• Open Access

Antiphospholipid Antibodies in Dogs with Immune Mediated Hemolytic Anemia, Spontaneous Thrombosis, and Hyperadrenocorticism

Authors


  • All work was completed at the Colorado State University Veterinary Medical Center and at the Colorado State University Department of Pathology.

  • This study was presented in part in abstract form at the joint 59th Annual Meeting of the American College of Veterinary Pathologists and the 43rd Annual Meeting of the American Society for Veterinary Clinical Pathology held November 15–19, 2008 in San Antonio, TX.

Corresponding author: C. Olver, Department of Microbiology, Immunology and Pathology, Colorado State University, 200 Lake Street, Fort Collins, CO 80523; e-mail: christine.olver@colostate.edu.

Abstract

Background

The pathophysiology of thrombus formation in canine IMHA and other diseases remains unclear. Antiphospholipid antibodies (aPL) are an important cause of thrombosis in humans and might cause thrombosis in dogs.

Hypothesis

Dogs with IMHA, spontaneous thrombosis, and hyperadrenocorticism will have increased levels of aPL and lupus anticoagulants (LA), compared with healthy and sick dogs.

Animals

Thre aPL were measured in healthy controls (n = 40–45); sick dogs without thrombosis (n = 86); IMHA (n = 37); spontaneous thrombosis (ST, n = 11); and hyperadrenocorticism (n = 17). Four groups of dogs were also tested for the presence of LA: healthy controls (n = 40); sick dogs without thrombosis (n = 13); IMHA (n = 13); and ST (n = 5).

Methods

Prospective cohort study. Dogs were tested for aPL by an ELISA and for LA by the dilute Russell's Viper venom time (dRVVT). Median values were compared by Kruskal–Wallis (aPL) or ANOVA (LA), and an odds ratio for development of thrombosis in dogs positive for aPL was calculated.

Results

aPL are uncommon in healthy dogs. A total of 13/86 sick dogs without thrombosis, 7/37 dogs with IMHA, 1/11 dogs with ST, and 3/17 dogs with HAC were positive for protein binding-dependent aPL. There was no significant difference in the number of dogs positive for aPL for any of the study groups, and there was no increased risk for thrombosis in dogs positive for aPL. No dogs had LA.

Conclusions

Our preliminary research does not support a strong role for aPL for the development of thrombosis in dogs with IMHA and other thombotic diseases, although future studies are warranted.

Abbreviations
ABP

adult bovine plasma

aCL

anticardiolipin antibodies

aPE

antiphosphatidylethanolamine antibodies

aPL

antiphospholipid antibodies

aPS

antiphosphatidylserine antibodies

APS

antiphospholipid syndrome

BSA

bovine serum albumin

LA

lupus anticoagulant

OD

optical density

PL

phospholipid

tPA

tissue plasminogen activator

β2GP1

beta-2 glycoprotein 1

There is a strong association between immune-mediated hemolytic anemia (IMHA) and the formation of thromboemboli.[1-6] Case fatality rates in dogs with IMHA have been reported to be as high as 70%, and death or euthanasia is often related to thromboembolic complications.[4, 5] Thrombi are present on postmortem examination in up to 80% of dogs with IMHA,[2] but the true incidence of thromboembolic disease in dogs is likely underestimated owing to rapid postmortem clot dissolution and the low sensitivity of noninvasive ante mortem diagnostic tests.[1, 7, 8] Pulmonary thromboemboli, secondary to IMHA are most commonly described in the literature, although thrombi also occur in the heart, liver, spleen, kidney, and pituitary gland.[2, 9]

The physiologic processes leading to thrombus formation in dogs with IMHA have not been elucidated, although several mechanisms have been proposed. Endothelial factors such as vasculitis and antiendothelial antibodies do not appear to be major contributing factors to thrombus formation in dogs,[3, 10] whereas platelet activation and hypercoagulability contribute to the pathophysiology.[11, 12] Antiphospholipid antibodies (aPL), extensively described in humans, have also been suggested to be a cause for thrombus formation in dogs with IMHA.[2, 3, 12] Antiphospholipid antibodies are immunoglobulins (IgG, IgM, or IgA) that target either phospholipids (PL) or, more commonly, proteins that bind to phospholipids, such as β2 glycoprotein1 (β2GP1).[13] The most commonly studied and clinically relevant aPL in humans are anticardiolipin antibodies (aCL) and lupus anticoagulants (LA), which are a subset of aPL that prolong phospholipid-dependent coagulation assays. Other antiphospholipid antibodies likely play a role.[13]

The antiphospholipid syndrome (APS) in humans is an autoimmune disease that is clinically characterized by recurrent thrombosis or recurrent pregnancy loss.[13-17] The pathophysiology of thrombus formation is not known, but might be because of anti-tPA antibodies,[18] annexin A5 resistance,[19] increased protein C resistance,[20, 21] or complement activation.[22] The diagnosis of APS in humans is made by the presence thrombosis or recurrent pregnancy loss, and by one of the following: the presence of aPL, the presence of LA, or the presence of aβ2GP1 antibodies detected at least twice, 12 weeks apart.[13-17, 23]

