Evaluation of platelet activation in canine immune-mediated haemolytic anaemia


  • A. E. Ridyard,

    1. Division of Veterinary Clinical Sciences, Royal (Dick) School of Veterinary Studies, The University of Edinburgh, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author
  • D. J. Shaw,

    1. Division of Veterinary Clinical Sciences, Royal (Dick) School of Veterinary Studies, The University of Edinburgh, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author
  • E. M. Milne

    1. Division of Veterinary Clinical Sciences, Royal (Dick) School of Veterinary Studies, The University of Edinburgh, Easter Bush Veterinary Centre, Roslin, Midlothian EH25 9RG
    Search for more papers by this author


Objectives: To establish whether heightened platelet activation is a common feature of canine immune-mediated haemolytic anaemia, and to evaluate the hypothesis that platelet activation plays a role in the pathogenesis of thromboembolism.

Methods: Using whole-blood flow-cytometric analysis, the proportion of activated platelets and platelet-leucocyte aggregates in blood samples from 14 dogs with immune-mediated haemolytic anaemia and 14 healthy dogs was calculated. General linear models with binomial errors were used to compare groups. Results from the immune-mediated haemolytic anaemia-affected dogs were then correlated with established risk factors for thromboembolism in canine immune-mediated haemolytic anaemia, D-dimer concentration and antithrombin activity.

Results: There was a strong correlation between platelet activation and severe thrombocytopenia, with heightened platelet activation being observed predominantly in severely thrombocytopenic dogs.

Clinical Significance: Dogs with immune-mediated haemolytic anaemia, particularly those with concurrent severe thrombocytopenia, are likely to have heightened platelet activation, which may play a role in the pathogenesis of thromboembolism.


Circulating platelets play a pivotal role in haemostasis. Direct interaction with damaged endothelium and coagulation factors leads to activation, aggregation and formation of the haemostatic plug at the site of vascular damage. Increasingly, platelets are implicated in the pathogenesis of thrombosis in a variety of human diseases, e.g. atherosclerosis, cardiac disease, sepsis and neoplasia (Schmitz and others 1998). This reflects the intimate link between activated platelets, the vascular endothelium and circulating leucocytes (Lindmark and others 2000, Wagner and Burger 2003, Barnard and others 2005, Li 2008).

Immune-mediated haemolytic anaemia (IMHA) is a common haematological disorder in the dog, with cocker and springer spaniels being over-represented (Burgess and others 2000, Warman and others 2008). Mortality rates of up to 70% are reported, with thromboembolism (TE) being cited as a significant cause of death (Scott-Moncrieff and others 2001, Carr and others 2002). This is despite the fact that thrombosis is otherwise very uncommon in dogs. Although coagulation abnormalities have been confirmed in IMHA (Mischke 1998, Scott-Moncrieff and others 2001, Carr and others 2002), the aetiology of TE in IMHA remains unclear. Recently re-ported heightened platelet activation in dogs with IMHA (Weiss and Brazzell 2006) and the evidence that aspirin therapy improves clinical outcomes (Weinkle and others 2005) suggest that platelet activation may be involved.

In human medicine, platelet activation is assessed by flow cytometric quantification of platelet-bound fibrinogen, platelet microparticles (PMPs) and P-selectin expression by platelets (Abrams and others 1990, Schmitz and others 1998). The detection of increased numbers of platelet-leucocyte aggregates (PLAs) has also been proposed as a sensitive indirect marker of platelet activation (Michelson and others 2001).

The primary aim of this study was to determine whether heightened platelet activation is common in canine IMHA, using whole blood (WB) flow cytometry. An additional aim was to evaluate whether platelet activation plays a role in the pathogenesis of TE in IMHA, by correlating markers of platelet activation with known risk factors for TE in IMHA (McManus and Craig 2001, Carr and others 2002), and with laboratory markers of TE and hypercoagulability, namely D-dimer concentration (Nelson and Andreasen 2003) and antithrombin (AT) activity (Green 1988), respectively.

