<%BANNER%>

Diagnosis and Epidemiological Investigation of H3N8 Canine Influenza Virus.

Permanent Link: http://ufdc.ufl.edu/UFE0042780/00001

Material Information

Title: Diagnosis and Epidemiological Investigation of H3N8 Canine Influenza Virus.
Physical Description: 1 online resource (166 p.)
Language: english
Creator: ANDERSON,TARA CREEL
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: CANINE -- DIVA -- EPIDEMIOLOGY -- FACTORS -- H3N8 -- HEMAGGLUTINATION -- HI -- ILI -- ILLNESS -- IMMUNOASSAY -- INFLUENZA -- INHIBITION -- LIKE -- MODELING -- NS1 -- RISK -- SEROLOGY -- VIRUS
Veterinary Medicine -- Dissertations, Academic -- UF
Genre: Veterinary Medical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Canine influenza A virus subtype H3N8 (H3N8 CIV) was recognized in 2004 as a respiratory pathogen for dogs. The virus has since affected thousands of dogs in 34 states, and has established endemicity in several metropolitan areas of the U.S. This PhD research encompassed two main objectives. The first objective was to develop tools for serological diagnosis of H3N8 CIV. The diagnostic performance of the H3N8 CIV hemagglutination inhibition (HI) assay was assessed and test parameter variations were evaluated. Study results promote standardization of the H3N8 CIV HI assay to ensure accurate diagnosis of infected dogs. Introduction of a H3N8 CIV vaccine confounds future diagnosis, however, because antibodies generated in response to infection or vaccination cannot be distinguished in the HI assay. Development of a test strategy that differentiates infected from vaccinated animals (DIVA) was therefore pursued. Uninfected and vaccinated dogs could not be distinguished from infected dogs in NS1 tests, however, supporting the conclusion that a NS1 immunoassay is not a viable DIVA strategy for H3N8 CIV. The second overall objective of this research was to conduct epidemiological investigations of canine influenza to better inform prevention and control strategies. To determine if H3N8 CIV was infecting dogs prior to 2004, a survey of archived canine sera was conducted. Antibodies to canine and equine H3 proteins were identified in dogs as early as 1999. Next, a national serosurvey for H3N8 CIV infection in dogs with influenza-like illness was conducted to identify risk factors for infection. Analyses revealed that dogs 1 year of age and older, dogs housed in boarding kennels or animal shelters, and dogs in the northeastern and western regions of the country were at highest risk for seropositivity. Based on identification of communal facilities as risk factors for infection, mathematical modeling was conducted to explore the impact of several parameters on H3N8 CIV transmission in shelters. Indirect virus transmission parameters and shelter capacity had the greatest influence on the time course of epidemics and the number of dogs affected. These factors can be targeted in future modeling studies to determine what intervention strategies would avert or mitigate epidemics.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by TARA CREEL ANDERSON.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Gibbs, Edward P.
Local: Co-adviser: Crawford, Patti C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042780:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042780/00001

Material Information

Title: Diagnosis and Epidemiological Investigation of H3N8 Canine Influenza Virus.
Physical Description: 1 online resource (166 p.)
Language: english
Creator: ANDERSON,TARA CREEL
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: CANINE -- DIVA -- EPIDEMIOLOGY -- FACTORS -- H3N8 -- HEMAGGLUTINATION -- HI -- ILI -- ILLNESS -- IMMUNOASSAY -- INFLUENZA -- INHIBITION -- LIKE -- MODELING -- NS1 -- RISK -- SEROLOGY -- VIRUS
Veterinary Medicine -- Dissertations, Academic -- UF
Genre: Veterinary Medical Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Canine influenza A virus subtype H3N8 (H3N8 CIV) was recognized in 2004 as a respiratory pathogen for dogs. The virus has since affected thousands of dogs in 34 states, and has established endemicity in several metropolitan areas of the U.S. This PhD research encompassed two main objectives. The first objective was to develop tools for serological diagnosis of H3N8 CIV. The diagnostic performance of the H3N8 CIV hemagglutination inhibition (HI) assay was assessed and test parameter variations were evaluated. Study results promote standardization of the H3N8 CIV HI assay to ensure accurate diagnosis of infected dogs. Introduction of a H3N8 CIV vaccine confounds future diagnosis, however, because antibodies generated in response to infection or vaccination cannot be distinguished in the HI assay. Development of a test strategy that differentiates infected from vaccinated animals (DIVA) was therefore pursued. Uninfected and vaccinated dogs could not be distinguished from infected dogs in NS1 tests, however, supporting the conclusion that a NS1 immunoassay is not a viable DIVA strategy for H3N8 CIV. The second overall objective of this research was to conduct epidemiological investigations of canine influenza to better inform prevention and control strategies. To determine if H3N8 CIV was infecting dogs prior to 2004, a survey of archived canine sera was conducted. Antibodies to canine and equine H3 proteins were identified in dogs as early as 1999. Next, a national serosurvey for H3N8 CIV infection in dogs with influenza-like illness was conducted to identify risk factors for infection. Analyses revealed that dogs 1 year of age and older, dogs housed in boarding kennels or animal shelters, and dogs in the northeastern and western regions of the country were at highest risk for seropositivity. Based on identification of communal facilities as risk factors for infection, mathematical modeling was conducted to explore the impact of several parameters on H3N8 CIV transmission in shelters. Indirect virus transmission parameters and shelter capacity had the greatest influence on the time course of epidemics and the number of dogs affected. These factors can be targeted in future modeling studies to determine what intervention strategies would avert or mitigate epidemics.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by TARA CREEL ANDERSON.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Gibbs, Edward P.
Local: Co-adviser: Crawford, Patti C.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2013-04-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042780:00001


This item has the following downloads:


Full Text

PAGE 1

1 DIAGNOSIS AND EPIDEMIOLOGICAL INVESTIGATION OF H3N8 CANINE INFLUENZA VIRUS By TARA CREEL ANDERSON A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Tara Creel Anderson

PAGE 3

3 To my husband and parents thank you for all of your love and support

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my supervisory committee, Dr. Paul Gi bbs, Dr. Cynda Crawford, Dr. Bill Castleman, Dr. Jorge Hernandez, Dr. Cynthia Lord, and Dr. Ayalew Mer gia for their insight, guidance, and encouragement throughout the PhD program. I also appreciated the support of Dr. Mary Peoples Sheps and Dr. Nabih Asal as I completed my concurrent MPH program. I thank the UF Graduate S chool for financial support, through an alumni fellowship as well as supplemental retention awards, in order to pur sue my graduate studies. In addition, I was grateful to receive a two yea r fellowship training grant from Mor ris Animal Foundation as well as additional salary support from the American Kenne l Club Canine Health Foundation I thank Dr. Charles encouragement and assistance durin g my graduate program and for much appreciat ed supplemental funding. I am also grateful to Dr. John Dame for providing excellent lab space in order to complete my research I thank Dr. James Coleman, Dr. Corinne Bromfield, Ms. Sara Salomon, Mr. Peter Nade au, Mr. Michael Crandall, Ms. Sylvia Tucker, Ms. Patty Dingman, and Mr. Joe Pohedra for all of their invaluable assistance throughout these projects. Individuals at several other institutions have contributed greatly to my PhD program. At the Centers for Disease Control and Prevention I thank Dr. Ruben Donis, Dr. Jackie Katz, Dr. Xiu Hua Lu, Ms. Justine Pompey, and Dr. Vi c Veguilla for their training in influenza laboratory techniques, assistance with diagnostic test evaluation, and for providing several H 3N8 CIV isolates. At Cornell University I thank Dr. Ed Dubovi a nd Ms. Nancy Cavender Zylich for their training in influenza laboratory techniques, assistance with data for epidemiological analyses, and for providing several H3N8 CIV and EIV isolates. At Co lorado State Un iversity I thank Dr. Gabrie le Landolt and Ms. Susi

PAGE 5

5 Bennett for training in influenza molecular techniques and for providing several H3N8 CIV isolates. At the University of Kentuc k y I thank Dr. Tom Chambers for providing several H3N8 EIV isol ates. At the Istituto Zooprofilattico Sperimentale de lle Venezie I thank Dr. Ilaria Capua for the opportunity to work in her laboratory on a research project and to serve as rapporteur at an avian influenza conference Both were excellent experiences. I a lso thank Dr. Paola De Benedictis Dr. William Dundon, and the rest of the lab for their collaboration on the project and their hospitality. Many individuals have inspired me throughout my education, and I likely would not be where I am today without thei r influence and encouragement. I thank Ms. Henry, Mr. from East Lake High School for their outstanding preparation and guidance. I thank Dr. Katie Sieving from the UF Department of Wildlife Ecology and Conserv ation for introducing m e to the challenges of research. I thank Dr. Don Forrester for his mentorship during undergrad and vet school and for being an excellent role model, both personally and professionally. His wildlife diseases course introduced me to tely set me on a fascinating career path. Also, I thank Mr. Garry Foster for his friendship and assistance in the lab during my first research projects. I cannot express enough gratitude for my family and friends who have given me so much love and suppor t over the years. My parents have encouraged all of my pursuits and I am grateful for the sacrifices they have made to help me reach my potential. I look forward to giving back now that this milestone has been achieved. And finally I thank my dear husband Jeremy, who has shared this journey with me for almost 20 years. His unconditional love is my greatest blessing.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 TA BLE OF CONTENTS ................................ ................................ ................................ .. 6 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 12 LIST OF A BBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 Research O bjectives ................................ ................................ ............................... 16 Diagnostic Test Development ................................ ................................ ........... 17 Specific aim 1 (Chapter 3): To assess the diagnostic performance of the H3N8 CIV HI assay ................................ ................................ ................. 17 Specific aim 2 (Chapter 4): To develop a DIVA test strategy for H3N8 CIV ................................ ................................ ................................ .......... 18 Epidemiological Investigations ................................ ................................ ......... 18 Specific aim 1 (Chapter 5): To investigate potential H3N8 canine influenza like virus circulation in racing greyhounds and shelter dogs prior to 2004 ................................ ................................ ............................ 18 Specific aim 2 (Chapter 6): To determine the prevalence of and risk factors for H3N8 CIV seropositivity in a population of U.S. dogs with influenza like illness ................................ ................................ ................ 18 Specific aim 3 (Chapter 7): To explore viral and/or population parameters that promote H3N8 CIV transmission in shelter facilities ..... 19 2 REVIEW OF THE LITERATURE ................................ ................................ ............ 20 Influenza Viruses ................................ ................................ ................................ .... 20 Virology ................................ ................................ ................................ ............ 20 Epidemiology ................................ ................................ ................................ .... 21 Determinants of Interspecies Transmission ................................ ...................... 24 H3N8 Canine Influenza Virus in the United States ................................ ................. 26 Molecular and Antigeni c Characterization ................................ ........................ 27 Epidemiology ................................ ................................ ................................ .... 28 Clinical Signs and Pathogenesis ................................ ................................ ...... 31 Diagnosis ................................ ................................ ................................ .......... 32 Prevention and Control ................................ ................................ ..................... 34

PAGE 7

7 Influenza A Viruses in Dogs around the World ................................ ....................... 35 Summary ................................ ................................ ................................ ................ 38 3 DIAGNOSTIC PERFORMANCE OF THE H3N8 CANINE INFLUENZA VIRUS HEMAGGLUTINATION INHIBITION ASSAY ................................ .......................... 40 Materials and Methods ................................ ................................ ............................ 42 Assessing the Diagnostic Performance of the Standard H3N8 CIV HI Assay .. 42 Evaluating the Ef fect of Test Parameter Variations on Diagnostic Performance of the H3N8 CIV HI Assay ................................ ....................... 44 Results ................................ ................................ ................................ .................... 47 Assessing the Diagnostic Perf ormance of the Standard H3N8 CIV HI Assay .. 47 Evaluating the Effect of Test Parameter Variations on Diagnostic Performance of the H3N8 CIV HI Assay ................................ ....................... 48 Discussion ................................ ................................ ................................ .............. 50 4 DEVELOPMENT OF A DIVA TEST STRATEGY FOR H3N8 CANINE INFLUENZA VIRUS ................................ ................................ ................................ 66 Materials and Metho ds ................................ ................................ ............................ 67 Virus Preparation ................................ ................................ .............................. 67 Amplification and Cloning of the NS1 Gene ................................ ..................... 68 Expression and Purification of the NS1 Protein ................................ ................ 70 SDS PAGE and Western Blot Assays for NS1 Protein Analysis ...................... 72 Synthetic NS1 Peptide ................................ ................................ ..................... 73 Serum Samples ................................ ................................ ................................ 74 ELISA for Detection of NS1 Antibodies in Canine Sera ................................ .... 74 Western Blot Assays for Detection of NS1 Antibodies in Canine Sera ............. 76 Results ................................ ................................ ................................ .................... 77 NS1 Protein Expression and Purification ................................ .......................... 77 ELISA to Detect NS1 Antibodies in Canine Sera ................................ .............. 78 Western Blot Assays to Detect NS1 Antibodies in Canine Ser a ....................... 79 Discussion ................................ ................................ ................................ .............. 80 5 SEROLOGICAL EVIDENCE OF H3N8 CANINE INFLUENZA LIKE VIRUS CIRCULATION IN U.S. DOGS PRIOR TO 2004 ................................ .................... 93 Materials and Methods ................................ ................................ ............................ 94 Study Design ................................ ................................ ................................ .... 94 Serum Samples ................................ ................................ ................................ 94 Viruses ................................ ................................ ................................ ............. 95 HI Assays ................................ ................................ ................................ ......... 95 Data Analysis ................................ ................................ ................................ ... 97 Results ................................ ................................ ................................ .................... 98 Discussion ................................ ................................ ................................ .............. 99

PAGE 8

8 6 PREVALENCE OF AND RISK FACTORS FOR H3N8 CANINE INFLUENZA VIRUS SEROPOSITIV ITY IN A POPULATION OF U.S. DOGS WITH INFLUENZA LIKE ILLNESS ................................ ................................ ................. 105 Materials and Methods ................................ ................................ .......................... 106 Animals ................................ ................................ ................................ ........... 106 Study Design ................................ ................................ ................................ .. 107 Laboratory Procedures ................................ ................................ ................... 107 Data Collection ................................ ................................ ............................... 107 Statistical Analyses ................................ ................................ ........................ 109 Results ................................ ................................ ................................ .................. 109 H3N8 CIV Seroprevalence in the Study Population ................................ ....... 109 Risk Factors Associated with H3N8 CIV Seropositivity ................................ .. 110 Discussion ................................ ................................ ................................ ............ 11 0 7 SIMULATION STUDIES OF H3N8 CANINE INFLUENZA VIRUS TRANSMISSION IN SHELTER FACILITIES ................................ ........................ 117 Materials and Methods ................................ ................................ .......................... 119 Mode l Structure ................................ ................................ .............................. 119 Model classes ................................ ................................ .......................... 120 Model parameters ................................ ................................ .................... 121 Model equ ations ................................ ................................ ....................... 125 Parameter Estimation ................................ ................................ ..................... 127 Sensitivity Analysis ................................ ................................ ......................... 129 Dat a Analysis ................................ ................................ ................................ 130 Results ................................ ................................ ................................ .................. 133 Discussion ................................ ................................ ................................ ............ 135 8 SUMMARY, RESEARCH CO NCLUSIONS, AND ISSUES FOR FUTURE INVESTIGATORS ................................ ................................ ................................ 150 Summary ................................ ................................ ................................ .............. 150 Research Conclusions ................................ ................................ .......................... 151 Diagnostic Test Development ................................ ................................ ......... 151 Specific aim 1 (Chapter 3): Diagnostic performance of the H3N8 CIV HI assay ................................ ................................ ................................ .... 151 Specific aim 2 (Chapter 4): Development of a DIVA test strategy for H3N8 CIV ................................ ................................ .............................. 151 Epidemiological Investigations ................................ ................................ ....... 152 Spec ific aim 1 (Chapter 5): Serological evidence of H3N8 canine influenza like virus circulation in dogs prior to 2004 .............................. 152 Specific aim 2 (Chapter 6): Prevalence of and risk factors for H3N8 CIV s eropositivity in a population of U.S. dogs with ILI ................................ 152 Specific aim 3 (Chapter 7): Simulation studies of H3N8 CIV transmission in shelter facilities ................................ ............................ 153 Issues for Future Investigators ................................ ................................ .............. 153

PAGE 9

9 Diagnostic Test Development ................................ ................................ ......... 153 Epidemiological Investigations ................................ ................................ ....... 154 LIST OF REFERENCES ................................ ................................ ............................. 155 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 166

PAGE 10

10 LIST OF TABLES Table page 3 1 ROC analysis of antibody titers determined for 614 convalescent sera in the standard H3N8 CIV HI assay using the MN antibody titers as the gold standard. ................................ ................................ ................................ ............ 57 3 2 H3N8 CIV agglutination of avian and mammalian red blood cells under different incubation conditions. ................................ ................................ ........... 58 3 3 HI antibody titers for sera from 11 dogs infected with H3N 8 CIV tested with 0.5% chicken red blood cells (chRBC), 0.5% turkey red blood cells (tRBC), or 1% canine red blood cells (cnRBC). ................................ ................................ ... 59 3 4 HI antibody titers for untreated or RDE treated sera fr om 11 dogs infected with H3N8 CIV and 15 uninfected dogs. ................................ ............................. 60 3 5 HI antibody titers for sera from 14 dogs infected with H3N8 CIV tested with live or inactivated virus. ................................ ................................ ...................... 61 3 6 HI antibody titers for sera from 46 dogs infected with H3N8 CIV and tested with different H3N8 CIV isolates in the standard HI assay. ................................ 62 3 7 HI a ntibody titers for sera from 46 dogs infected with H3N8 CIV and tested with different H3N8 EIV strains in the standard HI assay. ................................ .. 64 4 1 Primers for PCR amplification and cloning of the H3N8 CI V NS1 gene. ............ 84 4 2 Checkerboard titration results for the H3N8 NS1 antibody ELISA. Serial two fold dilutions of recombinant NS1 protein were incubated with goat anti influenza NS1 antibody ................................ ................................ ....................... 89 4 3 Checkerboard titration results for the H3N8 NS1 antibody ELISA. Serial two fold dilutions of uninfected and H3N8 CIV infe cted dog sera were incubated with varying NS1 protein concentrations ................................ ........................... 90 4 4 The H3N8 NS1 antibody ELI SA was performed with varying concentrations of the synthetic NS1 34 49 peptide incubated with uninfected and H3N8 CIV infected dog sera ................................ ................................ ............................... 91 5 1 Annual prevalence of H3 antibodies among 520 racing greyhounds sampled between 1999 to 2004. ................................ ................................ ..................... 102 5 2 Univariable conditional logistic regression analyses of potential risk factors for H3 seropositivity among 481 racing greyhounds sampled from 1999 to 2004. ................................ ................................ ................................ ................ 103

PAGE 11

11 5 3 Comparison of prevalence estimates of H3 seropositive and seronegative greyhounds ba sed on years raced. ................................ ................................ .. 104 6 1 Univariable conditional logistic regression analyses of potential risk factors for H3N8 CIV seropositivity in a population of 1,268 pet and shelter dogs with ILI. ................................ ................................ ................................ .................... 115 6 2 Multivariable conditional logistic regression analyses of potential risk factors for H3N8 CIV seropositivity in a population of 1,268 pet and shelter dogs with ILI. ................................ ................................ ................................ .................... 116 7 1 Parameters, ranges, and distributions used for simulations of H3N8 CIV introduction into shelter facilities. ................................ ................................ ...... 141 7 2 Descriptive statistics for specified outcomes of 498 H3N8 CIV introduction simulations classified as epidemic. ................................ ................................ ... 144 7 3 Results of stepwise regression on epidemic duration and time of epidemic peak. ................................ ................................ ................................ ................. 145 7 4 Results of stepwise regression on total number of infected dogs at the epidemic peak, as well as accumulated infectious clinical (I C ) and subclinical dogs (I SC ) at the steady state (SS). ................................ ................................ ... 146

PAGE 12

12 LIST OF FIGURES Figure page 4 1 ) plasmid for cloning of the H3N8 CIV NS1 gene. 84 4 2 SDS PAGE electrophoresis of plasmid preparations (1, 2, 3, 18, 19, 20) digested with NcoI and XhoI restriction enzymes to release the NS1 gene ................................ ...................... 85 4 3 SDS PAGE electrophoresis of total protein extracted from recombinant H3N8 CIV NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) and stained with Coomassie blue. ................................ ................................ ......................... 85 4 4 SDS PAGE electrophoresis of soluble and insoluble fractions of a lysed recombinant NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) cell pellet and stained with Coomassie blue. ................................ ............................ 86 4 5 We stern blot of soluble and insoluble fractions of a lysed recombinant NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) cell pellet and stained with an affinity purified antibody to influenza NS1 protein. ................................ 87 4 6 Western blot of purified H3N8 CIV NS1 protein detected using goat anti influenza NS1 IgG antibody and a HRP conjugated mouse anti goat IgG antibody. ................................ ................................ ................................ ............. 88 4 7 H3N8 CIV NS1 western blot with sera from uninfected SPF dogs and dogs experimentally infected with H3N8 CIV. ................................ ............................. 92 4 8 H3N8 CIV NS1 western blot with sera from uninfected, infected, and vaccinated dogs. ................................ ................................ ................................ 92 7 1 Model schematic for simulations of H3N8 CIV introduction into shelter facilities. ................................ ................................ ................................ ............ 140 7 2 Results of typical endemic settling time checks, used to standardize the criteria for determining the steady state point. ................................ .................. 143 7 3 Example of a simulation dominated by direct virus transmission at the peak of the epidemic. ................................ ................................ ................................ 147 7 4 Example of a simulation dominated by indirect virus transmission at the peak of the epidemic.. ................................ ................................ ............................... 148 7 5 Example of a simulation equally due to direct and indirect virus transmission at the peak of the epidemic.. ................................ ................................ ............ 149

PAGE 13

13 LIST OF ABBREVIATION S CIV Canine influenza virus DIVA Differentiate infected from vaccinated animals EIV Equine influenza virus ELIS A Enzyme linked immunosorbent assay H3N8 Hemagglutinin subtype 3 Neuraminidase subtype 8 HA Hemagglutination HI H emagglutination inhibition ILI Influenza like illness MN Microneutralization NS1 Nonstructural protein SEIR Susceptible, exposed, infectio us, r ecovered model

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DIAGNOSIS AND EPIDEMIOLOGICAL INVESTIGATION O F H3N8 CANINE INFLUENZA VIRUS By Tara Creel Anderson May 2011 Chair: E.P.J. Gibbs Cochair: P.C. Crawford Major: Veterinary Medical Sciences Canine influenza A virus su btype H3N8 (H3N8 CIV) was recognized in 2004 as a respira tory pathogen for dogs The v irus has since affected thousands of dogs in 34 states and has established endemicity in several m etropo litan areas of the U S. This PhD research encompassed two main objectives. The fir st objective was to develop tool s for serological diagnosis of H3N8 CIV. The diagnostic performance of the H3N8 CIV hemagglutination inhibition (H I) assay was assessed and test parameter variati ons were evaluated. Study results promote standardization of the H3N8 CIV HI assay to ensure accurate diagnosis of infec ted dogs. Introduction of a H3N8 CIV vaccine confounds future diagnosis however, because antibodies generated in response to infection or vaccination cannot be distinguished in the HI assay Development of a test strategy that differentiates infected from vaccinate d animals (DIVA) was therefore pursued Uninfected a nd vaccinated dogs could not be distinguished fro m infected dogs in NS1 tests however, supporting the conclusion that a NS1 immunoassay is not a viable DIVA strategy for H3N8 CIV. The second overall obje ctive of this research was to conduct epidemiological investigations of canine influenza to better inform prevention and control strategies. To

PAGE 15

15 determine if H3N8 CIV was infe cting dogs prior to 2004, a survey of archived canine sera was conducted Antibodi es to canine and equine H3 proteins were identified in dogs as early as 1999 Next, a national serosurvey for H3N8 CIV infection in dogs wi th influenza like illness was conducted to identify risk factors for infection. A nalyses reveal ed that dogs 1 year o f age, dogs housed in boarding kennels or animal shelters and dogs in the northeastern and western regions of the country w ere at highest risk for seropositivity Based on identification of communal facilities as ri sk factors for infection mathematical m odeling was conducted to explore the impact of several parameters on H3N8 CIV t ransmission in shelter s. I ndirect virus transmission par ameters and shelter capacity had the greatest influence on the tim e course of epidemics and the number of dogs affected These factors can be targeted in future modeling studies to determine what interventio n strategies would avert or mitigate epidemic s

PAGE 16

16 CHAPTER 1 INTRODUCTION Canine influenza A virus subtype H3N8 (H3N8 CIV) is a recently emerged highly contagious respirato ry pathogen for dogs that originated from interspecies transmission of H3N8 equine influenza virus to dogs prior to 2004, followed by virus adaptation to facilitate dog to dog transmission. 20 Since its em ergence, t his novel virus has in fected thousands of dogs in 34 states. 8,9,20,28,29,42,44,98,103,110,111 Dogs housed in communal settings such as racing greyhound kennels, boarding kennels, and animal she lters have been most affected. Due to efficient horizontal transmission and continued adaptation to the dog, H3N8 CIV is now considered endemic in several metropolitan areas of the country. 28,29 Risk factors for infection and the epide miology of H3N8 CIV transmission in different dog populations, however, are not fully understood. 28,29 Past and present influenza epidemics and pandemics have demonstrated that establishment of an influenza A virus in a new host species is not only important to the health and welfare of the new host, but is also a public health concern. It is prudent to study the epidemiology of influenza A viruses in all established hosts in order to more completely understand the exposure and health risks of these pathogens, and the potential for virus transmission to new hosts. Research Objectives The overall objectives of this PhD research were: 1) to develop diagnostic tools for serological diagnosis of dogs infected with H3N8 C IV, and 2) to conduct epidemiological investigations of H3N8 CIV in order to better inform prevention and control strategies.

PAGE 17

17 Diagnostic Test Development To adequately study the epidemiology of canine influenza, key diagnostic tests should be developed an d assessed. Although a variety of tests are available for diagnosis of H3N8 CIV infections, serological assays are often necessary to confirm diagnosis. 28,29 The hemagglutination inhibition (HI) assay is the curren t test of choice for diagnosis of influenza infections in animals. 134,135 HI assays are performed using techniques modified for the particular host species. There is no accepted standard HI assay for detection of CIV H3 antib odies, and variations in test parameters used by different laboratories may lead to inaccurate diagnosis. Further, t he availability of an inactivated whole virus vaccine for H3N8 CIV has confounded the diagnosis of infected dogs because the HI assay cannot distinguish antibodies induced by infection vs. vaccination. Development of a test strategy for differentiation of infected from vaccinated animals (DIVA) is vital for correct diagnosis of infection, evaluation of vaccine efficacy, and for future studies on the epidemiology of H3N8 CIV. The research presented in Chapters 3 and 4 address objective 1. The specific aims and hypotheses for each chapter are as follows: Specific aim 1 (Chapter 3): To assess the diagnostic performance of the H3N8 CIV HI assay It was hypoth esized that a standard H3N8 CIV HI assay would be a highly sensitive and specific test once a cutoff antibody titer for seropositivity was established. Further, it was hypothesized that high sensitivity and specificity would require use of red b Gal receptors, inactivation of nonspecific inhibitors of hemagglutination in canine serum, and use of live H3N8 CIV isolates in the HI assay.

PAGE 18

18 Specific aim 2 (Chapter 4): To develop a DIVA test strategy for H3N8 CIV Ba sed on studies in other species, it was hypothesized that immunoassays that detect antibodies to the H3N8 CIV NS1 protein would satisfy the DIVA principle for correct identification of infected versus vaccinated dogs. Epidemiological Investigations It is unknown if H3N8 CIV was infecting dogs prior to 2004, causing undiagnosed or misdiagnosed respiratory disease. Risk factors for H3N8 CIV infection have also not been defined, either for individual dogs or populations of dogs. Three epidemiological studies were performed to better understand when H3N8 CIV first emerged in dogs, what dog related (intrinsic) and exposure factors increase risk for infection, and what facility and population factors contribute to influenza epidemics in communal settings such as animal shelters I dentification of intrinsic and exposure risk factors for infection, and facility or population factors that promo te virus transmission, would greatly contribute to formulation of prevention and control strategies to protect t he health and welfare of dogs. The research presented in Chapters 5, 6, and 7 address objective 2. The specific aims and hypotheses for each chapter are as follows: Specific aim 1 (Chapter 5): To investigate potential H3N8 canine influenza like virus circulation in rac ing greyhounds and shelter dogs prior to 2004 It was hypothesized that a H3N8 canine influenza like virus was circulating in racing greyhounds prior to 2004, and that this virus was associated with respiratory disease outbreaks of unknown etiology in U.S. greyhounds in the 1990s and 2003. Specific aim 2 (Chapter 6): To determine the prevalence of and risk factors for H3N8 CIV seropositivity in a population of U.S. dogs with influenza like illness It was hypothesized that H3N8 CIV would be a common diagnosi s in a study population of U.S. dogs with influenza like illness ( ILI ) and that geographic location and

PAGE 19

19 communal housing settings would be significantly associated with H3N8 CIV seropositivity Specific aim 3 (Chapter 7): To explore viral and/or populati on parameters that promote H3N8 CIV transmission in shelter facilities It was hypothesized that H3N8 CIV transmission in shelter facilities would be driven by the high frequency of population turnover, and that indirect (fomite) modes of transmission betwe en infected and susceptible dogs would facilitate greater viral transmission than direct modes of transmission This was an exploratory, theoretical study using mathematical modeling.

