A diagnostic algorithm for detection of antibodies to influenza A viruses in dogs in Italy (2006-2008).

Abstract

Since 2004, there have been several reports of Influenza A virus (FLUAV) infection in dogs. Dogs have been infected with equine influenza H3N8, avian influenza H3N2 and H5N1, and the pandemic H1N1 virus. Because of recent avian and equine influenza outbreaks in Italy, the objectives of the present study were to estimate the level of exposure of Italian dogs to influenza A viruses and to assess a diagnostic algorithm for detection of FLUAV exposure in dogs. Sera collected from 6,858 dogs from 2006 to 2008 were screened in a competitive enzyme-linked immunosorbent assay (cELISA) for antibodies to the highly conserved influenza A nucleoprotein. Samples positive in the cELISA were confirmed by testing in hemagglutination inhibition (HI) and fluorescent antibody test (FAT). Two seropositive dogs had antibodies to H3 hemagglutinin proteins, consistent with exposure to recent canine and equine subtype H3N8 viruses. Using a Bayesian model, the sensitivity and specificity of the cELISA were estimated as 93.98% (probability intervals [PI]: 81.67-99.08%) and 98.71% (PI: 98.43-98.96%), respectively. After accounting for the imperfect sensitivity and specificity of the cELISA, the Bayesian posterior prevalence of FLUAV exposure among tested Italian dogs was 0.5% (PI: 0.1-1.4%). The study results indicate that screening with a cELISA for influenza A nucleoprotein antibody, followed by confirmatory testing with HI and/or FAT, is a highly sensitive and highly specific approach for diagnosing FLUAV exposure in dogs.

Full text

Page 1

http://vdi.sagepub.com/
Investigation
Journal of Veterinary Diagnostic
http://vdi.sagepub.com/content/22/6/914
The online version of this article can be found at:

DOI: 10.1177/104063871002200610
2010 22: 914J VET Diagn Invest
Cynda Crawford and Ilaria Capua
Paola De Benedictis, Tara C. Anderson, Andrés Perez, Elisabetta Viale, Crispina Veggiato, Silvia Tiozzo Caenazzo, P.
2008)
−
A Diagnostic Algorithm for Detection of Antibodies to Influenza A Viruses in Dogs in Italy (2006


Published by:

http://www.sagepublications.com
On behalf of:




Official Publication of the American Association of Veterinary Laboratory Diagnosticians, Inc.
can be found at:
Journal of Veterinary Diagnostic Investigation

Additional services and information for





http://vdi.sagepub.com/cgi/alerts
Email Alerts:

http://vdi.sagepub.com/subscriptions
Subscriptions:



http://www.sagepub.com/journalsReprints.nav
Reprints:


http://www.sagepub.com/journalsPermissions.nav
Permissions:
What is This?

- Nov 1, 2010Version of Record >>
at UNIV HAWAII on February 1, 2012 vdi.sagepub.comDownloaded from

