Publication Date – 27/08/2018
Research Authors –
- Kim B Stevens,
- Dan O’Neill,
- Rosanne Jepson,
- Laura Phillipa Holm,
- David John Walker and
- Jacqueline Martina Cardwell
Seasonal outbreaks of cutaneous and renal glomerular vasculopathy (CRGV) have been reported annually in UK dogs since 2012, yet the aetiology of the disease remains unknown. The objectives of this study were to explore whether any breeds had an increased or decreased risk of being diagnosed with CRGV, and to report on age and sex distributions of CRGV cases occurring in the UK. Multivariable logistic regression was used to compare 101 dogs diagnosed with CRGV between November 2012 and May 2017 with a denominator population of 446,453 dogs from the VetCompass database. Two Kennel Club breed groups—hounds (odds ratio (OR) 10.68) and gun dogs (OR 9.69)—had the highest risk of being diagnosed with CRGV compared with terriers, while toy dogs were absent from among CRGV cases. Females were more likely to be diagnosed with CRGV (OR 1.51) as were neutered dogs (OR 3.36). As well as helping veterinarians develop an index of suspicion for the disease, better understanding of the signalment risk factors may assist in the development of causal models for CRGV and help identify the aetiology of the disease.
Cutaneous and renal glomerular vasculopathy (CRGV) is a disease of unknown aetiology variably associated with clinically relevant acute kidney injury (AKI). Also sometimes referred to as ‘Alabama Rot’, CRGV cases typically present with ulcerated skin lesions, most often affecting the distal limbs, although lesions have also been reported to affect the face, nasal planum, oral cavity, tongue, ventrum and flanks. Common biochemical and haematological features have included mild to moderate hyperbilirubinaemia, anaemia and moderate to severe thrombocytopenia.1
A previous case series1 indicated that cases presenting with skin ulceration typically progress within a range of 1 to 9 days (median 4 days) to develop AKI, azotaemia and in many confirmed cases, acute renal failure with oligoanuria. Mortality rate in those cases that progress to oligoanuria is high, with a confirmatory diagnosis of CRGV only being made at postmortem examination. However, suspected cases have been identified that appear less severely affected and where renal recovery may occur, although lack of a viable antemortem diagnostic test precludes definitive diagnosis in these cases.
The histopathological lesions identified in the renal parenchyma of patients with CRGV are supportive of a thrombotic microangiopathy (TMA).1 In human medicine, TMAs are considered a complex group of diseases that can involve both hereditary and acquired contributing factors to the development of clinical disease.2 Hereditary factors that have been identified include genetic mutations in ADAMTS13, which results in the condition known as thrombotic thrombocytopenic purpura (TTP), complement factors, metabolic factors (methylmalonic aciduria and homocystinuria type C protein) and diacylglycerol kinase-ε, an abnormality of which results in a prothrombotic state. Acquired forms of TMA may be associated with autoantibody inhibition of ADAMTS13, Shiga toxin exposure (Shiga toxin-haemolytic uraemic syndrome), drug-mediated immune or toxic reactions, or complement-mediated.2 To date, however, preliminary investigations, including evaluation for Shiga toxin1 and other infectious aetiologies, have not been able to elucidate an underlying aetiology for CRGV, and therefore epidemiological studies are required to better understand the risk factors that may indicate pathogenesis in this condition.
CRGV has been reported in kennelled and racing greyhounds in the USA (n=1683; n=184), a single greyhound in the UK5 and in a great dane in Germany.6 In contrast to these few isolated incidents, the UK outbreaks have involved multiple breeds including the English springer spaniel, flat-coated retriever, whippet, border collie, Jack Russell terrier, dobermann, labrador retriever, cocker spaniel, Staffordshire bull terrier, Hungarian vizsla, weimaraner, dalmatian, Tibetan terrier and crossbreds.1 The objectives of this study were therefore to explore whether any breeds had an increased or decreased risk of diagnosis with CRGV, and to report on the age and sex distributions of CRGV cases occurring in the UK. These results may assist in the validation of current and future proposed pathogenic mechanisms and also assist clinicians in developing their index of suspicion to achieve earlier diagnosis of this serious condition.