There has been minimal research in the veterinary literature regarding the detection of aPL or APS in dogs. There have been 2 reports on detection of aCL in healthy and sick dogs,[24, 25] 1 case report of a lupus type anticoagulant in a dog with thrombosis,[26] and 1 study investigating the presence of the LA via screening tests in dogs with IMHA.[12] The role of LA in dogs with IMHA was investigated and the authors concluded that APS did not play a significant role in the 20 dogs studied.[12] Because mixing studies and confirmatory tests were not performed in this study and aPL ELISAs were not utilized, APS could not be completely ruled out in these dogs. Because of the strong association between IMHA and thrombosis in dogs, and because an association between aPL and IMHA in humans has been shown,[27] we investigated the role of aPL in dogs with IMHA. The overall goal of this work was to determine the association between aPL and dogs with IMHA. In addition, we investigated the association between aPL and dogs with spontaneous thrombosis. We hypothesized that these groups of dogs may be candidates for having the canine equivalent of human APS and that they would have increased levels of aPL compared with healthy dogs and sick dogs without thrombosis. Dogs with hyperadrenocorticism (HAC) were also tested for the presence of aPL because they are known to be in a hypercoagulable state. Although hypercoagulability in dogs with HAC has been shown to be because of decreased antithrombin and increased levels of coagulation factors,[28] no investigation into the presence of aPL has been performed. There has been 1 publication that showed no relation between aPL and atherosclerosis in human patients with HAC.[29] We hypothesized that IMHA and ST would be associated with the presence of increased aPL compared with healthy controls, sick dogs without thrombosis, and dogs with HAC. In addition, we hypothesized that IMHA and ST would be associated with the presence LA compared with healthy controls and sick dogs without thrombosis.

Materials and Methods

Study Design

This was a prospective cohort study. Dogs with IMHA, ST, and HAC, and sick dogs without thombosis (4 groups) were compared for presence of aPL by an ELISA, and for LA by the dilute Russell's Viper venom time (dRVVT). Values were compared by Kruskal–Wallis (aPL) or ANOVA (LA), and an odds ratio for development of thrombosis in dogs positive for aPL was calculated. The individuals within the groups for aPL or LA differed, and these groups will be described below in the “animals” section.

Animals

aPL ELISA

Healthy controls: Between 40 and 45 (depending on the phospholipid tested) healthy, client-owned dogs that presented to the Colorado State University Veterinary Teaching Hospital (CSU VTH) for routine spay or neuter, routine dental cleaning, or elective orthopedic procedures plus 10 healthy, employee-owned dogs were used as healthy controls. The median age of these dogs was 5. There were 17 castrated males, 4 intact males, 23 spayed females, and 1 intact female. There were 11 mixed breed dogs, 7 Labrador Retrievers, 3 Australian Shepherds and Golden Retrievers, and 2 of each of the following: Australian Heeler, Border Collie, and German Shepherd Dog. The remaining dogs were represented by 1 each of various purebred dogs. CBC and biochemistry panels were performed to rule out underlying disease.

Experimental groups: These groups include sick dogs without thrombosis, dogs with IMHA, dogs with spontaneous thrombosis (ST), and dogs with HAC. Sick dogs without thrombosis: Ninety-one dogs were identified as sick dogs without thrombosis. Although these dogs were not found to have thromboembolic disease based on medical records, it was not possible to completely rule out small thrombi or venous thrombi, as these are difficult to diagnose. The median age of this group of dogs was 9. There were 38 castrated males, 6 intact males, 44 spayed females, and 3 intact females. There were 28 mix breed dogs, 10 Labrador Retreivers, 8 Golden Retreivers. 4 German Shepherd Dogs, and 2 of each of the following breeds: Boston Terrier, Standard Poodle, Australian Shepherd, Husky, Bernese Mountain Dog, and Rottweiler. The remaining dogs were represented by 1 each of various purebred dogs. Five dogs were excluded from this group for insufficient samples. IMHA group: Criteria for inclusion in the IMHA group consisted of regenerative or nonregenerative anemia (PCV < 40%), and one or more of the following: moderate to marked spherocytosis, positive saline agglutination, or positive Coombs’ test. These dogs were considered to have primary IMHA because of lack of concurrent disease, but extensive diagnostic testing to rule out secondary IMHA was not performed in all dogs. Thirty-seven dogs fit the criteria for diagnosis of IMHA. The median age of these dogs was 8. There were 12 castrated males, 2 intact males, 17 spayed females, and 1 intact female. There were 6 mixed breed dogs, 3 Cocker Spaniels, and 2 Standard Poodles. The remaining dogs were comprised of 1 each of various purebreds. The breed and sex was not recorded for 5 of the dogs. Dogs with IMHA were confirmed to have thromboembolic disease by detection of 1 or more thrombi on ultrasound examination or postmortem examination. Dogs in any group were suspected of having thromboembolic disease by 1 of the following criteria: Sudden onset of respiratory distress, increased Aa gradient on blood gas analysis, sudden onset of unexplained neurologic disease (1 patient), or sudden onset of marked limb edema (1 patient). Spontaneous thrombosis: Criteria for ST consisted of detection of 1 or more thrombi on ultrasound examination or postmortem examination, and no evidence of underlying disease. Eleven patients were identified that fit the criteria for ST. The median age of these dogs was 7.1. There were 5 castrated males, 2 intact males, and 4 spayed females. There were 3 Labrador Retrievers, 3 Staffordshire Terriers, and 1 mix breed dog. The remaining dogs were comprised of 1 each of various purebreds. Hyperadrenocorticism: Criteria for inclusion in the HAC group consisted of an ACTH stimulation test compatible with HAC (post ACTH stimulation >29 μg/dL), and 1 or more clinical or biochemical abnormalities compatible with HAC, including polyuria and polydipsia, excessive panting, endocrine alopecia, pendulous abdomen, or increased serum alkaline phosphatase. Dogs were excluded if they had treatment with anticoagulants, before blood collection. Seventeen dogs were identified that fit the criteria for HAC. The median age of these dogs was 10. There were 5 castrated males, 1 intact male, and 11 spayed females. There were 3 mixed breed dogs and 2 Shetland Sheepdogs. The remaining dogs were comprised of 1 each of various purebred dogs.

