Available online at www.sciencedirect.

com

Vaccine 25 (2007) 8500–8507

Factors influencing the antibody response of

dogs vaccinated against rabies
Lorna J. Kennedy a,∗ , Mark Lunt b , Annette Barnes c ,
Lorraine McElhinney d , Anthony R. Fooks d ,
David N. Baxter e , William E.R. Ollier a
a Centre for Integrated Genomic Medical Research, University of Manchester,
Stopford Building, Oxford Road, Manchester M13 9PT, UK
b ARC Epidemiology Unit, University of Manchester, UK
c Faculty of Veterinary Science, University of Liverpool, UK
d Veterinary Laboratories Agency, Weybridge, Surrey, UK
e Greater Manchester Health Protection Unit, Peel House, Eccles, Greater Manchester, UK
Received 13 June 2007; received in revised form 11 September 2007; accepted 7 October 2007
Available online 26 October 2007

Abstract
Since 2000, there has been a legal requirement in the UK that dogs and cats should have an effective rabies vaccination with demonstrable
sero-conversion if their owners wish to avoid quarantine on re-entry to the UK. In 2002, 10,483 rabies titres were determined on dogs at the
VLA. Statistical analyses assessed the efficacy of each vaccine within different dog breeds. Animal size, age, breed, sampling time and vaccine
had significant effects on pass rates and median titres. Our data suggests that a general relationship between animal size and level of antibody
response exists and smaller sized dogs elicited higher antibody levels than larger breeds of dog. It was not however, only the magnitude
of response immediately following vaccination but also the duration of immunity that varied between breeds of dog. Another observation
was that young animals, less than 1-year of age, generated a lower antibody response to rabies vaccination than adults. Considerably higher
failure rates were also observed for different vaccines tested. Regression analysis revealed that two vaccines performed equally well, and
significantly better than the others tested. The variation in antibody response relating to length of interval of sampling following vaccination is
not unexpected and presumably relates to the response kinetics for primary vaccination. These data need to be placed in perspective in order
to minimise the risk of rabies being re-introduced into a rabies-free country, especially in the consideration of removing the requirement for
serological testing for rabies vaccinated dogs that participate in pet travel schemes.
© 2007 Elsevier Ltd. All rights reserved.

Keywords: Rabies; FAVN; Vaccine; Dogs; Pet travel scheme

1. Introduction virus infects the central nervous system causing encephalopa-

thy followed by rapid death. Prevention of rabies in humans
Rabies is a viral zoonosis transmitted to man, mainly can be achieved by vaccination either by pre or post exposure
by rabid dogs, although infection is also possible from bats prophylaxis.
and other mammals. There are an estimated 55,000 human In the UK, the pet travel scheme (PETS) was introduced
deaths annually worldwide from rabies (http://www.who.int/ to enable pets (primarily dogs and cats) to be brought into the
rabies/en/) and 94% of these are transmitted by rabid dogs [1]. UK without a lengthy quarantine. (http://www.defra.gov.uk/
Rabies virus belongs to the order Mononegavirales, which animalh/quarantine/pets/index.htm). It also allows pets to be
have non-segmented, negative stranded RNA genomes. The taken outside the UK and re-enter without quarantine (e.g. to
be taken on holiday with its owners). An important part of this
∗ Corresponding author. Tel.: +44 161 275 7316; fax: +44 161 275 1617. scheme is the requirement to ensure protection from rabies
E-mail address: Lorna.Kennedy@manchester.ac.uk (L.J. Kennedy). by vaccination, which in the UK is achieved using inactivated

0264-410X/$ – see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.vaccine.2007.10.015
 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507 8501