Page 2

A diagnostic algorithm for detection of antibodies to influenza A
viruses in dogs in Italy (2006–2008)
Paola De Benedictis, Tara C. Anderson, Andre ´s Perez, Elisabetta Viale, Crispina Veggiato,
Silvia Tiozzo Caenazzo, P. Cynda Crawford, Ilaria Capua1
Abstract.
Dogs have been infected with equine influenza H3N8, avian influenza H3N2 and H5N1, and the pandemic
H1N1 virus. Because of recent avian and equine influenza outbreaks in Italy, the objectives of the present
study were to estimate the level of exposure of Italian dogs to influenza A viruses and to assess a diagnostic
algorithm for detection of FLUAV exposure in dogs. Sera collected from 6,858 dogs from 2006 to 2008 were
screened in a competitive enzyme-linked immunosorbent assay (cELISA) for antibodies to the highly
conserved influenza A nucleoprotein. Samples positive in the cELISA were confirmed by testing in
hemagglutination inhibition (HI) and fluorescent antibody test (FAT). Two seropositive dogs had antibodies
to H3 hemagglutinin proteins, consistent with exposure to recent canine and equine subtype H3N8 viruses.
Using a Bayesian model, the sensitivity and specificity of the cELISA were estimated as 93.98% (probability
intervals [PI]: 81.67–99.08%) and 98.71% (PI: 98.43–98.96%), respectively. After accounting for the imperfect
sensitivity and specificity of the cELISA, the Bayesian posterior prevalence of FLUAV exposure among tested
Italian dogs was 0.5% (PI: 0.1–1.4%). The study results indicate that screening with a cELISA for influenza A
nucleoprotein antibody, followed by confirmatory testing with HI and/or FAT, is a highly sensitive and highly
specific approach for diagnosing FLUAV exposure in dogs.
Since 2004, there have been several reports of Influenza A virus (FLUAV) infection in dogs.
Key words:
antibody test; Influenza A virus; Italy; serology.
Dogs; enzyme-linked immunosorbent assay; hemagglutination inhibition; fluorescent
Introduction
Since the recognition of canine Influenza A virus
(FLUAV) subtype H3N8 (CIV H3N8) in the United
States, there has been renewed interest in the potential
role of dogs in the ecology of influenza A viruses. In
other countries, serological surveys14,15of dogs for
evidence of exposure to CIV H3N8 have been
conducted with largely negative results; however,
additional transmissions of influenza A viruses from
multiple species to dogs have been reported. Retro-
spective studies implicated equine FLUAV subtype
H3N8 as the cause of isolated respiratory outbreaks
in dogs in the United Kingdom in 2002 and 200312,20
and in Australia,13where dozens of dogs in contact
with infected horses developed influenza-like illness
during the 2007 equine influenza outbreak.
During the spread of highly pathogenic avian
FLUAV subtype H5N1 throughout Asia, a dog in
Thailand died after ingesting an H5N1-infected
duck.1,25A subsequent serosurvey5revealed that
25% of 629 asymptomatic dogs had been exposed to
the virus in Thailand. The transmission of an entire
avian FLUAV subtype H3N2 to dogs in South Korea
was identified in 2007,23and additional studies16,24
support the sustained transmission of this avian-
origin H3N2 virus between South Korean dogs. Since
its appearance in April 2009, pandemic H1N1 has
also been reported in dogs in both the United States
and China [ProMED-mail: 2009, PRO/AH/EDR.In-
fluenza pandemic (H1N1) 2009, animal (40): USA
(NY) canine. Available at http://www.promedmail.
org/pls/apex/f?p52400:1000,archivenumber 20091222.
4305. Accessed July 17, 2010; ProMED-mail: 2009,
PRO/AH/EDR.Influenza pandemic (H1N1) 2009,
animal (30): China, canine. Available at http://www.
From the Istituto Zooprofilattico Sperimentale delle Venezie,
World Animal Health Organization/Food and Agriculture Orga-
nization (OIE/FAO) and the National Reference Laboratory for
Avian Influenza and Newcastle Disease, OIE Collaborating
Centre for Diseases at the Human Animal Interface, Legnaro,
Padova, Italy (De Benedictis, Viale, Veggiato, Tiozzo Caenazzo,
Capua); the University of Florida, College of Veterinary
Medicine, Gainesville, FL (Anderson, Crawford); the Center for
Animal Diseases Modeling and Surveillance, School of Veterinary
Medicine, University of California, Davis, CA (Perez); and the
Consejo Nacional de Investigaciones Cientı ´ficas y Te ´cnicas,
Facultad de Ciencias Veterinarias, Universidad Nacional de
Rosario, Argentina (Perez).
1Corresponding Author: Ilaria Capua, Istituto Zooprofilattico
Sperimentale delle Venezie, OIE/FAO and National Reference
Laboratory for Avian Influenza and Newcastle Disease, OIE
Collaborating Centre for Diseases at the Human Animal
Interface, Viale dell’Universita ` 10, 35020 Legnaro, Padova, Italy.
icapua@izsvenezie.it
J Vet Diagn Invest 22:914–920 (2010)
914
at UNIV HAWAII on February 1, 2012vdi.sagepub.comDownloaded from