Lupus Anticoagulant Assays

For the lupus anticoagulant assays, the case inclusion criteria and sampling handling were the same, but the individuals within each group were different from those included in the aPL ELISA. One exception was that dogs with hyperadrenocorticism were not included because only serum was available. Healthy controls: Forty normal dogs were used as controls. The median age of these dogs was 6 years, although the ages were not available for 14 dogs (all were adult). There were 21 castrated males, 1 intact male, 14 spayed females, and 4 intact females. Breeds were 21 Jack Russell Terriers, 7 Golden Retrievers, 6 mixed breed, 2 Beagles, and 1 each of various other pure bred dogs. Although Jack Russel Terriers were overrepresented in this group, there were no statistical differences in dRVVT screen, dRVVT confirm, PT or aPTT when Jack Russel Terriers were compared with the other breeds (data not shown). Sick dogs without thrombosis: Thirteen dogs were identified as sick dogs without thrombosis. The median age of these dogs was 8. There were 6 castrated males and 7 spayed females. There were 3 mixed breed dogs, 3 Labrador Retrievers, and the remaining dogs were comprised of 1 each of various purebreds. IMHA: Thirteen dogs fit the criteria for IMHA and had plasma samples available for dRVVT testing. The median age of these dogs was 8. There were 4 castrated males, 1 intact male, 7 spayed females, and 1 intact female. There were 3 mixed breed dogs and 2 Cocker Spaniels, The remaining dogs were comprised of 1 each of various purebreds. Spontaneous thrombosis: Five dogs fit the criteria for ST and had plasma samples available for dRVVT testing. The median age of these dogs was 7.6. There were 3 castrated males and 2 spayed females. There was 1 mixed breed dog and the remaining dogs were comprised of 1 each of various purebreds (Australian Shepherd, Labrador Retriever, Jack Russell Terrier, Golden Retriever, and Staffordshire Terrier).

Blood Sampling

Whole blood for the aPL ELISA was collected in serum separator tubes, tubes containing no anticoagulant, or tubes containing sodium citrate as an anticoagulant. Previous validation studies performed in the lab showed no difference in detection of aPL with serum or citrated plasma (no difference in results comparing each PL tested in ELISA with matched serum or citrated plasma from 4 different dogs and paired t-tests, data not shown, publication pending). Blood was centrifuged at 4200 × g for 10 minutes twice, and serum or plasma was harvested and stored at −80°C for up to 6 months until tests were performed. Samples were batched for analysis. Whole blood for CBCs was collected into tubes containing EDTA as an anticoagulant. Samples were processed within 30 minutes of receipt. Blood smears were evaluated by medical technologists (routine evaluation of leukocyte differential and erythrocyte morphology as part of the CBC) and reviewed by a clinical pathologist. For the lupus anticoagulant assays, whole blood was drawn from a jugular vein with the first stick only and placed directly in tubes containing 3.2% sodium citrate at a 9 : 1 ratio whole blood: sodium citrate. Platelet-poor plasma was prepared from the whole blood collected in 3.2% sodium citrate by double centrifugation (to ensure platelet and platelet particle removal, personal communication, John McIntyre, HLA-V Vascular Biology Laboratory, Beech Grove, IN) for 10 minutes at 4200 × g. Platelet counts were not performed, although samples were evaluated microscopically to ensure that the majority of platelets had been removed. Samples were frozen at −80°C for up to 3 months until coagulation assays were performed.

