1 1 96/0889 Canine Cys 2 4 2 1 1 1 1 1 1 1 1 4 2 3 1 4 1 2 2 2 1 2 2 2 1 04-608 Feline Oth 2 2 2 1 1 1 1 1 1 2 2 2 2 4 1 2 1 1 1 N 1 1 1 1 1 09/1650 Canine Oth 2 2 1 2 1 1 1 1 1 1 1 2 1 5 2 2 1 3 1 N 1 1 1 1 1 96/1185 Canine Cys 2 5 1 1 N N N N N 2 1 3 2 2 1 4 1 2 1 N 1 1 1 2 1 96/1212 Canine Uro 2 2 2 1 1 1 1 1 2 1 1 1 N 4 1 2 1 1 1 N 1 1 2 1 1

246

96/1293 Canine Cys 2 1 1 1 1 1 1 1 1 1 2 1 N 1 1 3 2 3 1 N 1 1 1 1 1 13-417 Canine Neo 1 1 1 1 2 2 2 1 1 1 1 1 N 3 1 2 1 1 N N N N N 1 1 97/0102 Canine Cys 2 2 1 2 1 1 1 2 2 1 2 1 N 2 1 4 2 3 1 N 1 1 1 1 1 14/0645 Canine Oth 2 2 1 2 1 1 1 1 1 1 2 1 N 1 1 1 1 N 1 N 1 2 2 1 2 97/0137 Canine Cys 2 4 2 2 1 1 1 1 1 1 1 1 N 1 1 4 1 3 1 N 2 1 1 1 1 97/0165 Feline Neo 2 2 1 2 1 1 2 1 1 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 1 97/0270 Canine Cys 2 2 1 2 1 1 1 1 1 1 2 2 2 3 1 3 1 3 1 N 1 2 2 1 1 99/0805 Canine Oth 1 1 1 1 1 1 1 1 1 1 1 1 N 2 2 2 1 1 1 N 1 1 1 1 2 97/1037 Canine Cys 2 2 2 1 1 1 1 1 1 1 1 3 2 3 1 4 1 2 1 N 1 1 1 1 1 97/1240 Canine Cys 2 5 2 2 N N N N N 1 2 4 3 2 1 4 1 2 1 N 1 2 2 1 2 97/1529 Canine Cys 2 2 1 2 1 1 1 1 1 1 1 1 N 2 1 3 1 1 1 N 1 1 1 1 1 98/0183 Canine Cys 2 5 2 2 N N N N N 1 1 2 1 3 1 3 1 2 1 N 1 1 1 1 1 98/0574 Feline Cys 2 2 1 2 1 1 1 1 1 1 1 1 N 3 1 3 1 1 1 N 1 1 2 1 1 98/0624 Canine Norm 2 4 1 2 1 1 1 1 1 1 2 1 N 1 1 1 1 N 1 N 1 1 1 1 1 00/0427 Feline Oth 2 4 1 1 1 1 1 N 1 1 1 1 N 3 1 2 1 1 1 N 1 2 1 1 1 98/0666 Canine Neo 1 1 1 1 1 2 2 1 1 1 1 2 2 4 1 4 1 2 N N N N N 1 1 98/1183 Feline Cys 2 5 2 2 N N N N N 1 1 4 3 3 1 4 1 2 2 2 1 1 1 2 1 98/1359 Feline Uro 2 4 2 1 1 1 1 1 1 1 2 4 2 3 1 3 1 3 2 1 1 1 1 1 1 98/1578 Feline Norm 2 5 1 2 N N N N N 1 1 1 N 1 1 1 1 N 1 N 1 1 1 1 2 98/1760 Canine Norm 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 2 1 1 1 N 1 1 1 1 1 98/1761 Canine Cys 2 1 1 1 1 2 1 2 1 1 1 2 1 2 1 4 1 3 1 N 1 1 1 1 1 04/1207 Canine Oth 2 3 2 1 1 1 1 2 1 1 1 4 2 2 1 3 1 2 1 N 1 2 2 1 2 98/2032 Canine Uro 2 1 1 1 1 1 1 1 2 1 2 1 N 2 1 3 1 1 1 N 1 1 1 1 1 99/0449 Canine Cys 2 1 1 1 1 1 1 1 1 1 1 1 N 1 1 3 2 3 1 N 1 1 1 1 1 99/0645 Feline Norm 2 2 1 2 1 1 1 1 1 1 1 2 2 4 2 2 1 1 1 N 1 N N 1 2 99/0666 Canine Norm 2 5 1 2 N N N N N 1 1 1 N 3 2 2 1 1 1 N 1 1 1 1 2 98/0640 Canine Oth 2 2 1 2 1 1 1 1 1 1 1 2 3 1 1 2 1 1 1 N 1 1 1 1 2 99/0875 Canine Cys 1 4 2 1 1 1 1 1 1 1 1 4 2 3 1 4 1 2 N N N N N 1 1 99/1169 Canine Cys 2 5 2 1 N N N N N 1 1 2 2 3 1 4 1 2 2 2 1 1 2 2 1

247

99/1246 Canine Uro 2 4 2 2 1 1 1 1 1 1 1 1 N 2 1 2 2 3 1 N 1 1 1 1 1 96/1057 Feline Oth 2 2 2 2 1 1 1 2 1 1 1 2 3 3 1 1 1 N 1 N 1 2 2 1 2 99-1666 Feline Neo 2 1 1 1 2 2 2 1 3 1 1 1 N 1 1 3 2 1 1 N 1 1 1 1 1 99-3006 Canine Cys 2 1 1 1 1 1 1 1 2 1 1 2 1 3 2 3 1 3 1 N 1 1 1 1 1

Cys cystitis; Dx Diagnosis; N N/A, not applicable; Neo neoplasia; Norm normal bladder; Oth other diagnosis; Uro urolithiasis; Unkn unknown diagnosis.

248

Appendix 5: Significant predicted probabilities (P≤0.05) and 95% confidence intervals for diagnostic outcomes by histopathology finding patterns for canine cases submitted to UQVLS between 1994 and 2016 and collected from MUSVLS and southeast Queensland veterinary

clinics from 2016-2019.

neut_sm_ inflamm inflann sm_haem

ulcer Normal/Other P value Cystitis P value Neoplasia P value Urolithiasis P value

uroth_ lymph_ aggregates uroth_ No No No No 1 0.71 (0.53, 0.9) <0.01 0.08 (0.00, 0.15) 0.04 0.21 (0.05, 0.36) 0.01 No No No No 2 0.65 (0.42, 0.88) <0.01 0.14 (0.01, 0.28) 0.04 0.21 (0.03, 0.39) 0.02 No No No No 3 0.62 (0.33, 0.91) <0.01 0.30 (0.04, 0.57) 0.02 No No No Yes 1 0.32 (0.02, 0.62) 0.03 - - 0.50 (0.21, 0.80) <0.01 No No No Yes 2 0.26 (0.02, 0.50) 0.04 0.46 (0.16, 0.76) <0.01 No No No Yes 3 0.57 (0.22, 0.91) <0.01 No No Yes No 1 0.38 (0.15, 0.62) <0.01 0.27 (0.08, 0.47) 0.01 0.31 (0.09, 0.53) 0.01 No No Yes No 2 0.29 (0.07, 0.52) 0.01 0.41 (0.16, 0.67) <0.01 0.27 (0.05, 0.49) 0.02 No No Yes No 3 0.22 (0, 0.43) 0.05 0.69 (0.44, 0.93) <0.01 No No Yes Yes 1 0.34 (0.08, 0.61) 0.01 0.47 (0.17, 0.76) <0.01 No No Yes Yes 2 0.49 (0.18, 0.79) <0.01 0.38 (0.08, 0.67) 0.01 No No Yes Yes 3 0.78 (0.56, 1.00) <0.01 No Yes No No 1 0.33 (0, 0.66) 0.05 0.23 (0.00, 0.45) 0.05 0.44 (0.13, 0.75) 0.01 No Yes No No 2 0.35 (0.05, 0.65) 0.02 0.39 (0.08, 0.69) 0.01 249