vaccines. Until the introduction of PETS, the rabies-free sta- analysis. The six vaccine manufacturers were Fort Dodge,
tus of the UK was effectively controlled by long quarantine Intervet, Merial, Schering-Plough, Virbac and other (mainly
periods [2]. There are a number of companies who manu- unknown). For the analysis these were coded A–F, but not in
facture rabies vaccines and the standard method of testing to this order.
check whether a dog has adequate immunological protection
is by measuring virus neutralising antibodies (VNA) using 2.1. Statistical methods
the fluorescent antibody virus neutralisation test (FAVN) [3],
which measures both IgM and IgG, performed at officially We considered the effect, on both the success of the vacci-
recognised laboratories. A serum titre of 0.5 IU/ml and above nation and the titre, of the following variables: vaccine source,
of rabies virus-specific antibodies is considered adequate pro- vaccine batch, age, sampling time (i.e. the number of days
tection against rabies. A titre below this level is considered between the date of vaccination and the date of a sample
a vaccination “failure”, leaving the dog less likely to be pro- being taken for antibody titre testing), size and breed. Logistic
tected from the rabies virus [2,4]. It is becoming increasingly regression was used when considering success, linear regres-
apparent that not all dogs respond equally to vaccination. sion when considering the titre (after log-transforming the
A proportion fails to develop a protective immune response, data to obtain a normal distribution). Size, breed, vaccine
leaving them susceptible to disease [5]. By contrast, some ani- source and vaccine batch were fitted as categorical vari-
mals appear to “over” respond to vaccines potentially leading ables. To allow for the possibility of non-linear associations
to adverse reactions [6,7], such as Arthus type 3 hypersen- between continuous predictors and outcomes, polynomial
sitivity reactions when re-vaccinated. Variations in vaccine terms were included in the models, with additional terms
response may in part be a genetically pre-programmed event being added until the last term added was not statistically
and this is compatible with anecdotal reports of immune significant. The predicted values from these regression equa-
response differences between breeds. A previous analysis of tions were compared to the mean values at each age (for ages
samples submitted for antibody testing, demonstrated that ≤15, for which there were sufficient numbers of dogs to do
vaccine type, sampling interval, age and origin of animal all so) to assess the goodness of fit of the models.
significantly affected the failure rate [8]. To investigate the effect of vaccine batch on failure rate,
The immune response is probably the best example of a a logistic regression model was fitted with an indicator vari-
complex continuous trait where multiple genetic, environ- able for each batch. The most similar batch coefficients were
mental and lifestyle factors all contribute and interact to constrained to be equal, and the process continued until any
produce the phenotype measured. One strong genetic fac- further simplification of the model led to a statistically sig-
tor known to influence immune response to vaccination is nificant reduction in the fit of the model.
the Major Histocompatibility Complex (MHC). We have All analyses were performed using Stata (StataCorp,
previously identified that dog leukocyte antigen (DLA) poly- 2003). To get an idea of the distribution of the titres, a kernel
morphism is related to both autoimmune susceptibility and density estimator, a form of smoothed histogram, was used
infectious disease susceptibility [9–13]. Our intention is to [14]. Box and whisker plots were produced for each breed
examine DLA gene polymorphism in rabies vaccination. and vaccine separately, if there were at least 16 dogs of that
However, before attempting to investigate the relationship breed vaccinated with each vaccine. These plots consist of
between canine immune response to vaccination and gene boxes showing the median, the 25th percentile and the 75th
polymorphisms, it is necessary to identify the factors which percentile. The whiskers extend to the most distant obser-
correlate with antibody production in a primary response. A vation which is less than 1.5× the inter-quartile range. Any
dataset relating to a cohort of dogs vaccinated against rabies individual observation outside this range is plotted separately.
has provided this opportunity.

3. Results
2. Material and methods
Data were assessed both in terms of success/fail rates, plus
Serological and basic demographic data from a cohort absolute and median titres for all the analyses conducted.
of 10,483 dogs tested at the Veterinary Laboratory Agency The majority of animals had been vaccinated with vaccines
(VLA) for rabies antibody during 2002 were available for A (n = 5272) and B (n = 4313). The remaining animals were
analysis. The dataset included dog breed, sex, age, vaccine vaccinated with vaccines C (n = 481), D (n = 184), E (n = 115)
manufacturer, vaccine batch used, date of vaccination, date and F (n = 116).
of sampling for antibody titre and antibody titre. All testing The overall failure rate for the different vaccines varied
for rabies antibodies was undertaken at the Veterinary Lab- from 0.01 to 0.20, (Fig. 1). While all four batches of vaccine
oratories Agency (VLA), Weybridge UK using the FAVN C had higher failure rates than vaccines A or B, one particular
test. A titre of less than 0.5 IU/ml was considered a vacci- batch had a failure rate of 32.9% (data not shown), which was
nation failure. The dataset was assessed for errors, such as only used for 70 dogs. Only data for vaccines from A and B
mistyped dates, and these records were excluded from further were included in further analysis due to the vastly smaller
8502 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507