Materials and methods

Case selection

As part of routine diagnostic evaluation, blood samples were collected from 20 dogs with idiopathic IMHA presented to the University of Edinburgh Hospital for Small Animals between March 2006 and October 2008. Diagnosis of IMHA was based on standard criteria: the presence of anaemia (PCV<30%) and one or more of spherocytosis, positive slide agglutination and positive Coombs’ test. Coombs’ testing was performed in all cases by the Clinical Immunology Laboratory at the University of Bristol. Cases were excluded from the final dataset if they had been on immunosuppressive therapy with glucocorticoids for more than 2 days before presentation (n=4), or if anaemia was persistently non-regenerative (n=2). The final IMHA group comprised 14 dogs, which were classified as primary idiopathic IMHA based on the exclusion of underlying conditions known to result in secondary IMHA, and response to immunosuppressive therapy. Diagnostic evaluation in these cases included complete blood count (CBC) (n=14), serum biochemical profile (SBP) (n=14), abdominal ultrasonography (n=13), thoracic radiography (n=10), screening by PCR for Borrelia burgdorferi and Anaplasma phagocytophila (n=7), splenic aspirate cytology (n=2) and bone marrow aspirate cytology (n=2). Fourteen dogs having blood samples collected for a reason unrelated to the study (e.g. general health assessment before general anaesthesia or blood donation) formed a healthy control group; these dogs were judged to be healthy based on clinical examination and normal CBC/SBP.

Following collection into plastic syringes by standard venepuncture, blood was immediately transferred into tri--potassium ethylene diamine-N,N′, N′-tetra-acetic acid (K3EDTA; 0·5 or 1 ml plastic tubes, Teklab Ltd, Sacriston, Durham), 3·8% sodium citrate (9:1 blood:sodium citrate; 0·5, 1·0 or 2·0 ml plastic tubes, Teklab Ltd), plain (6 ml plastic tubes, BD Biosciences, Oxford, UK) and fluoride oxalate (0·5 ml plastic tubes, Teklab Ltd) tubes for routine haematology, coagulation screening and serum biochemical analysis, respectively. A 50 μl aliquot of citrate-anticoagulated WB was used for evaluation of platelet activation markers and a 125 μl aliquot of K3EDTA-anticoagulated WB for evaluation of PLAs. All samples were processed within 30 minutes of collection, to minimise ex vivo increases in platelet activation indices (preliminary studies, data not shown).

Flow cytometry studies

Platelet activation markers Platelets were identified by labelling with fluoroscein isothiocyanate (FITC) or phycoerythrin (PE)-conjugated antihuman CD61 antibody (Clone VIPL2, BD Biosciences). Platelet-bound fibrinogen was detected using FITC-conjugated antihuman fibrinogen antibody (Dako UK Ltd, Ely, UK). P-selectin expression was detected using PE-conjugated antihuman CD62P antibody (Clone AC1.2, BD Biosciences). Isotype (negative) controls were FITC or PE-conjugated mouse IgG1, κ (Clone MOPC-21, BD Biosciences) for CD61 and CD62P, FITC-conjugated rabbit IgG (AbD Serotec, Oxford, UK) for fibrinogen. Activated platelets (positive control) were generated in vitro by incubation of WB with bovine thrombin (final concentration 0·3 U/ml; Trinity Biotech Plc, Bray, Ireland), in the presence of 1 mM GPRP (Merck Chemicals Ltd, Nottingham, UK) to prevent fibrin polymerisation.

To evaluate platelet-bound fibrinogen and P-selectin expression, 5 μl of WB was incubated with 5 μl of anti-CD61 antibody and 5 μl of either antifibrinogen antibody or anti-CD62P antibody, respectively, in 35 μl FACS flow buffer (BD Biosciences) for 30 minutes at room temperature in the dark. Samples were diluted with 800 μl FACS flow buffer before FACS analysis. Isotype (negative) and positive control runs were included in analysis of each sample.