No Yes No No 3 0.66 (0.31, 1.02) <0.01 No Yes No Yes 1 0.27 (0.00, 0.54) 0.05 0.63 (0.33, 0.93) <0.01 No Yes No Yes 2 0.4 (0.06, 0.75) 0.02 0.53 (0.18, 0.88) <0.01 No Yes No Yes 3 0.78 (0.48, 1.07) <0.01 No Yes Yes No 1 0.47 (0.18, 0.77) <0.01 0.4 (0.12, 0.69) 0.01 No Yes Yes No 2 0.62 (0.32, 0.92) <0.01 0.3 (0.02, 0.58) 0.04 No Yes Yes No 3 0.88 (0.72, 1.00) <0.01 No Yes Yes Yes 1 0.47 (0.14, 0.79) 0.01 0.48 (0.15, 0.81) 0.01 No Yes Yes Yes 2 0.61 (0.28, 0.94) <0.01 0.35 (0.03, 0.67) 0.03 No Yes Yes Yes 3 0.89 (0.73, 1.00) <0.01 Yes No No No 1 0.69 (0.45, 0.94) <0.01 0.18 (0, 0.36) 0.06 Yes No No No 2 0.63 (0.37, 0.89) <0.01 0.15 (0.00, 0.30) 0.05 0.18 (0, 0.36) 0.05 Yes No No No 3 0.57 (0.31, 0.83) <0.01 0.31 (0.09, 0.54) 0.01 Yes No No Yes 1 0.39 (0.06, 0.72) 0.02 Yes No No Yes 2 0.26 (0.01, 0.50) 0.04 0.36 (0.06, 0.67) 0.02 Yes No No Yes 3 0.49 (0.17, 0.80) <0.01 Yes No Yes No 1 0.28 (0.02, 0.53) 0.03 0.22 (0.03, 0.42) 0.03 Yes No Yes No 2 0.22 (0.01, 0.44) 0.04 0.35 (0.10, 0.6) 0.01 Yes No Yes No 3 0.15 (0.00, 0.30) 0.05 0.53 (0.27, 0.8) <0.01 0.27 (0.00, 0.54) 0.05 250

Yes No Yes Yes 1 0.56 (0.07, 1.00) 0.03 Yes No Yes Yes 2 0.46 (0.00, 0.92) 0.05 Yes No Yes Yes 3 0.43 (0.05, 0.82) 0.03 0.50 (0.07, 0.92) 0.02 Yes Yes No No 1 0.25 (0.01, 0.49) 0.04 0.39 (0.07, 0.71) 0.02 Yes Yes No No 2 0.38 (0.08, 0.68) 0.01 0.34 (0.05, 0.63) 0.02 Yes Yes No No 3 0.67 (0.37, 0.98) <0.01 Yes Yes No Yes 1 0.29 (0.00, 0.57) 0.05 0.54 (0.20, 0.88) <0.01 Yes Yes No Yes 2 0.42 (0.09, 0.76) 0.01 0.45 (0.10, 0.79) 0.01 Yes Yes No Yes 3 0.74 (0.46, 1.00) <0.01 Yes Yes Yes No 1 0.44 (0.15, 0.73) <0.01 0.30 (0.03, 0.58) 0.03 Yes Yes Yes No 2 0.60 (0.31, 0.88) <0.01 Yes Yes Yes No 3 0.80 (0.59, 1.00) <0.01 Yes Yes Yes Yes 1 0.37 (0.04, 0.70) 0.03 Yes Yes Yes Yes 2 0.52 (0.18, 0.86) <0.01 Yes Yes Yes Yes 3 0.71 (0.37, 1.00) <0.01 SM = submucosa. Variables: urothelial ulceration, lymphoid aggregates, submucosal inflammation of neutrophilic type, urothelial inflammation, and submucosal haemorrhage (1 – no SM haemorrhage, 2 – haemorrhage present in up to 25% of the SM, 3 – present in 26-50% of the SM). Blank cells are those outputs that were not significant (P ≤ 0.05).

251

Significant predicted probabilities (P≤0.05) and 95% confidence intervals for diagnostic outcomes (1-5) by histopathology finding patterns for feline cases submitted to UQVLS between 1994 and 2016 and collected from MUSVLS and southeast Queensland veterinary clinics from 2016-

2019.

roth Normal/other P value Cystitis P value Neoplasia P value Urolithiasis P value

ulcer lymph_ aggregates neut_sm_in flamm uroth_ inflann sm_haem u No No No No 1 0.3655 (0.42, 0.68) <0.01 0.12 (0.05, 0.19) <0.01 0.32 (0.2, 0.44) <0.01

No No No No 2 0.48 (0.3, 0.65) <0.01 0.2 (0.09, 0.32) <0.01 0.31 (0.16, 0.46) <0.01

No No No No 3 0.45 (0.19, 0.7) <0.01 0.43 (0.19, 0.67) <0.01

No No No Yes 1 0.18 (0.01, 0.35) 0.03 0.18 (0.06, 0.3) 0.01 0.59 (0.4, 0.78) <0.01

No No No Yes 2 0.29 (0.11, 0.46) <0.01 0.53 (0.32, 0.73) <0.01

No No No Yes 3 0.62 (0.36, 0.88) <0.01 No No Yes No 1 0.22 (0.09, 0.35) <0.01 0.31 (0.18, 0.44) <0.01 0.37 (0.23, 0.51) <0.01 0.1 (0.01, 0.18) 0.02

No No Yes No 2 0.17 (0.04, 0.29) 0.01 0.46 (0.29, 0.63) <0.01 0.31 (0.16, 0.46) <0.01

No No Yes No 3 0.72 (0.54, 0.9) <0.01

No No Yes Yes 1 0.31 (0.14, 0.47) <0.01 0.44 (0.25, 0.63) <0.01 0.21 (0.04, 0.38) 0.02

No No Yes Yes 2 0.45 (0.25, 0.64) <0.01 0.36 (0.17, 0.55) <0.01 0.16 (0.01, 0.3) 0.03

No No Yes Yes 3 0.7 (0.49, 0.92) <0.01 0.18 (0, 0.37) 0.05

No Yes No No 1 0.26 (0.09, 0.44) <0.01 0.53 (0.3, 0.76) <0.01

No Yes No No 2 0.39 (0.16, 0.62) <0.01 0.45 (0.21, 0.69) <0.01

252

No Yes No No 3 0.73 (0.47, 1) <0.01

No Yes No Yes 1 0.27 (0.08, 0.47) 0.01 0.66 (0.45, 0.87) <0.01

No Yes No Yes 2 0.4 (0.16, 0.64) <0.01 0.55 (0.3, 0.8) <0.01

No Yes No Yes 3 0.78 (0.53, 1.03) <0.01

No Yes Yes No 1 0.48 (0.28, 0.67) <0.01 0.43 (0.23, 0.62) <0.01

No Yes Yes No 2 0.62 (0.42, 0.83) <0.01 0.31 (0.12, 0.51) <0.01

No Yes Yes No 3 0.88 (0.74, 1.01) <0.01

No Yes Yes Yes 1 0.44 (0.23, 0.64) <0.01 0.47 (0.25, 0.68) <0.01

No Yes Yes Yes 2 0.58 (0.37, 0.8) <0.01 0.35 (0.14, 0.56) <0.01

No Yes Yes Yes 3 0.85 (0.69, 1.02) <0.01 Yes No No No 1 0.48 (0.24, 0.71) <0.01 0.11 (0.02, 0.21) 0.02 0.25 (0.08, 0.42) <0.01 0.16 (0, 0.32) 0.04 Yes No No No 2 0.42 (0.2, 0.64) <0.01 0.2 (0.06, 0.33) <0.01 0.24 (0.08, 0.41) <0.01 0.14 (0, 0.28) 0.05

Yes No No No 3 0.35 (0.14, 0.56) <0.01 0.37 (0.18, 0.57) <0.01 0.2 (0.03, 0.38) 0.03

Yes No No Yes 1 0.13 (0, 0.26) 0.05 0.34 (0.08, 0.61) 0.01 0.41 (0.07, 0.75) 0.02

Yes No No Yes 2 0.22 (0.04, 0.4) 0.01 0.33 (0.09, 0.56) 0.01 0.35 (0.06, 0.64) 0.02

Yes No No Yes 3 0.38 (0.12, 0.64) <0.01 0.46 (0.15, 0.76) <0.01

Yes No Yes No 1 0.16 (0.02, 0.29) 0.02 0.15 (0.01, 0.29) 0.03 0.59 (0.32, 0.86) <0.01

Yes No Yes No 2 0.27 (0.08, 0.45) 0.01 0.14 (0.02, 0.27) 0.02 0.5 (0.23, 0.77) <0.01

Yes No Yes No 3 0.38 (0.14, 0.61) <0.01 0.54 (0.27, 0.8) <0.01

253

Yes No Yes Yes 1 0.09 (-0.03, 0.22) 0.13 0.78 (0.54, 1.00) <0.01

Yes No Yes Yes 2 0.17 (-0.01, 0.34) 0.06 0.71 (0.44, 0.98) <0.01

Yes No Yes Yes 3 0.23 (0.01, 0.46) 0.04 0.74 (0.49, 0.98) <0.01

Yes Yes No No 1 0.27 (0.08, 0.46) 0.01 0.44 (0.2, 0.69) <0.01

Yes Yes No No 2 0.41 (0.18, 0.63) <0.01 0.38 (0.15, 0.6) <0.01

Yes Yes No No 3 0.69 (0.45, 0.93) <0.01

Yes Yes No Yes 1 0.25 (0.05, 0.45) 0.01 0.49 (0.22, 0.76) <0.01

Yes Yes No Yes 2 0.38 (0.14, 0.62) <0.01 0.42 (0.17, 0.67) <0.01

Yes Yes No Yes 3 0.65 (0.36, 0.94) <0.01

Yes Yes Yes No 1 0.35 (0.14, 0.56) <0.01 0.25 (0.07, 0.43) 0.01 0.36 (0.12, 0.61) <0.01