Fig. 1. Percentage of failures by vaccine.

numbers in the other groups. On average vaccine B resulted

in a five times higher antibody titre than vaccine A. This can
be seen in Fig. 2, which represents kernel density estimates
for the titres produced by the two vaccines.
Batch variation within vaccines A and B was also con-
sidered. Logistic regression indicated that the 18 different
vaccine A batches could be clustered into 2 different groups,
with a mean failure rate of 3% in one group and 9% in
the other. Vaccine B showed no significant between-batch
(n = 22) variation (data not presented).
Fig. 3a shows the proportion that failed by age. This graph
fits the predicted values from a logistic regression equation
and the observed proportion failing at each age. Since small Fig. 3. (a) Percentage failure by age and (b) average titre by age.
differences in vaccination failure rates might have a dispro-
portionately large effect where the total number in that age
group is small, dogs aged over 15 years of age (n = 49) were more likely to fail using both vaccines (p < 0.04 for vaccine
excluded from the analysis. A, p < 0.001 for vaccine B), but young dogs were only more
When vaccinated dogs were divided into three age groups: likely to fail when using vaccine A (p < 0.001 vs. p < 0.3).
young (<1 year), adult (1–7 years) and old (>7years), logistic Adult dogs also had statistically significantly higher titres
regression showed that both young and old dogs were sig- than young and old dogs, (see Fig. 3b). The highest titres
nificantly more likely to fail than adult dogs. Old dogs were occurred in dogs approximately 3–4 years of age. Gender

Fig. 2. Log of antibody response for vaccines A and B. Fig. 4. Percentage of failures by sampling time.
 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507 8503

Weight was not recorded in the dataset, and consequently

an estimated weight and height category (small, medium
and large) was assigned to each dog based on breed stan-
dards. These were combined into five size categories: small,
small–medium, medium, medium–large and large. Data were
then analysed for variation attributable to size category. Fail-
ure rate increased with increasing size (p < 0.007), apart from
the small and medium–small groups, which were in reverse
order, (Fig. 5), as did median antibody titres decreased (data
not shown). These differences were significant except for the
lowest two groups.
Fig. 6 shows the distributions of titres using vaccines from
both vaccines A and B for all breeds that had at least 16
dogs vaccinated with either of the two vaccines. The black
vertical line shows the threshold for the vaccination being
Fig. 5. Proportion of dogs failing rabies vaccination according to size cate-
considered a success. As previously mentioned, titres were
gory. higher for vaccine B than vaccine A and this was seen in all
breeds. However, the titre observed is smaller in the larger
had no effect on failure rate or titre, but, interestingly, neu- dogs. Although most of the failures are in the larger breeds
tering increased the titre for both vaccines by 22% (data not (e.g. German shepherd dogs and Labradors), there are also
shown). A plot of the proportion that failed against sampling some smaller breeds with important failure rates (Cavalier
time shows that the least failures occur at 30 days (Fig. 4). King Charles spaniels and west highland white terriers).
When animals were divided into three sampling time groups: We also calculated an average titre for each breed, and
short (<20 days) normal (20–50 days) and long (>50 days), then divided the titres into quartiles. If size accounted for
analysis showed that the usual 28 day sampling time gave all the variation, then the lowest quartile should include
significantly lower failure rates (p < 0.00001) than short or the large breeds and the highest quartile should include
long sampling times. only small breeds. Breeds were listed by quartile and size,

Fig. 6. Box and whisker plots of titres of vaccines A and B for all breeds with at least 16 dogs vaccinated with each of the 2 vaccines.
8504 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507