Laboratory Methods

Antiphospholipid Antibody ELISA

A modified indirect ELISA procedure was performed (validation manuscript in preparation). Briefly, ELISA plates1 were coated with phospholipid (CL, PE, PS)2 diluted 1 : 3 in chloroform:methanol and allowed to dry in the dark at room temperature for 1.5 hours. A nonspecific binding plate coated with 1 : 3 chloroform:methanol only was treated exactly as the phospholipid-coated plates. Plates were blocked with 100 μL of 10% bovine serum albumin (BSA) in TBS per well. Plates were washed 1 time in filter sterilized Tris buffered saline2 (pH 7.4), containing 0.5% Tween (TBS-T).b Dog plasma was diluted 1 : 100 in 10% adult bovine plasma (ABP) or 1% BSA in TBS-T, and 50 μL of each sample was applied to wells in triplicate. Both ABP and BSA were used as diluents to determine if there was a difference in aPL binding with addition of phospholipid-binding proteins (ABP), versus no addition of binding proteins (BSA). A blank triplicate well containing diluent only and no primary antibody was also included in the phospholipid-coated ELISA plate. Plates were washed 8 times between each step with 200 μL TBS-T. A volume of 50 μL horseradish peroxidase conjugated sheep anti-dog IgG3 diluted 1 : 10 000 in TBS was added to each well. All incubations occurred at room temperature for 1 hour. Antibody binding was detected by by TMB microwell peroxidase substrate4 and absorbance at 450 nm was measured as optical density (OD).5 The final OD for each sample was determined by the mean OD of the triplicate wells. The mean blank OD and the mean OD for the corresponding nonspecific binding plate wells were subtracted from each sample final OD to give the adjusted OD. All ELISA results are reported as the adjusted OD. Normal dog blood incubated with hemin was used as a positive control, as described by McIntyre et al.[30-33] The ELISA was validated with these positive control values, and comparisons of OD readings between serum and citrated plasma, fresh and frozen samples, and samples refrigerated for 0–4 days were tested. There was no effect on OD readings in serum versus plasma, fresh or frozen (−80°C), or refrigerated storage of samples (manuscript in preparation).

Anti-β2GP1 Antibody ELISA

A modified indirect ELISA was performed. Briefly, ELISA plates1 were coated with 5% (w/v) human (the only available source) β2GP16 diluted in 0.01 M sodium bicarbonate buffer and allowed to incubate overnight at 4°C. Plates were washed 8 times between each step with 200 μL of PBS2 Tween.2 Plates were blocked with 100 μL 2.5% BSA in PBS per well. Dog plasma was diluted 1 : 100 in PBS, and 50 μL was applied to each well. A blank triplicate well containing diluent only and no primary antibody was also included in the β2GP1-coated ELISA plate. A volume of 50 μL horseradish peroxidase conjugated rabbit anti-dog IgG3 diluted 1 : 10 000 was added to each well. All incubations were performed at room temperature for 1 hour. Antibody binding was detected by TMB microwell peroxidase substrate (KPL),4 and absorbance at 450 nm was measured as OD.5 The final OD for each sample was determined by the mean OD of the triplicate wells. The mean blank OD was subtracted from each sample final OD to give the adjusted OD. All ELISA results are reported as adjusted OD. Normal dog blood incubated with hemin was used as a positive control as described by McIntyre et al.[30-33]

Dilute Russell's Viper Venom Time

An algorithm outlining the correct procedure for detection of LA by the dRVVT is shown in Figure 1. The dRVVT test was performed in 3 steps: (1) dRVVT screen, (2) dRVVT confirm, and (3) mixing study with 1 : 1 ratio of dog plasma to pooled plasma from healthy dogs. The presence of LA is detected by the dRVVT assay, which uses the AMAX Accuclot screen and confirm test kits7 on an automated coagulation analyzer.7 The test was run according to package insert instructions. Although information on the exact concentration of phospholipids in these reagents is proprietary, the AMAX Accuclot screen test contains a low concentration of phospholipids, whereas the AMAX Accuclot confirm test contains a high concentration of phospholipids. Mixing assays were performed with a 1 : 1 dilution of plasma with pooled normal dog plasma. Positive and negative control samples were provided with the AMAX Accuclot test kits and were tested, before each sample run to ensure that the instrument and reagents were working properly.

Figure 1.

Algorithm for detection of the lupus anticoagulant (LA). All steps must be completed to accurately detect LA.

Prothrombin Time (PT) and activated Partial Thromboplastin Time (aPTT)

PT was evaluated by AMAX ALEXIN reagents7 for the automated coagulation analyzer. The aPTT was evaluated by AMAX ThromboMAX reagents7 for the automated coagulation analyzer, according to manufacturer's instructions.

Statistical Analysis

The healthy control dogs were used to establish the normal reference interval for aPL for each phospholipid (CL, PE, PS) and β2GP1. Data points were considered to be outliers if the difference between the extreme data point and the next highest data point was equal to or greater than 1/3 the range of values.[34] For each phospholipid and β2GP1, the adjusted OD was recorded and the data were tested for normal distribution by the D'Agostino and Pearson omnibus normality test. Owing to high numbers of dogs having zero or low aPL adjusted OD, none of the data were normally distributed, therefore the reference interval was reported as the multiple of the median. The multiple of the median measures, how far a test result deviates from the median value of the reference interval. For example, if the median OD for anticardiolipin antibodies in normal dogs is 0.09, the OD for each patient will be divided by 0.09 to see how many multiples of the median it deviates from the normal patient median value. This method is commonly used in human medical tests that have highly skewed normal values and removes center to center variation in results so that values can be compared across different laboratories.[34]