Yes Yes Yes No 2 0.5 (0.28, 0.73) <0.01 0.2 (0.05, 0.36) 0.01 0.26 (0.04, 0.49) 0.02

Yes Yes Yes No 3 0.68 (0.39, 0.96) <0.01

Yes Yes Yes Yes 1 0.24 (0.04, 0.44) 0.02 0.2 (0.02, 0.39) 0.03 0.55 (0.25, 0.85) <0.01

Yes Yes Yes Yes 2 0.37 (0.14, 0.61) <0.01 0.18 (0.03, 0.34) 0.02 0.44 (0.16, 0.72) <0.01

Yes Yes Yes Yes 3 0.51 (0.17, 0.85) <0.01 0.45 (0.1, 0.8) 0.01 SM = submucosa. Variables: urothelial ulceration, lymphoid aggregates, submucosal inflammation of neutrophilic type, urothelial inflammation, and submucosal haemorrhage (1 – no SM haemorrhage, 2 – haemorrhage present in up to 25% of the SM, 3 – present in 26-50% of the SM). Blank cells are those outputs that were not significant (P ≥ 0.05 for significance).

254

Appendix 6: Bladder histology photomicrographs of cases used in the pathologist agreement study, Chapter 4. All images were taken at the sub gross view (fit to screen) and the second image at four times zoom in the Aperio eSlide Manager software. Due to technical issues there is unfortunately no image available for case 03/0244B, feline urothelial carcinoma.

Figure 8-1: Case 02/0691A, canine, urothelial carcinoma.

255

Figure 8-2: 04/1068A, canine, urothelial carcinoma.

256

Figure 8-3: Case 05/1429A, feline, urothelial carcinoma.

257

Figure 8-4: Case 07/2196, canine, mild cystitis.

258

Figure 8-5: Case 08/1399D, feline, cystitis secondary to lower motor neuron bladder.

259

Figure 8-6: Case 08/1885J, canine, urolithiasis.

260

Figure 8-7: Case 10/0259B, canine, bladder wall leiomyosarcoma.

261

Figure 8-8: Case 11/0220A, canine, urolithiasis.

262

Figure 8-9: Case 12/0516, feline, suspected FIC due to clinical urethral obstruction, no uroliths on necropsy, and lack of urine culture results.

263

Figure 8-10: Case 13/0197A, feline, suspected FIC due to clinical signs of stranguria, no uroliths on necropsy, and lack of urine culture results.

264

Figure 8-11: Case 14/0707A, feline, suspected FIC due to clinical signs of urethral obstruction, no uroliths on necropsy, and lack of urine culture results.

265

Figure 8-12: Case 14/0993B, canine, chronic follicular cystitis.

266

Figure 8-13: Case 15/0893E, canine, normal bladder.

267

Figure 8-14: Case 15-017E, feline, metastatic epitheliotropic lymphoma.

268

Figure 8-15: 16/0034I, canine, normal bladder.

269

Figure 8-16: Case 17/0311, canine, normal bladder.

270

Figure 8-17: Case 17/0398, canine, normal bladder with submucosal lymphoid follicles.

271

Figure 8-18: Case 17/0503A, feline, urolithiasis with urethral obstruction and bladder rupture.

272

Figure 8-19: Case 17/1335A, feline, normal bladder.

273

Figure 8-20: Case 17/1336A, feline, normal bladder.

274

Figure 8-21: Case 17/1337A, feline, urolithiasis + urinary tract infection (positive bacterial culture).

275

Figure 8-22: Case 18/0073A, canine, urolithiasis.

276

Figure 8-23: Case 19/0528A, feline, suspected FIC due to urethral obstruction with no uroliths at necropsy and no urine culture results.

277

Figure 8-24: Case 19/0729A, canine, urolithiasis and urinary tract infection (positive urine culture).

278

Appendix 7: A sample of the Microsoft Excel worksheet from the pathologist agreement study. Worksheet 1, first read with no signalment or history.

Primary type of Primary type of Urothelial Submucosal Submucosal Submucosal Detrusor Are organisms Slide ID submucosal detrusor Morphological diagnosis Etiological diagnosis Comments ulceration oedema haemorrhage inflammation inflammation visible inflammation inflammation

1234 Please Select

1235 Please Select

1236 Please Select ^ Only complete these slides from the eSlideManager folder. Slides for this study have the description 'Concordance study'. Please assess only bladder tissue.

Please complete spreadsheets in the order First, Second and Third (canine or feline first). Please select answers FOR EVERY BOX from the drop down lists. Provide a free form answer for morphological diagnosis and comments. Comments can be about the case or the testing procedure. Change the zoom at the bottom right of the page if required.

279

Worksheet 2, with animal signalment and history provided. When the history cell was selected, the entire text could be read in the text bar above the spreadsheet.

Primary type of Urothelial Submucosal Submucosal Submucosal Primary type of submucosal Detrusor Are organisms Morphological Etiological Slide ID Species History detrusor Comments ulceration oedema haemorrhage inflammation inflammation inflammation visible diagnosis diagnosis inflammation Productive cough, BAL neutrophils and macrophages. Weight 1234 Canine loss, no response of Please Select Bladder tumour on U/S. Blood in urine, constipation. 1235 Feline Please Select Abnormal kidney and bladder wall tumour. Nephrectomy and 1236 Canine bladder wall resection Please Select Bladder mass.

1237 Feline Please Select For this second sheet, complete as per the first sheet, but now with a summarised history submitted with the slide. Double click on the history box to see longer histories, or single click and expand the text box immediately above the spreadsheet. Please select answers FOR EVERY BOX from the drop down lists. Provide a free form answer for morphological diagnosis and comments. Please elaborate on 'other' diagnoses in the comment box. Comments can be about the case or the testing procedure.

280

Worksheet 3 (with predictive tool).

Submucosal Neutrophilic Amount of Normal/ Urothelial Urothelial Normal/ Cystitis Neoplasia Urolithiasis Urolithiasis Your # Slide ID lymphoid submucosal submucosal Other Cystitis CI Neoplasia CI Comments ulceration inflammation Other CI Probability Probability Probability CI diagnosis aggregates inflammation haemorrhage Probability

1 1234 yes no no yes moderate not_sig not_sig 0.26 0.01 - 0.5 0.36 0.06 - 0.67 not_sig not_sig 2 1235 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A 3 1236 #N/A #N/A #N/A #N/A #N/A #N/A #N/A #N/A

^Please select ^These cells (columns H-O) will provide diagnostic outcome probabilities and confidence intervals (CI). Please select your diagnosis from the drop down ^Please select yes or no from the drop down menu mild, moderate Please interpret AFTER all columns C-G have been entered. It is normal for these cells to begin saying #N/A. menu. Please elaborate on 'other' diagnoses in (columns C-F) or severe not_sig = not significat the comments box.

This sheet applies diagnostic outcome probabilities derived from logistic regression analysis of histological features of 338 canine and feline urinary bladder slides. Scroll to the right to see a graphical representation of these probabilities, AFTER columns C-G have been completed. See further instructions below spreadsheet.

As the pathologist selected the options for histological scoring from the drop-down lists, the probabilities (with 95% confidence intervals) for each disease appeared in the disease boxes. On the same worksheet was a graphical representation of the probability tool, using a chart that had been formatted to display the probabilities from the table.