Page 3

promedmail.org/pls/apex/f?p52400:1000, archive num-
ber 20091128.4079. Accessed July 17, 2010]. Clearly,
these examples confirm the susceptibility of dogs to
influenza virus infection.
Serological studies11,15–17,21,23of influenza A expo-
sure in dogs have been conducted by means of
enzyme-linked immunosorbent assay (ELISA) and/or
hemagglutination inhibition (HI) assay as screening
and confirmatory tests, respectively. The sensitivity
(Se) and specificity (Sp) of these assays under
unknown field conditions have not been reported
for canine sera. Several ELISA kits for influenza A
nucleoprotein have recently been developed. Most
commercially available kits are indirect ELISAs,
which have been validated for select species. Others
are based on a competitive ELISA (cELISA) format
and could theoretically be used for detection of serum
antibodies to avian or mammalian influenza A viruses
in any species. No ELISA kits, however, have been
developed, validated, or authorized for use with
canine sera. The HI assay is universally considered
the gold standard for serological diagnosis of
influenza A viruses in animals, and it generates
indirect evidence of the viral hemagglutinin subtype
to which an animal has been exposed. However, its
reliability is dependent on a variety of factors,
including the receptor affinity of the virus (a-2,3 or
a-2,6); the quality, concentration, and origin of the
erythrocyte suspension used in the assay; and the
homology of antigen used with the circulating
virus.6,19In addition, the HI assay is a time-con-
suming and labor-intensive test that has the major
diagnostic characteristic of being subtype specific
and, thus, is not suitable for a serosurvey aimed at
detection of multiple virus subtypes.
Since 2000, several FLUAV outbreaks have oc-
curred in poultry (H7N1, H7N3, H5N2) and horses
(H3N8) in Italy,7,8,18providing ample opportunities for
dogs to be exposed to these viruses by direct or indirect
contact. A recent study,22which investigated the
potential exposure of 637 dogs and cats to influenza
A viruses in northeastern Italy using an in-house
ELISA, did not identify any seropositive animals. To
date, however, there have not been any studies that
have comprehensively investigated the potential expo-
sure of dogs to influenza A viruses in Italy. The
objectives of the present study were to estimate the
level of exposure of Italian dogs to influenza A viruses
and to assess a diagnostic algorithm for detection of
FLUAV exposure in dogs using a combination of
serological tests, including cELISA, HI, and fluores-
cent antibody test (FAT). A Bayesian analysis was
used to determine the Se and Sp of the cELISA for
identifying FLUAV exposure in dogs with unknown
potential for contact with infected animals.
Materials and methods
Estimating the level of exposure of Italian dogs to
influenza A viruses
Samples.
Italy from 2006 to 2008 were originally submitted to the
Istituto Zooprofilattico Sperimentale delle Venezie (IZSVe;
Legnaro, Padova, Italy) for assessment of rabies vaccine
efficacy. Serum samples from 6,858 dogs that were at least
1 year of age and represented all 20 regions of Italy were
selected for testing for antibodies to influenza A viral
proteins (Fig. 1). Serum samples had been stored at 220uC
prior to testing. The selected samples included 2,041 dogs
from 2006; 2,038 dogs from 2007; and 2,779 dogs from
2008. The total sample size is equivalent to accepting an
error in the point estimates of 0.15%, assuming an expected
prevalence of 0.4%, a 95% confidence level, and a
population size that approximates to infinity. For example,
if the true prevalence was 0.4% and the diagnostic test was
perfect, this sample size would provide detection of a
prevalence value between 0.25% and 0.55% with a 95%
level of confidence.
A panel of 97 convalescent serum samples collected from
racing greyhounds naturally infected with A/canine/FL/04
(H3N8) served as reference sera in the study. These samples
were previously confirmed11as seropositive for antibodies
to the H3 protein in HI and virus neutralization assays. The
reference sera served as positive controls in all assays. Sera
collected from specific-pathogen-free dogs housed in a
barrier research facility served as negative controls.
Diagnostic tests.
Testing for antibody to influenza A
nucleoprotein was performed using a commercially avail-
able cELISA.aIn addition to its high throughput format,
this assay was selected as the initial screening test because it
detects exposure to any FLUAV strain, regardless of the
subtype. The test was performed according to the
manufacturer’s instructions. Samples were diluted at a
1:10 ratio in the provided dilution buffer. Positive and
negative control sera provided in the kit as well as a
reference positive and negative canine serum sample were
included in each test plate.
All Italian dog sera that had positive or doubtful results
in the cELISA (competition percentage greater than 45%)
were tested in a FAT, as an influenza type A–specific
confirmatory test. An in-house FAT was developed to
detect antibodies to influenza A viral proteins in infected
cells. Briefly, Madin–Darby canine kidney cells were
cultured in Eagle minimum essential medium with 10%
fetal calf serum and 1% L-glutamine.bA 50-ml volume of
cell culture suspension (1.5 3 104cells/well) was dispensed
into 96-well platescand incubated for 24 hr at 37uC (62uC)
in the presence of 5% CO2. Plates were washed with Earle
balanced salt solutiondto remove residual calf serum. The
wells were inoculated with 1 multiplicity of infection of A/
canine/FL/04 (H3N8) diluted in Eagle minimum essential
medium containing 1% L-glutamine and 0.01% trypsin.d
When the beginning of cytopathic effects appeared at 24 hr
postinfection, the cells were fixed with 80% acetone (v/v in
sterile distilled water). As a result of the small amount of
available sera, samples were tested in singlet. A reference
Serum samples collected from pet dogs in
Serological detection of Influenza A virus exposure in dogs915
at UNIV HAWAII on February 1, 2012vdi.sagepub.comDownloaded from