Table 1
Breeds listed as quartile by titre against size
Breed Quartile by titre Large Medium–large Medium Medium–small Small
Leonberger 1 17
Bernese Mountain dog 1 19
Bullmastiff 1 23
Akita (Japanese) 1 25
Great Dane 1 35
Rhodesian Ridgeback 1 48
Dobermann 1 71
Rottweiler 1 101
Setter (Red) 1 52
Lurcher 1 93
Retriever (Golden) 1 311
German shepherd dog 1 441
Collie (Rough) 1 25
Collie (Bearded) 1 49
Labrador (Retriever) 1 943
Bulldog 1 25
Basset hound 1 29
Sheepdog (Shetland) 1 32
Terrier (Scottish) 1 34
Malamute (Alaskan) 2 13
Bouvier des Flandres 2 17
Newfoundland 2 28
Setter (Gordon) 2 19
Setter (English) 2 31
Greyhound 2 47
Shar Pei 2 10
Husky (Siberian) 2 18
Spaniel (Welsh Springer) 2 20
Pointer (German) 2 34
Vizsla (Hungarian) 2 35
Sheepdog (Old English) 2 37
Terrier (Airedale) 2 39
Dalmatian 2 70
Retriever (Golden) 2 94
Boxer 2 157
Collie cross 2 249
Collie (Border) 2 434
Beagle 2 26
Spaniel (Cocker) 2 337
Terrier (Norfolk) 2 18
Terrier (Patterdale) 2 24
Spaniel (Cavalier King Charles) 2 191
Weimaraner 3 92
Mongrel 3 1,417
Spaniel (English Springer) 3 320
Terrier (Bull) 3 35
Terrier (West Highland White) 3 338
Pointer 4 30
Whippet 4 37
Terrier (Scottish) 4 1
Corgi 4 20
Terrier (Tibetan) 4 52
Dachshunda 4 61 58
Poodlea 4 128 59
Terrier (Staffordshire Bull) 4 178
Spaniel (King Charles) 4 19
Terrier (Lakeland) 4 19
Pug 4 20
Terrier (Maltese) 4 23
Pekingese 4 26
Papillon 4 27
Chihuahua 4 38
Terrier (Cairn) 4 48
Lhasa Apso 4 81
Schnauzer (Miniature) 4 99
Shih Tzu 4 108
Bichon Frise 4 110
Terrier (Border) 4 124
Terrier (Yorkshire) 4 284
Terrier (Jack Russell) 4 506
Breeds in bold text are not in the expected order.
a Poodles and Dachshunds are shown as a single group but there are medium–large and miniature (toy) varieties, which were not differentiated in our analysis.

and it is clear that some breeds were not in their expected were classed in the lowest quartile of titre. Similarly three
position (see breeds in bold text in Table 1). For example, small breeds (Cavalier King Charles spaniel, Patterdale terrier
two small breeds (Shetland sheepdog and Scottish terrier) and Norfolk terrier) plus two medium small breeds (Cocker
plus two medium–small breeds (Basset hound and Bulldog) spaniel and Beagle) were in the second quartile. These can
 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507 8505