281

This is a sample of one of the hidden worksheets containing the probabilities.

uroth_ lymph_agg sm_inflam_ uroth_in sm_hae NormaOther NormOther ulcer regates type_2 flamm m NormaOther Cystitis Neo Uro Canine sig prob Prob CI CystitisProb Cystitis CI NeoProb Neo CI Uro prob Uro CI no no no no mild 0.71 (0.53, 0.9) 0.08 (0, 0.15) 0.21 (0.05, 0.36) not_significant no-no-no-no-mild 0.71 0.53 - 0.90 0.08 0 - 0.15 0.21 0.05 - 0.36 not_sig not_sig no no no no moderate 0.65 (0.42, 0.88) 0.14 (0.01, 0.28) 0.21 (0.03, 0.39) not_significant no-no-no-no-moderate 0.65 0.42 - 0.88 0.14 0.01 - 0.28 0.21 0.03 - 0.39 not_sig not_sig no no no no severe 0.62 (0.33, 0.91) 0.3 (0.04, 0.57) not_significant not_significant no-no-no-no-severe 0.62 0.33 - 0.91 0.3 0.04 - 0.57 not_sig not_sig not_sig not_sig no no no yes mild 0.32 (0.02, 0.62) not_significant 0.5 (0.21, 0.8) not_significant no-no-no-yes-mild 0.32 0.02 - 0.62 not_sig not_sig 0.5 0.21 - 0.8 not_sig not_sig no no no yes moderate not_significant 0.26 (0.02, 0.5) 0.46 (0.16, 0.76) not_significant no-no-no-yes-moderate not_sig not_sig 0.26 0.02 - 0.5 0.46 0.16 - 0.76 not_sig not_sig no no no yes severe not_significant 0.57 (0.22, 0.91) not_significant not_significant no-no-no-yes-severe not_sig not_sig 0.57 0.22 - 0.91 not_sig not_sig not_sig not_sig no no yes no mild 0.38 (0.15, 0.62) 0.27 (0.08, 0.47) 0.31 (0.09, 0.53) not_significant no-no-yes-no-mild 0.38 0.15 - 0.62 0.27 0.08 - 0.47 0.31 0.09 - 0.53 not_sig not_sig no no yes no moderate 0.29 (0.07, 0.52) 0.41 (0.16, 0.67) 0.27 (0.05, 0.49) not_significant no-no-yes-no-moderate 0.29 0.07 - 0.52 0.41 0.16 - 0.67 0.27 0.05 - 0.49 not_sig not_sig no no yes no severe 0.22 (0, 0.43) 0.69 (0.44, 0.93) not_significant not_significant no-no-yes-no-severe 0.22 0 - 0.43 0.69 0.44 - 0.93 not_sig not_sig not_sig not_sig no no yes yes mild not_significant 0.34 (0.08, 0.61) 0.47 (0.17, 0.76) not_significant no-no-yes-yes-mild not_sig not_sig 0.34 0.08 - 0.61 0.47 0.17 - 0.76 not_sig not_sig no no yes yes moderate not_significant 0.49 (0.18, 0.79) 0.38 (0.08, 0.67) not_significant no-no-yes-yes-moderate not_sig not_sig 0.49 0.18 - 0.79 0.38 0.08 - 0.67 not_sig not_sig no no yes yes severe not_significant 0.78 (0.56, 1) not_significant not_significant no-no-yes-yes-severe not_sig not_sig 0.78 0.56 - 1.00 not_sig not_sig not_sig not_sig no yes no no mild 0.33 (0, 0.66) 0.23 (0, 0.45) 0.44 (0.13, 0.75) not_significant no-yes-no-no-mild 0.33 0 - 0.66 0.23 0 - 0.45 0.44 0.13 - 0.75 not_sig not_sig no yes no no moderate not_significant 0.35 (0.05, 0.65) 0.39 (0.08, 0.69) not_significant no-yes-no-no-moderate not_sig not_sig 0.35 0.05 - 0.65 0.39 0.08 - 0.69 not_sig not_sig

282

Appendix 8: Database search strings for systematic review. Database Keywords, BPS Keywords, FIC Pubmed®, MESH terms ("Cystitis, Interstitial"[Majr]) Felis"[Mesh] AND AND (("Biomarkers"[Majr]) OR "Cystitis"[Mesh] AND ("Cytokines"[Majr])) ("Biomarkers"[Mesh] OR "Cytokines"[Mesh]) Pubmed® advanced "interstitial cystitis" OR "bladder “feline idiopathic cystitis” OR search pain syndrome" AND “feline interstitial cystitis” AND "biomarker" OR "cytokine" “biomarker*” OR “cytokine*” Web of Science® “interstitial cystitis" OR "bladder “feline idiopathic cystitis” OR pain syndrome" AND “feline interstitial cystitis” AND "biomarker" OR "cytokine” “biomarker*” OR “cytokine*” NOT “mouse” OR “rat*” OR “mice” OR “rodent” Scopus® “interstitial cystitis" OR "bladder “feline idiopathic cystitis” OR pain syndrome" AND “feline interstitial cystitis” AND "biomarker" OR "cytokine” “biomarker*” OR “cytokine*” NOT “mouse” OR “rat*” OR “mice” OR “rodent” Congress Library® Search conducted in ‘subject’ and “feline idiopathic cystitis” OR also ‘title’. “feline interstitial cystitis” AND “interstitial cystitis" OR "bladder “biomarker*” OR “cytokine*” pain syndrome" AND "biomarker" OR "cytokine” NOT “mouse” OR “rat*” OR “mice” OR “rodent” Google Scholar "bladder pain syndrome" OR “feline idiopathic cystitis” OR "interstitial cystitis" AND “feline interstitial cystitis” AND biomarker OR cytokine “biomarker*” OR “cytokine*” ~"interstitial cystitis" AND (biomarker OR cytokine) - treatment -rat -mouse -mice - model -rodent -murine CAB abstract NA “feline idiopathic cystitis” OR “feline interstitial cystitis” AND “biomarker*” OR “cytokine*” Veterinary Information NA FIC biomarker cytokine Network NA – not applicable to human studies. CAB abstract - Centre for agriculture and bioscience international. The asterisk * was used to include both singular and pleural forms of that keyword.

283

8.1.1 Bladder pain syndrome database search strings

1. Pubmed® mesh terms 18/2/19

("Cystitis, Interstitial"[Majr]) AND (("Biomarkers"[Majr]) OR ("Cytokines"[Majr]))

Imported 20 to EndNote

2. Pubmed® advanced search terms 18/2/19

(("interstitial cystitis" OR "bladder pain syndrome")) AND ("biomarker" OR "cytokine")

Imported 24 to EndNote

3. Web of science® 19/2/19

TOPIC: (“interstitial cystitis” OR “bladder pain syndrome”) AND TOPIC: (biomarker OR cytokine) NOT TOPIC: (“mouse” OR “rat*” OR “mice” OR “rodent”)

Imported 54 to EndNote

4. Scopus® 19/2/19

(TITLE-ABS-KEY ("interstitial cystitis” OR” bladder pain syndrome”) AND TITLE-ABS- KEY (“biomarker” OR” cytokine”) AND NOT TITLE-ABS- KEY (“mouse” OR” rat*” OR” rodent” OR” mice”)) AND PUBYEAR > 2009

Imported 50 to EndNote

5. Congress library® 19/2/19

Two searches

• Subject containing “interstitial cystitis” OR “bladder pain syndrome” AND subject contains “biomarker” OR “cytokine” NOT title containing “rat” OR “mouse” OR “rodent” OR “mice”

• Title containing “interstitial cystitis” OR “bladder pain syndrome” AND title contains “biomarker” OR “cytokine” NOT title containing “rat” OR “mouse” OR “rodent” OR “mice”

284

Imported 39 to EndNote

6. Google scholar 19/2/19

Two searches

• allintitle: ("bladder pain syndrome" OR "interstitial cystitis") AND (biomarker OR cytokine)

• ~"interstitial cystitis" AND (biomarker OR cytokine) -treatment -rat -mouse -mice - model rodent -murine

22 imported into EndNote

Total BPS studies = 209

3 additional studies identified during the FIC search (Grundy 2018, Patnaik 2017, Kim 2018) = 212

First screen:

Removed 115 duplicates

Identified 41 narrative papers/reviews

Identified 16 for which full text was not able to be accessed

Total BPS studies to undergo secondary screening = 40 studies.

8.1.2 Feline idiopathic cystitis database search strings

1. Pubmed® mesh terms 21/2/19

Initially used Majr term restriction but got no results.

"Felis"[Mesh] AND "Cystitis"[Mesh] AND ("Biomarkers"[Mesh] OR "Cytokines"[Mesh])

Imported 5 to EndNote

2. Pubmed® advanced search 21/2/19

285

((“feline idiopathic cystitis” OR “feline interstitial cystitis”)) AND (“biomarker*” OR “cytokine*”)

Imported 3 to EndNote

3. Web of science® 21/2/19

(“feline interstitial cystitis” OR “feline idiopathic cystitis” AND TOPIC:( “biomarker*” OR “cytokine*”)

Imported 3 to EndNote

4. Scopus® 25/2/19

(TITLE-ABS—KEY (“feline interstitial cystitis” OR “feline idiopathic cystitis”) AND TITLE-ABS- KEY (“biomarker*” OR “cytokine*”))

Imported 3 to EndNote

5. Congress library® 25/2/19

Any (title/abstract/keywords) contains “feline interstitial cystitis” OR “feline idiopathic cystitis” AND contains “biomarker*” OR “cytokine*”

Imported 6 to EndNote

6. Google scholar 25/2/19

("feline idiopathic cystitis" OR "feline interstitial cystitis") AND (biomarker OR cytokine), 2009- 2018, English only

Exported 28 to EndNote

7. CABI (Centre for agriculture and bioscience international) VetMed Resource

("feline idiopathic cystitis" OR "feline interstitial cystitis") AND (biomarker OR cytokine)

Exported 3 to EndNote

8. Veterinary Information Network (VIN)

‘FIC biomarker cytokine’

286

Exported 3 to EndNote

Total FIC studies = 54

First screen:

Removed 24 duplicates

Identified 8 narrative papers/reviews

Identified 14 for which full text was not able to be accessed

Total FIC studies to undergo secondary screening = 8 studies.