Materials and methods
Study area, period and design
This research was based on a retrospective case–control study involving dogs with a confirmed diagnosis of CRGV in the UK between November 2012 and May 2017 (103 cases). The denominator population comprised all dogs under veterinary care in the UK during 2013 that were participating in the VetCompass programme and that are taken to represent the demography of the wider population of UK dogs that are registered for veterinary care from which the cases were derived. Because the cases were not extracted directly from the denominator population, this case–control study design cannot reliably report the incidence of CRGV but can usefully explore risk factor analysis.7
Identification of cases
Cases were compiled by two investigators (DJW and LPH), with 70 (68 per cent) from first-opinion practice and 33 (32 per cent) from referral centres. A confirmed diagnosis of CRGV was based on the presence of compatible clinical signs (including skin lesions), laboratory diagnostics (including progression to azotaemia, AKI±oligoanuria, hyperbilirubinaemia, anaemia and thrombocytopenia) and renal histopathology documenting findings compatible with TMA. Renal histopathology was available either in isolation or as part of a full postmortem examination, and in most cases dermal pathology was also available. The need for renal histopathology to confirm diagnosis precluded the inclusion of any dogs surviving suspected CRGV.
Identification of dog denominator data
The ‘VetCompass Denominator of Dogs under Veterinary Care in the UK during 2013’ (also known as dog denominator) population included all dogs under primary veterinary care at clinics participating in the VetCompass programme during 2013. Dogs under veterinary care were defined as those with either (1) at least one electronic patient record (EPR) (VeNom diagnosis term, free-text clinical note, treatment or bodyweight) recorded during 2013 or (2) at least one EPR recorded both before and after 2013. The VetCompass programme collates de-identified EPR data from primary care veterinary practices in the UK for epidemiological research.8 Collaborating practices can record summary diagnosis terms during episodes of care from an embedded VeNom code list.9
Data fields extracted from the VetCompass data set for the purpose of this study included a unique animal identifier together with (where available) breed, date of birth, sex, neuter status and partial postcode. The breed data recorded in the EPR were mapped to a standardised listing of breed terms. These breed lists were further mapped to classify breeds by purebred status, Kennel Club (KC) recognition of the breed and KC breed group. Neuter status described the status of the dog at the final EPR, while age was calculated from date of birth and described age at the final date under veterinary care during 2013 (December 31, 2013). Signalment and partial postcode of all cases were compared with the denominator dogs to ensure that none of the cases were duplicated as controls.
The CRGV case and dog denominator control data sets were combined to form the final data set, which was checked for unlikely values and missing data. Observations with missing data for three variables were removed from the data set as follows: breed (0.4 per cent of controls (n=2009); no cases), sex (0.5 per cent of controls (n=2310); no cases) and age (1.4 per cent of controls (n=6117); 1.9 per cent of cases (n=2)). However, of the 72,344 observations that lacked data on neutered status, 15 (15 per cent) were CRGV dogs. Rather than lose a quarter of the case data, the missing observations were labelled as ‘not recorded’, thus creating a neutering status variable comprising three levels: male, female and not recorded. Three variables were derived from breed and included (1) common breed name, (2) purebred versus crossbred versus designer dog (ie, a planned hybrid with a specific hybrid name, eg, cockapoo10) and (3) the UK KC breed groups: hounds, terriers, gun dogs, working, utility, pastoral, toy and not KC-recognised.
Descriptive statistics were derived for all variables for both the study population as a whole, and separately for CRGV dogs and the dog denominator population. Univariable logistic regression modelling was used to evaluate the associations between each variable and being a CRGV case, together with unadjusted odds ratios (ORs) and 95 per cent confidence intervals (CIs). The ‘common breed’ variable included only those breed types that appeared among the CRGV cases. Crossbred and terrier were chosen as the reference values for common breed and breed groups, respectively, as both were large categories. Age was categorised into four groups based on quartiles to create the variable age group, and a test for linear trend was used to determine whether the variable age should be included in the model in continuous (age) or categorical (age group) format. Those variables achieving a univariable P value <0.2 were taken forward for multivariable logistic regression modelling. Retention of variables in the final model was determined using a backward stepwise approach based on the likelihood ratio test (LRT). Model fit was assessed using Akaike’s information criteria. All statistical analyses were performed in STATA SE 14 and a P value of ≤0.05 was considered significant.
Description of study population
The 446,554 dogs comprising the study population had a median age of 4.4 (interquartile range (IQR): 5.90 years; range 0.1–24.7 years) and 51.8 per cent were male (n=231,450). Neutered dogs comprised 45.5 per cent (n=203,313) of the study population, 38.4 per cent (n=171,493) were entire and the status of 16.1 per cent (n=71,748) was not recorded. Three-quarters of the study population were purebreds (75.2 per cent; n=335,807), while 3.0 per cent were designer dogs (n=13,602). The most common KC breed groups were gun dogs (16.1 per cent; n=72,105), terriers (13.1 per cent; n=58,362) and toy dogs (12.6 per cent; n=56,431), while working dogs (4.9 per cent; n=22,001) and hounds (3.5 per cent; n=15,646) were the least represented.