Page 4

Figure 1.
.100 samples; light blue dots: 10–100 samples; green dots: ,10 samples. The actual number of samples is indicated for each dot.
Map showing the distribution and number of serum samples collected from dogs in Italy from 2006 to 2008. Purple dots:
916
De Benedictis et al.
at UNIV HAWAII on February 1, 2012vdi.sagepub.comDownloaded from

Page 5

positive and negative canine serum sample was tested on
each plate. Positive and negative reference sera were also
tested in wells containing uninfected cells to control for
nonspecific reactions. Starting at a dilution of 1:8, 10 serial
log2dilutions of each canine serum sample were prepared in
phosphate buffered saline (PBS). Briefly, 50 ml of each
serum dilution was added to the wells and the plates were
incubated for 30 min at 37uC and then washed 3 times with
PBS. Specific anti–influenza A antibodies in tested sera
were detected with fluorescein isothiocyanate (FITC)–
conjugated rabbit anti-dog immunoglobulin Gbdiluted
1:5,000 in Evan Blue solution as the contrast dye. A 50-ml
volume of the FITC-conjugated detection antibody was
added to each well and incubated for 30 min at 37uC,
followed by 3 washes with PBS. Using a fluorescent
microscope, serum samples were considered positive for
influenza A antibodies based on the presence of green foci
in the cellular membrane and cytoplasm on a brown–red
background. The endpoint FAT titer was defined as the
reciprocal of the last dilution of serum that produced
fluorescent foci.
All Italian dog sera that had positive or doubtful results
in the cELISA (competition percentage greater than 45%)
were also tested in subtype-specific HI assays. The HI
assays were performed using 9 different influenza A viruses
as antigens: H1N1 (A/swine/Italy/711/06); H2N3 (A/duck/
Germany/1215/73); H3N8 (A/canine/Florida/04, A/equine/
NM/99, and A/passerine/Italy/6000/00); H5N3 (A/duck/
Italy/775/04); H7N3 (A/turkey/Italy/2962/V03); H9N2 (A/
mallard/Italy/3817-34/05); and H10N1 (A/ostrich/South
Africa/01). These subtypes were selected as prototypes of
viruses likely circulating in poultry and mammals during
the period of canine serum collection. Each virus was
propagated in 10-day-old embryonated chicken eggs. The
viruses in the harvested allantoic fluids were inactivated
with 0.0005% b-propiolactoneeand stored at 280uC or
were lyophilized and stored at 220uC.
Prior to testing, the Italian canine serum samples and
control sera were incubated with receptor destroying
enzyme (RDEb; 1 part serum to 3 parts RDE) for 16 hr
at 37uC, followed by heat-inactivation at 56uC for 30 min.
The HI assays were performed using 4 hemagglutinating
units/25 ml of virus, 8 log2serial dilutions of serum samples
in PBS starting at 1:8 ratio, and 0.5% (mammalian-origin
virus) or 1% (avian-origin virus) suspensions of chicken
erythrocytes in PBS. The 2 different erythrocyte suspen-
sions were used based on World Organization for Animal
Health28–30standard procedures established for use of
avian- and mammalian-origin viruses in HI assays. The HI
titers were estimated visually after incubating at room
temperature for 45 min with 0.5% erythrocytes or for
30 min with 1% erythrocytes. The endpoint HI titer was
defined as the reciprocal of the last dilution of serum that
completely inhibited hemagglutination. The cutoff titer for
seropositivity was 32, based on previous testing of sera
from specific-pathogen-free dogs and dogs experimentally