PAGE 20

20 CHAPTER 2 REVIEW OF THE LITERATURE Influenza Viruses Virology Inf lu enza viruses are enveloped viruses composed of segmented, negative sense, single stranded RNA. 95 Three Influenza virus genera are classified in the Family Orthyomyxoviridae. Influenza B and Influenza C viruses primarily affect humans, although reports of infections in other species have been documented in the literature. 38,73 ,74,93 Influenza A viruses cause epidemic and pandemic disease in a variety of avian and mammalian species. These viruses are a significant cause of morbidity and mortality in humans, domestic poultry, pigs, horses, and now dogs. 20,28,29,98,136 Influenza A viruses are comprised of a host derived lipid envelope and an eight segment g enome, which encodes up to 1 2 known proteins. 19,88,95,133 The two surface proteins, the hema gglutinin (HA) and neuraminidase (NA), are used for virus subtyping. There are currently 16 HA subtypes (H1 H16) and 9 NA subtypes (N1 N9). 95 The HA protein binds sialic acid (SA) receptors on host epithelial cell surfaces, requires cleavage by host proteases to allow fusion activity, and is the major viral antigen. Through its sialidase activity the NA protein free s new virions budding from the host cell surface, which promotes virus spread. The remaining viral proteins include M2, M1, NP, PA, PB1, PB2, PB1 F2, N40, NEP (NS2) and NS1 19,95,133 M2, a membrane protein, is res ponsible for ion channel activity and assembly. The M1 or matrix protein interacts with viral ribonucleoproteins and surface proteins, and is involved in nuclear export and budding. The NP or nucleoprotein is involved in RNA binding, synthesis, and nucle ar

PAGE 21

21 import. The PA, PB1, and PB2 proteins comprise the viral polymerase complex PB1 F2 is a proapoptotic protein that is not expressed by all influenza viruses, and the function of the recently described N40 protein is not fully understood 19,133,142 The NEP or nuclear export protein functions in the export of viral ribonucleoprotein s. The NS1 or nonstructural protein is a multifunctional protein that inhibits the host immune response and is involved in several aspect s of viral replication. 39,95 Epidemiology Aquatic birds are the global reservoirs of influenza A viruses, and serve as the source of novel virus subtypes for all susceptible species. 132 Previous interspecies transmissions and adaptations to the new hosts have resulted in endemic influenza virus lineages in domestic poultry, pigs, horses, and humans. 136 Although influenza A virus evolution is complex, two major mechanisms of virus evolution are typically associated with generation of epidemic and p an demic viruses Antigenic drift occurs when progeny viruses acquire point muta tions in the HA or NA proteins. In these situations unaltered epitopes can still be recognized b y the immune system of previously infected individuals however, therefore only mil d disease will occur These viruses result in influenza epidemics. 136 Antigenic shift occurs either when two infl uenza viruses infect the same host cell and reassort their segmented genomes, or when a virus is directly transmitted from one species to another. Antigenic shift results in major antigenic changes, such that immunity from previous infections is not protec tive. These viruses result in pandemics. 136 Transmission of influenza A viruses from the avian reservoir to other birds or mammals is usua lly the result of reassortment events or the direct transmission of an entire influenza A virus. 67,132 It has been noted that influenza A viruses transmit

PAGE 22

22 relatively frequently from the avi an reservoir to other species yet do not typically establish permanent lineages in these new hosts. 131 Other endemic influenza A virus hosts, inculding domestic poultry, pigs, horses, and humans may also serve as a source of novel virus subtypes for other birds or mammals. 136 Sporadic influenza A virus infections (of avian and mammalian origin) have been reported in recent years in several mammals, such as stone martens, felids, and dogs; 1,56,114,115 however to date only the recently emerging H3N8 and H3N2 canine influenza viruses have established stable lineages capable of horizontal transmis sion in this new endemic mammalian host. 20,66,98,103,112 The complex epidemiology of influenza A viruses highlight s the potential impact of these viruses on both animal and public health. In avian species, influenza A viruses replicate in the respiratory and intestinal tracts, wh ich results in heavy virus shedding in the saliva, nasal secretions, and feces. Although infections in wild birds are typically subclinical, fecal contamination of water allows transmission to other species primarily through the fecal oral route. 130,132 Influenza A virus infections in domestic poultry can be mild and localized or systemic, and are also typically transmitted to other species by the fecal oral route. Influenza infections in mammals are primarily trans mitted through respiratory secretions, and result in localized or systemic disease. 136 Endemic H1N1, H3N2, and H1 N2 influenza A virus subtypes circulate seasonally in humans worldwide, and antigenic drift results in yearly influenza epidemics. Avian and swine influenza viruses are currently considered the greatest threat for generation of pandemic viruses in humans. 136 In domestic poultry the H5 and H7 influenza A virus subtypes are a significant animal health concern, since th ey can cause highly

PAGE 23

23 pathogenic disease with mortality rates of 90 100%. Outbreaks of multiple low pathogenic subtypes, including low pathogenic H5 and H7 viruses, however also occur. 136 Whether low or high ly pathogenic, avian influenza outbreaks among domestic poultry are considered a public health threat. Reassortment events between avian and human influenza A virus es resulted in the H2N2 and H3N2 pandemics of 1957 and 1968, respectively. 67,132 And direct interspecies transmissions of H5N1 avian influenza viruses to humans, first noted in 1997, have been a source of great pan demic concern and planning. 37,100,118 The role of swine in influenza A virus ecology is also complex. Multiple influenza A viruses, including H1N1, H3N2, and H1N2 subtypes currently circulate in pigs. Genetic analy ses of these viruses reveal there are North American and Eurasian swine virus lineages, and that double and triple reassortants of avian, swine, and human viruses are in circulation. 136 There have been reports of sporadic swine influenza virus infections in people, however the true extent of influenza virus transmission between pigs and people is poorly understood. 81 As a timely example, the origin of the 2009 novel H1N1 pandemic virus is unknown. The vi rus contains swine, avian, and human genes, and therefore reasserts the hypothesized role of swine as a mixing vessel of avian and human influenza A viruses with resultant pandemic potential. 99,136 Although there are no documented interspecies transmissions of equine influenza viruses (EIV) to people, influenza A viruses are significant respiratory pathogens of horses. Influenza like illness (ILI) had been described in horses for centuries before the discovery of t he H7N7 and H3N8 EIV in 1956 and 1963, respectively. 136 Interestingly, the H7N7 viruses have not been detected in horses since 1979, and are believed to be

PAGE 24

24 extinct. 132 H3N8 EIV was first detected in horses in Miami, Florida. Over time H3N8 EIV isolates evolved into genetically and antigenica lly distinct, co circulating American and Eurasian lineages; and the American lineage further evolved into the Argentina, Florida, and Kentucky sublineages. 22,61 The Florida and Kentucky sublineages were reportedly evolving in parallel in the U.S., such that these viruses alternately circulated in the equine population. 62 Currently, Florida sublineage clade 1 viruses predominate in North America and clade 2 viruses predomina te in Europe. 15 Due to frequent movement of horses fo r competition and bre eding, these viruses have resulted in sporadic EIV outbreaks worldwide. 23 Until recently there had been no documented eq uine influenza virus transmissions to other species, and it was suggested that the horse was an 132 However, the d iscovery of the entirely equine origin H3N8 CIV in the United States in 2004, 20 limited H3N8 EIV transmissions to dogs in the U.K. and Australia, 21,55,91 and the 2009 report of a s poradic H3N8 EIV transmission to pigs in China, 123 once again highlight the complex ecology of influenza A viruses. Determinants of Interspecies Transmission The above examples emphasize the importance of influenz a A viruses to both animal and public health, and reinforce the need for greater understanding of their classical and molecular epidemiology in all affected species. Significant research efforts have focused on the molecular determinants of influenza A vir us interspecies transmission, with particular effort placed on determinants of host range restriction and viral pathogenicity. 7,60,63,89,128 The requirements for interspecies transmission have been detailed, focusi ng on the interactions between the original and the new host, interactions between the new host and the novel virus, and host host interactions within

PAGE 25

25 the new recipient species. 60 F or a successful interspecies tran smission to occur, there must be contact between the two host species, the virus must be able to replicate in the new host, and the virus must be transmissible between individuals in the new host population. Methods of virus entry into the host cell, viral replication within the new host, viral evasion of the host immune response, transmission of virus to another individual within the new host population, and subsequent sustained transmission of the virus between in dividuals in the new host population are all important events for interspecies transmission to occur. 60,128 The host cell receptor specificity of the influenza A virus HA is considered a critical determinant of host range restriction. 48,60,89,119,120,128 The HA protein binds SA receptors on host epithelial cell surfaces. The SA species present, its linkage to galactose, and its distribution in the tissues of the host species are all important factors. The primary S A species are N acetyl neuraminic acid (NeuAc) and N glycolyl neuraminic acid (NeuGc), while the primary linkages of the SA species to galactose are SA Gal. 120 A vian and equine influenza viruses bind SA Gal receptors on the epithelial cells of the intestinal (avian) and respiratory (avian and equine) tracts; w hile human influenza viruses bind SA Gal receptors on the tracheobronchial epithelium 120 As an example of host range restriction, lectin binding assays have shown that NeuGc Gal receptors are d ominant in the equ ine trachea and e xperimental infections in horses have further shown these receptors are essential for viral replication. This may explain why equine viruses have reportedly not been transmitted to humans, since humans do not have NeuGc Gal receptors i n the respiratory tract. 120 In contrast, it is believed that successful direc t transmission of H5N1 avian influenza viruses to humans is

PAGE 26

26 associated with the presence of NeuAc Gal receptors o n particular target cells of the human re spiratory tract Experiments have shown that human influenza viruses bind nonciliated tracheal epithelial cells with SA Gal receptors, while avian viruses bind ciliated human tracheal epithel ial cells with SA Gal receptors 77 Regardless of these findings, it is important to note that single amino acid substitutions have been shown to change viral receptor specificity from SA Gal to SA Gal i n several HA protein subtypes, which increases potential t ransmission to and between species 119 Several other molecular determinants have been reviewed that influence viral host range restriction and/or pathogenicity. 7,60,63,89,128 It has be en noted that the NA also has SA Gal or SA Gal receptor specificity. Since NA releases virus from the infected cell, the combination of HA and NA are therefore needed for efficient replication. 89 Regarding a specific mutation affect ing pathogenicity, the glutamic acid to lysine mutation at position 627 (E627K) of the H5N1 avian influenza virus PB2 protein has been associated with increased replication and virulence in mammals. 60,89 The influe nza virus NS1 protein inhibits host immunity by counteracting the innate interferon antiviral response. Another particular concern with H5N1 avian influenza viruses is that the NS1 protein also increases pathogenicity by inducing proinflammatory cytokine g enes, resulting in cytokine dysregulation. 37,128 Ultimately, multiple viral and host factors are likely involved in interspecies transmission, and the specific molecular determinants required to overcome bar riers t o transmission are not well understood. 128 H3N8 Canine Influenza Virus in the United States Prior to 2004 the dog was not considered an endemic host of influenza A viruses. Previous studies demonstrated the dog was susceptible to influenza A virus infection (either naturally or experimentally), or reported serological evidence of transient natural

PAGE 27

27 infection of dogs with human influenza viruses. 18,46,51,53,92,106,121 However, there was no evidence of sustained horizontal influenza A virus transmission in the canine study populations. The direct transmission of an entire H3N8 EIV from horses to dogs, and its subsequent establishment of t he H3N8 CIV lineage in dogs in the U.S., was an unusual event. 20,98,103 Molecular and Antigenic Characterization Virologic, pathologic, and serologic evidence have demonstrated that H3N8 CIV is an emerging respirat ory pathogen in dogs. 20,28,29,42,98,103,141 Molecular analyses of virus isolates indicate the interspecies transmission of an entire H3N8 EIV to the dog. 20,42,98,103 Althou homology with 2002 and 2003 EIV isolates of the Florida sublineage clade 1, canine influenza viruses have formed a distinct monophyletic group and H3N8 CIV isolates are diverging from their equine ancestors. 20,98,103 Five signature amino acid substitutions in the CIV H3 protein distinguish all sequenced canine from equine viruses (N54K, N83S, W222L, I328T, N483T). 20 Analyses of H3N8 CIV isolates collected from 2005 through 2008 show the viruses continue to evolve, such that the or iginal 2003 and 2004 viruses have been Nine additional amino acid substitutions have been identified in all of the enzootic isolates including three in HA (L118V, K261N, G479E), three in PB2 ( S107N, A221V, I292T), two in NA (I62L, V147I) and one in PB1 (V200I). 98,103 The potential significance of these amino acid substitutions has been described. 20,98,103 Briefly, of the CIV H3 mutations, the change at position 54 is of unknown functional significance. The substi tution of serine for asparagine at position 83 has been previously associated with antigenic drift. Position 222 is adjacent to the

PAGE 28

28 receptor binding pocket of the HA protein, therefore the substitution of leucine for tryptophan at this location could be a significant modification required for adaptation to canine SA receptors. This is further supporte d by a recent H3N8 CIV receptor binding study indicating reduced binding of the virus to NeuGc Gal receptors. 139 Position 328 is near the HA cleavage site, therefore the substitution of threonine for isoleucine could be involved in its recogniti on by canine proteases. The substitution of threonine for asparagine at position 483 results in the loss of a glycosylation site in the HA2 subunit that is preserved in all other HA subtypes. 20,98 Although signifi cant antigenic drift has not yet occurred, 8,103 continued surveillance of circulating H3N8 CIV is necessary in order to evaluate any additional adaptations that could influence pathogenicity for the dog, transmissi on to other species, or the efficacy of diagnostic tests or vaccines. A recent study addressing the microevolution of 2008 2009 H3N8 CIV isolates in two New York City animal shelters has supported the original conclusion that the adaptation of the virus to dogs occurred early in the process of emergence or prior to transfer, since evidence of positive selection was not detected. 42 A study focused on H3N8 CIV intrahost evolutionary dynamics has demonstrated that potentially advantageous mutations are readily generated during single virus infections, however, which could result in antigenic escape or host range variants. 43 Epidemiology H3N8 CIV was first isolated in January 2004 from racing greyhounds affected with severe respiratory disease at a North Florida track. 20 The outbreak was characterized by an explosive onset of coughing and nasal discharge, which affected approximately 80% of the dogs. Six dogs also died of hemorrhagic pneumonia. Thousands of greyhound dogs, at tracks in multiple states, were subsequently affected during multiple

PAGE 29

29 respirat ory disease outbreaks from 2004 to 2006. Additional virus isolations, as well as substantial serological evidence, indicated that H3N8 CIV was also associated with these outbreaks. 20,141 H3N8 CIV was first detecte d serologically in non greyhound dogs in a Florida shelter in August 2004, a nd the first virus isolation occurred in 2005. 20,98 Since that time the University of Florida and Cornell University have been conducting surveillance of dogs with ILI to determine the location of H3N8 CIV infected dogs in the U.S. and to better define risk groups for infection. H3N8 CIV surveillance activities conducted by these institutions and others have revealed a widespread geographic distribution of infected pet and shelter dogs in 34 sta tes 8,9,20,28,29,42,44,98,103,110,111 The virus is sustained in the U.S. canine popul ation by horizontal transmission, which occurs through direct transmission (direct contact between a susceptible and infected dog, or large droplet transmission through coughing and sneezing between dogs in close proximity) and indirect transmission (aerosols generated by coughing and sneezing, and fomites). Dogs in shelter faci lities, boarding facilities, veterinary clinics, and other communal settings are considered at highest risk for exposure. Outbreaks have occurred in greyhound racetrack kennels, boarding kennels, shelter facilities, veterinary hospitals, and canine day car e centers across the nation. 9,20,28,29,42 44,98,141 H3N8 CIV outbreaks reach epidemic proportions in these communal facilities, and have resulted in depopulation in shelters and closure of boarding kennels and cani ne day care centers H3N8 CIV is considered endemic in metropolitan areas of Florida, New York, Colorado, and Pennsylvania, and infections have occurred year round. 28,29,103

PAGE 30

30 Knowledge of the prevalence of H3N8 CIV in the U.S. and risk factors associated with virus exposure in dogs is limited. In a study conducted in a university hospital in Iowa, 4 of 731 (0.5%) pet dogs with or without ILI were H3N8 CIV seropositive, indicating prior viral exposure. 110 In a study conducted in a university hospital in Colorado, 9 of 250 (3.6%) pet dogs with or without ILI were seropositive for H3N8 CIV. 8 In a study of pet dogs participating in a flyball tournament in Pennsylvania, 3 of 100 (3%) dogs without ILI were seropositive for H3N8 CIV. 111 Another study conduct ed in a Philadelphia shelter found that 31 of 74 (42%) tested dogs with or without ILI were H3N8 CIV seropositive. 44 In these studies, attending canine day care within the last six months pet dogs living with other dogs, and the number of days housed in a shelter facility were associated with H3N8 CIV seropositivity. 8,44,111 It is unknown how an H3N8 EIV was originally transmitted to dogs, although aerosol transmission or ingestion of infected horse tissues are proposed hypotheses. 20,21 Natural infection of dogs during the recent H3N8 EIV outbreak in Australia, 55 and an experimental study in Japan, 137 have demonstrated the feasibility of close contact transmission of EIV to the dog. Experimental infection s of hors es with H3N8 CIV indicate horses ar e susceptible to infec tion, developing both clinical signs of respiratory disease and pathological lesions ( Long MT, Gibbs EPF, Crawford PC, et al., Comparison of virus replication and clinical disease in horses inoculated with equine or canine influenza viruses. Immunobiology of Influenza Virus Infection: Approaches for an Emerging Zoonotic Disease Athens, GA, July 29 31, 2007 ) However, a recent study in Japan demonstrated that the virus does not replicate well in horses. 139 These discrepant results could be due to differences in study methodology. Experimental infections have

PAGE 31

31 demonstrated that BALB/c mice are susceptible to H3N8 CIV infection, and may serve as a useful model f or some pathogenesis studies. 17 A recent study in chickens, turkeys, and Pekin ducks revealed that these species are not susceptible to infection with H3N8 CIV 78 At this time, there is no evidence of natural H3N8 CIV transmission to any other species, including humans. Clinical Signs and Pathogenesis H3N8 CIV causes acute ILI in dogs. Although thousands of dogs have been infected with H3N8 CIV, the majority of the U.S. canine population has neither been exposed to nor vaccinated against the virus; therefore most dogs are susceptible to infection due to lack of immunity. 20,26,28,29,49 The h allmark of H3N8 CIV outbreaks is the explosive onset of ILI in most dogs in a facility within a short period of time. 20,28,29,98 Based on exper imental infections and outbreak investigations, nearly all nave dogs e xposed to the virus become infected. 20,26,28,29,49,98 About 80% of H3N8 CIV infected dogs exhibit some degree of clinical disease, while 20% are subclinically infected. The incubation period is 2 to 4 days, which i s typical for influenza vi rus infection in other species. 20,26,28,29,49 H3N8 CIV i nfected dogs shed virus for 7 to 10 days with peak shedding on days 2 to 4. 20,26,28,29,49 Since the incubation period overlaps with peak virus shedding, dogs can spread the virus before clinical signs appear. In addition, subclinical dogs also shed virus; therefore, all exposed dogs must be considered an infectious risk, whether or not they are showing clinical signs of respiratory infection. Once virus replication and shedding ceases, the dog is no longer contagious to other dogs. In one reported transmission study between H3N8 CIV infected and nave dogs, 75% of the contact dogs contracted the virus ; 49 however the transmissibility of H3N8 CIV has been questioned. 28,42

PAGE 32

32 Similar to influenza in horses and huma ns, H3N8 CIV is a disease with high morbidity and low mortality. Two clinical syndromes have been defined: 1) a milder form consisting of cough and nasal discharge wi th full recovery in 2 weeks, and 2) a more severe form with pneumonia, prolonged recovery, and increased risk for mortality. 17,20 Severe infections with hemorrhagic pneumonia wer e noted in early outbreaks, but this appear s to be unique to racing greyhounds. 20,29, 98,141 H3N8 CIV replicates in epithelial cells lining the airways of the respiratory tract, causing epithelial erosion and intense inflammation ( rhinitis, tracheitis, bronchitis, bronchiolitis) 17,20 The virus als o replicates in canine al veolar macrophages, inducing tumor necrosis factor alpha expression and cell death. 102 E pithelial cell necrosis and exposure of the basement membrane predisposes the dog to secondary bacterial infections. Many dogs develop purulent nasal discharge due to s econdary bacterial infections, and these pathogens also contribute to the persistent cough. 17,20 Although most dogs recover in 2 3 weeks, some progress to a secondary bacterial pneumonia that can be life threatenin g Diagnosis Virus isolation is the gold standard for diagnosis of H3N8 CIV infection, as with other species. 136 Nasal swabs collected at the onset of clinical illness can be cultured in embryonated eggs or MDCK cells to isolate virus. 20,28,29,98,103 Recovered virus is then available for molecular and antigenic characteriza tion. False negative results are a concern with virus isolation, however, since the clinical sample must be collected during the virus shedding period. In addition, the sample must be properly stored until testing to preserve a sufficient amount of live vi rus for inoculation of cultures. For these reasons virus isolation often yields false negative results. 28,29,103 A more rapid diagnosis of H3N8 CIV during the acute phase of infection can be obtained through detect ion of viral

PAGE 33

33 nucleic acids in nasal swabs using RT PCR (using primers for H3N8 CIV or contemporary EIV matrix and hemagglutinin genes). 20,28,29,71,98,103 RT PCR is the most sensitive antigen detection method for H3 N8 CIV, with 75% of RT PCR positive samples yielding a virus isolate in culture 28,29,103 Although RT PCR does not require the presence of live virus in the sample, however, it is still prone to false negative resu lts because sample collection must overlap with the virus shedding period. 28,29 Due to concerns with false negative results in other diagnostics, detection of virus specific antibodies in the serum is often necess ary for confirmation of H3N8 CIV infection. 28,29 Ideally, a blood sample is collected during the acute period of illness followed by another sample 2 to 4 weeks later. The paired serum samples are tested for antibo dies that neutralize virus infection of cells (microneutralization assay, or MN) or that inhibit hemagglutination by the virus (hemagglutination inhibition assay, or HI). Seroconversion, or a four fold rise in antibody titer, is diagnostic of recent active virus infection. Initial comparisons of MN and HI test results, based on testing of sera from SPF and H3N8 CIV experimentally infected dogs, indicated that these tests correlate well. 20 Currently the HI assay is c onsidered the test of choice for diagnosis of influenza A viruses in animals, 134,135 and is most commonly used in diagnostic laboratories. 28,29 This test detects antibodies to the HA protein. Specific attachment of antibodies to the viral HA protein interferes with virus binding to SA receptors on red blood cells (RBC), thereby inhibiting hemagglutination. The HI assay reliably detects antibodies to the HA protein as early as 7 days po st infection. 20,26,28,29 T herefore, t he ideal approach for diagnosis of H3N8 CIV infection in individual dogs is collection of nasal swabs during the acute phase of disease, along with paired

PAGE 34

34 acute and convalescen t serum samples. 20,28,29,98,103 Nasal swabs can be used for virus isolation and molecular antigen detection using RT PCR, while serology serves as an ideal confirmatory test. High throughput serological surveillanc e of dogs for exposure to influenza A viruses can be conducted with commercially available competitive ELISAs, which detect antibodies to the highly conserved influenza A nucleoprotein. 24,30 Positive ELISA results should then be confirmed in a second test such as the subtype specific HI assay 24,30 Prevention and Control Vaccination is an important prevention and control measure for influenza A virus infection in all specie s, 14 and vac cination of animal populations is considered an important tool for preventing novel influenza A virus transmission from animal reservoirs to humans. Whole virus and recombinant DNA vaccines interfere with virus infection by inducing formation of antibodies against the HA protein, which mediates viral attachment to SA receptors on host epithelial cells. In addition to vaccination, biosecurity and infection control measures are useful tools for managing influenza outbreaks. In the case of H3N8 CIV, biosecurit y practices (e.g., changing clothes and washing hands after handling affected dogs) in conjunction with isolation of sick and exposed dogs and effective disinfection should reduce virus transmission to other dogs. 9, 28,29 In June 2009 the USDA approved the first H3N8 CIV vaccine. 27 Similar to most influenza virus vaccines in use for other species, it contains inactivated whole virus. The H3N8 CIV vaccine is recommended for d ogs at risk of exposure (i.e., not as a core vaccine) and is given in two doses, two to four weeks apart. An annual booster is currently recommended. The vaccine is intended as an aid in the control of disease

PAGE 35

35 associated with H3N8 CIV and may not prevent i nfection; however, vaccination significantly reduces the severity and duration of clinical illness, the extent of pathology in the lungs, and the amount and duration of virus shedding. 27 Canarypox vectored and equi ne herpes virus vectored H3N8 CIV vaccines have also been described in the literature, 52,107 but have not yet received USDA approval. Two concerns for H3N8 CIV vaccination are highlighted by observations of EIV va ccination. Although horses are routinely vaccinated for H7N7 and H3N8 influenza subtypes, the complex evolution of H3N8 EIV has resulted in multiple EIV outbreaks worldwide due to vaccine failure (i.e., antigenic mismatch between vaccine strains and circul ating viruses) 75,90,138 Although significant antigenic drift has not yet been observed with H3N8 CIV, 8,98,103 this warrants further monitoring to ensure vaccine efficacy. An additional concern is that vaccination interferes with interpretation of routine serological assays such as the HI assay, since HA antibodies produced in re sponse to natural infection can not be distinguished from those produced due to vaccination. This has been addressed with avian, 5,31,117,124,127,143 equine, 10,94 and swine influenza 54 through development of DIVA (differentiate i nfected from vacc inated animals) test strategies and should be pursued for H3N8 CIV. Influenza A Viruses in Dogs around the World Since the discovery of H3N8 CIV in the U.S., several studies have evaluated canine populations in other countries for serolog ical evidence of H3N8 CIV exposure. A serological survey of 225 dogs from Ontario, Canada revealed that H3N8 CIV is either not present or is currently rare in the province despite its close proximity to affected states in the U.S. 59 The one dog that was seropositive to H3 antigen was a g reyhound originally from Florida. A study from New Zealand identified no H3N8 CIV seropositive

PAGE 36

36 dogs out of 251 tested. 57 A study conducted in northern Italy identified no H3N8 seropositive dogs out of 224 tested; 101 however a second study that tested sera collected from 6,858 dogs throughout Italy identified two dogs with antibodies to H3N8 CIV and EIV antigens. 24 One of these dogs moved to Italy from Florida three weeks before sample collection. The other seropositive dog was raised on a horse farm that previously experienced an equine influenza outbreak. In both cases the owners did not observe ILI in their dogs. 24 Since the recognition of equine origin H3N8 CIV, additional interspecies transmissions of H3N8 EIV isolates to dogs have been reported. In the United Kingdom, a retro spective study revealed that H3N8 EIV caused a respiratory outbreak among English foxhounds in 2002. 21 This was confirmed through serology, immunohis tochemistry, and RT PCR. Lectin binding studies performed by these investigators indicated that canine tracheal epithelial cells contain primarily SA Gal receptors. Serological evidence of a second spill over of H3N8 EIV to dogs, in the spring of 2003, has also been reported. 91 During the 2007 H3N8 EIV outbreak in Australia, dozens of dogs in contact with infected horses developed ILI Multiple cases were confirmed serologically and limited RT PCR sequencing indicated the HA M, and NA genes were identical to the circulating EIV, although no EIV isolates were obtained from infected dogs. 55 There is currently no evidence of sustained circulation of H3N8 EIV in the canine pop ulations of either the United Kingdom or Australia. Since 2006, interspecies transmissions of avian influenza A viruses (AIV) to dogs have also occurred. In Thailand, a fatal case of H5N1 AIV was reported in a dog that ingested an infected duck. 2,114 The virus isolate contained the multiple basic amino acid

PAGE 37

37 cleavage site and the 20 amino acid NA stalk deletion characteristic of these highly pathogenic AIV isolates, and also had the E627K PB2 mutation associated wit h enhanced viral replication in mammals. Serologi cal testing indicated that 160 of 629 (25%) Thai dogs had antibodies to H5N1 AIV, but sustained horizontal transmission has thus far not been reported in the country. 16 Experimental infections of dogs with H5N1 AIV have shown that dogs are susceptible to H5N1 infection and subclinically shed virus from the nose. 34 ,72 Virus labelling and histochemical examination, as well as lectin binding assays performed by these investigators, revealed that dogs have SA Gal receptors in their nasal mucosa, tracheal epithelium, bronchial epithelium, and alveoli. Although virus transmission to contact dogs or cats did not occur in one of these studies, both groups concluded that dogs may play a role in H5N1 virus adapt ation to mammals and contribute to spread to other species. 34,72 In South Korea, transmission of a H3N2 AIV to dogs was reported in 2007. Subsequent experimental and natural dog to dog transmission of H3N2 was rep orted, and serological evidence of infection has been detected in farmed and pet dogs in South Korea. 66,112,113 The H3N2 virus was isolated during outbreaks of severe respiratory disease in dogs at multiple facilit ies, and was reportedly traced to a dog purchased from a live animal market where chicken s duck s pheasant s rabbit s cats, pet dogs, and farmed dogs were sold. It was hypothesized that virus transmission occurred through feeding dogs infected poultry pro ducts or by aerosol transmission. isolates from Asia (South Korea, C hina, Hong Kong, Japan). Lectin binding assays indicated SA Gal receptors are found in canine bronchial and bronchiolar epithelial

PAGE 38

38 cells, and rarely on tracheal epithelial cells; while SA Gal receptors were not detected in canine tracheal, bronchial, or bronchiolar epithelial cells. 112 With virological and serological evidence of sustained dog to dog transmission, the H3N2 subtype represents a second influenza A virus considered an emerging respiratory pathogen in dogs. Finally, during the 2009 H1N1 pandemic several dogs were reportedly infected after contracting the virus from their owners. ProMED mail postings r eported canine cases in China and the U.S. ( P roMED mail. Influenza pandemic (H1N1) 2009, animal (30): China, canine ProMED mail 2009; 28 Nov: 20091128.4079 . Accessed 12 April 2011; P roMED mail. Influenza pandemic (H1N1) 2009, animal (40) USA (New York), canine. ProMED mail 2009; 22 Dec: 20091222.4305. . Accessed 12 April 2011) In a recent Ital ian study 7 o f 964 (0.7%) tested dogs had antibodies to a current H1N1 pandemic isolate in competitive ELISA, HI, and MN assays. 30 Summary The interspecies transmissions of H3N8 EIV and H3N2 AIV to dogs, as well as intermittent interspecies transfers of several other influenza viruses to dogs, have increased awareness regarding the potential role of canines i n the ecology of influenza A viruses. The H3 subtype is now the most broadly distributed HA subtype in mammals, with lineages in humans, horses, pigs, and dogs. 98 These events raise concern, in both the animal and public health communities, regarding the impact of influenza A viruses on the health and welfare of dogs and the possible transmission of influenza A viruses from dogs to other species. Experiences in the United States, the United Kingdom,

PAGE 39

39 Australia, Thail and, and South Korea demonstrate the importance of influenza A virus surveillance in canines, particularly during outbreaks in other species. 24

PAGE 40

40 CHAPTER 3 DIAGNOSTIC PERFORMAN CE OF THE H3N8 CANIN E INFLUENZA VIRUS HEMAGGLUTINATION INH IBITION ASSAY Canine influenza A subtype H 3N8 virus (H3N8 CIV) is a recently emerged respiratory pathogen for dogs. 20 Although this novel virus was initially identified in the racing greyhound population in the U.S., all nave dogs are susceptible to infec tion, including pet dogs and dogs housed in shelters. 20,26,28,29,49 Since the initial report in 2004, H3N8 CIV infections have been documented in thousands of dogs with widespread geographic distribution in 34 stat es. 8,9,20,28,29,42,44,98,103,110,111 To date, there is no evidence of H3N8 CIV circulation in other countries. 24,57,59,101 Similar to other mammalian influenza A viruses, H 3N8 CIV is efficiently transmitted between dogs, and can cause epidemics of influenza like illness (ILI) in communal housing settings such as racing greyhound kennels, boarding/training facilities, and animal shelters. 9,20,28,29,42,44,98,103,141 The incubation period for H3N8 CIV is 2 to 4 days and virus shedding begins prior to the development of clinical signs, with peak shedding occurring on day 4 and total virus shedding lasting 7 to 10 days. 20,26,29,49 Serum antibodies to the CIV hemagglutinin (H3 ) protein are detectable as 7 to 8 days after infection. 20,26,28,29 Molecular analyses of H3N8 CIV isolates indicate prior transmissio n of a Florida sublineage H3N8 equine influenza A virus (H3N8 EIV) from horses to dogs, followed by viral adaptation to the new canine host resulting in efficient dog to dog transmission. 20,42,98,103 H3N8 CIV isola tes form a monophyletic group that is molecularly and epidemiologically distinct from currently circulating H3N8 EIV in the U.S. 103 Signature amino acid changes in the H3 protein, p articularly around the receptor binding pocket, distinguish emerging CIV isolates from their EIV ancestors, and