Platelet-leucocyte aggregates Granulo-cytes and monocytes were identified using FITC-conjugated anticanine CD45 antibody and antihuman CD14 antibody, respectively (Clones YKIX 716.13 and TuK4, AbD Serotec). Isotype controls were FITC-conjugated rat IgG2b for CD45 and FITC-conjugated rat IgG2a for CD14 (Clones LO-DNP-11 and MRC OX-34, respectively, AbD Serotec).

To detect PLAs, 25 μl of WB was incubated with 5 μl of anti-CD61 antibody and 5 μl of either anti-CD45 or anti-CD14 antibody in 65 μl FACS flow buffer for 30 minutes at room temperature. A measured quantity of 800 μl of FACS lysing buffer (BD Biosciences) was added to lyse red blood cells and minimise coincident events. Isotype controls were included in the analysis of each sample.

Flow cytometric analysis

Data were collected with a flow cytometer (BD FACScalibur) and analysed using CellQuestPro (BD Biosciences). Logarithmic forward and side-angle scatter acquisition were used for initial identification of the platelet population and optimisation of acquisition settings, whereas linear acquisition was used for discrimination of leucocyte subsets. Electronic and fluorescent gating was used to identify populations of interest (Fig 1). Background fluorescence was evaluated using isotype control antibodies, and gates were established with more than 99·5% of the events falling within the first log-decade on the scatter plots. Color compensation was performed for each antibody pair to eliminate leakage between channels.

Figure 1.

Sequential gating and back-gating were used to define platelet and leucocyte subpopulations. Scatter plots of (a) logarithmic forward and side-angle scatter characteristics of whole blood; (b) CD61 fluorescence which identifies all platelet-related events (R1); (c) platelet subpopulations, platelets (R2), platelet aggregates (R4) and platelet microparticles (R4), defined using forward and side-angle scatter characteristics of CD61-positive events and (d) CD61-CD62P dual-positive events (right upper quadrant) representing platelets expressing P-selectin. Leucocytes were identified by CD45 fluorescence, with back-gating to forward and side-angle scatter characteristics then being used to identify the granulocyte cluster (e). Gating on the granulocyte region, the percentage of platelet- granulocyte aggregates was determined as the number of dual-positive events (upper right quadrant) (f)

To assess platelet-bound fibrinogen and P-selectin expression, 5000 or more CD61-positive events were acquired per sample. Platelet regions, namely all CD61-positive events (R1), platelets (R2), platelet aggregates (R3) and PMPs (R4), were identified as previously described (Moritz and others 2003) to enable subpopulation analysis (Fig 1). The proportion of platelet aggregates and PMPs within a sample was calculated, and using quadrant statistics, the percentage of events positive for both CD61 and fibrinogen or CD62P was also determined per region (Fig 1).

To detect platelet-granulocyte aggregates, 5000 or more CD45-positive events were collected. The granulocyte cluster was then identified using forward and side-scatter characteristics of CD45-positive events; the number of platelet-granulocyte aggregates (CD45-PLAs) in a sample was expressed as the percentage of CD45- positive events that were also positive for the platelet marker CD61 (Fig 1). To detect platelet-monocyte aggregates (CD14-PLAs), 1000 CD14-positive events were collected and the number of aggregates determined as above.

Clinical data

Age, sex and breed of all cases were recorded. Clinical data were collected from the IMHA group to enable correlations between platelet activation data and risk factors for TE, D-dimer concentration and AT activity. Risk factors for TE were defined as severe thrombocytopenia (<50×109/l), severe hyperbilirubinaemia (>85·5 μmol/l), hypoalbuminaemia (<26 g/l), increased alkaline phosphatase (ALP) (>60 IU/l) (Carr and others 2002), and moderate to severe leucocytosis (>28×109/l) (McManus and Craig 2001). Plasma D-dimer concentration more than 1000 ng/ml was used to identify those patients likely to have TE (Nelson and Andreasen 2003) and AT activity less than 70% of normal was also used to define patients at increased risk of thrombosis.