infected with canine FLUAV subtype H3N8.11
The agar gel immunodiffusion (AGID) assay is a
standard serological test for diagnosis of avian FLUAV
infection based on detection of antibodies to the highly
conserved viral matrix protein and nucleoprotein in the
nucleocapsid.2An AGID assay using A/canine/FL/04 as
the antigen was performed on the known positive reference
sera using standard procedures.2,28–30However, as a result
of poor performance (43/97 sera detected as positive),
further AGID testing was not pursued with the Italian dog
sera.
Sensitivity and specificity of the cELISA
A Bayesian approach was used to estimate the Se and Sp
of the cELISA. The Se and Sp of the HI and FAT could not
be estimated because of the low number of positive results
in the cELISA and the negligible discordant results between
the HI and FAT. The methodology for application of
Bayesian theory to inference of diagnostic test Se and Sp
has been described elsewhere.3,4Briefly, uncertainty re-
garding the true value of the Se and Sp of the test and the
disease prevalence (p) was modeled using probability
distributions referred to as prior distributions. Prior values
for the cELISA, based on the uncertainty for the Se and Sp
values, were modeled using data reported in peer-reviewed
literature9–11,15,27and at professional conferences (Rigoni M,
Viale E, Mancin M, et al.: 2009, Development of a
competitive ELISA system for the detection of influenza A
virus nucleoprotein antibodies and validation in five species
of poultry. In: Proceedings of the World Association of
Veterinary Laboratory Diagnosticians 14th International
Symposium, p. 174. Madrid, Spain). The prior distribution
for the value of p was estimated based on an author’s
(Capua) expert opinion. Alternatively, the prior distribution
of p was parameterized using a beta (1,1) distribution to
assess the robustness of the results to the prior uncertainty
on the true value of p. The prior beta distributions of these
test parameters were fitted using the betabuster software,
available at the Bayesian Epidemiologic Screening Tech-
niques webpage (http://www.epi.ucdavis.edu/diagnostictests/
betabuster.html). Data collected from the field were then
used to modify the prior distributions to produce new
distributions, referred to as posterior distributions. Posterior
distributions represent the uncertainty for the Se, Sp, and
p values for the cELISA given the researchers’ prior
uncertainty for these values and the test results. The Se and
Sp of the cELISA were estimated using the 6,858 canine sera,
assuming that samples were obtained from 1 single popula-
tion (i.e., dogs in Italy; 1 test, 1 population).3,4All
calculations were performed using the WinBUGS software,26
andresultswerecomputed for100,000iterations of themodel
after the initial 1,000 were discarded. The results of the
cELISA, HI, and FAT were securely stored in the FMD
BioPortal database (https://fmdbioportal.ucdavis.edu/),
which is a Web-based system for visualization and analysis
of infectious animal diseases.
Results
Estimating the level of exposure of Italian dogs to
influenza A viruses
A video displaying results from the 6,858 Italian
dogs tested is available at the FMD BioPortal Web
site (http://fmd.ucdavis.edu/ma/AI_Italy.avi). Of the
Serological detection of Influenza A virus exposure in dogs917
at UNIV HAWAII on February 1, 2012vdi.sagepub.comDownloaded from