PAGE 41

41 additional changes in endemic isolates indicate the virus continues to adapt t o dogs. 20,42,43,98,103 As for other influenza A viruses, H3N8 CIV can be diagnosed in clinical specimens by virus isolation or reverse transcription PCR (RT PCR) using subtype specific primers. 20,28,29,71,98,103 RT PCR is more sensitive than virus isolation, 28,29,103 however both assays can yield false negative results if clinical samples are collected outside the viral shedding win dow. 26,28,29 Therefore, serological assays are recommended for confirmation of H3N8 CIV infections, either as a diagnostic aid for current cases or as an epidemiological tool for surveillance and assessment of dise ase prevalence. 28,29 The hemagglutination (HA) and hemagglutination inhibition (HI) assays are classic influenza diagnostics, and are considered the tests of choice for virus surveillance and serological diagnosis of influenza A virus infections in animals. 134,135 In the HA assay, the viral hemagglutinin protein binds sialic acid (SA) receptors on the surface of red blood cells (RBC), resulting in hemagglutination. The assay is an impo rtant first step for virus hemagglutinin subtyping and virus preparation for the HI assay. Also, agglutination patterns for RBC from different species have been used to determine the SA receptor specificities of influenza A viruses 48 In the HI assay, serum antibodies specific for the hemagglutinin protein subtype interfere with virus binding to RBC, thereby inhibiting hemagglutination and indicating previous virus exposure. A variety of factors can influence the se nsitivity and specificity of the HI assay, including the matching of the SA receptor type on the RBC surface to that preferred by the viral hemagglutinin protein, the presence of nonspecific hemagglutinins and/or nonspecific inhibitors of hemagglutination in serum, and the binding specificity of serum antibodies for the

PAGE 42

42 particular hemagglutinin subtype of the test virus. 134,135 Although test conditions may require modification for detection of novel viruses or viruses that hav e undergone antigenic drift due to point mutations in the hemagglutinin protein, 79,116 standard HI conditions have been generally established for endemic influenza viruses in avian species, pigs, horses, and humans 134,135 To date, the diagnostic performance of the H3N8 CIV HI assay has not been formally evaluated. Further, the effect of test parameter variations on the performance of the assay has not been described. Discrepancies ex ist between diagnostic laboratories and research studies in the H3N8 CIV HI cutoff antibody titer selected for seropositivity as well as the test parameters used. The first objective of this study was to H3N8 CIV HI assay as performed in our previous reports documenting H3N8 CIV infections in dogs. 20,98 The second objective was to evaluate the effect of common test parameter variations on the diagnostic performance of the H3N8 CIV HI assay; including the use of different RBC species, canine serum treatment methods, live vs. inactivated virus, and different H3N8 CIV and EIV isolates. Materials and Methods All studies were approved by the University of Florida Instit utional Animal Care and Use Committee. Assessing the Diagnostic P erfo rmance of the Standard H3N8 CIV HI A ssay Comparison of the H3N8 CIV HI and serum micro neutralization assays To assess the diagnostic performance of the standard H3N8 CIV HI assay, HI and serum microneutralization (MN) test results were analyzed for 879 acute and/or convalescent sera collected from 619 dogs: 529 racing greyhounds, 20,98 50 shelter dogs 98 and 38 pet

PAGE 43

43 dogs involved in respiratory disease outbreaks in the U.S., as well as 2 beagles experimentally infected with the H3N8 A/ canine/Florida/43/2004 (CN/FL/04) isolate. 20 Sera collected from specific pathog en free (SPF) dogs 20 housed in a barrier research facility and normal ferret sera were used as negative controls for both assays, while polyclonal goat anti EQ/Miami/63 serum served as a positive control. HI and MN testing were conducted as previously described. 20,108 For the HI assay, serial two fold dilutions of serum pretreated with receptor destroying enzyme (RDE) were tested with 4 hemag glutinating units (HAU) of H3N8 C N/FL/04 virus and 0.5% turkey RBC in PBS. The endpoint antibody titer was defined as the reciprocal of the last serum dilution that completely inhibited hemagglutination. Seropositivity of a single sample was sera from SPF dogs and dogs experimentally infected with H3N8 CN/FL/04. 20 For the MN assay, the endpoint antibody titer was defined as the reciprocal of the highest serum dilution that gave 50% neutralization of 1 00 TCID 50 of H3N8 CN/FL/04. 20 Seropositivity of a single sample was 20,108 Intra assay and inter assay variability of the H3N8 CIV HI assay A subset of 12 serum samples previously tested in both HI and MN as says were re tested in the standard H3N8 CIV HI assay to assess intra assay and inter assay variability. This subset included 6 CIV seronegative samples and 6 CIV seropositive samples with HI antibody titers ranging fro m 32 to 1024. For intra assay testing each sample was tested 5 times in the same microtiter plate using the same virus and RBC preparations. For inter assay testing, each sample was tested in duplicate on three separate days using

PAGE 44

44 fresh virus and RBC preparations each day. Endpoint antibody titers were determined and defined as described above. Data analyses Statistical analyses comparing the H3N8 CIV HI and MN test results were conducted with standard software (XLSTAT Addinsoft, New York, NY, USA; SigmaStat Systat Software, San Jose, CA, USA analysis was used to determine correlation of HI and MN antibody titers for 879 sera. HI and MN test results for con valescent sera from 614 dogs (n=507 H3N8 CIV infected, n= 107 uninfected) were utilized in a receiver oper ating characteristic (ROC) analysis to determine the sensitivity, specificity, and positive and negative predictive values of the HI assay using the MN assay as the gold standard. H3N8 CIV seropositive samples negative samples (MN titer <40) were The recommended H3N8 CIV HI antibody titer cutoff for seropositivity was based on the optimum sensitivity and specificity values determ ined in the ROC analysis. Analytical results with a P value < 0.05 were considered significant. The percent coefficient of variation (% CV) was calculated for the intra assay and inter assay variability of the H3N8 CIV HI assay. Evaluating the E ffect of Te st Parameter Variations on D iagnostic P erformance of the H3N8 CIV HI A ssay Serum samples To determine the impact of common test parameter variations on the diagnostic performance of the H3N8 CIV HI assay, a subset of seronegative and seropositive convales cent sera tested in the standard H3N8 CIV HI and MN assays was re tested with different RBC species, serum treatment methods, and different H3N8 CIV and EIV isolates. Sera from SPF dogs were used as negative controls.

PAGE 45

45 Red blood cell sources Commercially available chicken, turkey, goose, horse, dog, sheep, and cow RBC collected with Alsevers as the anticoagulant ( Lampire Biological Laboratories, Pipersville, PA, USA ) were used in HA assays. The chicken, turkey, and dog RBC were also used in HI assays. All RBC were stored at 4C and used within 2 weeks of collection. Serum treatments Sera were either untreated or treated with RDE (Denka Seiken, Tokyo, Japan) or periodate ( Sigma Aldrich Corp., Saint Louis, MO, USA ) followed by heating at 56 C for 30 minutes to inactivate poten tial nonspecific inhibitors of hemagglutination in canine sera. For RDE treatment, serum samples were incubated with RDE (1 volume serum:3 volumes RDE) for 16 hours at 37 C prior to heat inactivation. Periodate treatment was conducted fo llowing a standard WHO protocol 134 that included, in a step wise fashion, the treatment of volume serum with 1 volume trypsin solution ( Sigma Ald rich Corp. ), inactivation at 56C for 30 minutes, incubation with periodate solution (3 volumes) for 15 minutes at room temperature (23 C), incubation with 3 volumes of a 1% glycerol ( Sigma Aldrich Corp. ) solution for 1 5 minutes at 23 C, and addition of 0.85% saline for a final serum dilution of 1:10. Viruses H3N8 CN/FL/04 and A/canine /Jacksonville/ 20 05 (CN/Jax/05) were kindly provided by Dr. R. Donis (Centers for Disease Control and Prevention, Atlanta, GA, USA) A/ca nine /FortCollins/ 3/ 20 06, A/canine /Weld/ 20 07, and A/canine /ColoradoSprings/ 20 09 were kindly provided by Dr. G. Landolt (Colorado State University, Fort Collins, CO, USA). A/canine /Pennsylvania / 96978/ 20 09 (CN/PA/09) and A/equine/NewYork/1999 ( EQ/NY/99 ) were kindly provided by Dr. E. Dubovi (Cornell University, Animal Health Diagnostic Center, Ithaca, NY, USA). H3N8 CN/FL/04

PAGE 46

46 previously inactivated by binary ethyleneimine ( BEI ) treatment, an alkylating agent that inactivates virus without altering virus protein s, was kindly provided by Dr. N. Lakshmanan (Intervet Schering Plough Animal Health, Millsboro DE, USA). A/equine/Kentucky/1991 ( EQ/KY/91 ) A/equine/Kentucky/1995 ( EQ/KY/95 ) and A/equine/Ohio/2003 ( EQ/OH/03 ) were kindly provided by Dr. T. Chambers (Unive rsity of Kentucky, Lexington, KY, USA). All equine viruses were cultured in embryonated chicken eggs as previously described. 20 The provided allantoic fluid aliquots were stored at 80 C until use. All canine iso lates were cultured in Madin Darby canine kidney (MDCK) cells using a standard protocol, 20 with the described modifications. Briefly, approximately 75% confluent MDCK cells grown in 75 cm 2 flasks were inoculated wi th a 1:750 dilution of virus stock in PBS. The flasks were incubated for 1 hour at 35 C in a humidified atmosphere containing 5% CO 2 and then washed with sterile PBS. Dulbecco modified Eagle medium (DMEM) supplemented with 1 g/mL TPCK (L 1 tosylamido 2 p henylethyl chloromethyl ketone) treated trypsin (Sigma A ldrich Corp. ), 0.85% bovine serum albumin (BSA), HEPES buffer, and penicillin streptomycin was added to the flasks. The inoculated cells were incubated at 35 C in a humidified atmosphere containing 5% CO 2 for 3 to 5 days, with daily evaluation for cytopathic effect. The culture supernatant was harvested, clarified by centrifugation at 2000 rpm for 10 minutes, and the amount of virus determ ined with the HA assay using 0.5% turkey red blood cells (Lampir e Biological La boratories ) in PBS as previously described. 20 The viral culture supernatant was stored at 80 C until future use.

PAGE 47

47 HA assay Prior to evaluating the effect of using different RBC species in the HI as say, the RBC binding preferences of H3N8 CN/FL/04 were assessed in HA assays Serial two fold dilutions of MDCK supernatant containing live H3N8 CN/FL/04 were incubated with an equal volume (50 L) of 0.5% chicken, 0.5% turkey, 0.5% goose, 1% horse, 1% dog 1% sheep, and 1% cow RBC prepared in PBS or PBS containing 0.5% bovine serum albumin (PBS/BSA). Incubations were performed in V bottom microtiter plates for 30 minutes or 1 hour at 23 C and 4 C. HAU, defined as the reciprocal of the last virus dilution t hat agglutinated RBC, were compared for the different RBC species and incubation conditions. The RBC binding preferences were evaluated based on the HAU and known SA receptors for each RBC species. RBC species with the highest HAU for H3N8 CN/FL/04 were se lected for use in subsequent HI assays. HI assay Serial two fold dilutions of untreated sera and sera treated with RDE or periodate were incubated in V bottom microt iter plate wells with 4 HAU/25 L of live H3N8 CIV or EIV isolates or inactivated CN/FL/0 4 for 30 minutes at 23 C. The serum/virus mixtures were then incubated at 23 C for 30 minutes with 0.5% chicken or turkey RBC in PBS, or for 60 minutes with 1% dog RBC in PBS/BSA. All samples were tested in duplicate. End point antibody titers and the cuto ff titer for seropositivity were defined as described above for the standard HI assay. For each serum sample, antibody titers generated by HI assays with different parameters were compared to the antibody titers determined by the standard HI assay. Resul ts Assessing the Diagnostic Performance of the Standard H3N8 CIV HI A ssay Comparison of the H3N8 CIV HI and MN assays correlation coefficient comparing the standard H3N8 CIV HI and MN assay antibody

PAGE 48

48 titers for 879 sera was 0.87 ( P <0.01 ). For the 642 seropositive samples, the geometric mean antibody titer (GMT) in the MN assay (543) was significantly ( P <0.01) higher than the GMT in the HI assay (381). ROC analysis of the HI antibody titers for 614 convalescent sera (83% seropositive) usi ng the MN titers as the gold standard resulted in an area under the curve (AUC) of 0.985 (95% CI = 0.98 to 0.99; P <0.01). The ROC generated sensitivity, specificity, positive and negative predictive values for HI antibody titers ranging from 4 to 2048 are shown in Table 3 1. Based on a sensitivity of 99.6% and specificity of 94.4%, the optimal cutoff antibod y titer for seropositivity was 32. At this cutoff, the positive predictive value (PPV) was 98.8% and the negative predictive value (NPV) was 98.1%. Intr a assay and inter assay variability of the H3N8 CIV HI assay The % CV for the intra assay H3N8 CIV HI variability testing was 4.1%. The % CV for the inter assay variability testing was 3.0%. For all replicates, the HI titers correctly classified the dogs as H3N8 CIV infected or unin fected using a cutoff titer of 32. Evaluating the Effect of Test Parameter Variations on Diagnostic P erfo rmance of the H3N8 CIV HI A ssay Red blood cell binding preferences of H3N8 CIV The HAU for H3N8 CN/FL/04 with various RBC species are summarized in Table 3 2. HAU were highest with chicken and turkey RBC, followed by goose and canine RBC. Incubation temperature and addition of BSA to the viral suspension had little effect on HAU for each RBC species. Use of different RBC spe cies in the H3N8 CIV HI assay To examine the effect of RBC source on HI assay performance, RDE treated sera from 11 H3N8 CIV infected dogs and 14 uninfected dogs were incubated with chicken, turkey, or canine RBC. Regardless of the RBC source, all of the samples from uninfected dogs had HI antibody

PAGE 49

49 shown in Table 3 3 the HI antibody titers for infected dogs were 2 to 8 fold higher when tested with chicken or turkey RBC c ompared to canine RBC; the GMT was significantly lower ( P <0.01) when canine RBC were used. An additional 46 serum samples from dogs infected with H3N8 CIV were RDE treated and tested with 0.5% chicken or turkey RBC in the H3N8 CIV HI assay. The HI antibody titers ranged from 32 to 1024. In general, there was good agreement in HI antibody titers between the two RBC sources. However 30% (14/46) of the samples had a 2 fold lower titer with chicken RBC compared to turkey, resulting in one dog bein g misclassifie d as seronegative Use of different serum treatments in the H3N8 CIV HI assay HI antibody titers for untreated sera from 11 H3N8 CIV infected dogs were similar to or 2 fold higher than titers for the same samples treated with RDE (Table 3 4). Although th ere were no false negative results for the infected dogs, untreated sera resulted in false positive results for 87% (13/15 dogs) of the uninfected dogs when chicken RBC were used in the HI assay instead of turkey. There were no false negative or false posi tive results for sera treated with RDE, regardless of the RBC source. F or uninfected dogs (n= 4) the HI antibody titers for periodate treated sera were similar to those obtained for the same samples treated with RDE. For H3N8 CIV infected dogs (n= 6) howeve r, the HI antibody titers for periodate treated sera were 2 fold lower than those obtained for the same samples treated with RDE Use of live or inactivated virus in the H3N8 CIV HI assay To evaluate whether inactivated H3N8 CIV can be used instead of liv e virus, sera from 14 H3N8 CIV infected

PAGE 50

50 and 14 uninfected dogs were tested in the HI assay with live H3N8 CN/FL/04 or virus inactivated with BEI. Overall, the HI antibody titers for sera from infected dogs were similar when live or inactivated CIV was used (Table 3 5). However, 3 of 4 samples with HI titers near the cutoff titer for seropositivity were false negative when inactivated virus was used in the assay. Use of different canine or equine viral isolates in the H3N8 CIV HI assay To evaluate consiste ncy in antibody titers whe n different H3N8 CIV isolates wer e used in the HI assay, sera from 46 dogs infected with H3N8 CIV between 2004 and 2005 were tested using 5 H3N8 CIV isolates acquired fro m dogs in different geographic regions from 2005 to 2009. Th e HI antibody titers for these samples ranged from 32 to 1024 when tested with CN/FL/04. The HI titers with the 5 isolates were equivalent to or two fold lower than the titers wi th CN/FL/04 (Table 3 6 ). There were no false negative results for samples test ed with the 5 isolates. To determi ne whether H3N8 EIV isolates could be reliably substituted for canine isolates in the HI assay, sera from the 46 infected dogs were also tested with 4 H3N8 EIV isolates acquired from horses in different geographic regions from 1991 to 2003. The HI antibody titers to EQ/OH/03 were equivalent to those for the CN/FL /04 isolate (Table 3 7 ). The HI titers to EQ/NY/99 were 2 to 4 fold lower than those with the CN/FL/04 isolate, resulting in misclassification of 4 infected dogs ( 8.7%) as uninfected. The HI titers to EQ/KY/95 and EQ/KY/91 were 4 to 8 fold lower than those for the CN/FL/04 isolate, and 8 infected dogs (17.4%) were misclassified as uninfected. Discussion This study demonstrated a high correlation between antibody t iters to H3N8 CIV determined in a serum micro neutralization (MN) assay and the standard HI assay

PAGE 51

51 described in our previous reports. 20,98 This standard H3N8 CIV HI assay is highly sensitive and specific when the cut off antibo dy titer for seropositivity is 32. The precision of the HI assay was demonstrated by coefficients of variation <5% for intra assay and inter assay replications. T he optimal sensitivity and specificity of the HI assay depended on treatment of ser u m samples with RDE use of 0.5% turkey RBC, and use of live H3N8 CIV isolates. Use of different H3N8 CIV isolates fro m dogs in different geographic regions from 2005 to 2009 and contemporary H3N8 EIV isolates from horses after 2000 did not alter the sensit ivity of the HI assay. The HI assay is considered the WHO/OIE test of choice for serological diagnosis of influenza A infection in animals. 134,135 The test parameters for the standard H3N8 CIV HI assay described in previous r eports included use of turkey RBC, RDE treated sera, live CN/FL/04 virus, and a seropositivity cutoff titer of 32 based on reactivity of sera from SPF dogs and experimentally infected dogs. 20,98 Under these condit ions, this study confirmed that diagnosis of H3N8 CIV infected and uninfected dogs by the HI assay correlated well with MN assay results. ROC analysis also confirmed that t he HI antibody titer cutoff of 32 resulted in the highest sensitivity (99.6%) and sp ecificity (94.4%). Lower cutoff values increased the number of false positives and decreased specificity. Sensitivity and specificity are fixed characteristics of diagnostic tests, whereas the predictive values of a positive or negative test are affected b y disease prevalence. While the PPV and NPV were high using a HI seropositivity titer cutoff of 32, these values were reflective of the high prevalence (83%) of seropositive dogs included in the tested population. However, a benefit of ROC analysis is that it allows an estimate of diagnostic accuracy that is independent of disease prevalence. 36 Therefore, i n locations

PAGE 52

52 where H3N8 CIV is known to circ ulate the HI assay is considered a reliable and sensitive test for serological diagnosis as long as the test antigen is homologous to the circulating virus. For large scale serological surveys seeking to estimate the prevalence of H3N8 CIV (or an other influenza A virus) seropositivity in locations where the status of virus circulation is unknown the HI assay has been recommended as a sensitive and specific confirmatory test of high throughput screening assays such as competitive nucleoprotein ELISAs. 24,30 Several variations in test parameters decreased the sensitivity and/or specificity of the standard H3N8 CIV HI assay, including use of RBC species other than turkey, lack of serum treatment to inactivate natur ally occur ring inhibitors of hemagglutination, use of periodate for serum treatment, use of inactivated virus, and use of H3N8 EIV isolates recove red from horses prior to 2000. Although there was good agreement in antibody titers for many samples when turkey or chic ken RBC was used, nearly one third of the sera had lower titers with chicken RBC. This suggests that truly infected dogs with HI antibody titers near the cutoff titer for seropositivity may be misdiagnosed when using chicken RBC. A variety of nonspecific hemagglutinins and inhibitors of hemagglutination have been identified in animal and human sera; limited studies with few samples have shown that canine serum may have low levels of nonspecific hemagglutinins for avian red blood cells and nonspecific inhib itors of hemagglutination for some influenza viruses 3,122 This study demonstrated that serum trea tment is required to inactivate nonspecific inhibitors of agglutination in canine serum. Nonspecific inhibitors of h emagglutination were demonstrated by higher antibody titers in untreated sera from

PAGE 53

53 infected dogs compared to titers obtained after RDE or periodate treatment and heat inactivation. Nonspecific inhibitors of hemagglutination were particularly evident when u ntreated sera were tested with chicken RBC rather than turkey RBC in the HI assay: the chicken RBC resulted in false positive reactions for nearly all of the truly uninfected dogs tested. Therefore, to decrease false positive reactions and maintain high sp ecificity, canine sera should be treated with RDE and heat inactivation prior to testing in the H3N8 CIV HI assay. Although use of inactivated H3N8 CIV in the HI assay would be advantageous to reduce biosafety risks to laboratory personnel and generation o f biohazardous waste, the use of inactivated virus reduced the sensitivity of the HI assay when testing sera from infected dogs with antibody titers near the cutoff titer for seropositivity. In addition, RBC streaming was at times more difficult to interpr et than with live virus. BEI treatment is supposed to preserve viral proteins, therefore it was unclear why BEI inactivated virus affected the sensitivity of the assay. One possible explanation is that this alkylating agent could cause cross linking of the H3 proteins on the viral surface that could hinder hemagglutination ; however this could not be evaluated by the investigators Further analysis of viral inactivation methods is warranted before inactivated H3N8 CIV could be reliably substituted for live virus in the HI assay. The use of multiple H3N8 CIV isolates recovered from dogs in 2005 to 2009 did not alter the sensitivity and specificity of the standard HI test. Molecular and serologic analyses indicate that significant antigenic change has not occ urred in H3N8 CIV isolates to date. 8,98,103 If antigenic drift does occur, then use of earlier isolates in HI testing of sera from dogs exposed to drifted viruses will compromise the sensitivity of

PAGE 54

54 the assay, leadi ng to misdiagnosis of truly infected dogs. This underscores the value of continual molecular monitoring of new isolates to maintain diagnostic test accuracy. Many diagnostic laboratories have H3N8 EIV isolates for routine HI testing, and some have attempte d to use these isolates for diagnosis of H3N8 CIV infection. Phylogenetic analyses of H3N8 CIV and EIV isolates have shown that the canine viruses are most closely related to contemporary EIV isolates recovered from horses since 2000. 20,42,98,103 This genetic relatedness likely explains why use of a 2003 EIV isolate in the CIV HI test did not affect sensitivity and specificity, but use of earlier EIV isolates resulted in decreased sensitivity and misdiagnosis of tru ly infected dogs. While avian, swine, and human H3 isolates were not evaluated in this study, it is assumed that their use in the HI assay would not identify dogs infected with H3N8 CIV based on the genetic disparity between the H3 proteins. 20 Our previous report demonstrated that HI antibody titers for dogs truly infected with H3N8 CIV were <16 when tested with human H3N2 viruses. 20 The binding preferences of H3N8 CIV were assessed in HA assays with a variety of avian and mammalian RBC. This is important not only to selection of the best RBC for use in the HI assay to maintain diagnostic accuracy, but the agglutination patterns for RBC from different species can also provid e insight into viral specificities for SA receptors. 48 This study demonstrated that H3N8 CIV preferred avian and canine RBC, which primarily h Gal and Gal receptors with N acetyl neuraminic acid (NeuAc) as the SA species. 41,48,140 There was less viral agglutination of horse and Gal receptors wi th N glycolyl neuraminic acid (NeuGc) as the SA species. 48 Gal and

PAGE 55

55 Gal receptors are required for viral entry and subsequent replication in the c ell. 120 The RBC agglutination patterns of H3N8 CIV suggest that the H3 protei Gal receptors like its EIV ancestor, but Gal receptors. This supports the findings of a recent report showing Gal receptor analogues. 139 These results Gal receptors on respiratory epithelial cells in dogs 21,72,80,112 suggest that the altered SA preferenc e of CIV compared to EIV may be an important adaptation of the horse virus to the dog. This adaptation may be associated with the predictable amino acid sub stitutions near the H3 receptor binding pocket that have become the molecular signature for H3N8 CIV 20,98,103 In conclusion, the HI assay is a highly sensitive and specific assay for serological diagnosis of H3N8 CIV infection in dogs. Implementation of the recommended cutoff titer for seropositivity of 32 and RBC, serum treatment with RDE, and use of a l ive H3N8 CIV isolate should minimize misdiagnosis of H3N8 CIV infected dogs. The best diagnostic approach for recent active infection is to perform the sta ndard HI test on both acute and convalescent serum samples to document at least a 4 fold increase in antibody titer over time. Moving forward it will be important to continue monitoring for antigenic drift in the CIV H3 protein in order to ensure that the H3N8 CIV used in the HI assay is closely related to that in circulation. In addition adequate vaccination history is now imperative to accurately interpret H3N8 CIV HI test results, since the assay cannot differentiate antibodies induced by infection from vaccination with the inactivated whole H3N8 CIV vaccine. 27

PAGE 56

56 This limits the use of the HI assay in future epidemiological studies of H3N8 CIV in the U.S. if vaccination status is unknown.

PAGE 57

57 Table 3 1. ROC analysis of antibody titers determined for 614 convalescent sera in the standard H3N8 CIV HI assay using the MN antibody titers as the gold standard. HI t iter a Sensitivity (%) (95% CI) b Specificity (%) (95% CI) PPV (%) c NPV (%) d True positives True negatives False positives False negatives 4 100.0 (99.1, 100.0) 0 (0, 4.3) 82.6 --507 0 107 0 16 99.6 (98.5, 100.0) 92.5 (85.7, 96.3) 98.4 98.0 505 99 8 2 32 99.6 (98.5, 100.0) 94.4 (88.0, 97.6) 98.8 98.1 505 101 6 2 64 97.4 (95.6, 98.5) 95.3 (89.2, 98.2) 99.0 88.7 494 102 5 13 128 90.9 (88.1, 93.1) 96.3 (90.4, 98.8) 99.1 69.1 461 103 4 46 256 66.9 (62.6, 70.8) 99.1 (94.3, 100.0) 99.7 38.7 339 106 1 168 512 39.6 (35.5, 44.0) 99.1 (94.3, 100.0) 99.5 25.7 201 106 1 306 1024 17.8 (14.7, 21.3) 100.0 (95.7, 100.0) 10 0.0 20.4 90 107 0 417 2048 0.6 (0.1, 1.8) 100.0 (95.7, 100.0) 100.0 17.5 3 107 0 504 a HI antibody titer = reciprocal of the last serum dilution that inhibited red blood cell agglutination by H3N8 CIV. b 95% confidence interval. c PPV = positive predicti ve value. d NPV = negative predictive value.

PAGE 58

58 Table 3 2. H3N8 CIV agglutination of avian and mammalian red blood cells under different incubation conditions. H3N8 CIV HAU c RBC s ource a RBC SA b 23 C X 30 min 4 C X 60 min PBS PBS/BSA PBS PBS/BSA Chicken 2,3/2,6 NeuAc 32 16 32 16 Turkey 2,3/2,6 NeuAc 16 16 16 16 Goose 2,3 /2,6 NeuAc 8 8 8 8 Dog 2,3/2,6 NeuAc 8 16 8 16 Horse 2,3 NeuGc 2 neg d 2 neg Sheep 2,3 NeuGc 2 1 2 2 Cow 2,3 NeuGc neg neg 1 1 a Species source of red blood cells (RBC) used in the H A assays. Avian RBC suspensions were 0.5%, and mammalian RBC suspensions were 1.0% in either PBS or PBS/BSA. b Predominant sialic acid (SA) linkages (2,3 and/or 2,6) and species ( NeuAc= N acetyl neuraminic acid ; NeuGc= N glycolyl neuraminic acid ) on spec ies specific RBC. c HAU = reciprocal of the last dilution of H3N8 virus that agglutinated RBC. d No RBC agglutination by undiluted H3N8 viral suspension.

PAGE 59

59 Table 3 3. HI antibody titers for sera from 11 dogs infected with H3N8 CIV tested with 0.5% chic ken red blood cells (chRBC), 0.5% turkey red blood cells (tRBC), or 1% canine red blood cells (cnRBC). Sera were pretreated with RDE and serial dilutions incubated with H3N8 CN/FL/04 virus prior to addition of RBC suspensions. All serum samples were tested in duplicate. H3N8 CIV HI antibody titer a Sample ID chRBC tRBC cnRBC 1 256 64 32 2 256 128 64 3 128 128 64 4 256 256 64 5 256 512 64 6 128 128 64 7 256 256 128 8 256 512 64 9 256 512 128 10 256 512 128 11 512 512 64 GMT b 240 256 73 c a HI a ntibody titer = reciprocal of the last serum dilution that inhibited red blood cell agglutination by H3N8 CIV. The cutoff t iter for sero positivity in the standard H3N8 CIV HI assay is 32. b Geometric mean antibody titer for 11 dogs. c The GMT for cnRBC is significantly different from chRBC and tRBC (P value <0.01).