CBCs and serum biochemical analysis were carried out in the R(D)SVS Clinical Laboratory using the Pentra 60 Impedance analyser (HORIBA ABX, Montpelier, France) and the IL-600 (Instrumentation Laboratory, Z0128, Milan, Italy), respectively. Blood smear evaluation accompanied all CBCs. D-dimer and AT activity were measured at the Animal Health Trust (Lanwades Park, Newmarket) using the Minutex D-dimer latex agglutination test (Trinity Biotech) and Dade Behring Berichrom (Marburg, Germany) Antithrombin III assay, with reference intervals of less than 250 ng/ml and 70 to 126%, respectively.

Outcome was recorded including whether the patient survived to discharge and whether any deaths were attributable to IMHA or TE.

Statistical analysis

To compare platelet-bound fibrinogen, PMP levels, P-selectin expression by platelets and the percentage of circulating PLAs between the IMHA and healthy dogs, standard general linear models with binomial errors (GLM be) with the number of positive and negative cells entered as the response variable and the group as the dependent variable were used. The association between markers of platelet activation and the presence or absence of risk factors for TE, D-dimer concentration and AT activity was evaluated with GLM be. A Spearman rank correlation test was carried out to look at the association between the percentage of PMP and the fibrinogen binding in R4. Analyses were carried out in R (version 2.7.0, © 2008 The R Foundation for Statistical Computing) and P<0·05 was taken to indicate statistical significance.

This study was approved by the University of Edinburgh Veterinary Ethical Review Committee.


Clinical data

In the IMHA group, the most common breeds were the cocker spaniel (n=7) and springer spaniel (n=3), with one each of golden retriever, bichon frise, Lhasa apso and crossbreed; the healthy group varied more with three springer spaniels, two each of cocker spaniels, Irish wolfhounds and newfoundlands, and one each of English bull terrier, German short haired pointer, golden retriever, Labrador retriever and West Highland white terrier. Of the IMHA dogs, 7 of 14 were neutered compared with 4 of 14 of the healthy dogs, and 8 of 14 IMHA dogs were female compared with 2 of 14 of healthy dogs. The mean age of IMHA dogs was 7·6 years (±2·1 sd) and of healthy dogs was 5·6 years (±3·4 sd).

Platelet activation in IMHA-affected versus healthy dogs

Platelet-bound fibrinogen There was no difference in fibrinogen binding in R1. However, there was an increase in the proportion of fibrinogen-positive events within the R2 platelet region in the IMHA group (12·9 versus 9·8%; P=0·047) (Table 1).

Table 1. Summary of the differences in fibrinogen binding, platelet microparticle number, platelet P-selectin (CD62P) expression and platelet-leucocyte aggregate number in IMHA compared with healthy dogs
 SubpopulationIMHA, % (95% CI)Healthy, % (95% CI)Analysis summary
  1. IMHA immune-mediated haemolytic anaemia; PMPs platelet microparticles.

FibrinogenR1 (all CD61-positive events)14·2 (9·1 to 22·0)12·6 (11·2 to 14·1)P=0·200
 R2 (normal platelets)12·9 (8·2 to 20·4)9·8 (8·7 to 11·0)P=0·047
 R3 (platelet aggregates)35·4 (27·4 to 45·8)39·3 (37·0 to 41·8)P=0·650
 R4 (platelet microparticles)10·9 (7·1 to 16·8)10·0 (7·6 to 13·2)P=0·330
PMPs 5·9 (4·0 to 8·7)2·6 (2·1 to 3·3)P<0·001
P-selectinR1 (all CD61-positive events)0·8 (0·5 to 1·5)0·5 (0·3 to 0·7)P=0·100
 R2 (normal platelets)0·4 (0·2 to 0·7)0·3 (0·2 to 0·5)P=0·520
 R3 (platelet aggregates)2·1 (0·2 to 23·1)0·9 (0·5 to 1·6)P=0·001
 R4 (platelet microparticles)0·4 (0·0 to 3·8)0·2 (0·0 to 4·9)P=0·196
CD45-PLAsGranulocytes2·4 (1·6 to 3·5)6·5 (5·2 to 8·1)P<0·001
CD14-PLAsMonocytes5·7 (4·0 to 8·3)8·4 (7·0 to 10·2)P=0·020