be considered to be poor responders. It is interesting to note mal encounters sufficient antigen to make a response, larger
that beagles are usually used for vaccine trials and registra- doses of antigen are not a major factor in increasing antibody
tion of vaccine products and yet there are clear differences production in primary responses. An alternative explanation
between responses in this breed and other pedigree dogs. could be that larger dogs are more likely to have deeper sub-
The choice of dog breed in vaccine immunity studies should cutaneous fat at popular sites for injection. Deposition and
therefore be a careful consideration. At the other end of the sequestration of antigen in fat is known to reduce the level of
spectrum the large breed Weimaraner was in the third quar- immune response [15,16] Any studies to clarify such poten-
tile, while Pointers (medium–large) and Whippets (medium) tial variability and standardize procedures would clearly be
were in the highest quartile, which can be considered as good beneficial in improving animal welfare.
responders. The breed identity supplied on sample submis- The relationship between antibody response and age is
sion forms may not always be accurate and a small proportion also an important consideration. The observation that young
of the dogs stated to be a certain breed will not be pedigree animals (<1 year) make a poorer immune response to rabies
animals and therefore would not necessarily conform to the vaccination than adults could be due to their immune systems
“breed standard”. We recognize that this may introduce some being less mature. A reduction in immune response in the
errors into the dataset, however, we do not consider this to be elderly is well documented in humans [17] even though over-
significant to alter the overall conclusions of the data. all immunoglobulin levels tend to rise with age, as does the
Of the variation in log titre observed, 19% was due to dif- prevalence of auto-antibodies [18]. A reduction in immune
ferences between vaccine A and vaccine B, and a further 8% regulation is thought to occur with age and this may explain
due to differences between breeds. Of the differences between why older dogs have a poorer response to rabies vaccination.
breeds, 5% could be attributed to differences in size between A policy of further booster vaccination may be justified for
the breeds, and 3% to other differences. These other differ- older dogs although this would need careful study i.e. a trial.
ences remain highly statistically significant (p < 0.0001). Another complicating factor is the different life expectancies
for different breeds, which makes the definition of “old” vary
for different breeds.
4. Discussion The variation in antibody response relating to length of
interval of sampling following vaccination is not unexpected
We have analysed a large dataset of dogs vaccinated and presumably relates to the response kinetics for primary
against rabies as part of the pet passport scheme. This analysis vaccination. Dogs iso-type shift from an IgM response to an
raises a number of important issues relating to canine vaccina- IgG as an immune response develops, and optimum measure-
tion in general and rabies vaccination in particular. Vaccines ment in an appropriate window of time should measure this
produced by different manufacturers have significantly dif- effect. Measurement at too early a stage will predominantly
ferent failure rates, and significantly different median titres only capture an IgM response developing but not be able to
of response. This is presumably due to their formulation and confirm whether an IgG response progresses. Measurement
the production differences between vaccines together with only at a later time point may reveal lower antibody levels,
the concentration and integrity of antigen content and the but this may not relate to a lack of immune protection as the
adjuvant used. Although such differences occur it is difficult total immunoglobulin measure may be proportionately more
to fully assess how clinically meaningful this is, if all com- accounted for by IgG.
mercially available vaccines generate sufficient immunity in This study clearly demonstrates that the standard time
most animals. frame for sampling following vaccination should be adhered
Interestingly, analysis of the most frequently used vac- to by owners and veterinarians if a true picture of rabies
cines (A and B) revealed a bimodal distribution of response immunity is to be determined. In Europe with quarantine and
with a small group of dogs making higher antibody levels. An PETS it would appear that rabies is under control. However on
explanation for this observation can only be speculated at. It a global level, rabies is still an important health issue. This
is possible that within the animals vaccinated there is a dis- study raises many variables that need further investigation
tinct group of dogs which make a higher antibody response and explanation. For example the time frame for vaccina-
due to their genetic profile or other characteristics such as tion and serological testing, the variable failure rates of some
size or age, since size, age of the dog and the time interval vaccines, the wide range of antibody responses and, interest-
between vaccination and sampling for rabies antibody titre ingly, the variability within and between breeds in response
all had significant effects on failure rates and median titres. to vaccination. We should therefore not become complacent
A general relationship between animal size and level of about the spread of rabies within Europe and further work is
antibody response clearly exists [8]. This could be explained clearly required. In some parts of the world rabies is not under
as a vaccine dose-effect, as it is unclear whether any standard control and a better understanding of the immune response
policy exists between veterinarians for adjusting vaccine dose to vaccination would help in devising strategies especially in
by body weight and whether small and large dogs receive the the feral populations. A clear correlation exists between the
same volume of injection. However this seems an unlikely reduction of human cases and the reduction of cases in the
explanation since so long as the immune system of an ani- domestic dog, consequently disease control in animals must
8506 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507