PAGE 60

60 Table 3 4. HI antibody titers for untreated or RDE treated sera from 11 dogs infected with H3N8 CIV and 15 uninfected dogs. Serial dilutions of the untreated or RDE treated sera were incubate d with H3N8 CN/FL/04 virus prior to addition of 0.5% turkey (tRBC) or chicken (chRBC) red blood cell suspensions. All serum samples were tested in duplicate. H3N8 CIV HI antibody titer b tRBC chRBC Sample ID CIV status a Untreated c RDE treated Untreat ed RDE treated 1 uninfected 4 4 32 4 2 uninfected 4 4 32 4 3 uninfected 4 4 64 4 4 uninfected 4 4 32 4 5 uninfected 4 4 32 4 6 uninfected 4 4 32 4 7 uninfected 4 4 32 4 8 uninfected 4 4 64 4 9 uninfected 4 4 32 4 10 uninfected 4 4 4 4 11 uninfec ted 4 4 32 4 12 uninfected 4 4 64 4 13 uninfected 4 4 32 4 14 uninfected 4 4 32 4 15 uninfected 4 4 4 4 16 infected 512 64 256 256 17 infected 512 128 256 256 18 infected 512 128 256 128 19 infected 256 128 128 128 20 infected 512 256 512 256 21 infected 512 256 512 256 22 infected 512 512 256 256 23 infected 1024 512 1024 256 24 infected 512 512 512 256 25 infected 1024 512 1024 256 26 infected 512 512 512 256 a H3N8 CIV infection status determined by the MN assay. b HI antibody titer = rec iprocal of the last serum dilution that inhibited red blood cell agglutination by H3N8 CIV. The cutoff titer for seropositivity in the standard H3N8 CIV HI assay is 32. c Sera were either untreated or treated with RDE prior to incubation with virus.

PAGE 61

61 Tabl e 3 5. HI antibody titers for sera from 14 dogs infected with H3N8 CIV tested with live or inactivated virus. Sera were pretreated with RDE and serial dilutions incubated with live or BEI inactivated H3N8 CN/ FL/04 virus prior to addition of 0.5% turkey red blood cells. All assays were conducted in duplicate. H3N8 CIV HI antibody titer a Sample ID Live virus b Inactivated virus c 1 32 4 2 32 4 3 64 4 4 64 64 5 128 128 6 128 128 7 256 128 8 256 256 9 256 256 10 512 512 11 512 512 12 512 512 13 10 24 512 14 1024 512 a HI antibody titer = reciprocal of the last serum dilution that inhibited red blood cell agglutination by H3N8 CIV. The cutoff titer for seropositivity in the standard H3N8 CIV HI assay is 32. b Live H3N8 CN/FL/04 virus harvested fr om MDCK cultures. c H3N8 CN/FL/04 virus ha rvested from MDCK cultures and inactivated by BEI treatment.

PAGE 62

62 Table 3 6. HI antibody titers for sera from 46 dogs infected with H3N8 CIV and tested with different H3N8 CIV isolates in the standard HI assay. Serial dilutions of RDE treated sera were incubated with H3N8 A/canine /Florida/ 43/20 04 (CN/FL/04), A/canine /Jacksonville/ 20 05 (CN/Jax/05), A/canine/Fort Collins/ 3/ 20 06 (CN/CO/06), A/canine /Weld/ 20 07 (CN/CO/07), A/canine /Colorado Springs/ 20 09 (CN/CO/09), and A/canine /Pennsylvania/ 96978/20 09 (CN/PA/09) virus prior to addition of 0.5% turkey red blood cell suspensions. All serum samples were tested in duplicate with each virus. The H3N8 CIV HI antibody titer is the reciprocal of the last serum dilution that inh ibited red blood cell agglutination. The cutoff titer for seropositivi ty in the standard HI assay is 32. H3N8 CIV HI antibody titer Sample ID CN/FL/04 CN/Jax/05 CN/CO/06 CN/CO/07 CN/CO/09 CN/PA/09 1 32 32 32 32 32 32 2 64 32 32 32 32 32 3 64 64 32 64 32 32 4 64 32 32 64 32 32 5 64 32 32 64 32 32 6 64 64 64 64 64 64 7 128 64 64 64 64 64 8 128 64 64 128 64 64 9 128 64 64 128 64 64 10 128 128 64 128 64 64 11 128 128 128 128 128 128 12 128 128 128 64 64 64 13 128 128 128 128 128 128 14 128 128 1 28 128 128 128 15 128 64 64 64 64 64 16 128 128 64 128 64 64 17 128 128 128 128 128 128 18 128 128 128 128 128 128 19 128 128 128 128 128 128 20 128 128 128 128 128 128 21 128 128 128 128 128 128 22 128 128 128 128 128 128 23 128 128 128 128 128 2 56 24 256 128 128 128 128 128

PAGE 63

63 Table 3 6. C ontinued H3N8 CIV HI antibody titer Sample ID CN/FL/04 CN/Jax/05 CN/CO/06 CN/CO/07 CN/CO/09 CN/PA/09 25 256 256 256 128 128 128 26 256 256 256 256 256 128 27 256 256 256 256 256 256 28 256 256 256 256 128 256 29 256 256 256 128 256 256 30 512 256 512 256 256 256 31 512 512 512 512 512 512 32 512 256 256 256 256 256 33 512 256 256 256 256 256 34 512 512 512 512 256 256 35 512 256 256 256 256 256 36 512 256 256 256 256 256 37 512 512 512 512 512 512 38 512 512 512 512 512 256 39 1024 1024 1024 1024 512 512 40 1024 512 1024 512 512 512 41 1024 1024 1024 1024 1024 1024 42 1024 1024 1024 1024 1024 1024 43 1024 1024 1024 1024 1024 1024 44 1024 512 1024 512 512 1024 45 1024 512 512 512 512 512 46 1024 1024 1024 1024 1024 1024

PAGE 64

64 Table 3 7. HI antibody titers for sera from 46 dogs infected with H3N8 CIV and tested with different H3N8 EIV strains in the standard HI assay. Serial dilutions of RDE treated sera were incubated with H3N8 A/canine /Florid a/ 43/20 04 (CN/FL/04), A/equine /Ohio/ 20 03 (EQ/OH/03), A/equine/New York/ 1999 (EQ/NY/99), A/equine /Kentucky/ 19 95 (EQ/KY/95), and A/equine /Kentucky/ 19 91 (EQ/KY/91) virus prior to addition of 0.5% turkey red blood cell suspensions. All serum samples were tested in duplicate with each virus. The H3N8 CIV HI antibody titer is the reciprocal of the last serum dilution that inhibited red blood cell agglutination. The cutoff titer for seropositivity in the standard HI assay is 32. H3N8 CIV HI antibody titer Sample ID CN/FL/04 EQ/OH/03 EQ/NY/99 EQ/KY/95 EQ/KY/91 1 32 32 16 8 8 2 64 64 16 16 8 3 64 64 32 16 16 4 64 64 16 8 8 5 64 64 16 16 16 6 64 128 32 16 16 7 128 128 32 16 16 8 128 128 32 16 32 9 128 128 64 32 16 10 128 128 64 64 64 11 128 128 64 32 32 1 2 128 128 64 32 32 13 128 128 64 64 64 14 128 256 64 32 32 15 128 128 32 32 32 16 128 128 32 32 32 17 128 256 64 32 32 18 128 256 64 64 32 19 128 128 64 64 64 20 128 256 64 64 64 21 128 256 64 64 64 22 128 256 64 64 64 23 128 256 64 64 64

PAGE 65

65 Table 3 7. C ontinued H3N8 CIV HI antibody titer Sample ID CN/FL/04 EQ/OH/03 EQ/NY/99 EQ/KY/95 EQ/KY/91 24 256 256 64 64 64 25 256 256 64 32 64 26 256 256 128 64 64 27 256 512 128 128 128 28 256 256 128 128 128 29 256 256 128 128 64 30 512 512 128 128 64 31 512 512 256 128 128 32 512 256 64 64 32 33 512 512 128 128 64 34 512 256 128 64 128 35 512 512 256 128 128 36 512 512 128 128 64 37 512 512 256 128 128 38 512 512 256 256 128 39 1024 1024 512 512 256 40 1024 1024 512 256 256 41 1024 1 024 1024 1024 1024 42 1024 1024 256 256 256 43 1024 1024 512 128 256 44 1024 1024 256 128 128 45 1024 1024 256 512 512 46 1024 1024 512 256 256

PAGE 66

66 CHAPTER 4 DEVELOPMENT OF A DIV A TEST STRATEGY FOR H3N8 CANINE INFLUENZ A VIRUS In June 2009, the USDA ap proved the first vaccine for canine influenza. The H3N8 canine influenza virus (H3N8 CIV) vaccine is intended as an aid in the control of disease, and is recommended for dogs at risk of exposure. 27 Similar to influ enza vaccines in other species, the H3N8 CIV vaccine contains inactivated whole virus that primarily induces antibodies to the hemagglutinin (H3) and neuraminidase (N8) surface proteins. Although an effective vaccine is beneficial for the prevention and co ntrol of H3N8 CIV, vaccine induced H3 antibodies interfere with the se rological diagnosis of infected dogs using the hemagglutination inhibition (HI) assay. The HI assay detects antibodies to specific hemagglutinin subtypes and is currently considered the confirmatory test of choice for H3N8 CIV infection. 28,29 The HI assay however cannot distinguish H3 antibodies induced by infection vs. vaccination, thereby creating a diagnostic dilemma for differentiating infecte d from vaccinated dogs based on seropositivity. This diagnostic dilemma has fueled the demand for alternative testing strategies that satisfy the principle of differentiation of infected from vaccinated animals (DIVA). There are no DIVA test strategies for H3N8 CIV, and a DIVA strategy should be developed to ensure correct identification of infected dogs in the face of vaccination. One DIVA testing strategy is based on the expected immune response to the inf luenza A nonstructural protein, NS1. The NS1 prot ein is a multifunctional protein that inhibits the host immune resp onse and is involved in several aspects of viral replication 39 This protein is produced in large quantities in virus infected cells, but is not pa ckaged into progeny virions released from infected cells. 58,64 Infected animals can mount an antibody response to the NS1 protein released from the intracellular

PAGE 67

67 cytoplasm of cells dying as a result of viral replic ation. Since inactivated whole virus influenza vaccines contain intact virions without NS1, vaccination should not induce antibodies to the NS1 protein. 117 Therefore, infected animals c an be differentiated from vaccinated animals based on the presence of NS1 antibodies, thereby satisfying the DIVA principle. NS1 antibody tests have been developed as a DIVA strategy for avian, 5,31,124,125,127,143 equine, 10,94 and swine 54 influenza. Based on these precedents, the objective of this study was to develop a NS1 antibody ELISA to differentiate H3N8 CIV infected from vacci nated dogs. The best DIVA strategy would be to initially test for CIV H3 antibodies using the HI assay. H3 seropositive dogs would then be tested in a NS1 antibody ELISA to differentiate between infection and vaccination. For this study, molecular techniqu es were used to produce recombinant H3N8 CIV NS1 protein to capture specific antibodies in an ELISA and western blot. Archived sera from known infected or vaccinated dogs were tested in the H3 HI assay and NS1 ELISA or western blot to evaluate the DIVA str ategy. Materials and Methods Virus Preparation The NS1 amino acid sequence in different H3N8 CIV isolates collected from 2004 to 2008 is well conserved. 20,98,103 The NS1 gene of H3N8 A/canine/Florida/43/2004 (CN/F L/04) was selected for cloning and protein expression. The NS1 gene (Gen Bank A ccession # DQ124153 ) contains 657 nucleotides, producing a protein of 219 amino acids (MW=24.7 kDal). The CN/FL/04 isolate was kindly provided by Dr. R. Donis (Centers for Disea se Control and Prevention, Atlanta, GA, USA). The virus was grown in MDCK cells using a

PAGE 68

68 standard protocol, 20 with the described modifications. Briefly, approximately 75% confluent MDCK cells grown in 75 cm 2 flasks were inoculated with a 1:750 dilution of virus stock in PBS. The flasks were incubated for 1 hour at 35 C in a humidified atmosphere containing 5% CO 2 and then washed with sterile PBS. Dulbecco modified Eagle medium (DMEM) supplemented with 1 g/mL TPCK ( L 1 tosylamido 2 phenylethyl chloromethyl ketone) treated trypsin (Sigma Aldrich Corp., St. Louis, MO, USA), 0.85% bovine serum albumin (BSA), HEPES buffer, and penicillin streptomycin was added to the flasks. The inoculated cells were incubated at 35 C in a humidified atmosphere containing 5% CO 2 for 3 to 5 days, with daily evaluation for cytopathic effect. The culture supernatant was harvested, clarified by centrifugation at 2000 rpm for 10 minutes, and the amount of virus determined with the hemagglutina tion (HA) assay using 0.5 % turkey red blood cells ( Lampire Biological Laboratories, Pipersville, PA, USA) in PBS as previously described. 20 The viral culture supernatant was stored at 80 C until future use. Ampli fication and Cloning of the NS1 Gene Total RNA was extracted from viral culture supernatant using the QIAamp Viral RNA Mini Kit ( QIAGEN Inc., Valencia, CA, USA) instructions. The RNA was reverse transcribed using SuperScript III Reverse Transcriptase (Invitrogen Corp., Carlsbad, CA, USA) and the cDNA amplified by PCR reactions using forward and reverse primers for the CN /FL/04 NS1 gene and Platinum Taq DNA Polymerase High Fidelity (Invitrogen Corp.) The 5' end of the NS1 fo rward primer contained the NcoI restriction enzyme cleavage site, while the 5' end of the reverse primer contained the XhoI restriction site (Table 4 1) to facilitate insertion of the NS1 amplicon into Exon 2 of the (Novagen EMD Chemi cals Inc.,

PAGE 69

69 Gibbstown, NJ, USA) (Figure 4 1). This vector codes for the fusion of a histidine tag on the 3' end of the cloned gene to facilitate purification of the expressed protein. The PCR product was run on a 1.0% agarose gel in Tris borate EDTA (TBE) b uffer with an exACTGene low range plus DNA ladder (Thermo Fisher Scientific Inc., Waltham, MA, USA). The NS1 PCR product was purified using the QIAquick PCR Purification Kit (QIAGEN Inc.) t and 1.1 vector were double digested separately with 20U NcoI and 40U XhoI in NE Buffer #3 (New England Biolabs, Ipswich, MA, USA) at 37C for 2 hours. The digested 1.1 vector was heat inactivated at 65 C for 20 minutes and dephosphorylate d with 5U calf intestinal alkaline phosphatase (New England Biolabs) at 37 C for 1 hour These products were run on a 1.0% agarose gel in TBE buffer with a HindIII Digest ladder (New England Biolabs) and an exACTGene mid range plus DNA ladder (Therm 1.1 vector were extracted from the gel using a QIAquick Gel Extraction Kit (QIAGEN Inc.) according to and were run on 1.0% agarose gels in TBE. The NS1 insert 1 .1 vector were ligated at 1:1, 1:5, and 1:7 molar ratios with 1L T4 DNA Ligase (New England Biolabs). The ligation reactions were incubated at room temperature (23 C) for 2 hours and then heat inactivated at 65 C for 5 minutes Ligation products were run on 1.0% agarose gels in TBE. (Novagen ) were transformed with the NS1 plated on Luria Bertani (LB) agar plates containing 50 g/mL ampic illin and incubated at

PAGE 70

70 37 C for 16 hours. Twenty colonies were selected and cultured overnight in LB broth supplemented with 1 g/mL ampicillin at 37 C in a shaking incubator at 250 rpm. Plasmid DNA mini preps were prepared from six cultures using the QIAp rep Spin Miniprep Kit (QIAGEN Inc.) digestion of the mini prep products was performed using the NcoI and XhoI restriction enzymes and the products were run on 1.0% agarose gels in TBE to verify the presence of the NS1 insert. Two of the plasmid DNA mini preps were sequenced by the University of Florida Interdisciplinary Center for Biotechnology Research (UF ICBR) sequencing core using the NS1 forward and reverse primers. The sequences were compared to that of the CN/FL/04 NS1 gene ( Gen Bank A ccession # DQ124153) using the National Center for Biotechnology Information Basic Local Alignment Search T ool (NCBI/BLAST, Bethesda, MD, USA) to verify proper orientation and lack of mutations. The sequences were also en tered into the Swiss Institute of Bioinformatics ExPASy translate tool (Lausanne, Switzerland) to verify that the amino acid codons and histidine tag on the carboxy terminus of the recombinant protein were in frame for proper translation. Expression and Pu rification of the NS1 Protein (Novagen ) were transformed with mini containing 50 g/mL ampicillin and 34 g/mL chloramphenicol. The bacteri a were Medium (Novagen ) supplemented with 50 g/mL ampicillin and 34 g/mL chloramphenicol at 37 C in a shaking incubator at 250 rpm for 24 hours For the first expression culture, 1 mL samples were collected at the initial time point (0) and then

PAGE 71

71 every two hours from 12 to 24 hours to monitor total protein production over time. The samples were centrifuged at 13,000 rpm for 5 minutes and the pellets lysed with BugBuster P rotein Extraction Reagent supplemented with benzonase nuclease, rlysozyme, and protease inhibitors (Novagen ) The total protein was analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) a s described below. Expression cultures were subsequently harvested at 24 hours by centrifugation at 3500 rpm for 30 minutes and the pellets lysed with BugBuster Protein Extraction Reagent. Attempts were made to purify recombinant NS1 protein from the sol uble fraction of the lysed culture pellets using nickel NTA (Ni NTA) His Bind resin, chromatography columns, and a Ni NTA buffer kit (Novagen ) according to Larger quantities of recombinant NS1 protein were purified from the insoluble fraction (inclusion bodies) of lysed pellets using a large scale purification process conducted under denaturing conditions as follows. Culture pellets were resuspended in lysis buffer containing 8M urea, 50mM Tris HCl, 2 mM mercaptoethanol, 1M NaCl, and 20% glycerol (final pH 8.0), and incubated at 4 C for 24 48 hours. The lysate was sonicated on ice using an Ultrasonics W225R sonicator set at 70% maximum output three to five times for 30 60 secon d intervals, then centrifuged in a J2 21 Beckman centrif u g e with a JA 20 rotor at 19,000 rpm for 1.5 hours. The resultant supernatant was passed through a 0.45m sterile filter to remove any particulate material and added dropwise to a renaturing wash buff er (wash buffer 1) containing 50mM NaH 2 PO 4 300mM NaCl, and 20mM imidazole (final pH 8.0). The renatured lysate was filter sterilized and loaded onto

PAGE 72

72 Ni NTA resin ( QIAGEN Inc. ) in a Bio Rad Laboratories Inc. (Hercules, CA, USA) column (10 by 2.5 cm) attach ed to an Econo line ( Bio Rad Laboratories Inc. ) gradient purification pump system set at 0.65 mL/min After flow through of the lysate, the column was washed with three volumes of buffer (wash buffer 2) containing 50mM NaH 2 PO 4 300mM NaCl, 20mM imidazole, and 2% Triton X 114 (final pH 8.0), followed by five volumes of wash buffer 1 The recombinant NS1 protein was eluted in fractions with an elution buffer containing 50mM NaH 2 PO 4 300mM NaCl, and 250mM imidazole (final pH 8.0). Fractions were analyzed by SD S PAGE and western blot as described below. Fractions containing the recombinant NS1 protein were pooled, filter sterilized, and dialyzed with wash buffer 1 to remove imidazole. The purified NS1 protein was concentrated by ultrafiltration in an Amicon stir red cell using a 10 kDal molecular weight cut off membrane (Millipore, Billerica, MA, USA), exchanging the buffer for PBS (pH 7.4). The protein concentration was determined using a Quick Start Bradford Protein Assay (Bio Rad Labo ratories Inc. ) according t was analyzed by western blot as described below and by mass spectrometry at the UF ICBR to assess purity. SDS PAGE and Western Blot Assays for NS1 Protein Analysis Total protein from expression cultures or purifie d NS1 protein was denatured in Rad Laboratories Inc.) supplemented with C for 5 minutes. Extracted total protein from untransformed Rosetta 2 E. coli (Novagen ) served as a negative control for protein expression gels. Total protein samples (5 L/well) or purified NS1 protein (0.5 to 1 g/well) Standards (Bio Rad Laboratories Inc.) were run on pre cast 12% Tris HCl R eady Gels

PAGE 73

73 using Tris Glycine SDS (TGS) buffer (Bio Rad Laboratories Inc.) at 200V for 35 minutes. Proteins were visualized by incubating gels with Bio Western blots were performed to evaluate the antigenicity of the purified NS1 protein. Unstained SDS PAGE gels were transferred to Immun Blot PVDF membranes at 110V for 70 minutes using Tris Glycine (TG) buffer (Bio Rad Laboratories Inc.). Blots were blocked with 5% (wt/vo l) nonfat dry milk in PBS for 2 hours at 23 C, and washed three times with wash buffer (150mM NaCl in sterile water with 0.05% Tween 20). Blots were incubated for 1 hour at 23 C with an affinity purified goat IgG anti influenza A NS1 antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) diluted 1:200 in PBS. Following three washes with wash buffer blots were incubated for 1 hour at 23 C with a mouse anti goat IgG antibody conjugated to horseradish peroxidase (HRP) (Santa Cruz Biotechnology Inc.) d iluted 1:10,000 in PBS. Foll owing three washes with wash buffer, antigen antibody reactions were visualized by adding 3, 3', 5, 5' tetramethylbenzidine ( TMB) peroxidase substrate with membrane enhancer (Kirkegaard & Perry Laboratories Inc., Gaithersburg, instructions. Blots were rinsed in water to stop color development. Synthetic NS1 Peptide A synthetic H3N8 CIV NS1 peptide containing 16 amino acids (NS1 34 49 = DRLRRDQKSLRGRGST) was synthesized by RS Synthesis (Louisv ille, KY, USA). This sequence is conserved between the H3N8 CIV isolates and is comparable to a NS1(B) avian influenza sequence determined by another study to be hydrophilic and antigenic. 124 This peptide was used in an ELISA as described below.

PAGE 74

74 Serum Samples Sera from 20 specific pathogen free (SPF) dogs involved in a H3N8 CIV vaccine trial were kindly provided by Dr. N. Lakshmanan ( Intervet Schering P lough Animal Health, Millsboro, DE USA). These included sera fr om dogs that were vaccinated and subsequent ly challenged with H3N8 CIV (n= 10), and dogs that were unvaccinated and experiment ally infected with H3N8 CIV (n= 10). Sera were collected from the vaccinated and challenged dogs prior to vaccination (day 0), 21 da ys after the first vaccination, 14 days after the second vaccination, and 10 days after challenge with H3N8 CIV; therefore, 40 samples were available for testing from these 10 dogs. Sera from unvaccinated dogs experimentally infected with H3N8 CIV were col lected 14 to 18 days after infection. Serum samples from uninfected dogs (n=9) and dogs naturally infected with H3N8 CIV (n=59) were also included in the study. Samples from infected dogs included acute phase sera (<10 days postinfection; n=18 dogs) and c onvalescent phase sera (>21 days postinfection; n=41 dogs). ELISA for Detection of NS1 Antibodies in Canine Sera The NS1 ELISA was initially tested with serial dilutions of the recombinant NS1 protein and goat anti NS1 antibody. The wells in EIA/RIA flat bottom microtiter plates ( Corning Costar Cambridge, MA, USA ) were coated with two fold serial dilutions of the recombinant NS1 protein in 0.05M carbonate bicarbonate coating buffer (pH 9.6) resulting in protein concentrations rangi ng from 4 g/well to 3 2 ng/well A no antigen control column was included in each plate. Plates were sealed and incubated overnight at 4 C, washed three times with wash buffer (150mM NaCl in sterile water with 0.05% Tween 20), and blocked with PBS containing 5% nonfat dry milk for 1 hour at 23 C.

PAGE 75

75 After washing three times, a 1:200 dilution of affinity purified polyclonal goat IgG anti influenza A NS1 antibody (Santa Cruz Biotechnology Inc ) in PBS/5% milk was added to the wells. A no antibody control row was included in the pla te. Plates were incubated at 23 C for 1 hour. After washing eight times, HRP conjugated mouse anti goat IgG antibody (Santa Cruz Biotechnology Inc.) diluted 1:10,000 in PBS/5% milk was added to the wells. Following incubation at 23 C for 1 hour, the plates were washed eight times, instructions for approximately 30 minutes at 23 C. Reactions were stopped with TMB stop solution ( Kirkegaard & Perry Laboratories Inc.). Endpoint absorbance values were read at 450 nm in an automatic microtiter plate reader. For checkerboard titrations with canine sera, microtiter wells were coated with two fold serial dilutions of the recombinant NS1 protein in coating buffer with NS1 protein con centrations ranging from 4 g/well to 4 ng/well. A no antigen control colu mn was included in each plate. Canine sera were two fold serially diluted in PBS/5% milk in the wells, starting at a dilution of 1:200. A no serum control row was included in each pl ate. Affinity purified HRP conjugated sheep anti dog IgG heavy chain antibody (Bethyl Laboratories Inc., Montgomery, TX, USA) was diluted 1:10,000 in PBS/5% milk and added to the wells. Plates were developed and read as previously described. An ELISA was p erformed with the synthetic NS1 34 49 peptide as described above with the following modifications. Wells were coated with 1 or 2 g of the NS1 peptide diluted in coating buffer. Canine sera were diluted 1:100 or 1:400 (tested in duplicate) and the HRP conj ugated sheep anti dog IgG heavy chain detection antibody was diluted 1:5,000 in SuperBlock (PBS) Blocking Buffer (Thermo Fisher Scientific Inc.).

PAGE 76

76 Western Blot Assays for Detection of NS1 Antibodies in Canine Sera The reactivity of canine sera to the H3 N8 CIV NS1 protein was initially assessed in western blot format according to the general procedure described above, but optimized with the following modifications. The ideal NS1 concentration was 0.5 g to 1 g/well and proteins were transferred to 0.2 m nitrocellulose membranes (Bio Rad Laboratories Inc.). SuperBlock (PBS) Blocking Buffer was used for the blocking buffer to eliminate nonspecific background staining. Canine sera, positive and negative control antibodies, and detection antibodies were dil uted in SuperBlock with 0.05% Tween 20. Blots were incubated with a 1:200 dilution of canine serum, a 1:10,000 dilution of HRP conjugated sheep anti dog IgG heavy chain antibody or a 1:10,000 dilution of HRP conjugated goat anti dog IgG Fc antibody (Life Span Biosciences Inc., Seattle, WA, USA) The positive control combination was goat anti influenza NS1 antibody and HRP conjugated mouse anti goat IgG. The negative control combination was polyclonal goat anti Histone H2B IgG antibody (Santa Cruz Biotechno logy Inc.) and HRP conjugated mouse anti goat IgG. Blots were developed with TMB peroxidase substrate and membrane enhancer as previously described. Western blots were also performed with canine sera that had been pretreated by one of two methods: 1) incub ation with 1% (wt/vol) E. coli acetone powder (Sigma Aldrich Inc.) at 4 C for 30 minutes, and centrifuged at 13,000 rpm for 10 minutes; 40 or 2) preadsorption with E. coli acetone powder as described followed by incubation with receptor destroying enzyme (RDE; Denka Seiken, Tokyo, Japan) (1 part serum: 3 parts RDE) for 16 hours at 37 C and heat inactivation for 30 minutes at 56 C. To test large pa nels of sera by western blot, the recombinant NS1 protein (0.5 g/well) and Precision Plu electrophoresed on 4 12%

PAGE 77

77 NuPAGE Novex precast gels ( Invitrogen Corp.) ac instructions. Gels were transferr ed via the i Blot Dry Blotting System to 0.2 m nitrocellulose membranes ( Invitrogen Corp.) Blots were blocked with SuperBlock for 1 hour at 23 C and washed three times with wash buffer. The remaining incubations and wash steps were as previously described. Canine sera and positive and negative control antibodies were diluted 1:500 in SuperBlock with 0.05% Tween 20. Secon dary detection antibodies (HRP conjugated sheep anti dog IgG Fc for canine sera and HRP conju gated mouse anti goat IgG for controls) were diluted 1:5,000 in SuperBlock with 0.05% Tween 20. Blots were developed with TMB peroxidase substrate and membrane enhancer as previously described. Results NS1 Protein Expression and Purification Restriction enzyme digestion of six plasmid DNA mini preps indicated that all contained the ~650bp H3N8 CIV NS1 gene insert (Fi gure 4 2). Sequence analysis of two plasmid DNA mini preps conf irmed that the NS1 gene was in proper orientation in both samples and that no mutations had occurred The amino acid codons and histidine tag were also in frame for proper translation. Recombinant NS1 protein expression was detectable at 12 hours and peaked by 18 hours (Figure 4 3). Initial attempts to purify substantial amounts o f recombinant NS1 protein from the soluble fraction of the protein extract were unsuccessful (Figure 4 4). Most of the recombinant protein was present in inclusion bodies in the insoluble fraction (Figures 4 4 and 4 5); therefore, large scale protein purif ication was performed under denaturing conditions. Once purified and refolded the antigenicity of the NS1 protein

PAGE 78

78 was verified by western blot ( Figure 4 6) and ELISA (Table 4 2) using an anti influenza A NS1 antibody. In both formats, the minimum amount of detectable NS1 protein was ~32 ng. Mass spectrometry indicated that the NS1 protein contained only one major contaminating E. coli protein, which had a molecular weight of 80 kDal. Since this protein was much larger than the NS1 protein (25 kDal), it was not considered a concern for immunodiagnostic testing. ELISA to Detect NS1 Antibodies in Canine Sera In the NS1 ELISA using goat anti NS1 as the primary antibody, there was a linear decline in absorbance values when the recombinant NS1 protein concentrati on decreased from 500 ng/well to 32 ng/well (Table 4 2) Checkerboard titrations with recombinant NS1 protein and canine sera revealed false positive reactivity with serum from uninfected/unvaccinated dogs (Table 4 3). Substitution of the secondary HRP con jugated sheep anti dog IgG heavy chain antibody with an alkaline phosphatase conjugated anti dog IgG heavy chain antibody did not resolve the false positive reactions (data not shown). A sandwich ELISA using anti influenza NS1 antibody as the coating agent followed by addition of NS1 protein to capture NS1 antibody in canine sera also did not eliminate false positive reactions with serum from uninfected/unvaccinated dogs (data not shown). To address poten tial binding of canine antibodies to contaminating E. coli proteins in the NS1 preparation as a cause of false positive reactions, the ELISA was performed with a synthetic 16 amino acid NS1 peptide (NS1 34 49 ) comparable to a NS1(B) avian influenza peptide determined by another study to be a major antibody bi nding region of NS1 proteins. 124 Use of the synthetic peptide did not resolve the false positive reactions for sera from uninfected/unvaccinated dogs (Table 4 4).

PAGE 79

79 Western Blot Assays to Detect NS1 Antibodies in Ca nine Sera When the optimized western blot was performed with canine sera, staining of the NS1 band was consistently observed with sera from uninfected/unvaccinated dogs. Substitution of anti dog IgG heavy chain with anti dog IgG Fc as the secondary antibod y did not eliminate false reactivity with uninfected/unvaccinated dog serum. Importantly, when the western blots were performed with only the secondary antibody (i.e., no serum present), no NS1 bands were observed, indicating that the false positive result s were likely attributable to a naturally occurring IgG antibody in canine serum that cross reacts with NS1 protein or contaminating E. coli protein. This was addressed by two serum pretreatment methods: preadsorption of sera with E. coli acetone powder an d preadsorption pl us treatment with RDE and heat inactivation. Neither serum treatment method resolved the false positive results for sera from uninfected/unvaccinated dogs. To determine if nonspecific antibody reactivity with NS1 was a variable or univer sal phenomenon among dogs, a lar ge panel of serum samples (n= 1 1 8 ) from 88 dogs with known H3N8 CIV infection or vaccination status was tested in the NS1 western blot in a blinded fashion. This panel included sera from 9 uninfected/unvaccinated SPF dogs, ac ute phase sera from 18 naturally infected dogs, convalescent sera from 41 naturally infected dogs, sera collected from 10 SPF dogs at different time points after H3N8 CIV vaccination and subsequent viral challenge, and sera from 10 SPF dogs 14 to 18 days a fter experimental H3N8 CIV infection. Regardless of infection or vaccination st atus, samples from all dogs appeared seropositive for NS1 antibody (Figures 4 7 and 4 8).