Platelet subpopulations While the mean proportion of aggregates did not differ (IMHA 6·3% versus healthy 8·3%; P=0·131), the proportion of PMPs was greater in IMHA dogs (5·9 versus 2·6%; P<0·001) (Table 1).

P-selectin expression There was no difference in P-selectin expression in R1 between IMHA and healthy dogs. However, there was a higher proportion of CD62P-positive platelets in the R3 aggregate region in the IMHA group (2·1 versus 0·9%; P=0·001) (Table 1).

Platelet-leucocyte aggregates IMHA-affected dogs had a lower proportion of both CD45-PLAs (2·4 versus 6·5%, P<0·001) and CD14-PLAs (5·7 versus 8·4%, P=0·020) (Table 1).

Relationship between platelet activation and risk factors for TE in dogs with IMHA

All IMHA-affected dogs had elevated ALP (median 178 IU/l, range 68 to 1522) and as such comparisons could not be performed. The median albumin concentration was 28·5 g/l (range 23·7 to 36·3). Only two IMHA dogs were hypoalbuminaemic (<26 g/l), precluding statistical analyses of its relationship with platelet activation. The median platelet count was 87×109/l (range 5 to 367), with 5 of 14 IMHA dogs severely thrombocytopenic (median 32×109/l, range 5 to 38). The median total bilirubin concentration was 29·2 μmol/l (range 1·9 to 338·3) and 4 of 14 had severe hyperbilirubinaemia (median 115·5 μmol/l, range 91·2 to 338·3). The median total white blood cell count was 29·5×109/l (range 9·1 to 72) and six IMHA dogs had moderate to severe leucocytosis (median 40·4×109/l, range 33·3 to 72·0).

Table 2 summarises the relationships between fibrinogen binding, PMPs, P-selectin expression and PLAs, and risk factors for TE. Increases were observed in fibrinogen binding in severely thrombocytopenic dogs, compared with the remaining IMHA-affected dogs in R3 and R4 (Fig 2a and b); a similar pattern was observed with PMP (Fig 2c). Interestingly, despite the relatively low number of CD62P-positive events overall, P-selectin expression was also increased in all the four platelet populations in the severe thrombocytopenic dogs (Fig 2d to g). The opposite pattern was observed in the CD45 granulocyte population (Fig 2h). Variation in platelet activation using any of the markers was not statistically associated with either hyperbilirubinaemia or moderate to severe leucocytosis.

Table 2. Summary of the univariate analysis of the relationship between the various markers of platelet activation and risk factors for thromboembolism and D-dimer concentration
Risk factors for thromboembolism
 SubpopulationSevere thrombocytopeniaHyperbilirubinaemiaModerate-severe leucocytosisD-dimer >1000 ng/ml
  Platelet countTotal bilirubinLeucocyte countD-dimer concentration
  >50×109/l, % (95% CI)<50×109/l, % (95% CI)P value<85·5 μmol/l, % (95% CI)>85·5 μmol/l, % (95% CI)P value<28×109/l, % (95% CI)>28×109/l, % (95% CI)P value<1000 ng/ml, % (95% CI)>1000 ng/ml, % (95% CI)P value
  1. PMPs platelet microparticles.