Dogs were categorized into the following groups: sick dogs without thrombosis, IMHA, ST, and HAC. Dogs were considered positive for aPL or anti-β2GP1 antibodies if their multiple of the median for the aPL or anti-β2GP1 ELISA were outside of the established reference interval for normal dogs. The numbers of dogs positive for aPL in each group were compared by a Chi-Square test. Multiples of the median values were then compared for each group by the Kruskal–Wallis test, followed by the Dunn's multiple comparison test to identify significant differences in median values for each phospholipid or β2GP1. To determine whether there was an association between the presence of aPL and thrombosis, median aPL values from IMHA dogs with and without thrombosis were compared for each phospholipid and β2GP1 by the Kruskal–Wallis test, followed by the Dunn's multiple comparison test. Fisher's exact test was used to determine whether there was an association between dogs with known and suspected thrombosis and the presence of aPL. An odds ratio with 95% confidence interval was calculated to predict whether aPL above the normal reference interval was associated with increased odds of developing thrombosis in dogs with IMHA. The level of significance was set at  .05. All statistical analyses were performed by a statistical software package.8 Data for dRVVT, PT, and aPTT were recorded and the data were tested for normal distribution by the D'Agostino and Pearson omnibus normality test. No outliers were identified. Data were normally distributed; therefore the mean ± 2 standard deviations was used to determine the reference interval. A ratio of dRVVT screen to confirm was also established for normal dogs.

The dRVVT screen versus confirm, and screen versus mixed plasma were compared by a one-tailed, paired t-test. The association between prolonged dRVVT screen and thrombosis, and prolonged aPTT and thrombosis were evaluated by a Fisher's exact test. To determine whether groups of dogs had increased median values for a given test, an ANOVA was used followed by the Dunn's multiple comparisons test. The level of significance was set at P ≤ .05. All statistical analyses were performed by a statistical software package.8

Results

aPL ELISA

Normal canine reference intervals for aCL, aPE, aPS, and anti-β2GP1 antibodies are reported in Table 1. There was no significant difference in the reference interval and median OD with BSA versus ABP as a diluent for the primary antibody. The multiple of the median reference intervals for dogs were similar, yet slightly higher than the multiple of the median values for normal human sera.[33] There were dogs that had aPL levels outside of the reference interval in each group, although no group had a statistically significant increase in the number of dogs outside of the reference interval (Table 2). For the sick dogs without thrombosis, 13/86 (17%) dogs had antibodies against 1 of the phospholipids with ABP as the diluent, and 7/86 (8%) had antibodies against one of the phospholipids with BSA as the diluent. Four out of 86 (5%) had antibodies against two of the phospholipids (aCL and aPE diluted in ABP). The dogs with aPL had a variety of illnesses (5: neoplasia, 3: degenerative, 2: congenital, 2: inflammatory, 2: autoimmune, 1: unclassified), and none of these dogs had documented evidence of thromboembolism or clinical suspicion of thromboembolism based on the medical records or necropsy findings when available. For dogs with IMHA, 7/37 (19%) had antibodies against one of the phospholipids with ABP as a diluent, and 5/37 (14%) had antibodies against one of the phospholipids with BSA as a diluent. There was 1 dog that had antibodies against 2 of the phospholipids (aPE and aCL in ABP). Twenty-three of the dogs with IMHA had sufficient follow-up data available to identify the presence or absence of thromboembolic disease. Of these dogs, 5/23 (22%) had confirmed thromboemboli (1: PTE, 1: hepatic vein, 1: Distal aorta, 2: multiple) and 4/23 (17%) had suspected thromboemboli, based on clinical signs. The remaining 14 dogs were either alive at the last recheck appointment, died, or were euthanized, and did not have evidence of thromboembolic disease clinically or on necropsy. One of the 5 dogs with confirmed thromboembolism and one of the 4 dogs with suspected thromboembolism had antibodies against phosphatidylserine (diluted in ABP). Four dogs with IMHA had antibodies against 1 or more phospholipids, but did not have evidence of thromboembolism. The odds ratio for risk of thromboembolism with the presence of aPL was 1.048 (95% CI 0.14–7.9). There was no significant difference in the median value of the multiple of the median for any of the phospholipids for dogs with and without thromboembolism.

Table 1. Reference intervals for antiphospholipid antibodies and anti beta 2 glycoprotein 1 antibodies in 40–45 healthy dogs
Phospholipid or ProteinPrimary Antibody DiluentMedian ODReference Interval (MOM)
  1. ABP: Adult bovine plasma, BSA: Bovine serum albumin.