287

Appendix 9: List of included studies Corcoran A T, Yoshimura N, Tyagi V, Jacobs B, Leng W, Tyagi P. Mapping the cytokine profile of painful bladder syndrome/interstitial cystitis in human bladder and urine specimens. World Journal of Urology 2013;31(1):241-6.

Jiang Y H, Peng C H, Liu H T, Kuo H C. Increased pro-inflammatory cytokines, C-reactive protein and nerve growth factor expressions in serum of patients with interstitial cystitis/bladder pain syndrome. PLoS One 2013;8(10):e76779.

Kuo H C, Liu H T, Tyagi P, Chancellor M B. Urinary Nerve Growth Factor Levels in Urinary Tract Diseases With or Without Frequency Urgency Symptoms. Luts-Lower Urinary Tract Symptoms 2010;2(2):88-94.

Lemberger S I, Deeg C A, Hauck S M, Amann B, Hirmer S, Hartmann K, et al. Comparison of urine protein profiles in cats without urinary tract disease and cats with idiopathic cystitis, bacterial urinary tract infection, or urolithiasis. Am J Vet Res 2011;72:1407-15.

Liu Hsin‐Tzu, Tyagi Pradeep, Chancellor Michael B, Kuo Hann‐Chorng. Urinary nerve growth factor level is increased in patients with interstitial cystitis/bladder pain syndrome and decreased in responders to treatment. BJU International 2009;104(10):1476-81.

Niimi A, Igawa Y, Aizawa N, Honma T, Nomiya A, Akiyama Y, et al. Diagnostic value of urinary CXCL10 as a biomarker for predicting Hunner type interstitial cystitis. Neurourology and Urodynamics 2018;37(3):1113-9.

Panboon Isadee, Asawakarn Sariya, Pusoonthornthum Rosama. Urine protein, urine protein to creatinine ratio and N-acetyl-β-D-glucosaminidase index in cats with idiopathic cystitis vs healthy control cats. Journal of Feline Medicine and Surgery 2017;19(8):869-75.

Parys M, Yuzbasiyan‐Gurkan V, Kruger JM. Serum cytokine profiling in cats with acute idiopathic cystitis. Journal of veterinary internal medicine 2018;32(1):274-9.

Tyagi P, Killinger K, Tyagi V, Nirmal J, Chancellor M, Peters K M. Urinary chemokines as noninvasive predictors of ulcerative interstitial cystitis. J Urol 2012;187(6):2243-8.

Vera P L, Preston D M, Moldwin R M, Erickson D R, Mowlazadeh B, Ma F, et al. Elevated Urine Levels of Macrophage Migration Inhibitory Factor in Inflammatory Bladder Conditions: A Potential

288

Biomarker for a Subgroup of Interstitial Cystitis/Bladder Pain Syndrome Patients. Urology 2018;116:55-62.

289

Appendix 10: List of excluded studies. Abernethy M G, Rosenfeld A, White J R, Mueller M G, Lewicky-Gaupp C, Kenton K. Urinary microbiome and cytokine levels in women with interstitial cystitis. Obstetrics and Gynecology 2017;129(3):500-6.

Bradley M S, Burke E, Grenier C, Amundsen C L, Murphy S K, Siddiqui N Y. A genome-scale DNA methylation study in women with interstitial cystitis/bladder pain syndrome. Neurourology and Urodynamics 2018;37(4):1485-93.

Choi D, Han J Y, Shin J H, Ryu C M, Yu H Y, Kim A, et al. Downregulation of WNT11 is associated with bladder tissue fibrosis in patients with interstitial cystitis/bladder pain syndrome without Hunner lesion. Scientific Reports 2018;8(1).

Di Capua-Sacoto C, Sanchez-Llopis A, O’connor E, Martinez A, Ruiz-Cerdá J L. Study of the apoptotic effect of urine as a diagnostic biomarker in patients with interstitial cystitis. Actas Urológicas Españolas (English Edition) 2016;40(9):570-6.

Di Capua-Sacoto C, Sanchez-Llopis A, O’Connor JE, Martinez-Romero A, Ruiz-Cerdá JL. Apoptotic effect as biomarker of disease, severity and follow-up in interstitial cystitis. Actas Urológicas Españolas (English Edition) 2018;42(4):262-6.

Furuta A, Yamamoto T, Suzuki Y, Gotoh M, Egawa S, Yoshimura N. Comparison of inflammatory urine markers in patients with interstitial cystitis and overactive bladder. International Urogynecology Journal 2018;29(7):961-6.

Gamper M, Regauer S, Welter J, Eberhard J, Viereck V. Are mast cells still good biomarkers for bladder pain syndrome/interstitial cystitis? Journal of Urology 2015;193(6):1994-2000.

Goo Y A, Tsai Y S, Liu A Y, Goodlett D R, Yang C C. Urinary proteomics evaluation in interstitial cystitis/painful bladder syndrome: a pilot study. Int Braz J Urol 2010;36(4):464-78; discussion 478- 9, 479.

Hauser Paul J, Vangordon Samuel B, Seavey Jonathan, Sofinowski Troy M, Ramadan Mohammad, Abdullah Shivon, et al. Abnormalities in Expression of Structural, Barrier and Differentiation Related Proteins, and Chondroitin Sulfate in Feline and Human Interstitial Cystitis. The Journal of Urology 2015;194(2):571-7.

290

Keay Susan, Nallar Shreeram C, Gade Padmaja, Zhang Chen-Ou, Kalvakolanu Dhan V. Oncosuppressor protein p53 and cyclin-dependent kinase inhibitor p21 regulate interstitial cystitis associated gene expression. Cytokine 2018;110:110-5.

Kim S W, Im Y J, Choi H C, Kang H J, Kim J Y, Kim J H. Urinary nerve growth factor correlates with the severity of urgency and pain. International Urogynecology Journal 2014;25(11):1561-7.

Kind T, Cho E, Park T D, Deng N, Liu Z, Lee T, et al. Interstitial Cystitis-Associated Urinary Metabolites Identified by Mass-Spectrometry Based Metabolomics Analysis. Sci Rep 2016;6:39227.

Kiuchi Hiroshi, Tsujimura Akira, Takao Tetsuya, Yamamoto Keisuke, Nakayama Jiro, Miyagawa Yasushi, et al. Increased vascular endothelial growth factor expression in patients with bladder pain syndrome/interstitial cystitis: its association with pain severity and glomerulations. BJU international 2009;104(6):826-31.

Kutlu O, Akkaya E, Koksal I T, Bassorgun I C, Ciftcioglu M A, Sanlioglu S, et al. Importance of TNF-related apoptosis-inducing ligand in pathogenesis of interstitial cystitis. Int Urol Nephrol 2010;42(2):393-9.

Lamb L E, Janicki J, Bartolone S N, Peters K M, Chancellor M B. Development of an interstitial cystitis risk score for bladder permeability. PLoS One 2017;12(10):e0185686.

Lemberger S I, Dorsch R, Hauck S M, Amann B, Hirmer S, Hartmann K, et al. Decrease of Trefoil factor 2 in cats with feline idiopathic cystitis. BJU Int 2011;107(4):670-7.

Liu H T, Tyagi P, Chancellor M B, Kuo H C. Urinary nerve growth factor but not prostaglandin E2 increases in patients with interstitial cystitis/bladder pain syndrome and detrusor overactivity. BJU International 2010;106(11):1681-5.

Liu S, Feng S, Luo D. Analysis of key genes and micro-RNA-mRNA regulatory networks in women with ulcerative interstitial cystitis/pain bladder syndrome. International Urogynecology Journal 2018.

Logadottir Y, Delbro D, Lindholm C, Fall M, Peeker R. Inflammation characteristics in bladder pain syndrome ESSIC type 3C/classic interstitial cystitis. International Journal of Urology 2014;21(S1):75-8.

Logadottir Y, Delbro D, Fall M, Gjertsson I, Jirholt P, Lindholm C, et al. Cytokine expression in patients with bladder pain syndrome/interstitial cystitis ESSIC type 3C. Journal of Urology 291

2014;192(5):1564-8.

Ma E, Vetter J, Bliss L, Lai H H, Mysorekar I U, Jain S. A multiplexed analysis approach identifies new association of inflammatory proteins in patients with overactive bladder. American Journal of Physiology - Renal Physiology 2016;311(1):F28-34.