FibrinogenR19·9 (5·4 to 8·3)25·0 (12·5 to 49·9)0·07416·8 (9·4 to 30)9·3 (4·4 to 19·6)0·06911·8 (6·7 to 20·8)181 (7·3 to 45·0)0·27213·1 (4·5 to 37·8)12·5 (5·5 to 28·4)0·766
 R29·1 (4·8 to 7·4)22·5 (10·5 to 48)0·22015·3 (8·4 to 27·9)8·5 (3·9 to 18·3)0·07410·4 (6 to 18)17·3 (6·7 to 45·0)0·09711·9 (4·1 to 348)107 (4·7 to 24·1)0·792
 R329·0 (19·6 to 2·9)48·5 (36·5 to 4·4)0·01437·6 (26·1 to 4·1)30·6 (22·6 to 1·4)0·28130·6 (21·3 to 3·9)43·1 (27·9 to 66·5)0·76232·3 (18·7 to 5·8)33·9 (19·1 to 60·4)0731
 R47·3 (4·4 to 2·2)20·2 (9·8 to 41·7)0·00112·9 (7·6 to 1·9)7·2 (2·5 to 20·7)0·44410·1 (5·9 to 17·3)12·1 (4·7 to 31·6)0·80910·8 (3·8 to 30·6)9·4 (5·3 to 16·8)0·963
PMPs 3·4 (2·7 to 4·3)7·0 (1·6 to 30·8)0·0245·3 (3 to 9·5)3·0 (1·1 to 8·1)0·3184·8 (2·4 to 9·6)4·2 (1·9 to 9·4)0·6866·5 (2·5 to 17·2)4·9 (2·9 to 8·1)0·376
P-selectinR10·5 (0·3 to 1)2·2 (1·2 to 4·1)0·0010·9 (0·4 to 2·1)0·6 (0·3 to 1·4)0·4370·7 (0·3 to 1·6)1·1 (0·4 to 3)0·5130·7 (0·2 to 2·2)1·0 (0·3 to 3·6)0·919
 R20·2 (0·1 to 0·5)0·8 (0·3 to 2·7)0·0230·4 (0·2 to 1)0·2 (0·1 to 0·5)0·2210·3 (0·1 to 0·6)0·4 (0·1 to 1·7)0·3830·3 (0·1 to 1)0·4 (0·1 to 1·2)0·610
 R31·2 (0·3 to 13·8)13·5 (3·4 to 53·3)0·0011·9 (0·6 to 14·6)8·7 (1·4 to 56)0·8661·4 (0·3 to 20·5)7·4 (2·6 to 21)0·7470·7 (0·1 to 21·6)5·3 (0·8 to 34·4)0·858
 R40·3 (0·1 to 2·3)1·7 (0·5 to 5·2)0·0010·6 (0·2 to 3·3)0·6 (0·2 to 2)0·9620·3 (0·1 to 2·7)1·2 (0·3 to 4·3)0·9850·3 (0·1 to 6·7)1·3 (0·3 to 6·6)0·645
CD45-PLAsGranulocytes3·1 (2 to 4·8)1·4 (0·7 to 3)0·0172·3 (1·4 to 4)2·4 (1·1 to 5·1)0·7912·1 (1·2 to 3·9)2·7 (1·4 to 5·2)0·5352·2 (0·9 to 5·1)2·3 (0·9 to 6)0·933
CD14-PLAsMonocytes9·9 (5·4 to 18·3)25·0 (12·5 to 49·9)0·5295·2 (3·2 to 8·5)7·4 (3·9 to 14·1)0·6296·8 (4·2 to 11·1)4·5 (2·2 to 9·2)0·2643·6 (2 to 6·6)7·9 (3 to 20·6)0·013
Figure 2.