PAGE 80

80 Discussion T he H3N8 CIV recombinant NS1 protein was successfully expressed in E. coli and purified for diagnostic immunoassays. The identity and purity of the recombinant NS1 protein was verified through sequencing and mass spectrometry analysis. The protein antigenicity was demonstrated in both ELISA and western blot formats using an anti influenza A NS1 antibody. However, ELISA and western blot assays using the recombinant NS1 protein to capture specific antibodies in sera from H3N8 CIV infected dogs lacked specificity based on false positive reactivity in sera from uninfected or unvaccin ated dogs and dogs vaccinated with the H3N8 CIV whole virus vaccine. Therefore, the NS1 ELISA and western blot evaluated in this study cannot differentiate between uninfected, infected, and vaccinated dogs. This finding is similar to that of another study conducted with sera from uninfected, infected, and vaccinated dogs tested in a NS1 western blot ( Dr. N. Lakshmanan, Intervet Schering Plough Animal Health, Millsboro, DE USA, personal communication). There are several possibilities for the lack of differ entiation between uninfected or vaccinated dogs and H3N8 CIV infected dogs by the immunoassays for NS1 antibody. The NS1 ELISA and western blots performed with secondary anti dog IgG antibody in the absence of canine serum were negative, demonstrating the necessity for canine IgG for seropositivity. Naturally occurring antibodies to ubiquitous E. coli may have recognized trace E. coli contamination in the NS1 protein preparation. This was c onsidered unlikely however, because pread sorption of canine sera wit h E. coli acetone powder and use of the synthetic NS1 peptide in the ELISA did not eliminate seroreactivity for uninfected or vaccinated dogs.

PAGE 81

81 Cross reactivity with avian influenza NS1 protein has been reported for sera from chickens infected with infecti ous bursal disease virus, avian reovirus, avian pneumovirus, and other avian pathogens. 127 Increased seroreactivity to NS1 protein was also reported for chickens in influenza free commercial flocks compared to SPF chickens, suggesting potential cross reactive antibodies in the commercial poultry. 127 Similar to chickens, antibodies to common canine pathogens induced by exposure or vaccination may have cross reactivity with ca nine influenza NS1 protein. In addition to H3N8 CIV, other pathogens associated with acute respiratory dise ase in dogs include canine distemper virus, canine adenovirus type 2, canine parainfluenza virus, canine respiratory coronavirus, canine herpesvirus, and the recently described canine pneumovirus. 28 Based on a BLAST search excluding influenza A NS1 proteins, n o proteins from these viruses or other pathogens (including canine parvovirus) were i dentified as closely related to NS1. Instead, a protein from the bay scallop (Gen Bank A cce ssion # EU623917.1) was most closely relate d to influenza A NS1 protein. T he significance of this similarity is unknown It is also uncertain if some degree of cross reactivity could still occur with the above listed canine pathogens despite their proteins not being closely related to the NS1 protein. It is important to note, h owever, that none of the uninfected SPF dogs had been exposed to or vaccinated against these pathogens yet their sera contained cross reactive antibodies to the NS1 protein in the imm unoassays. Although the histidine tag on the carbo xy terminus of the NS1 recombinant protein was valuable for purification, a peptide formed by union of the histidine residues with the NS1 protein may have been recognized by ubiquitous antibodies in canin e serum that

PAGE 82

82 are not specific for the recombinant protein. False positive r esults have been noted with histidine tagged recombinant proteins in ELISA testing by others ( Dr. W. Dundon, Istituto Zooprofilattico Sperimentale delle Venezie, Legnaro, Italy, per sonal communication). Identification of cross reactive antibodies attributable to the h is tidine tag on recombinant NS1 protein would require production of recombinant protein without the tag or removal of the tag after purification. However, the possibilit y of cr oss reactive antibodies to the h is tidine tag seems unlikely since use of a synthetic NS1 peptide in the ELISA did not eliminate seroreactivity in uninfected or vaccinated dogs. The inability of the NS1 immunoassays to differentiate uninfected or vac cinated dogs from infected dogs may also be due to the possibility that infected dogs produce very low amounts of antibody to the H3N8 CIV NS1 protein that are not detectable in the face of nonspecific reactions. Variability in NS1 antibody production has been reported between species and between influenza viruses. 5,10,31,125,127 Antibodies to the NS1 protein or its peptides are not produced in all influenza infected animals, and the emergence and persistence of NS1 antibodies can vary considerably. In studies with other species, antibodies to NS1 protein typically appeared early after infection and persisted for at least 1 to 5 weeks, but the amounts of antibody were relatively low. 5,10,31,54,94,124,125,127,143 Although the appearance and persistence of NS1 antibodies in influenza infected dogs is unknown, sera tested in this study were collected from naturally and experimentally infected dogs at several time points post infe ction, includin g <1 week, 14 to 18 days, and 3 to 4 weeks. The NS1 seroreactivity of these samples was indistinguishable from that obtained with samples from uninfected or

PAGE 83

83 vaccinated dogs, preventing any conclusions regarding the emergence and persistence of NS1 antibodies in canine sera. Overall, the specificity issues encountered in this study could not be resolved and supported the conclusion that an influenza A NS1 DIVA test is not a viable or reliable diagnostic strategy for differentiating dogs infec ted with H3N8 CIV from uninfected dogs or vaccinated dogs. Since a DIVA test strategy is necessary for correct identification of infected dogs in the face of vaccination future work should focus on development of alternative DIVA strategies. A successful strategy will not only allow correct identification of infected dogs with unknown vaccination history, but will also enable serological evaluation of vaccine efficacy in the field as well as future epidemiological studies of canine influenza.

PAGE 84

84 Table 4 1. Primers for PCR amplification and cloning of the H3N8 CIV NS1 gene. The primers contain clamp sequences for polymerase fidelity and restriction enzyme sites for gene ligation into the plasmid vector. Primer Target Sequence NS1 forward NS1 (1 26) CTC CTCCCATG CTCCTC = clamp CCATGG = NcoI restriction site NS1 reverse NS1 (657 627) CTCCTCCTCGAG CTCCTC = clamp CTCGAG = XhoI restriction site Figure 4 ) plasmid for cl oning of the H3N8 CIV NS1 gene. The gene was inserted into exon 2 (yellow box) at restriction enzyme sites for NcoI and XhoI.

PAGE 85

85 Figure 4 2. SDS PAGE electrophoresis of plasmid preparations (1, 2, 3, 18, 19, 20) digested with NcoI and XhoI restriction enz ymes to release the NS1 gene NS1 = Purified NS1 PCR product, pTriEx = ) vector DNA. Figure 4 3. SDS PAGE electrophoresis of total protein extracted from re combinant H3N8 CIV NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) and stained with Coomassie blue. To assess peak NS1 protein expression, s amples were collected from transformed cell cultures at the time of inoculation (0) and every two hours from 1 2 to 24 hours after autoinduction. Expression of the recombinant NS1 protein (MW=25 kDal) was apparent at 12 hours. E. coli = untransformed cells. MW = molecular weight marker.

PAGE 86

86 Figure 4 4. SDS PAGE electrophoresis of soluble and insoluble fractions of a lysed recombinant NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) cell pellet an d stained with Coomassie blue. E. coli = untransformed cells. SF = soluble fraction. SE = purified elution from soluble fraction. IW(1 4) = washes 1 4 of insoluble frac tion. IB = resuspended inclusion bodies. FW = final insoluble fraction wash. MW = molecular weight marker. Small amounts of recombinant NS1 protein are detected in the insoluble fraction washes; however, the majority is in the resuspended inclusion body fr action (IB).

PAGE 87

87 Figure 4 5. Western blot of soluble and insoluble fractions of a lysed recombinant NS1 plasmid transformed Rosetta 2 E. coli (Novagen ) cell pellet and stained with an affinity purified antibody to influenza NS1 protein. E. coli = un transformed cells. SF = soluble fraction. SE = purified elution from soluble fraction. IW(1 4) = washes 1 4 of insoluble fraction. IB = resuspended inclusion bodies. FW = final insoluble fraction wash. MW = molecular weight marker. Small amounts of recombi nant NS1 protein are detected in the insoluble fraction washes; however, the majority is in the resuspended inclusion body fraction (IB).

PAGE 88

88 Figure 4 6. Western blot of purified H3N8 CIV NS1 protein detected using goat anti influenza NS1 IgG antibody and a HRP conjugated mouse anti goat IgG antibody. Lanes 2 through 10 contain serial two fold dilutions of the purified protein representing from 1 g in lane 2 to 4ng in lane 10. The protein is detectable at 16 ng (lane 8).

PAGE 89

89 Table 4 2. Checkerboard titrat ion results for the H3N8 NS1 antibody ELISA. Serial two fold dilutions of recombinant NS1 protein were incubated with goat anti influenza NS1 antibody (1:200) followed by HRP conjugated mouse anti goat IgG (1:10,000). Absorbance values are reported for dup licate wells for each NS1 concentration. H3N8 CIV NS1 protein concentration 4 g 2 g 1 g 500 ng 250 ng 125 ng 62 ng 32 ng 0 3.8 3.8 3.8 3.5 2.6 1.5 0.7 0.4 0.1 3.9 3.9 3.9 3.7 3.1 1.8 1.0 0.5 0.1

PAGE 90

90 Table 4 3. Checkerboard titration results for th e H3N8 NS1 antibody ELISA. Serial two fold dilutions of uninfected and H3N8 CIV infected dog sera were incubated with varying NS1 protein concentrations followed by addition of HRP conjugated anti dog IgG heavy chain antibody diluted 1:10,000. Absorbance v alues are reported. H3N8 CIV NS1 protein concentration Serum 4 g 2 g 1 g 500 ng 250 ng 125 ng 62 ng 32 ng 16 ng 8 ng 4 ng 0 Uninfected 1:200 3.5 3.3 3.2 2.8 2.1 1.6 1.2 0.9 0.8 0.8 0.8 0.8 1:400 2.9 3.1 2.7 2.4 1.7 1.1 0.8 0.7 0.6 0. 5 0.5 0.5 1:800 2.7 2.5 2.7 1.7 1.2 0.8 0.6 0.5 0.3 0.3 0.3 0.3 1:1600 2.1 1.8 2.2 1.6 1.0 0.5 0.4 0.2 0.2 0.2 0.2 0.2 0 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 H3N8 CIV infected 1:200 2.1 2.2 1.6 1.1 0.9 0.7 0.6 0.5 0.5 0 .5 0.5 0.5 1:400 1.3 1.2 0.9 0.7 0.5 0.4 0.4 0.4 0.4 0.3 0.3 0.3 1:800 0.7 0.6 0.5 0.4 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.2 1:1600 0.4 0.4 0.3 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0.1 0.1 0 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

PAGE 91

91 Table 4 4. The H3N8 NS1 antibody ELISA was performed with varying concentrations of the synthetic NS1 34 49 peptide incubated with uninfected and H3N8 CIV infected dog sera followed by addition of HRP conjugated anti dog IgG h eavy chain antibody (1:5,000). Sera were tested in dupl icate, and a verage a bsorbance values are reported. NS1 peptide concentration Serum 2 g 1 g Uninfected 1:100 1.72 1.84 1:400 1.17 1.16 H3N8 CIV infected 1:100 1.60 1.47 1:400 1.30 1.01

PAGE 92

92 Figure 4 7. H3N8 CIV NS1 western blot with sera from uninfected SPF dogs and dogs experimentally infected with H3N8 CIV. NS1 protein concentration = 0.5 g per well. Canine sera and controls diluted 1:500. HRP conjugated anti goat I gG and HRP conjugated anti dog IgG Fc secondary antibodies diluted 1:5000. L anes: 1 = molecular weight marker; 2 = goat anti influenza NS1 antibody (positive control); 3 = goat anti histone antibody (negative control); 4, 6, 8, 10 = sera from SPF dogs; 5, 7, 9 = sera from experimentally infected dogs collected 14 days postinfectio n. Figure 4 8. H3N8 CIV NS1 western blot with sera from uninfected, infected, and vaccinated dogs. NS1 protein concentration = 0.5 g per well. Canine sera and controls diluted 1:500. HRP conjugated anti goat IgG and HRP conjugated anti dog IgG Fc sec ondary antibodies diluted 1:5000. Lanes: 1 = molecular weight marker; 2 = goat anti influenza NS1 antibody (positive control); 3 = goat anti histone antibody (negative control); 4 = uninfected SPF dog; 5, 6 = acute and convalescent sera from a dog infected with H3N8 CIV; 7 = dog vaccinated with one dose of H3N8 CIV vaccine; 8 = dog vaccinated with two doses of H3N8 CIV vaccine; 9 = vaccinated dog challenged with H3N8 CIV; 10 = dog experimentally infected with H3N8 CIV.

PAGE 93

93 CHAPTER 5 SEROLOGICAL EVIDENCE OF H 3N8 CANINE INFLUENZA LIKE VIRUS CIRCULATION IN U.S. DOGS PRIOR TO 2004 Canine influenza A virus subtype H3N8 (H3N8 CIV) was identified as a respiratory pathogen for dogs in 2004. 20 The virus was first isolated fr om racing greyhounds during a respiratory disease outbreak at a Florida track, and was subsequently identified as the cause of multiple respiratory disease outbreaks among tho usands of racing greyhounds in multiple states through 2006. 20,141 In 2005, dogs in Florida shelters were the first non greyhound cases of canine influenza to be confirmed by virus isolation, although serological evidence of canine influenza in non greyhound dogs had been reported in 2004. 20,98 Ongoing surveillance of U.S. dogs with influenza like illness (ILI) indicates widespread geographic distribution of H3N8 CIV infections in 34 sta tes and the virus has become endemic in metropolitan areas of the n ortheastern and western regions of the country. 8,9,20,28,29,42,44,98,103,110,111 Molecular and phylogenetic analyses of H3N8 CIV isolates recovered from dogs from 2003 to 2008 support the interspecies transfer of an entire H3N8 equine influenza A virus (H3N8 EIV) from the horse to the dog; 20,42,98,103 however, it is unknown when this transmission occurred or what population(s) of dogs were originally affected. Prior to the discovery of H3N8 CIV in 2004, f requent outbreaks of unknown etiology had occurred in racing greyhounds throughout the U.S. in the 1990s and 2003. 20,28,29 Limited testing of ar chived tissues and serum samples from racing greyhounds has suggested that a H3N8 CIV like virus may have circulated in this population as early as 2000. 20 Archived sera collected from 2000 to 2003 from 65 racing g reyhounds were tested in the hemagglutination inhibition (HI) assay with a 2004 isolate of H3N8 CIV and 9 dogs were seropositive indicating previous virus exposure. 20

PAGE 94

94 Further, e xamination of archived lung tissue fr om a greyhound that died of hemorrhagic bronchopneumonia during a respiratory disease outbreak at a Florida track in March 2003 yielded the earliest known H3N8 canine influenza isolate (A/canine/Florida/242 / 20 03). 20 Information on the prevalence of H3 antibodies in pet and shelter dogs prior to 2004 is also limited. To our knowledge, only one study has tested pet dog sera collected prior to 2004 for antib odies to CIV H3; and a ll 75 dogs from this Colorado study were seronegative. 8 The objective s of this study were 1) to determine the prevalence of CIV and EIV H3 antibodies in racing greyhound and shelter dog sera collected from 1984 to 2004 to further investigate when H3N8 viruses first infected dogs and 2) to analyze potential risk factors asso ciated with seropositivity This re trospective serosurvey provides additional information about the emergence of H3N8 CIV in the canine population. Materials and Methods Study Design A retrospective serosurvey of convenience samples collecte d from racing greyhounds and dogs housed in anima l shelters from 1984 to 2004 was conducted. Serum Samples Archived serum samples from 702 racing greyhounds were tested for antibodies to CIV and EIV H3 proteins. One set of samples (n=153 dogs) was collected in 1984 and 1985 from greyhounds located at tr acks and farms in north Florida These samples were originally collected for a heartworm antigen test study Another set of samples (n=549 dogs) was collected from 1999 to 2004 from greyhounds in multiple states as they came off the tracks and entered a co mmercial blood banking program ( Hemopet/Pet Life Line, Garden Grove, CA, USA). Available history for racing

PAGE 95

95 greyhounds was recorded using unique ear tattoos and a corresponding database housed at www.greyhound data.com. Serum samples from shelter dogs (n=1 ,568) were collected on entry into north Florida animal shelters from 1984 to 2004. Viruses Madin Darby canine kidney (MDCK) cell cultures were inoculated with H3N8 A/ canine/Florida/43/2004 (CN/FL/04) and the culture supernatant stored at 80 C prior to u se. H3N8 A/ equine/New York/1999 (EQ/NY/99), A/ equine/Kentucky/1995 (EQ/KY/95), and A/ equine/Kentucky/1991 (EQ/KY/91) isolates were cultured in embryonated chicken eggs and the allantoic fluids stored at 80 C prior to use CN/FL/04 was kindly provided by D r. R. Donis (Centers for Disease Control and Prevention, Atlanta, GA, USA) EQ/KY/91 and EQ/KY/95 were kindly provided by Dr. T. Chambers (University of Kentucky, Lexington, KY, USA), and EQ/NY/99 was kindly provided by Dr. E. Dubovi (Cornell University, A nimal Health Diagnostic Center, Ithaca, NY, USA). HI Assays HI assays were conducted as previously described. 20 Briefly, serum samples were pretreated with receptor destroying enzyme (RDE, Denka Seiken, Tokyo, Jap an) followed by heating at 56C for 30 minutes to inactivate nonspecific inhibitors of hemagglutination. Serial two fold dilutions of each sample were incubated in 96 well V bottom microtiter plates with 4 hemagglutinating units/25L of test virus for 30 m inutes at room temperature (23 C), followed by an additional incubation for 30 minutes at 23 C with an equal volume of 0.5% turk ey red blood cells in PBS. All sera were tested with the CN/FL/04 isolate and one of the three equine isolates. H3N8 EIV isolat es were included in the HI assays to address the possibility

PAGE 96

96 that dogs were initially infected by these (or similar) viruses prior to viral adaptation resulting in the currently circulating H3N8 CIV. This rationale is supported by molecular evidence that H3N8 EIV isolates have undergone significant antigenic drift since the 1980s 15,22,61,62 and all H3N8 CIV isolates analyzed to date are most closely related to the Florida sublineage of EIV circulating in U.S. hor ses since 2000. 20,98,103 Additionally, H3 antibodies in dogs infected with CIV since 2004 have reduced reactivity with EIV isolates obtained before 1999. 98 Based on this ra tionale, the 1984 to 1985 greyhound samples were tested with CN/FL/04 and EQ/KY/91, and the 1999 to 2004 greyhound samples were tested with CN/FL/04 and EQ/NY/99. To facilitate testing the large number of shelter dog samples, untreated sera were initially screened at 1:10 and 1:40 dilutions in the HI assay. Samples collected from 1984 to 1992 were screened using CN/FL/04 and EQ/KY/91, the 1993 to 1998 samples were screened with CN/FL/04 and EQ/KY/95, and the 1999 to 2004 samples were screened with CN/FL/04 and EQ/NY/99. Sera that were positive in the screening HI assay s were subsequently treated with RDE and hea t inactivation and re tested as previously described All greyhound and shelter dog s era were tested in duplicate. Sera collected from specific pat hogen free dogs housed in a barrier research facility served as negative controls in all assays. Convalescent serum samples from racing greyhounds naturally infected with H3N8 CIV served as positive controls. Their seropositivity was previously confirmed b y both HI and serum microne utralization (MN) assays. 20,98 Endpoint H3 antibody titers were defined as the reciprocal of the last serum dilution that completely

PAGE 97

97 inhibited hemagglutination. Seropositive dogs were def ined as those with endpoint 20 Data Analysis HI test results and other recorded data for the greyhounds (n=702) and shelter dogs (n=1,568) were managed in standard softwa re (Excel Microsoft, Redmond, WA, USA). For the greyhounds sampled from 1999 to 2004, the prevalence of H3 seropositive dogs was calculated for each year by dividing the number of seropositive dogs by the total number of dogs tested that year X 100. In a ddition, geometric mean antibody titers (GMT) were calculated for both CIV and EIV for each of these years. The Mann Whitney rank sum test was used to test for differences between CIV and EIV H3 viruses were not included in subsequent analyses. The following categorical variables were recorded for each racing greyhound sam pled between 1999 and 2004 using information in the online database: age at time of (1995 to 2004); and, number of months raced (1 the categorical variables were calculated in SPSS 18.0 (IBM Somers, NY, USA), and the association of each with H3 seropositivity was evaluated by logistic regressi on in Egret 2.0 .3 (Cytel Inc., Cambridge, MA, USA). After matching H3 seropositive and seronegative greyhounds by the year they were sampled, conditional logistic regression was used to test the null hypotheses that prior racing history, age at time of te sting, number of months raced, and total number of tracks were not associated with H3 seropositivity. Unconditional logistic regression was used to test the null hypothesis that

PAGE 98

98 individual race years were not associated with H3 seropositivity. Track locati on could not be analyzed by logistic regression. Results were reported as odds ratios (OR) with the 95% confidence interval (95% CI). Variables with P values < 0.05 were considered significant. Similar information was not available for the 1984 to 1985 grey hounds or the shelter dogs. Results None of the 153 racing greyhounds in Florida in 1984 and 1985 had detectable antibodies to CN/FL/04 or EQ/KY/91 H3. Of the 1,568 dogs in north Florida shelters from 1984 to 2004, one dog that entered in 2004 had H3 anti bodies (HI titer = 128 for both CN/FL/04 and EQ/NY/99). For the 549 racing greyhounds sampled from 1999 to 2004, 133 (24.2%) had antibodies to both CIV and EIV H3 proteins The H3 seroprevalence and CIV/EIV H3 antibody GMTs with titer ranges are summarize d for each test year in Table 5 1. H3 seropositive greyhounds were identified in each year from 1999 to 2004. The EQ/NY/99 H3 antibody titers in greyhounds tested from 1999 to 2001 (n=64) were significantly ( P <0.01) higher than H3 antibodies to CN/FL/04 i ndicating the virus to which these dogs were actually exposed may have been more closely related to its equine ancestor These seropositive greyhounds raced at tracks in AR, AZ, CO, CT, FL, IA, TX, and WI. There was no difference ( P =0.94) between the H3 an tibody titers to CN/FL/04 and EQ/NY/99 for seropositive greyhounds tested in 2002 to 2004 (n=69). However, the CIV and EIV H3 antibody titers in these dogs were significantly ( P <0.01) higher than the titers fo r dogs tested in 1999 to 2001. The greyhounds t hat were seropositive in 2003 and 2004 raced at tracks in AR, AZ, and FL.

PAGE 99

99 Associations of several variables with H3 seropositivity in greyhounds sampled from 1999 to 2004 are listed in Tables 5 2 and 5 3. For the 481 greyhounds that had accessible histori cal data in the online database, greyhounds that raced at some point from 1995 to 2004 were more likely ( P =0.01) to be seropositive than dogs that did not race and dogs that were 3 years of age or older at the time of testing were more likely ( P <0.01) to b e seropositive th an younger dogs For the 303 greyhounds that had a racing history, dogs that raced 6 months or longer were more likely ( P =0.01) to be seropositive than dogs that raced for less time, and d ogs that raced on 2 or more different tracks were m ore likely ( P =0.01) to be seropositive than those that raced on 1 track. Finally, although many dogs raced in multiple years, greyhounds that raced in 1998, 2002, and 2003 were more likely ( P 0.01) to be seropositive than dogs that raced in other years. Discussion Based on serological evidence, a H3N8 canine influenza like virus was circulating in the racing greyhound population from 1999 to 2004. H3 antibodies were not detected in racing grey hounds in 1984 and 1985, but the tested population was small an d restricted to one geographic are no large serum banks for greyhounds before 1999 that are available for testing; therefore, H3N8 exp osure prior to 1999 cannot be ruled out. Similar to findings of a serosurvey on a small number of Colorado pet dogs, 8 H3 antibodies were not detected in Florida shelter d ogs from 1984 to 2003. However it cannot be concluded that pet and shelter dogs were not infected by H3N8 viruses before 2004, since the tested populat ions in both serosurveys wer e restricted to one geographic location.

PAGE 100

100 Other than in 2002, the annual prevalence estimates of H3 seropositivity among racing greyhounds tested from 1999 to 2004 ranged from 10.3% to 40.5% The seropositive greyhounds in 1999 to 2001 had higher H3 antibody t iters to a 1999 equine virus than a 2004 canine virus suggesting the possibility that these dogs were infected by an ancestral canine influenza virus closely related to EIV. Without virus isolates for molecular analysis, ho wever, it is not possible to determine whether these seropositive greyhounds were infected by a progenitor virus more closely related to its EIV ancestor than contemporary CIV isolates. The higher CIV H3 antibody titers in dogs tested in 2002 to 2004 sugge st these dogs were infected by a contemporary virus homologous to CN/FL/04. Based on the results of this study, greyhounds that raced were more likely to be exposed to H3N8 viruses than dogs that did not, and dogs that were 3 years of age or older at the t ime of testing were more likely to be seropositive for H3 antibodies than younger dogs. Further, among greyhounds that raced, dogs that raced at 2 or more tracks dogs that raced for 6 months or longer and dogs that raced in 1998, 2002, and 2003 were more likely to be exposed to H3N8 viruses. While the clinical histories of the seropositive greyhounds are unknown, and frequent movement of dogs prevented analysis of particular tracks or geographic locations as potential risk factor s for seropositivity, the in 1998 1999 82,84,86,87 and 2002 2003. 85 Necropsies performed on 3 dogs that died during the 1998 1999 epidemic showed hemorrhagic pneumonia possibly caused by an unknown viral infection. 83 In addition H3N8 CIV was isolated from archived tissue from a dog that died of hemorrhagic pneumonia during an undiagnosed respiratory epidemic

PAGE 101

101 in 2003 20 However, i t is unknown whether dogs tested in this study were involv ed in these reported outbreaks. Without additional virus isolates from archived tissues, the role of a H3N8 canine influenza like virus in these epidemics is admittedly speculative.

PAGE 102

102 Table 5 1. Annual prevalence of H3 antibodies among 520 ra cing greyhounds sampled between 1999 to 2004. Sera were tested in the HI assay with CN/FL/04 and idered seropositive. Test year No. seropositive No. seronegative Seroprevalence (%) CIV H3 GMT a (range) EIV H3 GMT (range) 1999 43 116 27.0 35 (32 128) 49 (32 128) 2000 17 44 27.9 43 (32 128) 52 (32 128) 2001 4 35 10.3 38 (32 64) 45 (32 64) 2002 1 73 1.4 64 64 2003 51 75 40.5 72 (32 256) 72 (32 128) 2004 17 44 27.9 92 (32 512) 89 (32 256) a GMT = geometric mean titer.

PAGE 103

103 Table 5 2. Univariable conditional logistic regression analyses of potential risk factors for H3 seropositivity among 481 racing g reyhounds sampled from 1999 to 2004. Variable No. seropositive (%) (n = 126) No. seronegative (%) (n = 355) OR a 95% CI b P value c Raced No 30 (23.8) 148 (41.7) 1.00 Reference NA Yes 96 (76.2) 207 (58.3) 1.82 (1.12, 2.95) 0.01 Age at testing 2 yrs 32 (25.4) 135 (38.0) 1.00 Reference NA 94 (74.6) 220 (62.0) 2.15 (1.29, 3.59) <0.01 Number months raced 1 5 50 (52.1) 148 (71.5) 1.00 Reference NA 46 (47.9) 59 (28.5) 1.88 (1.12, 3.16) 0.01 Total no. tracks raced 1 41 (42.7) 128 (61.8) 1.00 Reference NA 55 (57 .3) 79 (38.2) 1.89 (1.12, 3.21) 0.01 a OR = odds ratio. b CI = confidence interval. c P values < 0.05 were considered significant.

PAGE 104

104 Table 5 3. Comparison of prevalence estimates of H3 seropositive and seronegative greyhounds based on years raced. Vari able No. seropositive (%) (n = 96) No. seronegative (%) (n = 207) OR a 95% CI b P value c Race year 1996 No 95 (32.3 ) 199 ( 6 7.7 ) 1.00 Reference NA Yes 1 (11.1 ) 8 (88.9 ) 0.26 (0.03, 2.12) 0.20 1997 No 79 (32.4 ) 165 (67.6 ) 1.00 Reference NA Yes 17 (28.8 ) 42 (71.2 ) 0.84 (0.45, 1.57) 0.59 1998 No 62 (27.7 ) 162 (72 .3) 1.00 Reference NA Yes 34 (43.0 ) 45 (5 7 .0 ) 1.97 (1.15, 3.36) 0.01 1999 No 82 (31.8 ) 176 (68.2 ) 1.00 Reference NA Yes 14 (31.1 ) 31 (68.9 ) 0.96 (0.48, 1.91) 0.92 2000 No 90 (32.5 ) 187 (67.5 ) 1.00 Reference NA Yes 6 (23.1 ) 20 (76.9 ) 0.62 (0.24, 1.60) 0.32 2001 No 85 (32.6 ) 176 (67.4 ) 1.00 Reference NA Yes 11 (26.2 ) 31 (73.8 ) 0.73 (0.35, 1.53) 0.41 2002 No 62 (27.1 ) 167 (72.9 ) 1.00 Reference NA Yes 34 (45.9 ) 40 (54.1 ) 2.28 (1.33, 3.93) <0.01 2003 No 49 (25.1 ) 146 (74.9 ) 1.00 Reference NA Yes 47 (43.5 ) 61 (56.5 ) 2.29 (1.39, 3.78) <0.01 2004 No 85 (31.8 ) 182 (68.2 ) 1.00 Reference NA Yes 11 (30.6 ) 25 (69.4 ) 0.94 (0.44, 2.00) 0.87 a OR = odds ratio. b CI = confidence interval. c P values < 0.05 were considered significant.