Crossbreds were the most common breed type comprising 37.7 per cent (n=97,146) of the study population, with labrador retrievers (12.8 per cent; n=32,938), Staffordshire bull terriers (12.5 per cent; n=32,134) and Jack Russell terriers (10.6 per cent; n=27,356) the most common specified breeds. Other relatively common breeds in the study population included cocker spaniels (6.1 per cent; n=15,671), German shepherd dogs (4.8 per cent; n=12,321) and border collies (4.7 per cent; n=12,165). Of those breeds represented among the cases, the least common were Hungarian vizslas (0.3 per cent; n=775), flat-coated retrievers (0.2 per cent; n=452), bearded collies (0.2 per cent; n=538), salukis (0.1 per cent; n=201) and Manchester terriers (0.05 per cent; n=126).
Distributions of breed, age, sex and neuter status of CRGV and denominator dogs
Following removal of missing data, the study population included 101 CRGV case dogs and 446,453 VetCompass denominator control dogs. The median age for CRGV dogs (4.0 years; IQR: 4.8 years; range 0.5–12 years) did not differ significantly from the denominator dog population (4.12 years; range 0.1–24.7 years; P=0.874). Compared with the denominator dogs which were evenly distributed between the four age groups, 34.7 per cent (n=35) of the CRGV dogs were aged between 1.73 and 4.11 years old. The smallest group of CRGV dogs comprised those aged less than 1.72 years old (15.8 per cent; n=16; P=0.010). CRGV dogs were more likely to be female (58.4 per cent; n=59) compared with denominator dogs (48.2 per cent; n=215,045; P=0.010). Similarly, CRGV dogs were more likely to be neutered (69.3 per cent; n=70) compared with denominator dogs (45.5 per cent; n=203 243; P<0.001). Proportions of purebred, designer and crossbred dogs were generally comparable between CRGV and denominator dogs (P=0.587) (table 1).
Two KC breed groups—gun dogs and hounds—comprised 60.4 per cent (n=61) of the CRGV cases. However, while gun dogs were the largest KC breed group for both CRGV and denominator dogs, proportions differed considerably (48.5 vs. 16.1 per cent, respectively; P<0.001). Likewise, hounds made up a far greater proportion of CRGV dogs than denominator dogs (11.9 vs. 3.5 per cent, respectively; P<0.001). Conversely, terriers were under-represented among CRGV dogs (CRGV: 4.0 per cent; denominator: 13.1 per cent) and, despite comprising 12.6 per cent of the denominator dogs (n=56,431), there were no toy dogs among those diagnosed with CRGV (table 1).
Of the five most commonly specified breeds in the study population (labrador retriever, Staffordshire bull terrier, Jack Russell terrier, cocker spaniel and German shepherd dog), three were under-represented among CRGV dogs: Staffordshire bull terriers (3.0 per cent (n=3) vs. 12.5 per cent (n=32,131); P=0.201), Jack Russell terriers (2.0 per cent (n=2) vs. 10.6 per cent (n=27,354); P=0.163) and German shepherd dogs (1.0 per cent (n=1) vs. 4.8 per cent (n=12,320); P=0.364). Conversely, breeds that were over-represented among CRGV dogs were generally the less common breeds such as English springer spaniels (10.9 per cent (n=11) vs. 2.1 per cent (n=5337); P<0.001), whippets (8.9 per cent (n=9) vs. 0.8 per cent (n=2126); P<0.001), flat-coated retrievers (6.9 per cent (n=7) vs. 0.2 per cent (n=445); P<0.001) and Hungarian vizslas (5.9 per cent (n=7) vs. 0.3 per cent (n=769); P<0.001) (table 1).
Common breed (P<0.001), KC breed group (P<0.001), neutered status (P<0.001), age group (P=0.017) and sex (P=0.010) were significantly associated with being a CRGV case in the univariable modelling. Owing to the collinearity between the derived breed variables, two multivariable models were built, one including common breed and the other including KC breed group, while keeping the remaining variables constant. Although the use of KC breed groups resulted in more robust ORs and 95 per cent CIs than when common breeds were used—because there were fewer categories and therefore more dogs in each—the use of specific breeds was considered more useful for veterinarians, and therefore the results of both models were presented. In addition, age group was not a significant risk factor in the multivariable models (LRT, P=0.06).