Boxplots of fibrinogen binding in (a) R3 and (b) R4; PMPs (c); and P-selectin expression in (d) R1, (e) R2, (f) R3 and (g) R4 as a proportion of the whole platelet population, and percentage of platelet-granulocyte aggregates (h), with IMHA dogs subdivided into two groups based on whether they had a platelet count of less than 50×109/l (severe thrombocytopenia). In all cases the shaded boxes are the interquartile range, the horizontal line the median and the dashed whiskers the range of data. R1=all CD61-positive events, R2=platelets, R3=platelet aggregates and R4=platelet microparticles (PMPs)

Platelet activation in thrombocytopenic IMHA dogs versus healthy dogs

When the platelet activation data from the five IMHA cases with concurrent severe thrombocytopenia were reanalysed in comparison with the healthy dogs, there were statistically significant increases in fibrinogen binding to platelets in all platelet regions, in the proportion of PMPs, and in P-selectin expression in R1, R2 and R3, but not in the PMP region (R4) (Table 3).

Table 3. Summary of the univariate analysis of the platelet activation in thrombocytopenic IMHA dogs compared with healthy dogs
 SubpopulationSeverely thrombocytopenic IMHA, % (95% CI)Healthy, % (95% CI)P value
  1. IMHA immune-mediated haemolytic anaemia; PMPs platelet microparticles.

FibrinogenR125·0 (18·1 to 34·5)12·6 (11·4 to 14·0)P<0·001
 R222·5 (15·8 to 32·0)9·8 (8·8 to 10·9)P<0·001
 R348·5 (42·4 to 55·3)39·3 (37·3 to 41·5)P<0·001
 R420·2 (14·4 to 28·3)10·0 (7·8 to 12·9)P<0·001
PMPs 8·6 (3·1 to 23·6)2·6 (2·1 to 3·3)P<0·001
P-selectinR12·3 (1·6 to 3·0)0·5 (0·3 to 0·7)P<0·001
 R20·8 (0·4 to 1·4)0·3 (0·2 to 0·5)P=0·012
 R313·5 (7·1 to 25·6)0·9 (0·5 to 1·6)P<0·001
 R41·7 (0·9 to 2·8)0·4 (0·04 to 3·6)P=0·472
CD45-PLAsGranulocytes1·4 (0·6 to 3·0)6·5 (5·2 to 8·1)P<0·001
CD14-PLAsMonocytes5·4 (1·5 to 19·5)8·4 (6·9 to 10·2)P=0·237

Relationship between platelet activation, D-dimer more than 1000 ng/ml and AT activity in dogs with IMHA

AT activity was within the reference interval for all the samples analysed. For four IMHA-affected dogs, no D-dimer value was obtained; in the remaining 10, all dogs had plasma D-dimer concentrations above the reference range; one at 250 to 500 ng/ml, five at 500 to 1000 ng/ml and four at more than 1000 ng/ml. For fibrinogen binding, PMPs and platelet P-selectin expression, there were no differences between those dogs with D-dimer less than or more than 1000 ng/ml (P>0·610; Table 2), nor were there differences in CD45-PLAs. However, there were more CD14-PLAs in dogs with D-dimer more than 1000 ng/ml (7·9 versus 3·6%; P=0·013). Only one case had concurrent severe thrombocytopenia and D-dimer more than 1000 ng/ml, although D-dimer concentration was only available for three of five severely thrombocytopenic dogs.


Eleven of the 14 IMHA cases survived to discharge. Of the three dogs that died during the initial period of hospitalisation, death was attributed to the sequelae of IMHA in all dogs, with PTE being suspected clinically in two dogs and DIC in the third. In one of the dogs with suspected PTE, in addition to severe thrombocytopenia, D-dimer concentration was more than 2000 ng/ml. Each of the remaining two dogs had two risk factors for TE, namely severe thrombocytopenia and marked leucocytosis in one, and severe hyperbilirubinaemia and marked leucocytosis in the other; unfortunately in these cases D-dimer and AT were not measured.