be effective [19]. The possibility that in 2008, the UK, under two major haplotypes respectively. There is some sharing of
an EU directive, will not have serological testing could there- DLA alleles between Doberman and Rottweiller. In contrast,
fore be premature given our current level of understanding. A Shih Tzu, which respond well to both vaccines also have a
risk assessment carried out by Defra also supports this stance fairly restricted profile of five DLA haplotypes, two of which
[20]. have only been found in Shih Tzu to date, and with little
Current differences exist between the UK and Sweden overlap for the alleles found in Dobermanns.
with their respective travel schemes in that serological test- From our preliminary analyses a wide range of antibody
ing for entry into the UK was set at an optimum of 28 days response exists for most breeds, even when only adult dogs
although it was still possible to take an earlier blood sam- tested in the optimum time frame are considered. This could
ple if deemed necessary with the understanding that the titre be explained by allelic variation of DLA class II loci in
might be low. In contrast, in Sweden the date of blood sam- breeds. It is possible that low and high levels of response are
pling was set at 120 days following vaccination to ensure associated with particular DLA polymorphisms. Although
that dogs with high titres still had a measurable titre after MHC genes are known to play a major role in determining
this time period. Hence, Sweden had a much higher failure the level of immune response, the response to an antigen
rate than the UK. As the blood sample was only taken from a is undoubtedly under complex genetic control. Other gene
single time point, it was argued by the competent authorities polymorphisms such as those in IL-10 and other TH1/TH2
in the UK that the titre level was only a marker of sero- regulatory cytokines may also contribute to this regulation.
conversion to the vaccine and not an indication of protection. This control of vaccine response will apply to all vaccines,
In addition, the vaccines used had proven efficacy and vaccine not just rabies.
potency. This assumed that the minimum antibody thresh- We now intend to investigate the role of DLA and other
old (0.05 IU/ml) was attained. If, during the 3-year period immuno-regulatory cytokine gene polymorphisms in deter-
before a booster vaccination was required, it was assumed mining canine response to vaccination.
that memory B- and T-cells had been primed and an anamnes-
tic response would generate VNA in response to a viral
challenge. Acknowledgements
In the US vaccinated dogs and cats regularly succumb to
natural rabies, probably from a wildlife reservoir [21]. This ARF and LMcE were funded by the UK Department
would suggest that there is an increased risk, if vaccinated for Environment, Food and Rural Affairs (Defra) by grant
non-responders are allowed into rabies-endemic areas. SEV3800. We are grateful to Mrs. Trudy Goddard for carry-
Interestingly, considerably higher failure rates were ing out the FAVN tests.
observed for vaccine A (17.1%) than vaccine B (3.5%) for
samples taken >50 days post vaccination. One interpretation
References
of this and the data suggested in Fig. 3a is that vaccine B
provokes better sero-conversion to IgG responses. Although [1] Perrin P, Jacob Y, Aguilar-Setien A, Loza-Rubio E, Jallet C,
vaccine A has relatively low fail rates at sampling times <20 Desmezieres E, et al. Immunization of dogs with a DNA vaccine induces
days and 20–50 days, this becomes higher at times >50 days protection against rabies virus. Vaccine 1999;18(5–6):479–86.
as presumably IgM levels diminish. [2] Fooks AR, McElhinney LM, Brookes SM, Johnson N, Keene V, Parsons
After these factors were excluded, residual failure rates G, et al. Rabies antibody testing and the UK pet travel scheme. Vet Rec
2002;150(14):428–30.
of 3.0% and 0.7% remain (vaccines A and B, respectively). [3] Cliquet F, Aubert M, Sagne L. Development of a fluorescent antibody
These data represents a high level of risk in the consideration virus neutralisation test (FAVN test) for the quantitation of rabies-
of removing the requirement for testing that rabies vaccina- neutralising antibody. J Immunol Meth 1998;212(1):79–87.
tion has been successful. [4] Burr PD, Snodgrass DR. Serological response to rabies vaccination for
Analysis restricted to only adult dogs with a normal sam- the pet travel scheme. Res Vet Sci 2002;72(Suppl 1):25.
[5] Ek-Kommonen C, Sihvonen L, Pekkanen K, Rikula U, Nuotio L. Out-
pling time revealed that size still has a significant effect break of canine distemper in vaccinated dogs in Finland. Vet Rec
on vaccine response, and accounts for approximately half 1997;141(15):380–3.
the differences between breeds. However, significant dif- [6] Tizard IR, Vaccination, Vaccines.Veterinary immunology. 5th ed. Lon-
ferences between breeds still exist. Importantly, even by don: W.B. Saunders; 1996.
taking breed and size into account, other factors must explain [7] Tjalve H. Adverse reactions to veterinary drugs reported in Sweden
during 1991–1995. J Vet Pharmacol Ther 1997;20(2):105–10.
some of the differences in response between dogs, including [8] Mansfield KL, Burr PD, Snodgrass DR, Sayers R, Fooks AR. Factors
failures. affecting the serological response of dogs and cats to rabies vaccination.
We know from our extensive DLA typing of over 4000 Vet Record 2004;154:423–6.
dogs, that some breeds have characteristic distributions of [9] Kennedy LJ, Davison LJ, Barnes A, Isherwood D, Ollier WER, Catch-
DLA polymorphisms and may also have very restricted pole B. Susceptibility to canine diabetes mellitus is associated with
MHC class II polymorphism. BSAVA annual congress 2003 scientific
DLA allele and haplotype profiles [22–24]. Two breeds that proceedings 2003. p. 593.
respond poorly to both vaccines are Doberman and Rot- [10] Kennedy LJ, Huson HJ, Leonard J, Angles JM, Fox LE, Wojciechowski
tweiller, both with low DLA variation, having just four and JW, et al. Association of Hypothyroid disease in Doberman Pinscher
 L.J. Kennedy et al. / Vaccine 25 (2007) 8500–8507 8507