Makino T, Kawashima H, Konishi H, Nakatani T, Kiyama H. Elevated urinary levels and urothelial expression of hepatocarcinoma-intestine-pancreas/pancreatitis-associated protein in patients with interstitial cystitis. Urology 2010;75(4):933-7.

Offiah I, Didangelos A, Dawes J, Cartwright R, Khullar V, Bradbury E J, et al. The Expression of Inflammatory Mediators in Bladder Pain Syndrome. Eur Urol 2016;70(2):283-90.

Ogawa T, Homma T, Igawa Y, Seki S, Ishizuka O, Imamura T, et al. CXCR3 binding chemokine and TNFSF14 over expression in bladder urothelium of patients with ulcerative interstitial cystitis. J Urol 2010;183(3):1206-12.

Richter B, Roslind A, Hesse U, Nordling J, Johansen J S, Horn T, et al. YKL-40 and mast cells are associated with detrusor fibrosis in patients diagnosed with bladder pain syndrome/interstitial cystitis according to the 2008 criteria of the European Society for the Study of Interstitial Cystitis. Histopathology 2010;57(3):371-83.

Rooney P, Srivastava A, Watson L, Quinlan L R, Pandit A. Hyaluronic acid decreases IL-6 and IL-8 secretion and permeability in an inflammatory model of interstitial cystitis. Acta Biomaterialia 2015;19:66-75.

Rubio-Diaz D E, Pozza M E, Dimitrakov J, Gilleran J P, Giusti M, Stella J L, et al. A candidate serum biomarker for bladder pain syndrome/interstitial cystitis. Analyst 2009;134(6):1133-7.

Schrepf A, O'Donnell M, Luo Y, Bradley C S, Kreder K, Lutgendorf S. Inflammation and inflammatory control in interstitial cystitis/bladder pain syndrome: Associations with painful symptoms. Pain 2014;155(9):1755-61.

Schwalenberg T, Stolzenburg J U, Ho T P, Mallock T, Hartenstein S, Alexander H, et al. Enhanced urothelial expression of human chorionic gonadotropin beta (hCGbeta) in bladder pain syndrome/interstitial cystitis (BPS/IC). World J Urol 2012;30(3):411-7.

Shahid M, Lee M Y, Yeon A, Cho E, Sairam V, Valdiviez L, et al. Menthol, a unique urinary volatile 292

compound, is associated with chronic inflammation in interstitial cystitis. Scientific Reports 2018;8(1).

Stella J, Croney C, Buffington T. Effects of stressors on the behavior and physiology of domestic cats. Applied Animal Behaviour Science 2013;143(2-4):157-63.

Tonyali S, Ates D, Akbiyik F, Kankaya D, Baydar D, Ergen A. Urine nerve growth factor (NGF) level, bladder nerve staining and symptom/problem scores in patients with interstitial cystitis. Advances in Clinical and Experimental Medicine 2018;27(2):159-63.

Treutlein Gudrun, Dorsch Roswitha, Euler Kerstin N, Hauck Stefanie M, Amann Barbara, Hartmann Katrin, et al. Novel potential interacting partners of fibronectin in spontaneous animal model of interstitial cystitis. PloS one 2012;7(12):e51391.

Walker Stephen J, Colaco Marc, Koslov David S, Keys Tristan, Evans Robert J, Badlani Gopal H, et al. Transcriptome analysis of bladder biopsy from interstitial cystitis/bladder pain syndrome patients. Genomics Data 2014;2(C):366-8.

Wen H, Lee T, You S, Park S H, Song H, Eilber K S, et al. Urinary metabolite profiling combined with computational analysis predicts interstitial cystitis-associated candidate biomarkers. J Proteome Res 2015;14(1):541-8.

293

Appendix 11: Characteristics of excluded studies. Study Year Substrate Test method Disease group Control group Biomarkers measured Reason for exclusion (4) 2017 Urine Immunoassay NHIC Healthy controls FGF-2, TGF-a, G-CSF, Flt-3L, Inadequate results reporting GM-CSF, CX3CL1, IFNa2, IFNγ, GRO, MCP-3, IL-12p40, MDC, PDGF-AA, PDGF-BB, IL-1, sCD40L, IL-1ra, IL-4, IL- 7, IL-8, IP-10, MCP-1, MIP-1B, CCL5, VEGF (42) 2018 Urine PCR IC/BPS Healthy controls MAPK pathway Did not differentiate types of IC/BPS + could be on oral treatment. Author replied: "We did not have any patients that had Hunner's lesion on cystoscopy, but only 50% of our patients had had a cystoscopy." Therefore, can't tell if had ulcer/non ulcer combined or not (77) 2018 Tissue PCR HIC and NHIC Healthy control WNT pathway genes, sonic No results values, no email reply hedgehog, other biomarkers- CCR2, MCP-1, NFKB, HB- EGF, NGF, ARF, eNOS, nNOS, iNOS, CHAT, CHT, OCT-1, SMRT-1 (101) 2016 Urine Flow cytometry IC/BPS Healthy controls NR Did not differentiate types of IC/BPS, included both HIC and NHIC (102) 2018 Urine Flow cytometry IC/BPS Healthy control Apoptotic effect of urine Insufficient data on control population, combined HIC/NHIC

294

Study Year Substrate Test method Disease group Control group Biomarkers measured Reason for exclusion (138) 2018 Urine Immunoassay HIC and NHIC OAB VEGF, IL1a, IL6, CCL2, CCL5, Controls not healthy CXCL1, CXCL8, CXCL10 + more that were not sig (40 in total - growth factors, cytokines, chemokines) (139) 2015 Serum IHC + serology HIC and NHIC Healthy controls Mast cells, lymphocytes, Most results categorical only, and and OAB tryptase, C-reactive protein can't compare. Insufficient tissue results reported for CRP. (147) 2010 Urine Mass spectrometry IC/BPS Healthy controls NR Did not differentiate types of IC/BPS (159) 2015 Tissue IHC FIC Healthy controls C6S, biglycan, decorin, E- Results categorical cadherin, Uroplakin, K20, ZO-1 (188) 2018 NR NR NR NR NR Cell culture only, not real people. (196) 2014 Urine NR IC/BPS NR NR Unclear if NHIC and HIC combined, or just NHIC. Emailed authors – no reply. (197) 2016 Urine Spectrophotometry NHIC Healthy controls Histidine, tartaric acid No results values. Disease group all NHIC - confirmed by author. (199) 2009 Tissue IHC IC/BPS NR VEGF, CD34, a-smooth muscle Unclear if NHIC and HIC actin combined, or only one type (219) 2010 Tissue IHC IC/BPS No controls TNF-related apoptosis-inducing Did not differentiate types of ligand (TRAIL) IC/BPS + no control group (220) 2017 Urine Immunoassay HIC and NHIC Healthy control GRO-a/CXCL1, IL-1RA, IL-6, Relying on patients' own IL-8, CXCL10, MCP/CCL2, understanding of whether they RANTES/CCL5, VEGF, PDGF- have HIC or NHIC. Controls BB selected by IC participants. (233) 2011 Tissue IHC, FIC Healthy controls TFF2 No results values and spectrophotometry, urine Western blot

295

Study Year Substrate Test method Disease group Control group Biomarkers measured Reason for exclusion (248) 2010 Urine ELISA NHIC NR NR No units for results (249) 2018 Other Other - analysis of HIC Healthy control Gene regulatory networks Genetic pattern, not so much mRNA profiles of biomarkers bladder tissue from biopsy (251) 2014 Tissue IHC and PCR HIC and NHIC Healthy controls NO (chemiluminescence), NoS, No results values IL-6, IL-17a, IL-10 (250) 2014 Tissue IHC and PCR HIC Women having IL4, IL6, IL10, IL17A, iNOS, Controls not healthy operation for TNF, TGF, IFNγ, IL17 stress urinary (IHC), mast cells incontinence (265) 2016 Urine Immunoassay IC/BPS OAB TNF, MIP-1B Did not differentiate types of IC/BPS (269) 2010 Urine ELISA and IHC IC/BPS Health control HIP/PAP Did not differentiate types of and IC/BPS tissue (309) 2016 Tissue PCR IC/BPS Stress urinary CCL21, FGF7, IL12A, CXCL1, Did not differentiate types of incontinence TNF + 91 others IC/BPS – authors confirmed both HIC and NHIC pooled together (310) 2010 Tissue PCR HIC UC, benign NR No exact measurements given; prostatic control groups not differentiated. hyperplasia (348) 2010 Tissue IHC IC/BPS Healthy controls YKL-40, tryptase, CD-68 Urine sample IC/BPS includes both NHIC and HIC together. Also the biopsy group all IC/BPS that had mastocytosis - bias (352) 2015 NR NR NR NR NR In vitro model, no real people