CLABP0.130–5.0
CLBSA0.170–5.0
PSABP0.110–6.8
PSBSA0.130–5.3
PEABP0.070–5.3
PEBSA0.030–6.0
β2GP1PBS0.240–4.8
Table 2. Percent of dogs with aPL ELISA optical density measurements outside of the reference intervals for the various phospholipids and β2GP1
 Sick Dogs without ThrombosisIMHAHACST
PS ABP2% (2/86)8% (3/37)6% (1/17)0
PS BSA5% (4/86)5% (2/37)12% (2/17)9% (1/11)
PE ABP11% (9/86)11% (4/37)6% (2/17)0
PE BSA4% (3/86)000
CL ABP7% (6/86)06% (2/17)9% (1/11)
CL BSA1% (1/86)3% (1/37018% (2/11)
β2GP12% (2/86)000

For dogs with ST, aCL antibodies were measured in 10 dogs and aPE, aPS, and anti-β2GP1 antibodies were measured in all 11 dogs. Within the ST group, 1/11 (9%) dogs had antibodies against one of the phospholipids with ABP as the diluent, and 3/11 (27%) had antibodies against one of the phospholipids with BSA as the diluent. One dog with spontaneous thrombosis had antibodies against 2 phospholipids (aCL and aPS in BSA). Within the hyperadrenocorticism group, 3/17 (18%) had antibodies against one of the phospholipids with ABP as the diluent, and 2/17 (12%) had antibodies against one of the phospholipids with BSA as the diluent. None of the dogs with HAC had antibodies against 2 or more of the phospholipids.

There was no significant difference in the number of dogs positive for aPL antibodies among the study groups. In addition, there was no significant difference among the number of dogs positive with BSA versus ABP as the diluent. Compared with healthy controls, median values for aCL and aPS IgG antibodies measured by ELISA were increased for dogs with IMHA with BSA as the diluent (≤ .05, Figs 2 and 3). Compared with healthy controls, dogs with spontaneous thrombosis had increased median values of aPS IgG antibodies and dogs with hyperadrenocorticism had increased median values for aCL anticardiolipin IgG antibodies, with BSA as the diluent (≤ .05, Figs 2 and 3).

Figure 2.

Scatter plot comparison of multiple of the median results for aCL diluted in BSA for dogs with various diseases and healthy control dogs. (NC: healthy control dogs, PC: sick dogs without thrombosis, IMHA: immune mediate hemolytic anemia, ST: spontaneous thrombosis, HAC: hyperadrenocorticism). *≤ .05, Kruskal–Wallis test followed by the Dunn's multiple comparison test.

Figure 3.

Scatter plot comparison of multiple of the median results for aPS diluted in BSA for dogs with various diseases and normal control dogs. NC: healthy control dogs, PC: patient control dogs or sick dogs without thrombosis, IMHA: immune mediate hemolytic anemia, ST: spontaneous thrombosis, HAC: hyperadrenocorticism. *≤ .05, Kruskal–Wallis test followed by the Dunn's multiple comparison test.

Anti-β2-Glycoprotein1 Antibodies

One sick dog without thrombosis had antibodies against β2GP1. Dogs with IMHA, spontaneous thrombosis, and hyperadrenocorticism did not have evidence of anti-β2GP1 antibodies (data not shown).

Dilute Russell's Viper Venom Time for Detection of Lupus Anticoagulants

The dRVVT screen and dRVVT confirm times for healthy dogs were 21.2 ± 2.5 seconds (mean, SD) and 15.0 ± 0.9 seconds (mean, SD), respectively. The PT and aPTT times for healthy dogs were 7.7 ± 0.5 seconds (mean, SD) and 12.6 ± 1.3 seconds (mean, SD). The ratio of dRVVT screen to confirm for healthy dogs was < 1.4. For dogs with IMHA, 8/13 (62%) had a prolonged dRVVT screen test. Seven of these 8 dogs also had a prolonged aPTT and increased D-Dimers. Of the 8 prolonged dRVVT screens, 6 corrected by the dRVVT confirm, with a ratio of screen: confirm <1.4. One of the 8 samples corrected with a ratio >1.4 and one of the samples did not correct at all with the dRVVT confirm assay (Fig 4). The sample that did not correct using the confirm assay did correct back into the normal reference interval when the mixing study was performed. The mean ratio of screen to confirm for dogs with IMHA was 1.35, which is normal. A mixing study of 1 : 1 plasma to pooled plasma from healthy dogs was performed in 7 of 8 dogs with prolonged dRVVT screen (1 dog did not have an adequate plasma sample for mixing studies). Of these 7 dogs, all dRVVT screen times returned to normal with addition of pooled plasma from healthy dogs (Fig 5). For the group of 5 dogs with spontaneous thrombosis, only 4 dogs had sufficient plasma samples to perform coagulation testing. One dog with ST had a prolonged dRVVT screen test (37.2 seconds) which corrected with a ratio of 1.4 with the dRVVT confirm test. The mixing study for this dog also corrected back into the reference interval. Three of the dogs had minimally prolonged dRVVT screen. Two of 3 of these dogs corrected with a ratio of 1.4 with the dRVVT confirm test, and all 3 corrected back into the normal reference interval with the mixing study.

Figure 4.

Comparison of dRVVT screen to dRVVT confirm test results for 8 dogs with IMHA that had prolonged dRVVT screen test results. (= .015, paired t-test).

Figure 5.

Comparison of dRVVT screen from 8 dogs with IMHA that had a prolonged dRVVT screen to a dRVVT screen with a 1 : 1 mixture of healthy dog plasma with sick dog plasma (= .004, paired t-test).