Page 6

2,041 Italian canine serum samples from 2006, 18
(0.88%) were positive for nucleoprotein antibody by
cELISA, and 2 (0.09%) samples were doubtful. For
the 2,038 samples from 2007, 23 (1.12%) were positive
by cELISA, and 14 (0.69%) were doubtful. For the
2,779 samples from 2008, 16 (0.58%) were positive by
cELISA, and 9 (0.32%) were doubtful. All 82
cELISA-positive and doubtful samples were retested
in the HI and FAT. Only 1 cELISA-positive serum
sample from 2006 and 1 cELISA-positive sample
from 2007 were also positive in both the HI and FAT.
The 2 samples were positive for antibody to the H3
proteins of A/canine/FL/04 (H3N8) and A/equine/
NM/99 (H3N8) viruses. The HI titer was 256 and
the FAT titer was 1,024 for the 2006 serum sample,
while the HI titer was 64 and the FAT titer was 256
for the 2007 serum sample. The remaining cELISA-
positive or doubtful samples had HI titers of less than
8. The apparent prevalence of FLUAV exposure for
the tested Italian dogs was, therefore, 0.03% (2/6,858
dogs).
For the reference serum panel from dogs naturally
infected with A/canine/FL/04 (H3N8) virus, 92 (95%)
tested positive in the cELISA, and 95 (98%) tested
positive in the FAT. None of the serum samples from
uninfected specific-pathogen-free dogs tested positive
in any of the assays.
Sensitivity and specificity of the cELISA
Eighty-two samples were considered positive or
doubtful in the cELISA and were therefore classified
as cELISA-positive samples for purposes of the
analysis. Two of these 82 samples were also positive
in both the HI and FAT, 79 were negative in both the
HI and FAT, 1 was HI negative and FAT positive,
and none were HI positive and FAT negative.
Sensitivity and Sp of the HI and FAT could not be
computed using Bayesian methods because of the low
proportion of cELISA-positive samples, along with
the negligible proportion of discordant results for the
HI and FAT. The Bayesian posterior Se and Sp
estimates for the cELISA, the Bayesian posterior
distribution of p after accounting for the imperfect
Se and Sp of the test, and their respective 95%
probability intervals are listed in Table 1.
Discussion
The present study identified 2 Italian dogs that had
serum antibodies to highly conserved influenza A
proteins and to H3 proteins from contemporary
canine and equine H3N8 viruses using cELISA, HI,
and FAT. Although evaluation of disease transmis-
sion was beyond the scope of this study, the potential
source of FLUAV exposure for the 2 seropositive
dogs was investigated. The dog that tested positive
from 2006 was a 5-year-old mixed-breed canine that
had lived in Miami, Florida, for 3 years before
moving to Sardinia, Italy, in 2006. The dog was bled
3 weeks after arrival in preparation for travel to the
United Kingdom and was likely infected by canine
FLUAV subtype H3N8 while in Florida. The dog
that tested positive from 2007 was a 4-year-old mixed-
breed canine from Lombardy, Italy. The dog had
never traveled outside of continental Europe; how-
ever, the owner confirmed that the dog was raised in
close contact with horses, including during 2003,
when there was an equine influenza outbreak in
Italy.18The owner did not report whether the dog
had been exposed to sick horses during the outbreak,
and the owners of both dogs never observed any
influenza-like illness in their dogs.
The current study evaluated a large-scale diagnostic
approach for detection of FLUAV antibodies in
canine sera. The proposed diagnostic algorithm
consists of an initial screening with a cELISA for
influenza A nucleoprotein antibody, followed by
confirmatory testing of positive samples with a FAT
for influenza A nucleocapsid antibodies and/or HI
assays for hemagglutinin subtype–specific antibodies.
This algorithm was supported by the Bayesian
analysis results for the cELISA and the generally
recognized advantages of the confirmatory HI and
FAT. The commercially available cELISA demon-
strated good Se (93.98%) and Sp (98.71%). In
addition, the high-throughput microtiter plate format
of the cELISA is ideal for screening large numbers of
Table 1.
assay for detection of antibodies to influenza A viruses in Italian dogs and of the virus prevalence in the population.*
Prior and Bayesian posterior estimates of the sensitivity and specificity of a competitive enzyme-linked immunosorbent
Prior estimates (%)Prior distributionsPosterior estimates (%)
Sensitivity
Specificity
Prevalence
96.11 (95% confidence .83.95%)
89.64 (95% confidence .76.89%)
0.4 (95% confidence ,1.2%)
Beta (27.133, 2.058)
Beta (29.503, 4.294)
Beta (3.170, 541.422){
93.97 (95% PI: 81.76–99.08%)
98.71 (95% PI: 98.43–98.96%)
0.5 (95% PI: 0.1–1.4%)
* PI 5 probability intervals.
{ The alternative parameterization p , beta (1,1) led to similar (,1% variation) posterior estimates of sensitivity, specificity, and
prevalence.
918
De Benedictis et al.
at UNIV HAWAII on February 1, 2012vdi.sagepub.com Downloaded from