296

Study Year Substrate Test method Disease group Control group Biomarkers measured Reason for exclusion (355) 2009 Serum Other - Infrared FIC and Healthy control Infrared microscopy spectra Study about testing for microspectroscopy, IC/BPS biomarkers, not actual marker LC-MS (363) 2014 NR NR IC/BPS NR NR Unclear if IC/BPS includes NHIC and HIC (364) 2012 Tissue PCR IC/BPS Healthy control hCGB Did not differentiate types of IC/BPS (371) 2018 Urine Spectrophotometry IC/BPS Healthy controls Menthol Did not differentiate types of IC/BPS. Did not describe patient sample sizes/ages (385) 2013 Serum FIC Healthy controls TNF-a was increased in FIC vs About response to stress, not healthy, at baseline. After baseline FIC vs controls stressful event - re measured cortisol, leukogram, expression of genes for IL-1B, IL-6, TNF-a. (403) 2018 Urine ELISA IC/BPS Cystoscopy of NGF Author confirmed - IC/BPS previous bladder includes both Hunner’s and non- tumour Hunner’s surveillance or LUTS (407) 2012 Tissue IHC, FIC Healthy controls Fibronectin, Ig kappa chain V No results values and spectrophotometry, region 3315, Ig gamma chain C urine Western blot region, Alpha-S1-casein, Caspase-14, Complement C4-A, Galectin-7, Fatty acid-binding protein (intestinal), Thioredoxin, NF-kB, MAPK (426) 2014 Tissue PCR IC/BPS normal Healthy controls NR Not specifically about capacity = biomarkers. Small pilot study NHIC only.

297

Study Year Substrate Test method Disease group Control group Biomarkers measured Reason for exclusion (438) 2015 NR NR IC/BPS Healthy controls NR Not about biomarkers, about predictive modelling using metabolite profiling NR = not reported, that is, the study was excluded prior to data extraction.

298

Appendix 12: Meta-analysis for all biomarkers for which a meta-analysis was possible (more than one study for that biomarker, n=5). Biomarker Subgroup Overall Z value P value Sensitivity analysis differences* significant significant? effect (P value ≤0.05) NGF no yes 3.48 <0.001 HIC alone = no heterogeneity, no/borderline significant effect (P = 0.06) NHIC alone = no heterogeneity, significant overall effect MIF yes no 1.25 0.21 HIC only = only one study so can’t compare. NHIC only = moderate heterogeneity (not significant), no overall effect CCL2 (MCP- yes no 0.62 0.54 HIC only = no heterogeneity, no overall 1) effect NHIC alone = significant heterogeneity, no overall effect CXCL10 yes no 1.70 0.09 HIC only = significant heterogeneity, no overall effect. NHIC only = significant heterogeneity, no overall effect. IL-6 yes no 0.80 0.42 HIC only = no heterogeneity, significant overall effect. NHIC only = moderate heterogeneity, no overall effect. IL-6 HIC no yes 3.76 <0.001 HIC only = no heterogeneity, significant only overall effect. *The subgroups were the different diseases (HIC/NHIC/FIC) and units (pg/ml, pg/mg Cr).

299

Appendix 13: Summary of results from the Review Manager analysis of all 45 outcomes (biomarkers).

Category Marker ID Overall Significant

significant overall effect

subgroup

differences dy

NHIC urine pg/ml urine NHIC pg/mg Cr urine HIC Cr pg/mg urine NHIC pg/ml FICserum tissue pg/mg NHIC protein pg/ serum NHIC FICother stu one Only Corcoran** HIC urine pg/ml urine HIC CC chemokine CCL2 (MCP-1) yes yes but not y y y y y no . accepted due to SG differences CCL3/MIP1a no no y y no . CCL5 (RANTES) no yes y y y no . CTACK (CCL27) NA* NA y y yes yes MCP-3/CCL7 yes no y y yes yes CXC CXCL1 (GRO) yes yes but not y y y y y no yes chemokine accepted due to SG differences CXCL10 yes yes but not y y y y no . accepted due to SG differences IL-8 (CXCL8) yes no y y no . MIG/CXCL9 no no y y yes no SDF1a/CXCL12 yes yes if y y y no yes remove urine and only have FIC 300

serum/ human tissue Cytokine Flt-3L NA no y yes . ICAM-1 no no y y yes yes IFN-a no no y y yes yes IFN-y NA no y yes . IL-12 no no y y yes yes IL-12p40 NA yes y yes . IL-13 NA no y yes . IL-16 yes no y y yes yes IL-18 yes yes but not y y y no yes accepted due to SG differences IL-1a no no y y yes yes IL-1B NA yes y yes . IL-1RA no no y y yes . IL-2Ra no no y y yes yes IL-3 no no y y yes yes IL-4 NA yes y yes . IL-6 yes yes but not y y y y y y no . accepted due to SG differences LIF no no y y yes no MCSF no no y y yes yes MIF yes yes but not y y y y no no accepted due to SG differences SCF no no y y y no yes 301

TNF-a yes no y y no . TNF-B no no y y yes yes TNFSF14/LIGHT no yes y y yes . TRAIL no no y y yes yes VCAM-1 yes no y y yes yes growth factor HB-EGF no yes y y yes . HGF yes no y y yes yes NGF no yes y y y y y y no yes SCGFB yes no y y yes yes VEGF no yes y y yes . other CRP NA yes y yes . (inflammatory protein) other PGE2 no no y y yes . (prostaglandin) other sFas NA no y yes . (apoptosis) other (enzyme) NAG NA no y yes . other Fibronectin NA yes y yes . (extracellular matrix) SG = subgroup

*NA = meta-analysis unable to be performed as no similar groups to compare.

**Corcoran: . = Corcoran study not involved; No = Corcoran study evaluated, inverse effect not present; Yes = Corcoran study evaluated, and urine biomarkers were higher in controls compared to disease (decreased in disease), and for tissue analysis the biomarkers were higher in disease than controls.

302

Appendix 14: Biomarker control genes synthesised by Bioneer Pacific. Biomarker Gene sequence Amp (bp) Nerve Growth CGCAGAGCCCGCAGCACCCCGGCCGGGGCAATAGCCGCGAGGGTGGCAGGGCAGACCCGCAAC 287 Factor beta ATCACTGTGGACCCCAAACTTTTTAAAAAGCGGCGACTGCGTTCACCCCGCGTGCTGTTCAGCA subunit (NGF CGCACCCCCCACCTGTGGCTGCAGATACTCAGGGTCTGGACTTGGAGGCGGGTGGTGCTGCCTC gene), mRNA CTTCAACAGGACTCACAGGAGCAAGCGGTCGTCGTCGCACCCTGTCTTCCACCGGGGGGAGTTC TCGGTGTGCGACAGCGTCAGCGTGTGGGTGGG Fibronectin 1 ACCATCCAAGTCATTGCCCTCAAGAACAACCAGAAGAGTGAGCCTCTGATTGGAAGGAAAAAG 300 (FN1), transcript ACAGATGAGCTTCCCCAACTGGTAACCCTTCCACACCCCAATCTTCACGGACCAGAGATCTTGG variant X1, ATGTTCCCTCCACAGTTCAAAAGACCCCTTTCATCAACCCTGGGTATGACACTGGAAACGGTAT mRNA TCAGCTTCCTGGCACTTCTGGTCAGCAGCCCAGTGTTGGGCAACAAATGATCTTTGAGGAGCAT GGTTTTAGGCGCACCACACCGCCCACAACGGCCACCCCTGTAAGG Tight junction TCTGCAGGCGTTGAGATGGGTGAGGCAGGGGCCAGGCCCTGGTTCTATTGCTGTGAGTCAACAA 480 protein 1 (TJP-1), CACAGGCAGAAAGGGAGCAGAGGACCAAGCAGCTCTTGGAAATGCCGCTGAGTTACAAGGAG transcript variant AGGATACTGAAGGAATTCAACATCAGCCAGGTGCTCCCACGTCTAGTATATGATGGCGTGTTCT X1, mRNA CCCTGAAGGAGTACAGAGAGATCCTCTCCTGGCATTGCCACCCACGGAGAGTGGAGTCCTTTTT TCTGAAGCTTTGCTCCAAAGGTCCAAAGGCTTTCTGTGCTTTCTGCTCTCATCTGGAGGAATTCT GTCCTTACCTGCTCACCTGCTTTTTCCTTTATTACCAAGAACAAACCTATAGGATCTTACAAGAG GCATCCACTGCAGAAGAAAAAGCAAGAGTTGGAACTCAGCCTGAACCTAGCGATGTCCTTGAG ACAGAAGATCAAGAACAAAGTCTTCAATTATTT Interleukin-8 ATGCCAGTGCATAAAAACTCATTCCACACCTTTCAATCCCAAATTAATCAAAGAACTGACAGTG 360 mRNA, complete ATTGACAGTGGCCCACACTGTGAAAACTCAGAAATCATTGTAAAGCTCGTCAATGGAAAAGAG cds GTGTGCCTGGACCCCAAGCAAAAGTGGGTGCAGAAGGTTGTGGAGATATTTTTGAAGAAAGCT GAGAAACAAAATGCATAAACAAACAAACAAACACATTCTCCACGGTTTCCAAGAATTCTTCAAT AAAGATGCCAATGAGACTTCAAGCAAGCAAATTCACTTCAGCACTTCATGCAGTGTGTGGGTCT GGTGTAGGGTTGGTTGCCAGATAAAATAGAGTATGCTCAGTT Uroplakin 3A CGCCAGGTGCGCGGACCCTGAGAGGCGCGGTGCCCGCTGGACCGCCCGCCCCGCACTCCCCGC 420 (UPK3A), GGCTGGTTCAAGGCCATGCTTCCGCTCTGGGCCCTGCTGGCCCTCGGCTGCCTGCGCCTCGGCTC transcript variant GGGTGTGAACCTCCAGCCCCAGCTGGCCAGTGTGACCTTTGCCACCAACAACCCGACCCTCACC X1, mRNA ACGGTGGCCTTGGAAAAGCCTCTCTGCGTGTTTGACAGCTCAGCGGCCCTCGATGGCACTTTTG AGGTCTACCTCTATGTCCTGGTCGACTCAGCCAGCTCCAGGAACGCCTCCGTGCAGGACGCCAA GACCCCGCTGAGCTCCACCTTCCAGCAGACAGAGGGGGGAAGGACGGGCCCCTATAAGGCAGC GGCCTTTGACCTGATCCCCTGTGGTGACCTGCCCAGC 303