PAGE 105

105 CHAPTER 6 PREVALENCE OF AND RI SK FACTORS FOR H3N8 CANINE INFLUENZA VIR US SEROPOSITIVITY IN A POPULATION OF U.S. D OGS WITH INFLUENZA LIKE ILLNESS Canine influenza A virus subtype H3N8 (H3N8 CIV) was first identified in 2004 as a novel respiratory pathogen for dogs. 20 Molecular analyses of H3N8 CIV isolates indicate prior transmission of a Florida sublineage H3N8 equine influe nza A virus from horses to dogs, followed by viral adaptation to the new canine host leading to efficient horizontal transmission and induction of influenza like illness ( ILI ) 20,42,98,103 The virus is now consider ed endemic in several U.S. communities and continues to adapt to dogs, such that H3N8 CIV is genetically and epidemiologically distinct from circulating equine viruses. 28,29,42,43,98,103 Similar to other influenza A viruses, H3N8 CIV can cause explosive epidemics of ILI upon introduction into group housing situations. Outbreaks have occurred in greyhound racetrack kennels, boarding kennels, shelter facilities, veterinary hospitals, and canine day care centers acros s the nation. 9,20,28,29,42 44,98,141 The incubation period for H3N8 CIV is 2 to 4 days, and serum antibodies to the H3 protein are detectable approximately 7 days after infection. 20,26,28,29 H3N8 CIV surveillance activities have revealed a widespread geographic distribution of infected pet and shelter dog s in 34 states ; 8,9,20,28,29,42,44,98,103,110,111 however, the virus is only cons idered endemic in metropolitan areas of Florida, New York, Colorado, and Pennsylvania. 28,29,103 Knowledge of the prevalence of H3N8 CIV in the U.S. and risk factors associated with virus exposure in dogs is limite d. In a study conducted in a university hospital in Iowa, 4 of 731 (0.5%) pet dogs with or without ILI were H3N8 CIV seropositive, indicating prior viral exposure. 110 In a study conducted in a university hospital i n

PAGE 106

106 Colorado, 9 of 250 (3.6%) pet dogs with or without ILI were seropositive for H3N8 CIV. 8 In a study of pet dogs participating in a flyball tournament in Pennsylvania, 3 of 100 (3%) dogs without ILI were seropositive for H3N8 CIV. 111 Another study conducted in a Philadelphia shelter found that 31 of 74 (42%) tested dogs with or wit hout ILI were H3N8 CIV seropositive. 44 In these studies, attending canine day care within the last six months, pet dogs living with other dogs, and the number of days housed in a shelter facility were associated with H3N8 CIV seropositivity. 8,44,111 It is currently unclear whether age, breed, sex, potential exposure settings, or geographic location are predisposing factors that can increase susceptibility to H3N8 CIV infection in dogs. Although previous studies have provided an epidemiolo gic framework for investigating risk factors associated with infection, these studies were limited to one geographic location and/or used small sample sizes, making the identification of risk factors associated with H3N8 CIV infection inconclusive. 8,44,111 The objectives of this study were to estimate the prevalence of H3N8 CIV seropositivity in a large population of dogs presenting with ILI in geographically diverse regions of the U.S., and to identify risk factors for seropositivity. Materials and Methods The Institutional Animal Care and Use Committees of both the University of Florida (UF) and Cornell University (CU) approved diagnostic testing for this study. Animals The study population included 1,268 pet a nd shelter dogs that were presented for evaluation of ILI (e g cough, nasal discharge, or pneumonia) in 42 states from 2005 through 2009.

PAGE 107

107 Study Design This investigation was a cross sectional study using a convenience sample of sera collected by veterina rians from pet and shelter dogs presenting with ILI from 2005 to 2009 and vol untarily submitted to the UF College of Veterinary Medicine and the CU A nimal H ealth D iagnostic C enter for serological diagnosis of H3N8 CIV infection. Laboratory Procedures The HI assay was conducted as previously described. 20 Briefly, sera were pretreated either with receptor destroying enzyme or periodate, then heated to 56 C for 30 minutes to inactivate nonspecific inhibitors of hemag glutination. Sera were serially diluted two fold in PBS and incubated with 4 hemagglutinating units/25L H3N8 CIV ( A/ canine/Florida/43/2004) in 96 well V bottom microtiter plates for 30 minutes at room temperature. An equal v olume of 0.5% turkey red blood cells (Lampire Biological Laboratories, Pipersville, PA, USA) in PBS was added to each well and incubated for 30 minutes at room temperature. The endpoint antibody titer was defined as the last dilution of serum that completely inhibited hemagglutination. The cutoff titer for seropositivity in this HI a ssay was previously defined as 32. 20 Sera from specific pathogen free dogs housed in a barrier research facility were used as negative controls, while sera from dogs with confirmed H3N8 CIV infection served as positive controls. 20,98 All dogs were sampled and tested prior to the introduction of the inactivated whole virus H3N8 CIV vaccine in June 2009; therefore, information on vaccination status was not requested. Data Collection For each dog, the following data were requested from the submitting veterinarian: clinical signs, date of illness onset, date of sample collection, age

PAGE 108

108 breed (pure, mixed), sex (male, female), potential exposure setting (shelter facility, boarding kennel, other exposure settings), and geographic location at the time of sample collection [U.S. regions = Northeast (CT, DE, MA, MD, M E, NH, NJ, NY, PA, RI, VT, WV), Southeast (AL, AR, FL, GA, KY, LA, MS, NC, SC, TN, VA), Midwest (IA, IL, IN, KS, MI, MN, MO, NE, OH, ND, SD, WI), Southwest (AZ, NM, OK, TX), West (AK, CA, CO, HI, ID, MT, NV, OR, UT, WA, WY)]. The date of illness onset and date of sample c ollection were requested to determine if the sample was an acute (<7 days Seronegative dogs with single samples collected less than 10 days after the date of illness onset were excluded from the study, since H3N8 CIV infection status could not be confirmed. To determine whether certain breeds were predisposed for H3N8 CIV infection, reported dog breeds were further classified according to one of 10 canine genomic spitz, herding, mastiff like, retriever, scent hound, sight hound, small terriers, spaniel, toy, or working dog). 126 The genomic categories of pure breed dogs and mixed breeds with one predominant b reed reported were coded for analysis. When a breed was not found in the genomic reference, it was classified P otential virus exposure settings reported by veterinarians at the time of sample collection were classified into three general categories: potential exposures (i.e., purchased from pet stores; visited dog parks, dog shows, or grooming facilities; exposed to another dog that may or may n ot have had ILI; and dogs presented to a veterinary hospital for ILI but the potential exposure setting was

PAGE 109

109 unknown); 2) dogs housed in animal shelter or rescue group facilities; and 3) pet dogs recently housed in commercial boarding kennels, veterinary ho spital boarding kennels, training or rehabilitat ion kennels, or canine day care Statistical Analyses A total of 1,268 pet and shelter dogs with ILI from 2005 through 2009 were included in the analyses. H3N8 CIV seroprevalence in the stud y population was calculated as the number of seropositive dogs div ided by the total number tested X 100. The 95% CI was estimated for the seroprevalence. Conditional logistic regression was used to model the odds of dogs being classified as H3N8 CIV seropositive as a funct ion of investigated exposure factors. Due to potential variability in H3N8 CIV circulation between years, dogs were matched in the analyses by the year they were sampled. Initial evaluation of potential risk factors for seropositivity was performed using u nivariable conditional logistic regression (Egret 2.0 .3 Cytel Inc., Cambridge, MA, USA). 45 Variables associated ( P 0.20) with the outcome of interest (H3N8 CIV seropositivity) were considered for inclusion in a multivariable conditional logistic regression model. A forward stepwise approach was used to identify variables associated with seropositivity using a 2 sided P values to enter and P values to remove of 0.05 and 0.10, respectively. Values for the final model were considered significant at P < 0.05. In the final model, the adjusted OR and 95% CI were reported. Results H3N8 CIV Seroprevalence in the Study Population The number of H3N8 CIV seropositive dogs was 618 out of 1,268 in the study population, representing an overall seroprevalence of 49% (95% CI = 46, 52%). The geometric mean antibody titer for these dogs was 302 (range = 32 to 4,096). Samples

PAGE 110

110 were submitted from dogs in 42 states; no dogs from IA, MS, MT, ND, NM, NV, RI, or WV were tested. Fewer than 10 dogs were tested in 24/42 (52%) states. Colorado (101 dogs), Florida ( 444 dogs), and New York ( 205 dogs) had the highest number of tested dogs in the West, S outheast, and Northeast respectively The frequency of s eropositive dogs by geographic region was: Northeast 62% (265/427 dogs); West 57% (144/252 dogs); Southeast 30% (192/490 dogs); Southwest 26% (11/43 dogs); and Midwest 11% (6/56 dogs). Risk Factors As sociated with H3N8 CIV Seropositivity Age, exposure setting, and geographic region were significantly ( P <0.01) associated with H3N8 CIV seropositivity in univariable conditional logistic regression analyses (Table 6 1). These 3 variables were retained as s ignificant risk factors for H3N8 CIV se ropositivity in the multivariable analyses (Table 6 dogs recently housed in boarding kennels or shelter facilities and dogs in the Northeast or West regions of the U.S. were at greatest risk for H3N8 CIV seropositivity Discussion This study represents the first national survey to estimate the seroprevalence of H3N8 CIV in a geographically diverse population of dogs with ILI, and to identify risk factors a ssociated with H3N8 CIV seropositivity The seroprevalence of H3N8 CIV in dogs with ILI dogs recently housed in boarding kennels or shelter facilities and dogs in the Northeast or West regions of the U.S. were a t greatest risk for H3N8 CIV seropositivity The high seroprevalence of H3N8 CIV in the study population suggests that this virus is a common respiratory pathogen in dogs in the U.S. However, this was a syndromic serosurvey limited to dogs with clinical signs suggestive of influenza like

PAGE 111

111 illness, and testing was not conducted to determine other causes of illness. Another limitation of th is study was the dependency on veterinarians to voluntarily collect samples from dogs with ILI for testing. The proportion of dogs with ILI that were not tested is unknown. In addition, diagnosis of H3N8 CIV was based so le ly on detection of antibodies and H3N8 CIV infection status could not be determined for many dogs due to improper timing of sample collection. Exclusion of dogs with improperly timed serum samples therefore reduced the number of seronegative dogs in the study Ultimately, the prevalence of H3N8 CIV in this study cannot be extrapolated to all dogs with ILI. E stimation of the true prevalence of H3N8 CIV in the U.S. would require random testing of large numbers of dogs independent of clinical status. The geographic regions of the U.S. with the highest H3N8 CIV seroprevalence were the Northeast, West, and Southeast. This should be interpreted with caution since each of these regions was overrepresented by a state from which large numbers of symptomatic dogs were tested. This may have occurred be cause the 3 states with highest representation ( New York, Colorado, and Florida respectively ) had diagnostic laboratories that actively solicited samples for H3N8 CIV testing. Although dogs in 42 states were tested, more than half of the states were repres ented by fewer than 10 dogs. This low sampling may have resulted from lack of interest of the veterinarian or dog owner in p ursuing testing and thus did not reflect the actual numbers of dogs with ILI in these states. The underrepresentation of many states and lack of testing in other states likely resulted in underestimation of H3N8 CIV prevalence in some geogr aphic regions Variability in H3N8 CIV sero prevalence may also reflect variability in virus circulation by year, location, and the susceptibility of exposed dogs.

PAGE 112

112 The H3N8 CIV seroprevalence in dogs from Colorado was higher in this study compared to a previous study conducted in that state. 8 One explanation is the different study designs and time periods used in each. The current study tested pet and shelter dogs with ILI from 2005 to 2009, while the previous stud y tested only pet dogs with or without ILI seen at a university teaching hospital from March to December 2009. The overall H3N8 CIV seroprevalence in this study was actually more comparable to the reported seroprevalence in a Philadelphia shelter (42%) tha n in a pet dog population tested over a two day period during a flyball tournament (3%). 44,111 Despite the limitations in interpretation of seroprevalence, do gs with ILI in the Northeast or West regions of the U.S were at significant ly greater risk for H3N8 CIV seropositivity This may be attributable to the frequently reported outbreaks of H3N8 CIV in boarding kennels, veterinary clinics, and shelters in these regions, particularly in New York City, Philadelphia, and the greater Denver area. 8,28,29,42 44 Although H3N8 CIV was first recognized in Florida as the causative agent for seve ral outbreaks from 2004 to 2007, 20,98 the Southe ast was not identified as a high risk region for H3N8 CIV seropositivity This unexpected finding may be due to lack of canine influenza virus detection in the other southeastern states, respiratory disease in Florida due to other canine respiratory pathog ens, or due to the fact that less than half of the Florida dogs were H3N8 CIV seropositive. In contrast, more than two thirds of the large numbers of dogs tested in New York (Northeast region) and Colorado (West region) were H3N8 CIV seropositive. Breed an d sex were not associated with H3N8 CIV seropositivity in this study population of dogs with ILI. Adult dogs however were significantly more likely to be

PAGE 113

113 seropositive than juveniles <1 year old, similar to another study that found older dogs had higher ris k (although the small sample size did not allow this association to be statistically significant). 8 Age was not a risk factor for seropositivity in another recent study evaluating the point prevalence of H3N8 CIV in pet dogs participating in a flyball tournament in Pennsylvania although the sample size was smal l 111 In the present study the association of seropositivity with adults was confounded in part by other factors, including high risk exposure settings and geographic locations. However, the association of H3N8 CIV seropositivity with the adult age group in dogs with ILI persisted after adjustment for these confounding factors in the multivariable analysis In addition, when age was evaluated as a potential risk factor for dogs in the exposure were still significantly m ore likely to be seropositive ( P <0.01) than dogs < 1 year of age. H3N8 CIV seropositivity for the majority of dogs in this study was based on testing of one serum sample, not paired acute and conv alescent samples. Therefore, the actual time ( or ag e at the time) of viral infection could not be determined for these dogs. The assoc i ation of H3N8 C IV infection with age warrants further study. Dogs housed in boarding kennels or shelters at the time o f ILI were at greater risk for H3N8 CIV seropositivity than dogs in other settings (e.g., pet stores, dog parks, groomers). The opportunity for virus transmission between infected and susceptible dogs via direct and indirect contact is very high in communa l settings containing high density/high turnover populations such as boarding kennels and shelters. 8,20,28,29,42,44,98 A recent study in a Philadelphia shelter determined that the risk for H3N8 CIV exposure doubled every 3 days that a dog stayed in the shelter. 44

PAGE 114

114 In summary, this syndromic survey found a high prevalence of H3N8 CIV seropositivity in 1,268 dogs with ILI in 42 states over a 5 year period. Communal housing situations such as boarding kennels or shelter facilities posed the greatest risk for H3N 8 CIV seropositivity In addition dogs located in the Northeast or West regions of the country where the virus has established endemicity in several metropolitan areas were at greater risk of infection These results justify the nee d for shelters, boarding kennels, and other communal dog housing facilities to formulate, implement and/or evaluate surveillance and biosecurity p rotocols to reduce the risk of virus transmission between dogs and avert large scale epidemics within these fa cilities.

PAGE 115

115 Table 6 1. Univariable conditional logistic regression analyses of potential risk factors for H3N8 CIV seropositivity in a population of 1,268 pet and shelter dogs with ILI. Variable Seropositive n (%) Seronegative n (%) OR g 95% CI h P value i Age a <1 yr 95 (19) 210 (35) 1.00 Reference NA 401 (81) 396 (65) 2.40 1.80, 3.19 <0.01 Sex b F 236 (46) 215 (47) 1.00 Reference NA M 282 (54) 243 (53) 1.07 0.82, 1.38 0.60 Breed c Pure 348 (68) 322 (66) 1.00 Reference NA Mixed 165 (32) 164 (34) 0.93 0.71, 1.23 0.64 Genomic categ ories d Ancient Spitz 22 (5) 26 (6) 1.00 Reference NA Herding 30 (7) 26 (6) 1.36 0.61, 3.03 0.44 Mastiff like 60 (13) 60 (14) 1.08 0.54, 2.16 0.80 Retriever 142 (31) 105 (24) 1.44 0.76, 2.73 0.25 Scent hound 53 (11) 41 (9) 1.42 0.70, 2.91 0.32 Si ght hound 10 (2) 16 (4) 0.98 0.35, 2.74 0.96 Small terriers 26 (6) 36 (8) 0.79 0.36, 1.71 0.55 Spaniel 22 (5) 27 (6) 0.95 0.42, 2.15 0.90 Toy 34 (7) 38 (9) 0.92 0.43, 1.95 0.84 Working dog 61 (13) 60 (14) 1.15 0.58, 2.29 0.68 Exposure setting e M ultiple f 57 (12) 182 (40) 1.00 Reference NA Shelter facilities 261 (56) 199 (44) 5.11 3.54, 7.37 <0.01 Boarding kennels 147 (32) 73 (16) 5.96 3.94, 9.00 <0.01 Geographic region Southwest 11 (2) 32 (5) 1.00 Reference NA Midwest 6 (1) 50 (8) 0.31 0 .10, 0.93 0.03 Southeast 192 (31) 298 (46) 1.73 0.84, 3.52 0.13 West 144 (23) 108 (16) 4.06 1.94, 8.49 <0.01 Northeast 265 (43) 162 (25) 4.67 2.26, 9.64 <0.01 a Age was provided for 496/618 seropositive dogs and 606/650 seronegative dogs. b Sex was pro vided for 518/618 seropositive dogs and 458/650 seronegative dogs. c Breed was provided for 513/618 seropositive dogs and 496/650 seronegative dogs. d Information adequate for genomic classification was provided for 400/618 seropositive dogs and 435/650 se ronegative dogs. e Exposure settings were provided for 465/618 seropositive dogs and 454/650 seronegative dogs. f Multiple = purchased from pet stores; visited dog parks, dog shows, or grooming facilities; exposed to another dog with or without ILI; and ta ken to a veterinary hospital for ILI but a potential exposure setting was unknown. g OR = odds ratio. h CI = confidence interval. i P values <0.05 were considered significant.

PAGE 116

116 Table 6 2. Multivariable conditional logistic regression analyses of potential risk factors for H3N8 CIV seropositivity in a population of 1,268 pet and shelter dogs with ILI. Variable OR a 95% CI b P value c Age <1 yr 1.00 Reference NA 1.49 1.01, 2.19 0.04 Exposure setting Multiple 1.00 Reference NA Shelter facilities 4.17 2.70, 6.44 <0.01 Boarding kennels 5.97 3.64, 9.78 <0.01 Geographic region Southwest 1.00 Reference NA Midwest 0.31 0.05, 1.84 0.19 Southeast 1.49 0.56, 3.93 0.41 West 4.19 1.49, 11.75 <0.01 Northeast 9.27 3.30, 26.07 <0.01 a OR = Adjusted odd s ratio. b CI = confidence interval. c P values <0.05 were considered significant.

PAGE 117

117 CHAPTER 7 SIMULATION STUDIES O F H3N8 CANINE INFLUE NZA VIRUS TRANSMISSI O N IN SHELTER FACILITIES Since its recognition in 2004 H3N8 CIV has been diagnosed as a common cause of acute respiratory disease in dogs, particularly those housed in communal settings. H3N8 CIV outbreaks have occurred in racetrack kennels, boarding kennel s, shelter facilities, veterinary hospitals, and canine day care centers across the nation. 9,20,28,29,42 44,98,141 The hallmark of H3N8 CIV has been explosive respiratory outbreaks, with high morbidity due to lack of prior immunity to this emerging pathogen. 20,28,29,98 However, it is currently unclear which viral and/or population parameters promote H3N8 CIV transmission in communal settings. Information is available regardi ng the basic biology of H3N8 CIV, however detailed epidemiological data focused on virus transmission rates and modes of transmission is limited. 20,26,27,49 Further, population parameters that may influence virus t ransmission in communal settings have not been defined. H3N8 CIV has significantly impacted the health and welfare of dogs in animal shelters, where some facilities have experienced recurrent outbreaks and some have resorted to extreme control measures su ch as depopulation. Outbreak management is particularly challenging for government shelters with animal control responsibilities, where ongoing admission of stray dogs and legal custody cases maintains a steady flow of susceptible dogs into the population. In addition, these shelters frequently have limited housing space for segregation of dogs based on H3N8 CIV infection status. Outside of these common challenges however, other presumably important factors vary between government shelters including total s helter capacity; intake, adoption, transfer, and euthanasia rates; and level of veterinary care. Considering the significant impact of H3N8 CIV outbreaks on shelter facilities, and the unique challenges faced by these

PAGE 118

118 organizations, it is valuable to explo re which viral and/or population parameters promote virus transmission in order to improve prevention and control recommendations. A variety of mathematical modeling approaches, including compartmental (deterministic and stochastic) and network models, ha ve been used to investigate influenza virus transmission in human, avian, swine, and equine populations. 25,32,33,65,68,69,96,109 The results of these studies have typically shown that transmission rates and contact between hosts are the most important parameters for virus spread, and that intervention strategies which include a combination of methods (vaccination, antiviral therapy, isolation and quarantine) are often most effective. More recently, several avian inf luenza models have also highlighted the importance of environmental transmission. 13,104,105 However it has been noted that host population organization and contact patterns vary greatly between species, making it d ifficult to directly apply models developed for one species to another. 6,25 Among current studies, models of equine influenza virus transmission among horses in training yards are most comparable to the scenario of H3N8 CIV transmission in shelters. 25,35,96,97 However, although the overall population sizes are similar, the population structure and movements of dogs in shelters are quite different from these equine population s. By comparison, d ogs in shelter facilities are highly managed populations. F actors such as high population density, high frequency of turnover in the population potentially lower contact rates between dogs, and potentially higher contact rates between d ogs and fomites (due to increased contact with handlers and objects in the environment) likely affect virus transmission.

PAGE 119

119 The objective of this study was to develop a mathematical model of H3N8 CIV transmission in shelter facilities in order to address th e following questions: 1) which viral and/or population parameters most significantly influence virus transmission in shelters, in terms of the time course and the numbers of dogs affected during CIV outbreaks; and 2) given the suspected importance of indi rect virus transmission in the epidemiology of H3N8 CIV in shelters, what is the relative contribution of direct vs. indirect transmission? Study results should provide a valuable foundation for future work, which will focus on development of more effectiv e prevention and control strategies. Materials and Methods Model Structure A compartmental, deterministic model was developed to simulate H3N8 CIV transmission in shelter facilities. Although stochastic models are recommended when dealing with small popu lations, since random events can significantly impact virus transmission (e.g., leading to virus die out), 35,97 a deterministic model was selected because of the limited data available for estimation of several par ameters. Deterministic models are well established in the literature and have been recommended over stochastic models in situations where parameter data is limited, in order to explore model behavior across a range of values in conjunction with sensitivity analysis. 4 The model was initially intended to investigate the effect of multiple viral and population parameters on the occurrence of H3N8 CIV outbreaks (i.e., epidemics) in shelters, including the time course, the numbers of dogs affected, and the relative contribution of direct and indirect virus transmission. However establishment of ende mic infection was recognized during the early stages of model development, therefore the

PAGE 120

120 viral and population parameters that promoted establishment of an endemic state were also investigated. A full sensitivity analysis and statistical analysis was comple ted. Model classes A modified SEIR model was developed with five dog classes and one fomites (X) class. The dog classes included s usceptible (S), latent ( L), i nfectious clinical (I C ), i nfectious sub clinical ( I SC ), and r ecovere d (R) dogs. All dogs in the s usceptible class were assumed to be nave to H3N8 CIV infection. The latent period (i.e., the time from virus exposure to virus shedding) rather than the incubation period (i.e., the time from virus exposure to development of clinical signs) was specified for the model. Although dogs exposed to H3N8 CIV begin shedding virus (i.e., become infectious) prior to the development of clinical signs, 20,26,27,49 class indicates that the relatively short clinic ally silent virus shedding period was included in the infectious classes ( in order to simplify the model ) Two infectious classes, infectious clinical and infectious subclinical dogs, were included in the model because studies of natural H3N8 CIV outbreak s as well as experimental infection studies have demonstrated that approximately 20% of H3N8 CIV infected dogs do not develop clinical signs. 20,26,27,49 It is important to separate clinical and subclinical dogs in order to explore their roles in H3N8 CIV transmission, and to provide a framework for future models focused on development of intervention strategies. It was assumed that infectious clinical and subclinical dogs are sufficiently different to track independ ently, since subclinical dogs would only be identified through diagnostic testing (virus isolation, RT PCR, serology) in clinical settings. In the model subclinical dogs were assumed to be half as infectious as clinical dogs, since coughing

PAGE 121

121 and sneezing pr omote virus transmission via direct contact and increase the amount of virus shed into the environment. A fomites class was included in the model because one of the goals of the study was to explore the relative contribution of direct vs. indirect virus tr ansmission on H3N8 CIV transmission in shelters. Direct transmission accounts for virus transmission from infected to susceptible dogs and indirect accounts for vi rus transmission via fomites as will be further defined when describing the model parameters Other than recent models seeking to understand the role of environmental transmission in avian influenza virus epidemics, 13,104,105 modes of transmission have not typically been separated in influenza models. Alt hough indirect virus transmission is poorly understood, and it ultimately is not possible to determine if an individual H3N8 CIV infection is due to direct or indirect transmission, it was hypothesized that this mode of transmission would be an important f actor in the epidemiology of H3N8 CIV in highl y managed shelter populations. The overall model s chematic is depicted in Figure 7 1. Model parameters Model parameters account for shelter capacity, population parameters that reflect typical dog movements i nto and out of model classes, and virus transmission factors. The model p arameters are listed and described in Table 7 1. Shelters were assumed to be at full capacity at the beginning of virus introduction simulations and the populations were assumed to be homogeneous. Further, it was assumed that there were no age or sex effects in dog movements. Since they were considered to be negligible over the time frame of a simulation, the per capita birth rate and the disease independent death rate were not inclu ded in the model. For simulations, a single latent dog ( initL ) was introduced into a shelter of a particular

PAGE 122

122 capacity ( initS ) at the initial time point. The total shelter capacity was therefore the sum of these values, however the initS parameter was consi dered an adequate approximation of the total shelter capacity since initL was set at one. This shelter capacity value ( initS ) was fixed throughout simulations since it was assumed that most shelters remain at capacity on a daily basis. Other population par ameters accounted for typical dog movements into and out of model classes, including immigration ( c ) due to shelter intakes, emigration ( m ) due to adoptions and transfers, and deaths due to euthanasia ( d ). After the introduction of one latent dog, immigrat ion was only allowed into the susceptible class. I mmigration was not dependent on the number of dogs in the class, but was expressed as a percentage of the total shelter capacity ( initS ) based on typical daily shelter intakes. To achieve a fixed shelter ca pacity throughout the simulations, immigration ( c ) and emigration ( m ) were considered independent variables, and euthanasia was calculated from these values based on E quation 7 1. ( 7 1) Emigration out of model classes was dependent on the number of dogs in those classes (i.e., the upper bound was the total number of dogs in the class at a given time), and was tracked in terms of percentages of dogs leaving each class per day It was assumed that emigration and euthanasia practices would be different for infectious clinical dogs compared to dogs in the other classes, therefore separate emigration ( m IC ) and euthanasia ( d IC ) parameters were established for this class. Emigration was allowed out of all classes of dogs, however it was set to zero for the i nfectious clinical class to mimic typical shelter practices. Euthanasia from the infectious clinical class was

PAGE 123

123 set at 1.5 d where d is calculated as described above, with the assumption that clinically ill dogs would be euthanized at a higher rate than dog s in the other classes. H3N8 CIV is transmitted by direct transmission (dog to dog contact, droplet transmission through coughing/sneezing) and indirect transm ission (aerosols and fomites) All sources of infection were grouped into one force of infection two terms were defined: one for direct virus transmission and one for indirec t virus transmission (Equation 7 2). ( 7 2) In the direct virus transmission term is the proportion of susceptible dogs that become infected when contact ing infectious dogs, or when contacting infectious droplets directly coughed or sneezed on them by infectious dogs. A devaluing parameter ( a ) is included to account for the reduced infectiousness of subclinical dogs as has been applied in other studies 68 Direct virus transmission was scaled by the total number of dogs in the shelter ( N ) following the conventional mass action mixing approach. For the indirect virus transmission term, is the proportion of susceptible dogs that become infected whe n contacting fomites ( X ); this includes contacting infectious virus on any con taminated environmental object (e.g., a food bowl or a handler) or aerosols. Indirect virus transmission was scale d by ( X + k ), where k is a scaling parameter representing th e number of noninfectious object s in the environment. This attempts to mimic the mass action mixing approach for fomites. This approach is based on recent modeling studies seeking to understand th e role of indirect transmission of several pathogens, including avian influenza virus, through environmental reservoirs such as lakes 13,50,104,105 Although shelters are obviously not aquatic environments, it was

PAGE 124

124 a ssumed that minimal infectious doses of H3N8 CIV would be deposited throughout the shelter environment by infectious (clinical and subclinical) dogs, which would represent a substantial source of infection for susceptible dogs. N early all nave dogs expose d to H3N8 CIV become infected due to lack of pre existing immunity, 20,26,27,49 therefore it was assumed that all dogs in the shelter would be susceptible to infection and that all dogs in the latent class would bec ome infectious. Further, it was assumed that a proportion of latent dogs would become infectious and develop clinical signs, while the remaining dogs would become infectious but remain subclinical. 20,26,27,49 This was accounted for by multiplying the per capita rate at which dogs become infectious ( f) by the percentage of dogs that become clinical ( q ) or the percentage of dogs that remain subclinical ( 1 q ). This approach was selected since there is no known differen ce in the rate at which clinical vs. subclinical dogs become infectious, and it has been used in the influenza modeling literature. 47,76 It was assumed that all infectious dogs (clinical and subclinical) serve as p otential sources of virus for susceptible dogs during the course of a simulation. However devaluing parameters were introduced for the decreased infectiousness of subclinical dogs ( a ) and for the decreased virus shedding into the environment by subclinical dogs ( b ) due to lack of clinical signs such as coughing and sneezing, which enhance virus spread. 68 Once virus shedding ceases, dogs are no longer contagious to other dogs; therefore the virus shedding period was expressed as the per capita rate at which infectious dogs recover. Severity of clinical disease and deaths due to H3N8 CIV were not considered in the model.