The odds of gun dogs (OR 9.69; 95 per cent CI 3.50 to 28.86; P<0.001) and hounds (OR 10.68; 95 per cent CI 3.44 to 33.13; P<0.001) being a CRGV case was between 9 and 11 times that of terriers. Pastoral dogs were also significantly more likely to be a CRGV case than terriers (OR 3.50; 95 per cent CI 1.01 to 11.96; P=0.046). As there were no toy dogs among CRGV cases, this breed group was dropped from the model. Specific breeds with increased odds of being a CRGV case compared with crossbreds included the flat-coated retriever (OR 84.48; 95 per cent CI 35.19 to 202.80; P<0.001), Hungarian vizsla (OR 40.98; 95 per cent CI 16.34 to 102.75; P<0.001), Manchester terrier (OR 41.41; 95 per cent CI 5.49 to 312.22; P<0.001), saluki (OR 27.46; 95 per cent CI 3.65 to 206.32; P=0.001), whippet (OR 22.43; 95 per cent CI 10.18 to 49.42; P<0.001), English springer spaniel (OR 11.41; 95 per cent CI 5.44 to 23.94; P<0.001) and bearded collie (OR 10.85; 95 per cent CI 1.45 to 81.34; P=0.020). Breeds with decreased odds of being a CRGV case compared with crossbreds were the Staffordshire bull terrier (OR 0.50; 95 per cent CI 0.15 to 1.70; P=0.268), German shepherd dog (OR 0.45; 95 per cent CI 0.06 to 3.38; P=440) and Jack Russell terrier (OR 0.37; 95 per cent CI 0.09 to 1.58; P=0.179) (table 2).
Female dogs were significantly more likely to be a case than male dogs (OR 1.51; 95 per cent CI 1.02 to 2.24; P=0.042), while the odds of neutered dogs being diagnosed with CRGV was 3.36 times that of entire dogs (95 per cent CI 1.93 to 5.85; P<0.001) (table 2).
This study is the first to investigate signalment risk factors for CRGV in UK dogs. Breed (P<0.001), KC breed group (P<0.001), neuter status (P=0.001) and sex (P=0.011) were shown to be significantly associated with confirmed diagnosis of the disease. Age group was not a significant risk factor. Two KC breed groups—gun dogs and hounds—were between 9 and 10 times more likely to be diagnosed with CRGV than terriers, while no toy dogs were diagnosed with the disease. Specific breeds showing increased odds of CRGV compared with crossbreds included Hungarian vizslas, flat-coated retrievers, whippets and English springer spaniels. Breeds with decreased odds included German shepherd dogs, Jack Russell terriers and Staffordshire bull terriers. Females and neutered dogs were also more likely to be diagnosed with CRGV.
Previous studies have suggested CRGV to be associated primarily with greyhounds,3–5 11 with a single instance reported of a great dane in Germany.6 While greyhounds did not have a significantly higher odds of CRGV diagnosis in this study (OR 1.65, P=0.629), the disease (as it is currently occurring in the UK) was instead associated with multiple breeds. Compared with crossbreds, specific breeds with increased odds of being a CRGV case included the flat-coated retriever (OR 84.48), Hungarian vizsla (OR 40.98), Manchester terrier (OR 41.41), saluki (OR 27.46), whippet (OR 22.43), English springer spaniel (OR 11.41) and bearded collie (OR 10.85) (table 2). Breeds with decreased odds of being a CRGV case, when compared with crossbreds, were the Staffordshire bull terrier (OR 0.50), German shepherd dog (OR 0.45) and Jack Russell terrier (OR 0.37). The UK KC classifies spaniels and retrievers as gun dogs, and salukis, whippets and Hungarian vizslas as hounds, which explains why these breed groups were much more likely to be diagnosed with CRGV than terriers. It is possible that these breed associations result from an inherent susceptibility among these breeds as a result of genetic or behavioural patterns, but it is also possible that the predisposition results from geographical confounding whereby these breeds may occur more commonly in areas with a high risk of CRGV occurrence. While CRGV has been reported from multiple locations across the UK, breed popularity varies throughout the country. A recent study by the UK KC, which analysed the breakdown of dog registrations by breed in 10 UK regions in 2016, suggested that different regions each have their own top 10 favourite breeds (http://www.telegraph.co.uk/pets/essentials/top-dog-breeds-across-the-uk/). In fact, English springer spaniels (second most likely breed diagnosed with CRGV) were among the top 10 favourite breeds in both South East and North West England—the two regions containing a high percentage of cases.