Page 7

canine sera. A limitation, however, is that the
Bayesian model used herein is not identifiable because
there are more parameters than degrees of freedom.
For that reason, prior information about the model
parameters is required (Table 1). Because no prior
information was available regarding the disease
prevalence in the assessed population, best-guess
(based on expert opinion) and noninformative pa-
rameterizations of p were alternatively assumed.
Results were robust for alternative parameterization
of p and were consistent with the presented results.
Although the Se and Sp of the HI and FAT could
not be estimated using Bayesian analysis in the current
study, cELISA-positive and doubtful samples should
be confirmed in a highly sensitive and specific assay.
The Se and Sp of the HI assay are generally high when
the test antigen is homologous to the circulating virus.
In the current study the HI assay was conducted with 7
hemagglutinin subtypes, including 3 H3N8 viruses
representing canine, equine, and avian isolates. These
viruses were selected as prototypes of those likely
circulating in poultry and mammals during the period
ofcanine serum collection. The 80 HI-negative samples
had titers of ,8, indicating that the Se of the assay was
not negatively affected by the defined cutoff titer for
seropositivity of 32. However, the authors acknowl-
edge that exposure to the remaining virus subtypes was
not definitively ruled out since sera were not tested
against all 16 influenza A hemagglutinin subtypes. The
FAT allows type-specific detection of influenza virus
exposure, and the assay is highly sensitive and specific;
however, major disadvantages of the FAT include the
necessary in-house development of the test and its
admittedly subjective interpretation.
Although the low seroprevalence (0.5% after
accounting for the imperfect Se and Sp of the
cELISA) in the present study indicates that dogs do
not currently play a significant role in FLUAV
ecology in Italy, the identified seropositive dogs
highlight 2 potential sources of FLUAV exposure
for dogs in Italy and abroad: the potential spread of
canine-specific influenza A viruses by international
movement of infected dogs and exposure of dogs to
other species infected by FLUAV. Based on interna-
tional observations, monitoring dogs during future
influenza A outbreaks in other species is recom-
mended. A diagnostic algorithm utilizing cELISA as
well as HI and/or FAT is a highly sensitive and
specific approach to assessing exposure to any
influenza A subtype in dogs, especially in those with
unknown exposure histories at the time of testing.
Acknowledgements
Dr. William G. Dundon and Giovanni Cattoli, IZSVe-
Italy, and Dr. Jeanet Van der Goot, CIDC–The Nether-
lands, are acknowledged for technical advice on this
project. In part, the European Commission, project
FLUTRAIN (Training and Technology Transfer of Avian
Influenza Diagnostics and Disease Management Skills) and
EPIZONE (Network of Excellence for Epizootic Disease
Diagnosis and Control), and the U.S. National Center for
Medical Intelligence funded the current project. Paola De
Benedictis and Tara C. Anderson contributed equally to
this article.
Sources and manufacturers
a.ID ScreenH Influenza A Antibody Competition Assay, ID
Vet, Montpellier, France.
Sigma-Aldrich, St. Louis, MO.
BD Bioscience, San Jose, CA.
GibcoH, Invitrogen Corp., San Diego, CA.
Ferak Berlin GmbH, Berlin, Germany.
b.
c.
d.
e.
References
1. 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.
2. Beard CV: 1970, Demonstration of type-specific antibody in
mammalian and avian sera by immunodiffusion. Bull W H O
42:779–785.
3. Branscum AJ, Gardner IA, Johnson WO: 2004, Bayesian
modeling of animal- and herd-level prevalences. Prev Vet Med
66:101–112.
4. Branscum AJ, Gardner IA, Johnson WO: 2005, Estimation of
diagnostic-test sensitivity and specificity through Bayesian
modeling. Prev Vet Med 68:145–163.
5. Butler D: 2006, Thai dogs carry bird-flu virus, but will they
spread it? Nature 439:773.
6. Capua I, Kajaste-Rudnitski A, Bertoli E, Vicenzi E: 2009,
Pandemic vaccine preparedness—have we left something
behind? PLoS Pathog 5:e1000482.
7. Capua I, Marangon S, Cordioli P, et al.: 2002, H7N3 avian
influenza in Italy. Vet Rec 151:743–744.
8. Capua I, Mutinelli F, Pozza MD, et al.: 2002, The 1999–2000
avian influenza (H7N1) epidemic in Italy: veterinary and
human health implications. Acta Trop 83:7–11.
9. Capua I, Terregino C, Cattoli G, et al.: 2003, Development of
a DIVA (Differentiating Infected from Vaccinated Animals)
strategy using a vaccine containing a heterologous neuramin-
idase for the control of avian influenza. Avian Pathol
32:47–55.
10. Chan PK, Ng KC, Chan RC, et al.: 2004, Immunofluores-
cence assay for serologic diagnosis of SARS. Emerg Infect Dis
10:530–532.
11. Crawford PC, Dubovi EJ, Castleman WL, et al.: 2005,
Transmission of equine influenza virus to dogs. Science
310:482–485.
12. Daly JM, Blunden AS, Macrae S, et al.: 2008, Transmission of
equine influenza virus to English foxhounds. Emerg Infect Dis
14:461–464.
13. Kirkland PD, Finlaison DS, Crispe E, Hurt AC: 2010,
Influenza virus infection in dogs—transmission during the
Australian equine influenza outbreak. Emerg Infect Dis
16:699–702.
14. 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.
Serological detection of Influenza A virus exposure in dogs919
at UNIV HAWAII on February 1, 2012vdi.sagepub.com Downloaded from