PAGE 125

125 Model equations Differential equations, representing the change over time for each model class during a simulated H3N8 CIV introduction, were programmed into MATLAB (MathWorks, Natick, MA, USA) computatio nal software Simulations were completed Equation 7 3 represents the change over time in the susceptible ( S ) class. Movement into this class is affected by immigration ( c(initS) ), representing the number of shelter intakes expressed as a percentage of the total shelter capacity. Movement out of the class depends on emigration ( m ) due to adoptions and transfers, deaths due to euthanasia ( d ), and the transition of susceptible dogs to the latent class due to t he force of infection ( S ) as previously described. ( 7 3) Equation 7 4 represents the change over time in the latent ( L ) class. Movement into the class is affected by infection of susceptible dogs ( S ) while movement out of the class depends on emigra tion ( m ), deaths due to euthanasia ( d ), and the transition of dogs to the infectious clinical ( fqL ) and infectious subclinical ( f(1 q)L ) classes (simplified to fL ). ( 7 4) Equation 7 5 represents the change over time in the infectious clinical ( I C ) clas s. Movement into the class is affected by the transition of a percentage of latent ( fqL ) dogs, while movement out of the class depends on deaths due to euthanasia ( d IC ), and the transition of dogs to the recovered class ( I C ) based on the per capita rate a t which

PAGE 126

126 i nfectious clinical dogs recover ( ). Emigration ( m IC ) from this class was set to zero and d IC was calculated as previously described ( 7 5) Equation 7 6 represents the change over time in the infectious subclinical ( I SC ) class. Movement into th e class is affected by the transition of a percentage of latent dogs ( f(1 q)L ); while movement out of the class depends on emigration ( m ), deaths due to euthanasia ( d ), and the transition of dogs to the recovered ( I SC ) class. ( 7 6) Equation 7 7 repres ents the change over time in the recovered ( R ) class. Movement into this class is affected by the transition of infectious clinical ( I C ) and infectious subclinical ( I SC ) dogs while movement out of the class depends on emigration ( m ) and deaths due to eu thanasia ( d ). ( 7 7) Equation 7 8 represents the change over time in fomites ( X ). The number of infectious H3N8 CIV doses ( ) produced per infectious dog is included in the equation, where b is the devaluing parameter for decreased virus shedding of subc linical dogs and represents the inactivation of virus on fomites over time (i.e., is the inverse of the length of time virus remains infectious). ( 7 8)

PAGE 127

127 Parameter Estimation The model parameters, values used, and sourc es of data are listed in Table 7 1. Parameter estimates were based on published literature and expert opinion. When applicable, r ate based transitions were chose n for mathematical simplicity. Although a fixed lag is often used for the transition from the laten t to infectious classes, a r ate based transition was selected because there is individual variation in this value. Latin hypercube sampling (LHS) was used for the full sensitivity analysis of the model in order to allow simultaneous variation in the parameter values; 11,70 therefore a probability distribution was selected for each independent parameter based on the available data. Triangular distributions were used when a most probable value was supported by the literature; however, the range of these parameter values was then based on the quality of available data (i.e., parameters with less supporting data were given wider ranges). When a most probable value was not available in the literature, parameters were assigned uniform distributions. considered most biologically plausible by the investigators, in order to explore the full range of potential values. Three parameters ( a m IC and b ) were set at constant values, and d and d IC were calculated; therefore these values were not included in the sensitivity analysis. The values of a and b were held constant at 0.5, which is the same value used in another study. 68 The value of m IC was set to 0, since it was assumed that infectious clinical dogs would not be allowed to emigrate The euthanasia parameters ( d and d IC ) were calculated as previously described, in order to maintain shelter capacity. T he remaining 11 independent model parameters were included in the sensitivity analysis. Three parameters ( f q and ) are well supported in the literature, therefore

PAGE 128

128 triang ular distributions with maximum, minimum, and most probable values were assigned 20,26,27,49 Virus inactivation ( ) is also well supported, 9,12,129 however a uniform distribution was assigned due to the variability in time to virus inactivation (based on the t ype of surface or medium). Values and ranges for s helter capacity a nd dog movement parameters ( initS c and m ) were determined by the investigators based on typical shelter si zes and dog movements. A range of 50 to 400 dogs was selected for shelter capacity ( initS ) and a uniform distribution was assigned since s helters vary in size across this range. The ranges for immigration ( c ) and emigration ( m ) were also assigned uniform distributions, since these percentages likely also vary across the full ranges selected. The selected values for c and m reflect the high turnover considered typical in government shelters. H3N8 CIV epidemic curve data are not available to estimate the transmission parameters ( ), and ultimately it is not feasible to distinguish infections due to direct or indirect virus transmission. These value s were therefore based on one published experiment and expert opinion. 49 Since the published experiment was focused on dog to dog transmission of H3N8 CIV, wh ere 75% of contact dogs contracted H3N8 CIV, the study findings primarily informed the values for the parameter in the direct transmission term. Since the transmissib ility of H3N8 CIV has been question ed, 28,42 however, a triangular distribution with a central val ue of 0.5 and a wide range ( 0.2 to 0.9) was selecte d This would allow exploration of potential extreme values of but many randomly generated values would be around the central value of 0.5. Since infections due to indirect virus transmission could not be identified in the published

PAGE 129

129 study, the paramet er of the indirect transmission term was given a wide range and a uniform distribution to explore the full range of possible values. T he remaining model parameters, k and were estimated by the investigators. The values and ranges were refined by runnin g the model program with extreme proposed values in order to gain a better understanding of their behavior in the model. As previously noted, the environmental fomites scaling parameter ( k ) was based on literature addressing indirect transmission of other pathogens and environmental transmission of avian influenza virus. 13,50,104,105 These studies however addressed indirect transmission in aquatic environments, therefore a reasonable range for k in a terrestrial en vironment was uncertain. When the model was run with a wide range of values for k it was found that a full spectrum of behavior was captured with values ranging from 10 2 to 10 4 Through this exploration it was also apparent that the simulation length shou ld be set at 100 days in order to capture the time course for the range of potential epidemics. A similar process was completed for Although published studies give an idea of virus excretion per dog, based on nasal swabs collected at single time points, 20,26,27,49 it is uncertain how much virus is truly sh ed into the environment by infectiou s dogs in a day. The concept of an infectious dose was adopted based on the cited indirect transmission studies, to avoid quantifying virus particles in the environment, and a range of 10 to 100 infectious doses produced per dog per day captured the full r ange of model behaviors. Uniform distributions were assigned to these parameters to explore the full range of possible values. Sensitivity Analysis A sample size of n = 500 simulation runs was selected, and the LHS scheme was programmed into MATLAB (MathW orks) Briefly, to generate unique parameter sets,

PAGE 130

130 the program divided each assigned probability distribution into n = 500 equal intervals. Within each interval a value was randomly selected and then back transformed to the original parameter axis, therefo re 500 values were generated for each independent parameter. The values for all 11 independent parameters were then randomly reassorted into 500 unique parameter sets, which were saved to an Excel file. Data Analysis The 500 virus introduction simulations were run in the model program using a 100 day time period and a one day time step. Each simulation began on day zero with the introduction of a latent dog. To assess potential virus die out in the deterministic model, the model was programmed to terminate a simulation if the sum of the latent, infectious clinical, and infectious subclinical dogs fell below 0.5. If this threshold was reached the simulation was restarted with these classes set to zero. The following outcomes of interest were analyzed. Occurr ence of an epidemic. Each simulation was characterized as epidemic or initS ) being infected ( I C + I SC dogs ) at the peak of infection Respiratory disease is considered common in animal shelters, however the level of baseline (i.e., typical) respiratory disease can vary greatly from one facility to another. Given this variability, a threshold of 10% of the population being infected (with approximately 8% exhibiting clinical signs) was con sidered a reasonable level to trigger recognition of an epidemic (i.e., respiratory disease beyond the normal level in the shelter). The percentage of simulations that resulted in an epidemic was calculated. The original intention was to perform logistic r egression analysis, assessing which of the 11 independent model parameters

PAGE 131

131 contributed to the occurrence of an epidemic; however this was not performed since only two of the simulations were classified as not epidemic. Establishment of endemic infection. Each simulation was also characterized as endemic or not endemic. Establishment of endemic infection in a simulation was defined as one or more dogs being infected ( I C + I SC dogs ) at the steady state point (i.e., the time point at which a potential epidemi c was classified as over and an endemic state was established). To standardize the determination of this steady state point, a subset of runs were evaluated in MATLAB ( MathWorks) the settling time based on four pot ential cutoffs. These simulations were evaluated at 100 day and 200 day simulation lengths, and the time points at which the number of infectious clinical dogs and the total number of infectious dogs settled at a tolerance of 2% or 3% were calculated for e ach. Across the examined runs, the 100 day and 2% tolerance cutoff provided the most consistent determination of steady state points that were also considered most biologically plausible. F igure 7 2 provides examples of these settling time checks and the outcomes of interest are identified The percentage of simulations that resulted in an endemic infection was calculated. The original intention was to perform logistic regression analysis, assessing which of the 11 independent model parameters contributed to the establishment of endemic infection; however this was not performed since none of the simulations were classified as not endemic. Time course of epidemics. For each simulation classified as epidemic, two time points were calculated: the total durati on and time of epidemic peak. The total duration of the epidemic, in days, was calculated from day 0 (when a latent dog was introduced) until the time when the number of infectious clinical dogs reached the steady state point.

PAGE 132

132 The range and mean duration o f simulated epidemics were summarized. Multiple stepwise regression was conducted in MATLAB ( MathWorks) function, to assess which of the 11 independent model parameters significantly affected the total duration of epidemics, with e ntry and removal levels set at P = 0.05 and P = 0.10 respectively. Stepwise regression was chosen in order to prioritize the 11 independent model parameters in terms of their effect on this outcome. The time of the epidemic peak was defined as the day at which the number of infected dogs ( I C + I SC dogs ) was at its maximum. The range and mean for the times of the epidemic peaks were summarized. Stepwise regression, assessing which of the 11 independent model parameters significantly affected this outcome, was co nducted as previously described. Numbers of dogs. The total number of dogs infected ( I C + I SC dogs ) at the peak of the epidemic was determined for each epidemic simulation. The range and mean number of infected dogs at the epidemic peaks were summarized, a nd stepwise regression assessing which of the 11 independent model parameters significantly affected this outcome was conducted as previously described. The accumulation of the total number of dogs in a shelter during an epidemic included all of the dogs i n the shelter at the beginning of the simulation (the initS and initL values) plus the dogs that entered the sh elter during the simulation (c(i nitS)) until the steady state point was reached. The range and mean number of total accumulated dogs during epide mics was described. T o address the number of dogs infected during the course of an epidemic, t he accumulation of the total number of infectious and sub clinical dogs during the epidemic

PAGE 133

133 were determined. These values were calculated by accumulating the app r opriate elements of Equations 7 5 and 7 6 respectively, from day 0 to the steady state point. The range and mean number of accumulated infectious c linical and subclinical dogs were summarized, and stepwise regression was conducted as previously described t o determine which of the 11 independent model parameters significantly affected each of these outcomes. Relative contribution of direct and indirect virus transmission The relative contribution of the direct and indirect virus transmission terms were de fined as the accumulation of the number of dogs that transitioned from the susceptible class to the latent class, due to the direct and indirect components of the force of infection (Equation 7 2 ), from day 0 until the epidemic peak The direct or indirect transmission term was considered the dominant driver of the epidemic if it was >55% of the force of infection, otherwise the terms were considered equal. The percentage s of epidemics that were due primarily to direct or indirect virus transmission, or tha t were due to both, w ere calculated. To visualize the full range of model behaviors, a plot was generated for each epidemic simulation depicting the epidemic threshold, the time of the epidemic peak, the steady state point, the number of infectious clinica l and subclinical dogs at these time points, and the force of infection divided between direct and indirect transmission. This also allowed the epidemic threshold and the 100 day simulation length to be further ev aluated. Results Of the 500 H3N8 CIV introd uction simulations, 498 (99.6% ) were classified as epidemic. In addition, all 500 simulations were classified as endemic. Examples of

PAGE 134

134 epi demic simulations are shown in F igures 7 2 through 7 5. Descriptive statistics (means and ranges) for the epidemic simu lation outcomes (total duration, time of the epidemic peak, and numbers of dogs aff ected) are summarized in Table 7 2. A large range was expected for these outcomes, since some parameters were quite wide in their distributions. It is notable that large num bers of infectious subclinical dogs accumulated during the course of epidemics, with an average of 89 (Range = 15 237) infectious subclinical dogs accumulating. The stepwise regression results for the epidemic time outcomes (total duration and time of the epidemic peak) are summarized in Table 7 3. For total epidemic duration, 9 of the 11 independent parameters entered the model and explained 71% of the variance. For the time of the epidemic peak, 9 parameters entered the model and explained 82% of the vari ance. Parameters primarily associated with indirect virus transmission ( k and ) were ranked as the top three contributors to these outcomes. The and parameters were negatively associated with these outcomes, while the k parameter was positively associated. The stepwise regression results for the number of dog outcomes (total number of infected dogs at the epidemic peak, total number of infectious clinical and infectious subclinical dogs that accumulated during the epid emics) are summarized in Table 7 4. For the total number of dogs infected at the epidemic peak, all 11 parameters entered the model and explained 97% of the variance. For the total number of infectious clinical dogs that accumulated during the course of epidemics, 9 parameters entered the model and explained 96% of the variance. For the total number of inf ectious subclinical dogs that accumulated during the course of epidemics, 5 parameters entered the model and

PAGE 135

135 explained 94% of the variance. The top ranked contributor was the initS parameter, which was positively associated with each of these outcomes. For the total number of dogs infected at the peak, parameters associated with indirect virus transmission were the next two greatest contributors. The parameter was positively associated, and the k parameter was negatively associated with this outcome. Of the 498 virus introduction simulations that resulted in an epidemic, 21% were primarily due to direct virus transmission, 63% were primarily due to indirect virus transmission, and 16% were due to both direct and indirect transmission as previously described. Figures 7 3, 7 4, and 7 5 show example plots for simulations driven primarily by direct transmission, indirect transmission, or equally by both, respectively. These graphs depict the total duration of the epidemic, the time of the epidemic peak, the steady state point, the number of infectious clinical and subclinical dogs at these points, and the force of infection divided between direct and indi rect transmission Discussion This is the first study to mathematically model H3N8 CIV transmission in shelter facilities. A unique feature of this model (for mammalian influenza models) was the separation of the force of infection into two terms, in orde r to account for both direct and indirect virus transmission. This approach was pursued because indirect virus transmission was hypothesized to be an important mode of H3N8 CIV transmission in highly managed shelter populations. There is precedent for this approach in the influenza modeling literature, in studies seeking to understand the epidemiological role of environmental transmission in avian influenza epidemics. 13,104,105 Lim itations to the study largely cente red on the limited or unavailable data for these and other model parameters. Investigators selected biologically plausible values in these situations,

PAGE 136

136 however ranges and distributions could not be narrowly defined. In addition, since only two simulations w ere not classified as epidemic and no simulations were classified as not endemic, it was not possible to conduct logistic regression analysis to assess which of the 11 independent model parameters were significantly associated with the occurrence of an H3N 8 CIV epidemic or the establishment of endemic infection. Even if the epidemic threshold was raised to 25% (i.e., of the shelter population being H3N8 CIV infected), only approximately 5% of the simulations would have been classified as not epidemic; there fore the selected threshold (10%) was not considered a contributor to this result. In this study, given the model structure and parameter ranges used, H3N8 CIV epidemics occurred in 99.6% of the simulations and endemic infection was establish ed in all 50 0 Simulated epidemic time s, as well as the overall numbers of dogs affected, varied widely based on the parameter values used in each simulation. Stepwise regression results indicated that multiple model parameters significantly influenced the se outcomes. However, several parameters consistently ranked in the top three contributors and patterns emerged Generally, as and increased (i.e., as more infectious virus doses were shed into the environment and a greater proportion of susceptible dogs became in fected after contact with fomites ) the total duration of the simulated epidemics decreased and the epidemics peaked earlier. As k increased (i.e., more noninfectious object s were present in the envi ronment to scale the fomites) the total duration of the ep idemic increased and the epidemic peak t ime occurred later. T he greatest contributor to outcomes involving numbers of infected dogs was the value of initS (i.e., as the size of the shelter increased the number of total infected dogs at the

PAGE 137

137 peak and the num ber of accumulated infectious clinical and subclinical dogs increased). Interestingly, however, parameters primarily associated with indirect virus transmission were the next greatest contributors to the total number of dogs infected at the peak; where the total number of dogs increased as increased, and decreased as k increased. In this study 63% of the epidemic simulations were dominated by indirect transmission at the epidemic peak. Based on the defined criteria, this meant that >55% of the latent dogs that accumulated from day 0 to t he epidemic peak were infected by indirect transmission in 63% of the simulations. Recent modeling studies have reported that environmental transmission is a significant driving factor in avian influenza epidemics, and that it is particularly important for persistence of infection; 13,104,105 however these studies focused on a feature unique to the biology of avian influenza, where virus is shed through the feces and is able to persist long term (i.e., months) under certain conditions in aquatic environmental reservoirs. This study attempted to capture indirect virus transmission by first quantifying the number of fomite s in the environment (Equation 7 8), and then scaling them in the indirect virus transmission term by a range of noninfectious o bjects (Equation 7 2). The study results indicate that indirect transmission may be more important than previously considered, and that separation of the transmission terms may shed light on targets for prevention and control s trategies. For all of these outcomes, it is important to note that no interventions (i.e., halting shelter intakes; isolation, sanitation, or vaccination) were applied to the model. Such interventions theoretically would have reduced the number of suscept ible animals and

PAGE 138

138 the amount of infectious virus in the simulations. In addition elements of the model structure (e.g., inclusion of a fomites class and separation of the force of infection into two transmission terms), and the parameter values and distribu tions used also influenced these result s. In order to more fully explo re the parameters that affect H3N8 CIV transmission in shelters, this study highlights the importance of collecting additional epidemiological and experimental data to refine the values, ranges, and distributions of several parameters. For example, additional data on the and components of the direct and indirect transmission terms respectively would be particularly valuable. A dditional work should also focus on how to define and parameterize virus shedding into the environment and the scaling of fomites in t he indirect trans mission term, since indirect virus transmission parameters such as and k significantly influenced several of the simulation outcomes. Despite these limitations the study results overall reflect what has been observed clinically during H3N8 CIV epidemic s in shelters (i.e., explosive epidemic s with high morbidity). The epidemic time outcomes and numbers of dogs affected were similar to what has been observed during H3N8 CIV shelter outbreak consultations conducted by the investigators. 20,98 The reported accumulation values highlighted the large number of dogs that can be infected during the course of H3N8 CIV epidemic s. T he accumulated number of infectious subclinical dogs further supports the current recommendatio n to quarantine all H3N8 CIV exposed dogs during an epidemic, since subclinical dogs can silently shed and spread virus Although beyond the scope of this study, t he establishment of endemic infection in all of the simulations raises important clinical que stions regarding which parameters might ultimately promote endemicity in shelters.

PAGE 139

139 In conclusion, b ased on the overall results of this study, indirect virus transmission and the number of dogs in a shelter (i.e., the shelter capacity) have the greatest im pact on H3N8 CIV transmission dynamics in shelter facilities. P revention and control strategies that reduce cont act between dogs and fomites, decrease the amount of infectious virus in the environment, and limit the number of incoming susceptible dogs will likely reduce the impact of epidemics in shelter facilities (and may prevent the establishment of endemic infection). Studies focused on modeling the efficacy of different intervention strategies, including isolation and vaccination, are being pursued to test these hypotheses.

PAGE 140

140 Figure 7 1. Model schematic for simulations of H3N8 CIV introduction into shelter facilities.

PAGE 141

141 Table 7 1. Parameters, ranges, and distributions used for simulations of H3N8 CIV introduction into shelter facilities. Parameters Symbol Distribution a Range b Center c Sources d Initial no. of susceptible dogs in the shelter population initS Uni 50 400 1 Initial latent dog introduced to begin simulation initL Con 1 1 Proporti on of dogs that become infected after contact by direct transmission routes Tri 0.2 0.9 0.5 1,2 Proportion of dogs that become infected after contact by indirect transmission routes Uni 0.2 0.8 1 Devaluing parameter for infectiousness of subclinical dogs a Con 0.5 3 Scaling parameter representing the p otential no. of noninfectious object s in the environment k Uni 10 2 10 4 1,4 P ercent of shelter capacity ( initS ) immigrating into the susceptible class at time (t) c Uni 0.05 0.1 1 P ercent of dogs in class (all but infectious clinical ) emigrating at time (t) m Uni 0 0.05 1 P ercent of dogs emigrating out of the infectious clinical class at time (t) m IC Con 0 1 Percent of dogs in class (all but infectious clinical ) euthanized at time (t) d Calc Equation 7 1 Percent of dogs euthanized from the infectio us clinical class at time (t) d IC Calc 1.5 d Per capita rate at which latent dogs become infectious (clinical or subclinical) f Tri 0.33 0.95 0.5 2,5,6,7 Percent of dogs that become infectious clinical q Tri 0.7 0.9 0.8 2,5,6,7 Per capita rate at wh ich infectious clinical and subclinical dogs recover Tri 0.1 0.2 0.15 2,5,6,7 No. of infectious H3N8 CIV doses produced per infectious clinical and subclinical dog Uni 10 1 10 2 2,5,6,7 a Probability distribution used in LHS analysis. Tri, triangular; Uni, uniform; Calc, calculated from other parameters (not in LHS analysis); Con, constant (not in LHS analysis). b Minimum maximum values. c Most probable value. d Literature used: 1, Investigator opinion; 2, Jirjis et al., 2010; 3, Longini et al., 2004 ; 4, Rohani et al., 2009; 5, Crawford et al., 2005; 6, Deshpande et al., 2009; 7, Deshpande et al., 2009

PAGE 142

142 Table 7 1 continued. Parameters Symbol Distribution a Range b Center c Sources d Devaluing parameter for decreased virus shedding of subclinical dogs b Con 0.5 3 Inactivation of infectious H3N8 CIV doses over time on fomites Uni 0.25 0.95 8,9 ,10 a Probability distribution used in LHS analysis. Tri, triangular; Uni, uniform; Con, constant (not in LHS analysis). b Minimum maximum values. c Most probable value. d Literatur e used: 3, Longini et al., 2004 ; 8, Brankston et a l., 2007; 9, Weber et al., 2008; Beeler, 2009.

PAGE 143

143 A B Figure 7 2. Results of typical endemic settling time checks, used to standardize the criteria for determining the steady state point. I C and I SC = Infectious clinical and subclinical dogs, 100 2% or 100 3%= 100 day simulation where the number of infectious dogs settles to within 2 or 3% tolerance, 200 2% or 200 3%= 200 day simulation with 2 or 3% tolerance. A, Example simulation where the steady state point was comparable regardless of the criteria applied; initS = 220, c = 0.09, m = 0.02, = 0.34, = 0.40, f = 0.44, q = 0.79, = 0.14, = 78, = 0.88, k = 9808 (run #147). B, Example simulation where the steady state point differed between the selected criteria; initS= 367, c= 0.05, m= 0.02, = 0.80, = 0. 36, f = 0.51, q = 0.76, = 0.14, = 83, = 0.79, k = 3646 (run #292)

PAGE 144

144 Table 7 2. Descriptive statistics for specified outcomes of 498 H3N8 CIV introduction simulations classified as epidemic. Outcome Mean Range Epidemic times (days): Total duration 38 (20 93) Epidemic peak 19 (6 59) Total number of dogs: Infected (clinical and subclinical ) at the epidemic peaks 88 (10 219) Accumulated (entering shelter) during epidemics 850 (175 2226) Accumulated infectious clinical during epidemics 358 (78 838) A ccumulated infectious subclinical during epidemics 89 (15 237)

PAGE 145

145 Table 7 3. Results of stepwise regression on epidemic duration and time of epidemic peak. Parameters as defin ed in Table 7 1. Outcome Step Entered Removed Coefficient Model R 2 F P value a E pidemic duration Intercept 85.6246 1 0.2038 0.2329 150.6229 <0.001 2 k 0.0019 0.4546 206.3597 <0.001 3 19.4341 0.5407 193.8725 <0.001 4 c 201.8824 0.5836 172.7888 <0.001 5 initS 0.0267 0.6230 162.6221 <0.001 6 16.7487 0.65 54 155.6615 <0.001 7 f 16.6987 0.6873 153.8814 <0.001 8 9.9345 0.7103 149.8748 <0.001 9 m 77.0380 0.7180 138.1224 <0.001 10 q 3.0075 0.6855 11 27.6981 0.0632 Time of epidemic peak Intercept 42.9029 1 k 0.0014 0.3203 233.80 62 <0.001 2 0.1300 0.5462 297.9519 <0.001 3 14.5381 0.6597 319.3555 <0.001 4 12.6420 0.7110 303.2770 <0.001 5 f 11.3414 0.7549 303.2277 <0.001 6 initS 0.0150 0.7940 315.5254 <0.001 7 5.5029 0.8166 311.6825 <0.001 8 c 36.36 31 0.8214 281.2515 <0.001 9 q 7.1921 0.8229 252.0641 <0.001 10 m 13.1223 0.1902 11 1.8382 0.7958 a P values <0.05 were considered significant.

PAGE 146

146 Table 7 4. Results of stepwise regression on total number of infected dogs at the epidemic pe ak as well as accumulated infectious clinical (I C ) and subclinical dogs (I SC ) at the steady state (SS) Parameters as defined in Table 7 1. Outcome Step Entered Removed Coefficient Model R 2 F P value a Total number infected dogs at peak Intercept 20.55 28 1 initS 0.4329 0.8440 2683.9579 <0.001 2 46.3329 0.8767 1761.1285 <0.001 3 k 0.0026 0.9043 1556.3925 <0.001 4 322.1694 0.9241 1501.8924 <0.001 5 0.2471 0.9420 1599.8543 <0.001 6 f 40.7496 0.9554 1757.0821 <0.001 7 27.36 41 0.9619 1770.7701 <0.001 8 c 245.3557 0.9676 1828.9364 <0.001 9 12.8175 0.9704 1777.8599 <0.001 10 m 125.7297 0.9718 1678.6584 <0.001 11 q 24.8676 0.9722 1548.6551 <0.001 Accumulated I C dogs at SS Intercept 415.4966 1 initS 1.55 38 0.9262 6230.4552 <0.001 2 q 559.1149 0.9456 4309.0315 <0.001 3 m 727.2562 0.9504 3158.4911 <0.001 4 0.3193 0.9520 2463.5350 <0.001 5 349.7988 0.9542 2051.7242 <0.001 6 30.1928 0.9555 1758.7330 <0.001 7 c 348.5366 0.9564 1538.8 567 <0.001 8 k 0.0016 0.9573 1372.0809 <0.001 9 f 23.7296 0.9577 1228.0226 <0.001 10 9.6745 0.3618 11 9.4473 0.2152 Accumulated I SC dogs at SS Intercept 360.3887 1 initS 0.3917 0.7682 1644.0073 <0.001 2 q 441.9163 0.9300 32 89.9189 <0.001 3 m 191.4053 0.9337 2320.0066 <0.001 4 0.0829 0.9360 1804.0266 <0.001 5 k 0.0006 0.9378 1485.2920 <0.001 6 c 1.8404 0.9585 7 6.4987 0.0673 8 2.9971 0.3084 9 f 5.2240 0.1777 10 46.5237 0.0618 11 3.2597 0.1986 a P values <0.05 were considered significant.

PAGE 147

147 A B Figure 7 3. Example of a simulation dominated by direct virus transmission at the peak of the epidemic. A) Number of infected dogs (I C = infectious clinical and I SC = infectious subcl inical) during the simulation, at the epidemic peak and the steady state point (SS); blue= total infectious dogs, green = clinical, red= subclinical. B) The force of infection ( lambda depicted in red) separated into direct (blue) and indirect (green) transmission. Parameter values: initS= 117, c= 0.05, m= 0.04, = 0.62, = 0.45, f= 0.43, q= 0.81, = 0.14, = 89, = 0.89, k= 9348 (run #306).

PAGE 148

148 A B Figure 7 4. Example of a simulation dominated by indirect virus transmission at the peak of the epidemi c. A) Number of infected dogs (I C = infectious clinical and I SC = infectious subclinical) during the simulation, at the epidemic peak and the steady state point (SS); blue= total infectious dogs, green = clinical, red= subclinical. B) The force of infection ( lambda depicted in red) separated into direct (blue) and indirect (green) transmission. Parameter values: initS= 72, c= 0.09, m= 0.002, = 0.40, = 0.53, f= 0.42, q= 0.71, = 0.12, = 70, = 0.29, k= 6216 (run #391).

PAGE 149

149 A B Figure 7 5. Example of a simulation equally due to direct and indirect virus transmission at the peak of the epidemic. A) Number of infected dogs (I C = infectious clinical and I SC = infectious subclinical) during the simulation, at the epidemic peak and the steady state point (SS) ; blue= total infectious dogs, green= clinical, red= subclinical. B) The force of infection ( lambda depicted in red) separated into direct (blue) and indirect (green) transmission. Parameter values: initS= 235, c= 0.06, m= 0.04, = 0.55, = 0.47, f= 0.5, q= 0.79, = 0.15, = 49, = 0.84, k= 8851 (run #433).