Breed preferences can be driven by multiple factors including body size: large dogs are more common in rural areas, while smaller dogs are generally preferred in urban areas (http://www.telegraph.co.uk/pets/essentials/top-dog-breeds-across-the-uk/). Similarly, it is logical that gun dogs and hounds may predominate in rural areas where owners may participate in countryside sports such as shooting and hunting. Further studies investigating the geographical distribution of breeds and breed groups in the UK would help to decompose the breed associations identified in this study and explore whether these breeds or breed groups are inherently more susceptible to developing CRGV, or whether areas with a higher risk of CRGV occurrence coincide with higher proportions of these breeds.
The potential reasons for the associations between CRGV and being female or neutered are less clear. It has previously been reported that being female is a risk factor for certain TMAs in human beings, including TTP,12 although for other TMA conditions this is not necessarily the case. There is no evidence that females in the CRGV cohort were pregnant or postpartum and indeed, although there was an association with female dogs there was also an association with neuter status.
The denominator used in this study represented a totally primary care population, whereas the cases included some referral cases (30 per cent) and therefore some referral bias may have been created during selection.13 In addition, the denominator population represented the spread of dogs under primary veterinary care during 2013, whereas the cases were recorded from 2012 to 2017. Breed popularity can wax and wane quite rapidly, so the 2013 denominator may not exactly represent the breed spreads for each year from 2010 to 2017. In addition, this study classified Jack Russell terriers as ‘not-KC recognised’, but since 2016 the KC has officially recognised this breed as belonging to the breed group ‘terriers’. As this study identified the breed to have a decreased odds of diagnosis (OR 0.37), future studies may find the terrier breed group to have an even lower risk of being diagnosed with CRGV than the current study depending on how Jack Russell terriers are classified (based on the period of interest and denominator population). CIs for the variables common breed and KC breed group were comparatively wide, most likely due to the small number of CRGV cases, and suggest that these results are less robust than those variables with narrower CIs. However, CIs for breeds with a decreased odds of being diagnosed with CRGV were considerably narrower and more robust, suggesting that greater confidence can be placed in the identification of breeds with a lower risk of being a CRGV case than those with an increased risk. A larger sample of CRGV dogs would allow for a more robust analysis.
CRGV was initially reported largely in the New Forest area of England, resulting in an increased interest and awareness of the disease in this area. If certain breeds are more popular in that area, then the results of this study may be biased towards those breeds. However, since seasonal outbreaks began in 2012, CRGV has been reported in other parts of the UK, and the disease has been widely publicised in national and local media, so that increased awareness is likely no longer confined to the New Forest area and therefore any potential bias arising from the New Forest focus is likely to have been mitigated over time.
In conclusion, the results of this study suggest that gun dogs and hounds have an increased risk of developing CRGV in the UK, while toy dogs and terriers appear to be the breed groups least at risk. Specific breeds with increased odds of CRGV included Hungarian vizslas, flat-coated retrievers, whippets and English springer spaniels. As well as helping veterinarians develop an index of suspicion for the disease, an understating of the breeds at risk may help to develop causal models for CRGV, and potentially play a role in identifying the aetiology of the disease. However, further studies investigating the distribution of specific breeds and breed groups in the UK, and the factors driving these distributions, would help to determine whether the high-risk breeds and breed groups identified in this study are indeed inherently more disposed to being diagnosed with CRGV or whether the results stem from an increased proportion of such breeds in areas of greater risk.
The authors acknowledge the Medivet Veterinary Partnership, Vets4Pets/Companion Care, Blythwood Vets, Vets Now and the other UK practices who collaborate in VetCompass.
Contributors KBS performed all analyses and wrote the first draft of the paper. LPH and DJW compiled the case data set. DON compiled the VetCompass dog denominator data set. All authors contributed substantially to the interpretation of data, drafting of the final manuscript and critical revision for important intellectual content. All authors approved the final version of the manuscript for submission.
Funding This study was generously funded by the New Forest Dog Owners Group (NFDog) and the Alabama Rot Research Fund (ARRF).
Competing interests None declared.
Ethics approval Ethics approval was granted by the RVC Ethics and Welfare Committee (reference number URN 2015 1369).
Provenance and peer review Not commissioned; externally peer reviewed.