Page 8

15. Kruth SA, Carman S, Weese JS: 2008, Seroprevalence of
antibodies to canine influenza virus in dogs in Ontario. Can
Vet J 49:800–802.
16. Lee C, Song D, Kang B, et al.: 2009, A serological survey of
avian origin canine H3N2 influenza virus in dogs in Korea.
Vet Microbiol 137:359–362.
17. Maas R, Tacken M, Ruuls L, et al.: 2007, Avian influenza
(H5N1) susceptibility and receptors in dogs. Emerg Infect Dis
13:1219–1221.
18. 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 repre-
sentatives of the Florida sub-lineage. Vet Microbiol 121:56–63.
19. Meijer A, Bosman A, van de Kamp EE, et al.: 2006,
Measurement of antibodies to avian influenza virus A(H7N7)
in humans by hemagglutination inhibition test. J Virol
Methods 132:113–120.
20. NewtonR,CookeA,EltonD,etal.:2007,Canineinfluenzavirus:
cross-species transmission from horses. Vet Rec 161:142–143.
21. 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.
22. 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.
23. Song D, Kang B, Lee C, et al.: 2008, Transmission of avian
influenza virus (H3N2) to dogs. Emerg Infect Dis 14:741–746.
24. Song D, Lee C, Kang B, et al.: 2009, Experimental infection of
dogs with avian-origin canine influenza A virus (H3N2).
Emerg Infect Dis 15:56–58.
25. Songserm T, Amonsin A, Jam-on R, et al.: 2006, Fatal avian
influenza A H5N1 in a dog. Emerg Infect Dis 12:1744–1747.
26. Spiegelhalter D, Thomas A, Best N, Gilks W: 1996, BUGS
0.05: Bayesian Inference Using Gibbs Sampling. MRC
Biostatistics Unit, Cambridge, UK.
27. Toffan A, Olivier A, Mancin M, et al.: 2010, Evaluation of
different serological tests for the detection of antibodies
against highly pathogenic avian influenza in experimentally
infected ostriches (Struthio camelus). Avian Pathol 39:11–15.
28. World Organization for Animal Health (OIE): 2008, Chapter
2.3.4, Avian influenza. In: Manual of diagnostic tests and
vaccines for terrestrial animals, 6th ed., vol. 1, pp. 1–20. OIE,
Paris, France.
29. World Organization for Animal Health (OIE): 2008, Chapter
2.5.7, Equine influenza. In: Manual of diagnostic tests and
vaccines for terrestrial animals, 6th ed., vol. 2, pp. 871–883.
OIE, Paris, France.
30. World Organization for Animal Health (OIE): 2008, Chapter
2.8.8, Swine influenza. In: Manual of diagnostic tests and
vaccines for terrestrial animals, 6th ed., vol. 2, pp. 1128–1138.
OIE, Paris, France.
920
De Benedictis et al.
at UNIV HAWAII on February 1, 2012vdi.sagepub.comDownloaded from