PAGE 150

150 CHAPTER 8 SUMMARY, RESEARCH CONCLUSIONS AND ISSUES FOR FUTURE INVESTIGATORS Summary T he overall objectives of this PhD research were to develop tools for accurate serological diagnosis of H3N8 CIV infection and to investigate the epidemiology of canine influenza in order to improve prevention and control strategies. Serological assays are often necessary to confirm diagnosis of H3N8 CIV infection, however a standard HI assay has not been generally accepted for canine influenza. To adequately study the epidemiology of H3N8 CIV, key diagnostic tests therefore needed to be developed and assessed. T he first research aim was to determine the diagnostic performance of the H3N8 CIV HI assay as conducted in our previous studies, and to evaluate the impact of common test parameter variations on diagnostic performance. The standard H3N8 CIV HI assay was then utilized in subsequent epidemiological studies. The introduction of the inactivated H3N8 CIV vaccine in June 2009 created a diagnostic dilemma, however, since the HI assay does not allow d ifferentiation of infected from vaccinated animals (DIVA). Although epidemiological investigations conducted for this research were not confounded by vaccination, a DIVA test strategy is not currently available for H3N8 CIV; therefore the second diagnosti c research aim was to develop a DIVA test strategy for canine influenza based on the reported differential immune response to th e influenza A NS1 protein. Regarding the epidemiology of H3N8 CIV, several gaps in knowledge were addressed through this resear ch. I t is unknown if H3N8 CIV was infecting dogs prior to 2004, causing undiagnosed or m isdiagnosed respiratory disease; therefore the first

PAGE 151

151 epidemiological study addressed this question by testing archived sera from racing greyhounds and shelter dogs for antibodies to canine and equine H3 proteins in the standard HI assay In addition, r i sk factors for H3N8 CIV infection have not b een defined for individual dogs; therefore the second study utilized a syndromic serosurvey to assess poten tial risk factors fo r infection including intrinsic factors, exposure settings and geographic location. Based on identification of communal settings as risk factors for infection and the realization that factors which promote H3N8 CIV transmission in shelters are unknow n, m athematical modeling was finally utilized to explore which factors promote virus transmission in shelter facilities. The research findings and conclusions are summarized below, followed by issues to be addressed by future investigators Research Conclusion s Diagnostic Test Development Specific aim 1 (Chapter 3): Diagnostic performance of the H3N8 CIV HI assay The results of this study indicate that the standard HI assay is a highly sensitive and specific assay for diagnosis of H3N8 CIV infection in dogs w hen the antibody cuto ff titer for seropositivity is 32. The optimal sensitivity and specifici ty depends on use of treatment with RDE and use of li ve H3N8 CIV isolates. Variation in these paramete rs results in misdiagnosis of infected dogs. Standarization of this assay should ensure accurate diagnosis of H3N8 CIV infected dogs between laboratories and research groups. Specific aim 2 (Chapter 4): Development of a DIVA test strategy for H3N8 CIV In t his study, uninfected and vaccinated dogs could not be differentiated from H3N8 CIV infected dogs in the NS1 immunoassays After multiple troubleshooting efforts

PAGE 152

152 the encountered specificity issues could not be resolved supporting the conclusion that an in fluenza A NS1 DIVA test is not a viable or reliable diagnostic strategy for H3N8 CIV An alternative DIVA strategy is necessary for continued accurate diagnosis of H3N8 CIV infected dogs and for future epidemiological studies of canine influenza in the fac e of vaccination. Epidemiological Investigations Specific aim 1 (Chapter 5): Serological evidence of H3N8 canine influenza like virus circulation in dogs prior to 2004 The evidence from this study suggests that an ancestral H3N8 canine influenza like virus was circulating in the racing greyhound population as early as 1999. Based on the results of this study, greyhounds that raced were more likely to be exposed to H3N8 viruses. Of greyhounds that raced, dogs that raced at 2 or more tracks, dogs that raced f or 6 months or longer, and dogs that raced in 1998, 2002, and 2003 were more likely to be seropositive to CIV and EIV H3 antigens. Although the clinical histories of the tested dogs are unknown, these results suggest that a H3N8 CIV like virus may have bee n associated with r espiratory disease epidemics that occurred at U.S. racetracks prior to 2003. Specific aim 2 (Chapter 6): Prevalence of and risk factors for H3N8 CIV seropositivity in a population of U.S. dogs with ILI This study identified a high prev alence of H3N8 CIV exposure in 1,268 dogs with ILI in 42 states over a 5 year period. Dogs 1 year of age, dogs housed in communal settings such as boarding kennels or shelter facilities and dogs in the No rtheast or West were at great est risk for H3N8 CIV seropositivity These results justify the need for shelters, boarding kennels, and other c ommunal dog housing facilities to formulate, implement and/or evaluate surveillance and biosecurity protocols to reduce the risk of

PAGE 153

153 virus transmission between dogs and avert or mitigate large scale epidemics within these facilities. Specific aim 3 (Chapte r 7): Simulation studies of H3N8 CIV transmission in shelter facilities The results of this study indicate that indirect virus transmission and the number of dogs in a shelter (i.e., the shelter capacity) have the greatest impact on H3N8 CIV transmission d ynamics in shelter facilities. Prevention and control strategies that reduce cont act between dogs and fomites, decrease the amount of infectious virus in the environment, and limit the number of incoming susceptible dogs will likely reduce the impact of ep idemics in shelter facilities and may prevent establishment of endemic infection. Issues for Future Investigators Diagnostic Test Development Moving forward it will be important to continue monitoring for antigenic drift in the CIV H3 protein in order to ensure that the H3N8 CIV used in the HI assay is closely related to that in circulation. Antigenic mismatches between circulating strains and test strains would compromise the sensitivity and specificity of the standard HI assay, resulting in misdiagnosis of infected dogs. Since the H3N8 CIV HI assay cannot differentiate antibodies induced by infection from vaccina tion with the inactivated H3N8 CIV vaccine, it is imperative to know whether a H3 seropositive dog was vaccinated to accurately interpret HI test results. This limits the use of the HI assay in future epidemiological studies of H3N8 CIV in the U.S. if vaccination status is unknown. Since the H3N8 CIV NS1 immunoassays were not a viable DIVA strategy for dogs, an alternative DIVA strategy based on d etection of antibodies to the influenza viral

PAGE 154

154 nucleoprotein (NP) is proposed. This NP DIVA strategy, coupling the NP ELISA with use of a recombinant CIV H3 vaccine, could resolve the diagnostic dilemma created by the currently used inactivated whole virus vaccine. With this approach, H3N8 CIV infected dogs could be differentiated from vaccinated dogs based on the presence of NP antibody: infected dogs would have antibody to both H3 and NP, while dogs vaccinated with the recombinant vaccine would only have a ntibodies to H3. Epidemiological Investigations Studies focused on modeling the efficacy of different intervention strategies, including isolation and vaccination, for averting or mitigating H3N8 CIV epidemics in shelters are being pursued in order to dev elop prevention and control recommendations.

PAGE 155

155 LIST OF REFERENCES 1 Amonsin A, Payungporn S, Theamboonlers A, et al.: 2006, Genetic characterization of H5N1 influenza A viruses isolated from zoo tigers in Thailand. Virology 344 :480 491. 2 Amonsin A, Songserm T, Chutinimitkul S, et al.: 2007, Genetic analysis of influenza A virus (H5N1) derived from domestic cat and dog in Thailand. Arch Virol 152:1925 1933. 3 Ananthanarayan R, Paniker C: 1960, Non specific inhibitors of in fluenza viruses in normal sera. Bull World Health Organ 22:409 419. 4 Arino J, Brauer F, van den Driessche P, et al.: 2006, Simple models for containment of a pandemic. J R Soc Interface 3:453 457. 5 Avellaneda G, Mundt E, Lee CW, et al.: 2010, Diffe rentiation of infected and vaccinated animals (DIVA) using the NS1 protein of avian influenza virus. Avian Dis 54:278 286. 6 Baguelin M, Newton JR, Demiris N, et al.: 2010, Control of equine influenza: scenario testing using a realistic metapopulation m odel of spread. J R Soc Interface 7:67 79. 7 Baigent SJ, McCauley JW: 2003, Influenza type A in humans, mammals, and birds: determinants of virus virulence, host range, and interspecies transmission. BioEssays 25:657 671. 8 Barrell EA, Pecoraro HL, To rres Henderson C, et al.: 2010, Seroprevalence and risk factors for canine H3N8 influenza virus exposure in household dogs in Colorado. J Vet Intern Med 24:1524 1527. 9 Beeler E: 2009, Influenza in dogs and cats. Vet Clin North Am Small Anim Pract 39:2 51 264. 10 Birch Machin I, Rowan A, Pick J, et al.: 1997, Expression of the nonstructural protein NS1 of equine influenza A virus: detection of anti NS1 antibody in post infection equine sera. J Virol Methods 65:255 263. 11 Blower SM, Dowlatabadi H: 1994, Sensitivity and uncertainty analysis of complex models of disease transmission: an HIV model, as an example. Int Stat Rev 62:229 243. 12 Brankston G, Gitterman L, Hirji Z, et al.: 2007, Transmission of influenza A in human beings. Lancet Infect Di s 7:257 265.

PAGE 156

156 13 Breban R, Drake JM, Stallknecht DE, Rohani P: 2009, The role of environmental transmission in recurrent avian influenza epidemics. PLoS Comput Biol 5:e1000346. 14 Bridges C, Katz J, Levandowski R, Cox N: 2008, Inactivated influenza vac cines. In : Vaccines, ed. Plotkin S, Orenstein, W, pp. 259 290. Saunders. 15 Bryant NA, Rash AS, Russell CA, et al.: 2009, Antigenic and genetic variations in European and North American equine influenza virus strains (H3N8) isolated from 2006 to 2007. Vet Microbiol 138:41 52. 16 Butler D: 2006, Thai dogs carry bird flu virus, but will they spread it? Nature 439:773. 17 Castleman WL, Powe JR, Crawford PC, et al.: 2010, Canine H3N8 influenza virus infection in dogs and mice. Vet Pathol 47:507 517. 18 Chang CP, New AE, Taylor JF, Chiang HS: 1976, Influenza virus isolations from dogs during a human epidemic in Taiwan. Int J Zoonoses 3:61 64. 19 Chen W, Calvo PA, Malide D, et al.: 2001, A novel influenza A virus mitochondrial protein that induces cell death. Nat Med 7:1306 1312. 20 Crawford PC, Dubovi EJ, Castleman WL, et al.: 2005, Transmission of equine influenza virus to dogs. Science 310:482 485. 21 Daly JM, Blunden AS, Macrae S, et al.: 2008, Transmission of equine influenza virus to eng lish foxhounds. Emerg Infect Dis 14:461 464. 22 Daly JM, Lai AC, Binns MM, et al.: 1996, Antigenic and genetic evolution of equine H3N8 influenza A viruses. J Gen Virol 77 ( Pt 4):661 671. 23 Daly JM, Newton JR, Mumford JA: 2004, Current perspectives on control of equine influenza. Vet Res 35:411 423. 24 De Benedictis P, Anderson TC, Perez A, et al.: 2010, A diagnostic algorithm for detection of antibodies to influenza A viruses in dogs in Italy (2006 2008). J Vet Diagn Invest 22:914 920. 25 de l a Rua Domenech R, Reid SW, Gonzalez Zariquiey AE, et al.: 2000, Modelling the spread of a viral infection in equine populations managed in Thoroughbred racehorse training yards. Prev Vet Med 47:61 77. 26 Deshpande M, Abdelmagid O, Tubbs A, et al.: 2009 Experimental reproduction of canine influenza virus H3N8 infection in young puppies. Vet Ther 10:29 39.

PAGE 157

157 27 Deshpande MS, Jirjis FF, Tubbs AL, et al.: 2009, Evaluation of the efficacy of a canine influenza virus (H3N8) vaccine in dogs following experime ntal challenge. Vet Ther 10:103 112. 28 Dubovi EJ: 2010, Canine influenza. Vet Clin North Am Small Anim Pract 40:1063 1071. 29 Dubovi EJ, Njaa BL: 2008, Canine influenza. Vet Clin North Am Small Anim Pract 38:827 835. 30 Dundon WG, De Benedictis P, Viale E, Capua I: 2010, Serologic evidence of pandemic (H1N1) 2009 infection in dogs, Italy. Emerg Infect Dis 16:2019 2021. 31 Dundon WG, Maniero S, Toffan A, et al.: 2007, Appearance of serum antibodies against the avian influenza nonstructural 1 pro tein in experimentally infected chickens and turkeys. Avian Dis 51:209 212. 32 Dushoff J, Plotkin JB, Viboud C, et al.: 2007, Vaccinating to protect a vulnerable subpopulation. PLoS Med 4:e174. 33 Ferguson NM, Cummings DA, Cauchemez S, et al.: 2005, Strategies for containing an emerging influenza pandemic in Southeast Asia. Nature 437:209 214. 34 Giese M, Harder TC, Teifke JP, et al.: 2008, Experimental infection and natural contact exposure of dogs with avian influenza virus (H5N1). Emerg Infect D is 14:308 310. 35 Glass K, Wood JL, Mumford JA, et al.: 2002, Modelling equine influenza 1: a stochastic model of within yard epidemics. Epidemiol Infect 128:491 502. 36 Greiner M, Pfeiffer D, Smith RD: 2000, Principles and practical application of t he receiver operating characteristic analysis for diagnostic tests. Prev Vet Med 45:23 41. 37 Guan Y, Poon LL, Cheung CY, et al.: 2004, H5N1 influenza: a protean pandemic threat. Proc Natl Acad Sci U S A 101:8156 8161. 38 Guo YJ, Jin FG, Wang P, et al .: 1983, Isolation of influenza C virus from pigs and experimental infection of pigs with influenza C virus. J Gen Virol 64 (Pt 1):177 182. 39 Hale BG, Randall RE, Ortin J, Jackson D: 2008, The multifunctional NS1 protein of influenza A viruses. J Gen Virol 89:2359 2376. 40 Harlow E, Lane D: 1988, Antibodies: a laboratory manual, p. 633. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

PAGE 158

158 41 Hashimoto Y, Yamakawa T, Tanabe Y: 1984, Further studies on the red cell glycolipids of various breeds of dogs. A possible assumption about the origin of Japanese dogs. J Biochem 96:1777 1782. 42 Hayward JJ, Dubovi EJ, Scarlett JM, et al.: 2010, Microevolution of canine influenza virus in shelters and its molecular epidemiology in the United States. J Viro l 84:12636 12645. 43 Hoelzer K, Murcia PR, Baillie GJ, et al.: 2010, Intrahost evolutionary dynamics of canine influenza virus in naive and partially immune dogs. J Virol 84:5329 5335. 44 Holt DE, Mover MR, Brown DC: 2010, Serologic prevalence of ant ibodies against canine influenza virus (H3N8) in dogs in a metropolitan animal shelter. J Am Vet Med Assoc 237:71 73. 45 Hosmer DW, Lemeshow S: 1989, Logistic regression for matched case control studies. In : Applied logistic regression, eds. Hosmer DWL emeshow S, pp. 187 215. Wiley, New York. 46 Houser RE, Heuschele WP: 1980, Evidence of prior infection with influenza A/Texas/77 (H3N2) virus in dogs with clinical parainfluenza. Can J Comp Med 44:396 402. 47 Hsu SB, Hsieh YH: 2008, On the role of as ymptomatic infection in transmission dynamics of infectious diseases. Bull Math Biol 70:134 155. 48 Ito T, Suzuki Y, Mitnaul L, et al.: 1997, Receptor specificity of influenza A viruses correlates with the agglutination of erythrocytes from different an imal species. Virology 227:493 499. 49 Jirjis FF, Deshpande MS, Tubbs AL, et al.: 2010, Transmission of canine influenza virus (H3N8) among susceptible dogs. Vet Microbiol 144:303 309. 50 Joh RI, Wang H, Weiss H, Weitz JS: 2009, Dynamics of indirectl y transmitted infectious diseases with immunological threshold. Bull Math Biol 71:845 862. 51 Kaplan MM, Payne AM: 1959, Serological survey in animals for type A influenza in relation to the 1957 pandemic. Bull World Health Organ 20:465 488. 52 Karaca K, Dubovi EJ, Siger L, et al.: 2007, Evaluation of the ability of canarypox vectored equine influenza virus vaccines to induce humoral immune responses against canine influenza viruses in dogs. Am J Vet Res 68:208 212. 53 Kilbourne ED, Kehoe JM: 1975, Demonstration of antibodies to both hemagglutinin and neuraminidase antigens of H3N2 influenza A virus in domestic dogs. Intervirology 6:315 318.

PAGE 159

159 54 Kim WI, Wu WH, Janke B, Yoon KJ: 2006, Characterization of the humoral immune response of experimentally infected and vaccinated pigs to swine influenza viral proteins. Arch Virol 151:23 36. 55 Kirkland PD, Finlaison DS, Crispe E, Hurt AC: 2010, Influenza virus infection in dogs transmission during the Australian equine influenza outbreak. Emerg Infect D is 16:699 702. 56 Klopfleisch R, Wolf PU, Wolf C, et al.: 2007, Encephalitis in a Stone Marten (Martes foina) after Natural Infection with Highly Pathogenic Avian Influenza Virus Subtype H5N1. J Comp Pathol 137:155 159. 57 Knesl O, Allan FJ, Shields S : 2009, The seroprevalence of canine respiratory coronavirus and canine influenza virus in dogs in New Zealand. N Z Vet J 57:295 298. 58 Krug RM, Etkind PR: 1973, Cytoplasmic and nuclear virus specific proteins in influenza virus infected MDCK cells. V irology 56:334 348. 59 Kruth SA, Carman S, Weese JS: 2008, Seroprevalence of antibodies to canine influenza virus in dogs in Ontario. Can Vet J 49:800 802. 60 Kuiken T, Holmes EC, McCauley J, et al.: 2006, Host species barriers to influenza virus inf ections. Science 312:394 397. 61 Lai AC, Chambers TM, Holland RE, Jr., et al.: 2001, Diverged evolution of recent equine 2 influenza (H3N8) viruses in the Western Hemisphere. Arch Virol 146:1063 1074. 62 Lai AC, Rogers KM, Glaser A, et al.: 2004, Alt ernate circulation of recent equine 2 influenza viruses (H3N8) from two distinct lineages in the United States. Virus Res 100:159 164. 63 Landolt GA, Olsen CW: 2007, Up to new tricks a review of cross species transmission of influenza A viruses. Anim Health Res Rev 8:1 21. 64 Lazarowitz SG, Compans RW, Choppin PW: 1971, Influenza virus structural and nonstructural proteins in infected cells and their plasma membranes. Virology 46:830 843. 65 Le Menach A, Vergu E, Grais RF, et al.: 2006, Key strat egies for reducing spread of avian influenza among commercial poultry holdings: lessons for transmission to humans. Proc Royal Soc B Biological Sciences 273:2467 2475. 66 Lee C, Song D, Kang B, et al.: 2009, A serological survey of avian origin canine H 3N2 influenza virus in dogs in Korea. Vet Microbiol 137:359 362.

PAGE 160

160 67 Lipatov AS, Govorkova EA, Webby RJ, et al.: 2004, Influenza: emergence and control. J Virol 78:8951 8959. 68 Longini IM, Jr., Halloran ME, Nizam A, Yang Y: 2004, Containing pandemic i nfluenza with antiviral agents. Am J Epidemiol 159:623 633. 69 Longini IM, Jr., Nizam A, Xu S, et al.: 2005, Containing pandemic influenza at the source. Science 309:1083 1087. 70 Lord CC, Day JF: 2001, Simulation studies of St. Louis encephalitis vi rus in south Florida. Vector Borne Zoonotic Dis 1:299 315. 71 Lu Z, Dubovi EJ, Zylich NC, et al.: 2010, Diagnostic application of H3N8 specific equine influenza real time reverse transcription polymerase chain reaction assays for the detection of Canine influenza virus in clinical specimens. J Vet Diagn Invest 22:942 945. 72 Maas R, Tacken M, Ruuls L, et al.: 2007, Avian influenza (H5N1) susceptibility and receptors in dogs. Emerg Infect Dis 13:1219 1221. 73 Manuguerra JC, Hannoun C: 1992, Natural infection of dogs by influenza C virus. Res Virol 143:199 204. 74 Manuguerra JC, Hannoun C, Simon F, et al.: 1993, Natural infection of dogs by influenza C virus: a serological survey in Spain. New Microbiol 16:367 371. 75 Martella V, Elia G, Decaro N et al.: 2007, An outbreak of equine influenza virus in vaccinated horses in Italy is due to an H3N8 strain closely related to recent North American representatives of the Florida sub lineage. Vet Microbiol 121:56 63. 76 Mathews JD, McCaw CT, McVernon J, et al.: 2007, A biological model for influenza transmission: pandemic planning implications of asymptomatic infection and immunity. PLoS One 2:e1220. 77 Matrosovich MN, Matrosovich TY, Gray T, et al.: 2004, Human and avian influenza viruses target d ifferent cell types in cultures of human airway epithelium. Proc Natl Acad Sci U S A 101:4620 4624. 78 McKinley ET, Spackman E, Pantin Jackwood MJ: 2010, The pathogenesis of H3N8 canine influenza virus in chickens, turkeys and ducks. Influenza Other Res pi Viruses 4:353 356. 79 Medeiros R, Escriou N, Naffakh N, et al.: 2001, Hemagglutinin residues of recent human A(H3N2) influenza viruses that contribute to the inability to agglutinate chicken erythrocytes. Virology 289:74 85.

PAGE 161

161 80 Muranaka M, Yamanaka T, Katayama Y, et al.: 2010, Distribution of Influenza Virus Sialoreceptors on Upper and Lower Respiratory Tract in Horses and Dogs. J Vet Med Sci 73:125 127. 81 Myers KP, Olsen CW, Gray GC: 2007, Cases of swine influenza in humans: a review of the lit erature. Clin Infect Dis 44:1084 1088. 82 National Greyhound Association: 1999, The Greyhound Review 27:63. 83 National Greyhound Association: 1999, The Greyhound Review 27:17 18. 84 National Greyhound Association: 1999, The Greyhound Review 27:20 85 National Greyhound Association: 2003, The Greyhound Review 31:75. 86 National Greyhound Association: 1999, The Greyhound Review 27:75. 87 National Greyhound Association: 1999, The Greyhound Review 27:19. 88 Nayak DP, Balogun RA, Yamada H, e t al.: 2009, Influenza virus morphogenesis and budding. Virus Res 143:147 161. 89 Neumann G, Kawaoka Y: 2006, Host range restriction and pathogenicity in the context of influenza pandemic. Emerg Infect Dis 12:881 886. 90 Newton JR, Daly JM, Spencer L Mumford JA: 2006, Description of the outbreak of equine influenza (H3N8) in the United Kingdom in 2003, during which recently vaccinated horses in Newmarket developed respiratory disease. Vet Rec 158:185 192. 91 Newton R, Cooke A, Elton D, et al.: 20 07, Canine influenza virus: cross species transmission from horses. Vet Rec 161:142 143. 92 Nikitin A, Cohen D, Todd JD, Lief FS: 1972, Epidemiological studies of A Hong Kong 68 virus infection in dogs. Bull World Health Organ 47:471 479. 93 Osterhaus AD, Rimmelzwaan GF, Martina BE, et al.: 2000, Influenza B virus in seals. Science 288:1051 1053. 94 Ozaki H, Sugiura T, Sugita S, et al.: 2001, Detection of antibodies to the nonstructural protein (NS1) of influenza A virus allows distinction between vaccinated and infected horses. Vet Microbiol 82:111 119. 95 Palese P, Shaw M: 2006, Orthomyxoviridae: The Viruses and Their Replication. In : Fields Virology, eds. Knipe D, Howley P, Griffin D, et al., 5th ed., pp. 1647 1689. Lippincott Williams & Wilk ins, Philadelphia.

PAGE 162

162 96 Park AW, Wood JL, Daly JM, et al.: 2004, The effects of strain heterology on the epidemiology of equine influenza in a vaccinated population. Proc Biol Sci 271:1547 1555. 97 Park AW, Wood JL, Newton JR, et al.: 2003, Optimising v accination strategies in equine influenza. Vaccine 21:2862 2870. 98 Payungporn S, Crawford PC, Kouo TS, et al.: 2008, Influenza A virus (H3N8) in dogs with respiratory disease, Florida. Emerg Infect Dis 14:902 908. 99 Peiris JS, Poon LL, Guan Y: 2009 Emergence of a novel swine origin influenza A virus (S OIV) H1N1 virus in humans. J Clin Virol 45:169 173. 100 Peiris JS, Yu WC, Leung CW, et al.: 2004, Re emergence of fatal human influenza A subtype H5N1 disease. Lancet 363:617 619. 101 Piccirillo A, Pasotto D, Moreno Martin A, Cordioli P: 2009, Serological Survey for Influenza Type A Viruses in Domestic Dogs (Canis Lupus Familiaris) and Cats (Felis Catus) in North Eastern Italy. Zoonoses Public Health 57:239 243. 102 Powe J, Castleman W: 2009, Canine influenza virus replicates in alveolar macrophages and induces TNF {alpha}. Vet Pathol 46:1187 1196. 103 Rivailler P, Perry IA, Jang Y, et al.: 2010, Evolution of canine and equine influenza (H3N8) viruses co circulating between 2005 and 2008. V irology 408:71 79. 104 Roche B, Lebarbenchon C, Gauthier Clerc M, et al.: 2009, Water borne transmission drives avian influenza dynamics in wild birds: the case of the 2005 2006 epidemics in the Camargue area. Infect Genet Evol 9:800 805. 105 Rohani P Breban R, Stallknecht DE, Drake JM: 2009, Environmental transmission of low pathogenicity avian influenza viruses and its implications for pathogen invasion. Proc Natl Acad Sci U S A 106:10365 10369. 106 Romvary J, Rozsa J, Farkas E: 1975, Infection of dogs and cats with the Hong Kong influenza A (H3N2) virus during an epidemic period in Hungary. Acta Vet Acad Sci Hung 25:255 259. 107 Rosas C, Van de Walle GR, Metzger SM, et al.: 2008, Evaluation of a vectored equine herpesvirus type 1 (EHV 1) vacc ine expressing H3 haemagglutinin in the protection of dogs against canine influenza. Vaccine 26:2335 2343. 108 Rowe T, Abernathy RA, Hu Primmer J, et al.: 1999, Detection of antibody to avian influenza A (H5N1) virus in human serum by using a combinatio n of serologic assays. J Clin Microbiol 37:937 943.

PAGE 163

163 109 Saenz RA, Hethcote HW, Gray GC: 2006, Confined animal feeding operations as amplifiers of influenza. Vector Borne Zoonotic Dis 6:338 346. 110 Seiler BM, Yoon KJ, Andreasen CB, et al.: 2010, Antib odies to influenza A virus (H1 and H3) in companion animals in Iowa, USA. Vet Rec 167:705 707. 111 Serra VF, Stanzani G, Smith G, Otto CM: 2011, Point seroprevalence of canine influenza virus H3N8 in dogs participating in a flyball tournament in Pennsyl vania. J Am Vet Med Assoc 238:726 730. 112 Song D, Kang B, Lee C, et al.: 2008, Transmission of avian influenza virus (H3N2) to dogs. Emerg Infect Dis 14:741 746. 113 Song D, Lee C, Kang B, et al.: 2009, Experimental infection of dogs with avian orig in canine influenza A virus (H3N2). Emerg Infect Dis 15:56 58. 114 Songserm T, Amonsin A, Jam on R, et al.: 2006, Fatal avian influenza A H5N1 in a dog. Emerg Infect Dis 12:1744 1747. 115 Songsermn T, Amonsin A, Jam on R, et al.: 2006, Avian influenz a H5N1 in naturally infected domestic cat. Emerg Infect Dis 12:681 683. 116 Stephenson I, Wood JM, Nicholson KG, Zambon MC: 2003, Sialic acid receptor specificity on erythrocytes affects detection of antibody to avian influenza haemagglutinin. J Med Vir ol 70:391 398. 117 Suarez DL: 2005, Overview of avian influenza DIVA test strategies. Biologicals 33:221 226. 118 Subbarao K, Klimov A, Katz J, et al.: 1998, Characterization of an avian influenza A (H5N1) virus isolated from a child with a fatal res piratory illness. Science 279:393 396. 119 Suzuki Y: 2005, Sialobiology of influenza: molecular mechanism of host range variation of influenza viruses. Biol Pharm Bull 28:399 408. 120 Suzuki Y, Ito T, Suzuki T, et al.: 2000, Sialic acid species as a determinant of the host range of influenza A viruses. J Virol 74:11825 11831. 121 Todd JD, Cohen D: 1968, Studies of influenza in dogs. I. Susceptibility of dogs to natural and experimental infection with human A2 and B strains of influenza virus. Am J Epidemiol 87:426 439. 122 Topciu V, Diosi P, Moldovan E, et al.: 1967, Serological survey concerning the presence of influenza haemagglutination inhibitors in normal dog sera. Acta Virol 11:55 57.

PAGE 164

164 123 Tu J, Zhou H, Jiang T, et al.: 2009, Isolation and molecular characterization of equine H3N8 influenza viruses from pigs in China. Arch Virol 154:887 890. 124 Tumpey TM, Alvarez R, Swayne DE, Suarez DL: 2005, Diagnostic approach for differentiating infected from vaccinated poultry on the basis of antib odies to NS1, the nonstructural protein of influenza A virus. J Clin Microbiol 43:676 683. 125 van der Goot JA, Engel B, van de Water SG, et al.: 2010, Validation of diagnostic tests for detection of avian influenza in vaccinated chickens using Bayesian analysis. Vaccine 28:1771 1777. 126 Vonholdt BM, Pollinger JP, Lohmueller KE, et al.: 2010, Genome wide SNP and haplotype analyses reveal a rich history underlying dog domestication. Nature 464:898 902. 127 Watson DS, Reddy SM, Brahmakshatriya V, Lup iani B: 2009, A multiplexed immunoassay for detection of antibodies against avian influenza virus. J Immunol Methods 340:123 131. 128 Webby R, Hoffmann E, Webster R: 2004, Molecular constraints to interspecies transmission of viral pathogens. Nat Med 1 0:S77 81. 129 Weber TP, Stilianakis NI: 2008, Inactivation of influenza A viruses in the environment and modes of transmission: a critical review. J Infect 57:361 373. 130 Webster RG: 1998, Influenza: an emerging disease. Emerg Infect Dis 4:436 441. 131 Webster RG: 2002, The importance of animal influenza for human disease. Vaccine 20 Suppl 2:S16 20. 132 Webster RG, Bean WJ, Gorman OT, et al.: 1992, Evolution and ecology of influenza A viruses. Microbiol Rev 56:152 179. 133 Wise HM, Foeglein A Sun J, et al.: 2009, A complicated message: Identification of a novel PB1 related protein translated from influenza A virus segment 2 mRNA. J Virol 83:8021 8031. 134 World Health Organization: 2002, WHO Manual on Animal Influenza Diagnosis and Survei llance. World Health Organization Global Influenza Programme, Geneva, Switzerland. 135 World Organization for Animal Health: 2008, Manual of diagnostic tests and vaccines for terrestrial animals, 6th ed. World Organization for Animal Health (OIE), Paris France.

PAGE 165

165 136 Wright P, Neumann G, Kawaoka Y: 2006, Orthomyxoviruses In : Fields Virology, eds. Knipe D, Howley P, Griffin D, et al., 5th ed., pp. 1691 1740. Lippincott Williams & Wilkins, Philadelphia. 137 Yamanaka T, Nemoto M, Tsujimura K, et al.: 2 009, Interspecies transmission of equine influenza virus (H3N8) to dogs by close contact with experimentally infected horses. Vet Microbiol 139:351 355. 138 Yamanaka T, Niwa H, Tsujimura K, et al.: 2008, Epidemic of equine influenza among vaccinated rac ehorses in Japan in 2007. J Vet Med Sci 70:623 625. 139 Yamanaka T, Tsujimura K, Kondo T, et al.: 2010, Infectivity and pathogenicity of canine H3N8 influenza A virus in horses. Influenza Other Respi Viruses 4:345 351. 140 Yasue S, Handa S, Miyagawa S et al.: 1978, Difference in form of sialic acid in red blood cell glycolipids of different breeds of dogs. J Biochem 83:1101 1107. 141 Yoon KJ, Cooper VL, Schwartz KJ, et al.: 2005, Influenza virus infection in racing greyhounds. Emerg Infect Dis 11: 1974 1976. 142 Zell R, Krumbholz A, Eitner A, et al.: 2007, Prevalence of PB1 F2 of influenza A viruses. J Gen Virol 88:536 546. 143 Zhao S, Jin M, Li H, et al.: 2005, Detection of antibodies to the nonstructural protein (NS1) of avian influenza viru ses allows distinction between vaccinated and infected chickens. Avian Dis 49:488 493.

PAGE 166

166 BIOGRAPHICAL SKETCH Tara Creel Anderson graduated with a Bachelor of Science in wildlife ecology and c on servation in 1999 and earn ed her Doctor of Veterinary Medici ne (DVM) in 2003, both from the University of Florida. After graduation she practiced general small animal medicine and surgery in Gainesville, Florida for two years. She returned to graduate school at the University of Florida in the summer of 2005 to pur sue a combined Master of Public Health/Doctor of Philosophy (MPH/PhD) program. During her studies she served on the joint DVM/MPH program co mmittee and was fortunate to work with several collaborators in influenza laboratories in the U.S. and Italy. She re ceived her MPH with a focus in epidemiology in 2007, and completed her PhD in veterinary medical s ciences in 2011.