Appendix 15: Protein BLAST homology reports, displaying the top protein sequence for feline and canine TJP-1, and E-cadherin.

304

305

Appendix 16: Nanodrop results for extracted RNA and reverse transcribed cDNA. Pathology Sample# Species Diagnosis RNA RNA RNA Reverse cDNA cDNA cDNA GAPDH Code ng/µL† 260/280† 260/230† transcribed ng/µL 260/280 260/230 band** 19/0528 1 Feline UTI 5.75 1.86 1.45 yes 1369.06 1.83 2.14 yes 08/1436 2 Canine UC 3.75 1.71 1.41 yes 1039.91 1.79 1.88 no 14/0993 3 Canine Cystitis 10.80 1.91 1.80 yes 1813.04 1.85 2.25 no 17/1334 4 Feline Normal 5.55 2.34 1.67 yes 1841.59 1.87 2.30 no 18/0653 5 Canine Urolithiasis 2.80 1.89 1.05 yes 1878.12 1.86 2.29 no 17/0311 6 Canine Normal 21.45 1.84 1.21 yes 1727.53 1.86 2.29 yes 09/0313 7 Feline UC 40.65 1.83 1.31 yes 1883.16 1.85 2.27 no 09/1608 8 Canine UC 2.10 0.81 0.21 yes 1328.36 1.83 2.04 no 97/1529 9 Canine Cystitis 0.70 15.90 0.26 yes 1557.65 1.85 2.21 no 98/574 10 Feline Stomach* 1.65 4.61 0.60 yes 1662.30 1.85 2.24 no 12/0445 11 Feline FIC 1.85 3.14 0.67 yes 1517.57 1.87 2.20 no 03/0244 12 Feline UC 166.70 1.87 2.01 yes 1825.38 1.86 2.26 no 12/0445 13 Feline Kidney* 3.40 1.65 1.19 yes 1767.76 1.85 2.24 no 08/1527 14 Feline UC 31.95 1.76 1.11 yes 1666.37 1.85 2.24 no 15/0893 15 Canine Normal 2.55 1.40 1.01 yes 1581.96 1.84 2.14 no 18/0914 16 Feline Normal 7.30 0.74 0.37 yes 1181.09 1.83 2.03 no 96/1337 17 Canine UC 51.60 1.83 1.96 yes 1578.84 1.85 2.20 no 08/1399 18 Feline UTI 5.25 1.85 0.95 yes 1615.43 1.85 2.24 yes 21/no 22 17/1336 19 Feline Normal 6.35 2.05 1.29 yes 1797.49 1.86 2.29 no 21/yes 22 07/2196 20 Canine Cystitis 17.15 1.70 1.07 yes 1761.80 1.86 2.29 no 11/0220 21 Canine Urolithiasis 1.00 5.55 1.41 yes 1734.88 1.86 2.31 no 05/1429 22 Feline UC 28.40 1.90 1.87 yes 1578.66 1.84 2.20 no 98/0574 23 Feline UTI 25.00 1.80 1.30 yes 1553.82 1.83 2.20 no 10/0592 24 Canine Urolithiasis 17.20 2.08 1.91 yes 1824.03 1.85 2.28 no 15/0893 25 Canine Small intestine* 115.40 1.93 1.74 yes 1799.89 1.85 2.28 no 95/1363 26 feline FIC 7.65 1.76 1.21 yes 1807.01 1.86 2.29 no 17/0398 27 Canine Normal 118.20 1.84 1.48 yes 1822.71 1.86 2.29 no 306

Pathology Sample# Species Diagnosis RNA RNA RNA Reverse cDNA cDNA cDNA GAPDH Code ng/µL† 260/280† 260/230† transcribed ng/µL 260/280 260/230 band** 14/0707 28 Feline FIC 21.35 1.60 0.85 yes 1276.39 1.84 2.11 no 17/0313 29 Canine Normal 85.60 1.92 1.73 yes 1355.77 1.84 2.12 no 07/0026 30 Canine Cystitis 8.10 1.67 1.10 yes 1487.09 1.85 2.19 no 17/1335 31 Feline Normal 59.30 1.84 1.47 yes 1795.61 1.86 2.29 yes 97/1529 32 Canine Kidney* 21.75 1.69 1.06 yes 1875.36 1.86 2.28 no 12/0516 33 Feline FIC 36.05 1.77 1.33 yes 1766.10 1.86 2.29 no 08/1885 34 Canine Urolithiasis 24.85 1.65 1.29 yes 1716.61 1.86 2.28 no 97/0102 35 Canine Cystitis 107.05 1.82 1.88 no NA NA NA NA 13/0197 36 Feline FIC 35.15 1.83 1.64 yes 1920.10 1.87 2.31 no 04/1068 37 Canine UC 47.55 1.91 1.87 yes 1534.82 1.83 2.12 no 11/0143 38 Feline UTI 20.50 1.70 1.19 yes 1947.82 1.86 2.26 no 18/0916 39 Feline Normal 22.60 1.62 0.93 yes 1541.90 1.85 2.18 yes 02/0691 40 Canine Lung 276.50 1.94 1.98 no NA NA NA NA 17/0397 41 Canine Normal 58.40 1.81 1.50 yes 1671.08 1.85 2.22 yes 21/no 22 98/1359 42 Feline FIC/Urolithiasis 29.60 1.72 1.26 yes 1702.60 1.88 2.32 no 16/0034 43 Canine Normal 16.90 1.47 0.80 no NA NA NA NA 96/1185 44 Canine Cystitis 78.65 1.97 1.98 no NA NA NA NA 18/0073 45 Canine Urolithiasis 51.65 1.73 1.04 no NA NA NA NA 07/0581 46 Canine UC 19.40 1.66 0.81 no NA NA NA NA 18/0009 47 Canine Urolithiasis 18.40 1.87 1.17 no NA NA NA NA 98/1183 48 Feline UTI 9.75 1.96 1.87 yes 1033.90 1.88 2.27 no 17/1337 49 Feline Urolithiasis 20.15 1.89 1.40 yes 1720.80 1.87 2.31 no 19/0729 50 Canine Urolithiasis 7.90 2.08 1.48 yes 1690.20 1.87 2.31 yes 19/0025 51 Feline Normal 73.77 2.14 1.50 yes 1661.10 1.86 2.24 yes 19/0024 52 Canine Normal 20.35 1.98 1.88 yes 1545.60 1.87 2.26 yes 19/0026 53 Feline Normal 76.15 2.11 1.97 yes 1521.40 1.88 2.28 yes 19/0528 54 Feline UTI + blocked 24.10 1.99 0.87 yes 1334.50 1.87 2.25 yes

307

Pathology Sample# Species Diagnosis RNA RNA RNA Reverse cDNA cDNA cDNA GAPDH Code ng/µL† 260/280† 260/230† transcribed ng/µL 260/280 260/230 band** 19/0729 55 Canine Urolithiasis + 9.60 1.81 1.39 no NA NA NA NA UTI *Additional tissues included as a reference for housekeeping genes and RNA extraction. FIC = feline idiopathic cystitis; NA = not applicable; UC = urothelial carcinoma; UTI = urinary tract infection. **Yes = a band was seen on conventional PCR at the appropriate size. 21/22 refers to experiments 21 and 22. †For RNA, two measurements were taken, and this table displays the average of these two measurements.

308