The ST and IMHA, dRVVT screen, confirm, and mixing assays, and PT and aPTT were evaluated by ANOVA followed by the Dunn's multiple comparison test. This showed that the dRVVT screen was elevated in dogs with IMHA, compared with healthy control dogs and sick dogs without thrombosis (Fig 6). Although the median dRVVT screen was elevated, all individual IMHA dogs with an elevated dRVVT screen test returned to within the reference interval for dRVVT screen when mixed 1 : 1 with pooled plasma from healthy dogs. The majority of these dogs also corrected by the dRVVT confirm with a normal ratio of <1.4.

Figure 6.

Scatter plot comparison of dRVVT screen results for the groups of dogs tested. NC: healthy control dogs, PC: sick dogs without thrombosis, IMHA: immune-mediated hemolytic anemia, ST: spontaneous thrombosis. Dogs with IMHA had a significantly increased median dRVVT screen compared with NC and PC dogs. *≤ .05.

Discussion

The results of this study show that aPL are uncommon in healthy dogs, but do occur in dogs with IMHA, ST, HAC, and various other illnesses. Detection of aCL, aPS, and aPE can be performed by a modified indirect ELISA. Although dogs with IMHA had increased levels of both aCL and aPS, the increased value was not outside of the reference interval for healthy dogs. In addition, there was abundant overlap in results for all groups of dogs, making this increase difficult to interpret in a clinical setting. One important factor to consider is that the increased levels of aCL and aPS were only seen when samples were diluted in BSA and not in ABP. ABP provides the necessary protein cofactors required for optimal phospholipid-binding and detection of phospholipid-binding protein dependent aPL. Because phospholipid-binding protein independent aPL are largely considered to be nonpathogenic in humans,[26] this increase might indicate that dogs with IMHA have increased levels of nonpathogenic aPL compared with healthy control dogs and sick dogs with and without thrombosis. These phospholipid-binding protein independent aPL are not typically associated with thromboembolic complications[35] and likely do not explain the increased incidence of thromboembolism in dogs with IMHA. The importance of the increase in phospholipid-binding protein independent aPL in this study remains unknown because of species differences between humans and dogs, and the paucity of information regarding canine aPLs in the literature. It has been hypothesized that autoantigens bind to membrane phospholipids in some dogs with IMHA.[36] It is possible that phospholipid-binding protein independent aPL might play a role in hypercoagulability in dogs, but further studies need to be performed to investigate this association. In addition, dogs were only tested for the presence of aPL at 1 point in time. Testing these dogs twice, 12 weeks apart would have allowed for evaluation of persistent aPL which are more important in people.[16]

When comparing dogs with IMHA that had suspected or confirmed thromboemboli with those without thromboemboli, there was no association between the presence of a thrombus and the presence of IgG aPL detected by ELISA. In addition, there was not an increased odds ratio for development of thrombosis for dogs with IgG aPL, although the number of dogs with thrombosis was small. Interestingly, the 2 dogs that had thrombosis and aPL, both had aPS. Antiphosphatidylserine IgG antibodies have recently been shown to be stronger predictors of thrombosis in human patients with APS.[37-39] Phosphatidylserine (PS) is a membrane phospholipid that participates in signaling leading to assembly of coagulation enzymes or to apoptosis, and if perturbed can allow for binding of proteins and autoantibodies.[13] Activation of PS could, therefore potentially contribute to hypercoagulability in dogs, but too few dogs were positive for aPS in the current study to draw strong conclusions from this observation.

These results support that there is not a strong association between IgG aPL and dogs with IMHA and that the presence of aPL is not the sole underlying mechanism for thromboembolism in dogs with IMHA, although aPL might lead to thrombosis in individual dogs. IgM and IgA aPL might be better predictors of thrombosis in people with APS.[40] Because only IgG aPL were evaluated in this study, the importance of IgM and IgA aPL in dogs with IMHA still needs to be evaluated. It has also been shown that people with 2 or more aPL in their serum are more likely to have clinical signs of APS than those with only 1 aPL in their serum.[39] In this study, only 1 dog with IMHA had 2 or more aPL present in serum. This dog was still alive 1 year after presentation for IMHA and had no clinical or historical evidence of thromboembolic disease.

Dogs with spontaneous thrombosis have a clinical presentation that is similar to humans with APS; therefore this group could represent the canine equivalent of APS. Dogs with ST had increased median levels of aPS compared with normal dogs and sick dogs without thrombosis, but the overall percentage of dogs with ST and aPL was very low. If this group of dogs truly represented the canine equivalent of APS, a higher percentage of aPL positive dogs would be expected. Although larger studies of dogs with ST may be needed, the low prevalence of aPL within this group suggests that there is not an association between aPL and dogs with ST.