dogs with a rare Major Histocompatibility Complex DLA class II hap- [18] Tomer Y, Shoenfeld Y. Ageing and autoantibodies. Autoimmunity
lotype. Tissue Antigens 2006;67:53–6. 1988;1(2):141–9.
[11] Kennedy LJ, Quarmby S, Happ GM, Barnes A, Ramsey IK, Dixon [19] Rupprecht CE, Hanlon CA, Hemachudha T. Rabies re-examined.
RM, et al. Association of canine Hypothyroidism with a common Lancet Infect Dis 2002;2(6):327–43.
Major Histocompatibility Complex DLA class II allele. Tissue Anti- [20] Wilsmore T, Hamblin C, Taylor N, Taylor W, Watson B. Qualita-
gens 2006;68:82–6. tive veterinary risk assessment of the introduction or rabies into the
[12] Ollier WER, Kennedy LJ, Thomson W, Barnes AN, Bell SC, Bennett United Kingdom. A report prepared for Defra (Department for the
D, et al. Dog MHC alleles containing the “human RA shared epitope” Environment, Food and Rural Affairs) 2006: http://www.defra.gov.uk/
confer susceptibility to canine rheumatoid arthritis. Immunogenetics animalh/diseases/notifiable/rabies/pdf/qra-rabies.pdf.
2001;53:669–73. [21] Eng TR, Fishbein DB. Epidemiologic factors, clinical findings, and
[13] Quinnell RJ, Kennedy LJ, Barnes A, Courtenay O, Dye C, Garcez vaccination status of rabies in cats and dogs in the United States
LM, et al. Susceptibility to visceral leishmaniasis in the domestic in 1988. National Study Group on Rabies. J Am Vet Med Assoc
dog is associated with MHC class II polymorphism. Immunogenetics 1990;197(2):201–9.
2003;55(1):23–8. [22] Kennedy LJ, Barnes A, Happ GM, Quinnell RJ, Bennett D, Angles
[14] Fox J. Describing univariate distributions. In: Fox J, Long JS, editors. JM, et al. Extensive interbreed, but minimal intrabreed, variation
Modern methods of data analysis. Newbury Park, CA: Sage Publica- of DLA class II alleles and haplotypes in dogs. Tissue Antigens
tions; 1990. p. 58–125. 2002;59(3):194–204.
[15] Ellis RW, editor. Hepatitis B vaccines in clinical practice. Marcel [23] Kennedy LJ, Barnes A, Happ GM, Quinnell RJ, Courtenay
Dekker; 1993. O, Carter SD, et al. Evidence for extensive DLA polymor-
[16] Keating GM, Noble S. Recombinant hepatitis B vaccine (Engerix-B): phism in different dog populations. Tissue Antigens 2002;60(1):43–
a review of its immunogenicity and protective efficacy against hepatitis 52.
B. Drugs 2003;63(10):1021–51. [24] Kennedy LJ, Carter SD, Barnes A, Bell S, Bennett D, Ollier WER, et
[17] Aspinall R. Longevity and the immune response. Biogerontology al. Interbreed variation of DRB1, DQA1 alleles and haplotypes in the
2000;1(3):273–8. dog. Vet Immunol Immunopathol 1999;69:101–11.