Dogs with hyperadrenocorticism had increased levels of aCL compared with normal dogs and sick dogs without thrombosis, although the median value did not exceed the normal reference interval for healthy dogs. Again the primary antibody was diluted in BSA indicating phospholipid-binding protein independent aCL. None of the dogs with HAC had evidence of thromboembolic disease. Based on these results, we conclude that there is not an association between aPL and dogs with hyperadrenocorticism. We wished to ensure that all dogs with Cushing's disease were included, and thus the inclusion criteria were quite broad. This may have resulted in a decreased median aPL compared with including dogs with more rigorous criteria. However, only 1 dog had an aPL value outside of the reference range, and thus we can stand by the conclusion that dogs with HAC, as we hypothesized, do not have increased levels of aPL.

Persistent LAs in humans have been shown to be stronger risk factors, than aCL for venous and arterial thrombosis in the APS.[41, 42] Other reports have shown a stronger association between venous thrombosis only and the presence of LA.[43] Because detection of LA is 1 of the 3 diagnostic criteria used to confirm the APS in people, it was necessary to include this test to evaluate APS as the potential underlying etiology of thrombosis formation in dogs with IMHA.

The laboratory detection of LA in humans lacks standardization and there is no specific test available.[44, 45] The detection of LA is based on the criteria set by the Scientific and Standardization Committee, Subcommittee on antiphospholipid antibodies/lupus anticoagulant of the International Society on Thrombosis and Hemostasis.[46] These criteria include (1) prolongation of 1 or more phospholipid-dependent coagulation assays, (2) evidence that the prolongation is not because of a coagulation factor deficiency, and (3) evidence that the prolongation is because of a phospholipid inhibitor.[46] These criteria were met in this study using 3 separate steps, respectively, by screening the plasma by a coagulation test with low concentrations of phospholipids in the reagent (dRVVT screen test), mixing the plasma 1 : 1 with normal dog plasma to rule out a coagulation factor deficiency, and then confirming the presence of a phospholipid inhibitor by adding back high levels of phospholipids (dRVVT confirm test). The ratio of dRVVT screen: confirm is also utilized to determine if LAs are present and is the best way to express the final result. A dRVVT screen:confirm ratio above the established cutoff helps confirm the presence of LA.[45] Normalization of prolongation with a 1 : 1 dilution of normal dog plasma indicates that a factor deficiency is present, rather than the lupus anticoagulant. The dRVVT is considered a highly specific test for detection of LA in humans.[47] Patients who are positive for LA with the dRVVT have been shown to have a higher risk for thrombosis compared with those positive for LA by the Kaolin Clotting Test.[48] Because of this evidence, we chose the dRVVT for detection of LA in this study.

Lupus anticoagulants were not identified in any of the dogs with IMHA or ST in this study. Although, 8 dogs with IMHA had a prolonged dRVVT screen test, mixing studies with pooled plasma from healthy dogs 1 : 1 suggested the presence of a factor deficiency rather than a coagulation inhibitor, such as the LA. This is most likely because of consumption of multiple coagulation factors attributable to thrombosis or DIC in these dogs. This is further supported by increased D-dimers and prolonged aPTT in seven of these dogs. Another less likely explanation of prolonged dRVVT corrected by mixing is that the mixing studies resulted in dilution of antibodies. Therefore, weak LA activity might not be detected. This situation has been documented in 600 human patients with history of thrombosis,[49] however, dilution of antibodies is unlikely considering that the ratio of dRVVT screen:dRVVT confirm was normal in all dogs with IMHA except 1 dog with ratio >1.4 : 1. This dog had no evidence of thromboembolism on physical examination and had a negative test for d-dimers, making thromboembolism unlikely. Performing mixing studies using a ratio of 1 : 4 would have been useful to rule out a dilutional effect, but was not performed in this group of patients. Lastly, aPTT has been shown to be slightly prolonged in frozen samples compared with fresh samples,[50] but samples from all groups were frozen and freezing would not explain the presence of increased d-dimers.

Overall, the results of this study show that there does not appear to be an association between the presence of aPL and dogs with IMHA and ST. Moreover, the presence of aPL does not appear to be associated with thromboembolic complications in this population of dogs with IMHA. Although dogs with HAC were evaluated for the presence of aCL, aPE, aPS, and anti-β2GP1 antibodies, this group of dogs was not evaluated for the presence of LA, therefore the same conclusions cannot be drawn regarding dogs with HAC. Although aPL might play a role in individual dogs with IMHA and thromboembolic disease, further research investigating the role of additional antibody isotopes and additional aPL are necessary to prove or disprove an association.

Acknowledgments

This work was funded by a Morris Animal Foundation grant D05CA-089. The authors acknowledge John McIntyre and Dawn Wagenknecht for their help with the development of antiphospholipid antibody ELISAs.

Footnotes

  1. 1

    Greiner Bio-One, Frickenhausen, Germany

  2. 2

    Sigma-Aldrich, Inc, St Louis, MO

  3. 3

    AbD Serotec, Oxford, UK

  4. 4

    KPL, Gaithersburg, MD

  5. 5

    EMAX precision microplate reader, Molecular Devices, Sunnyvale, CA

  6. 6

    Fitzgerald Industries International, Concord, MA

  7. 7

    Trinity Biotech plc, Bray Co, Wicklow, Ireland

  8. 8

    Graphpad Prism, version 5.00 for Windows, San Diego, CA

Ancillary