While there was limited evidence of global platelet activation, this study revealed changes within the platelet subpopulations and in dogs with severe thrombocytopenia. The lack of overall increase in platelet activation is in marked contrast to the previous work in IMHA dogs (Weiss and Brazzell 2006), where heightened P-selectin expression was commonly observed. This is likely to be due to differences in populations and methodology, especially our use of WB rather than platelet-rich plasma. In human medicine, the use of WB flow cytometry has superseded other methods as it minimises in vitro platelet activation (Ritchie and others 2000). The range in the number of platelets expressing P-selectin in our study (0·12 to 1·75% in healthy dogs and 0·11 to 4·36% in IMHA-affected dogs) is consistent with another recent canine study (Wills and others 2006) and in naturally occurring human diseases (Cassar and others 2003, Mimidis and others 2004, Yngen and others 2004, Chung and others 2007), where WB was used. Although our use of the AC1.2 antihuman CD62P antibody may have underestimated P-selectin expression due to lower affinity for canine P-selectin (Tarnow and others 2008), the absence of a global increase in fibrinogen binding suggests that low P-selectin expression is a true reflection of the activation status.

Following activation, platelets can undergo a variety of activation-related events including conformational changes to the platelet GpIIb/IIIa complex which leads to fibrinogen binding, formation of PMPs and expression of P-selectin on the platelet surface, with the potency of the initial stimulus determining which activation-related events occur. The observation in this study that fibrinogen binding by platelets, but not P-selectin expression, was increased in the R2 platelet region in IMHA may be explained by the fact that P-selectin expression appears to require exposure to a greater concentration of agonist than other activation-related events (Holmes and others 1999, Gerdsen and others 2005). This apparently discordant observation has also been reported in people with acute pulmonary embolism where conformational changes in GpIIb/IIIa have been seen in the absence of increased expression of platelet P-selectin (Chung and others 2007).

As platelet aggregation is mediated by fibrinogen, it is not surprising that the proportion of fibrinogen-positive events was high in the aggregate region in both IMHA and control groups compared with other regions. The small but significant increase in P-selectin expression in this region in the IMHA group is more difficult to explain although it may reflect platelet hyper-reactivity, at least in a subset of platelets, as observed in some human conditions associated with thrombosis (e.g. type I diabetes mellitus; Yngen and others 2004). An alternative explanation is that these events are platelet-erythrocyte aggregates (PEAs), not platelet-platelet aggregates, as has been observed in sickle cell anaemia (Wun and others 1997) where PEAs may contribute to vaso-occlusive disease. Interaction between metabolically active erythrocytes and platelets is known to augment platelet reactivity (Santos and others 1991, Valles and others 1991). Further evaluation using erythrocyte-specific markers is warranted.

PMP formation occurs following platelet activation and apoptosis (Piccin and others 2007). Increased levels of circulating PMPs have been demonstrated in a number of diseases associated with thrombosis (Hugel and others 2005, van der Zee and others 2006, Piccin and others 2007). In this study, not only were PMPs increased in IMHA-affected dogs, but interestingly, as with both fibrinogen binding and P-selectin expression, the percentage of PMPs was greatest in the severely thrombocytopenic IMHA cases. The clinical significance of these findings is unclear. PMPs have the potential to exhibit procoagulant activity, expressing binding sites for fibrinogen and expressing P-selectin. In people with immune-mediated thrombocytopenia (IMT), PMPs are thought to play a protective role by preventing haemorrhagic diatheses (Panzer and others 2006), whereas they may be responsible for the thrombotic complications of heparin-induced thrombocytopenia and atherosclerosis (Hughes and others 2000). PMP heterogeneity may influence biological effect (van der Zee and others 2006, Perez-Pujol and others 2007).

Evaluation of the association between risk factors for TE and platelet activation markers revealed a strong association between severe thrombocytopenia and fibrinogen binding, P-selectin expression and the proportion of PMPs. This is interesting as severe thrombocytopenia is also a risk factor for negative outcome in dogs with IMHA (McManus and Craig 2001, Carr and others 2002). The high clinical index of suspicion for TE in two dogs with severe thrombocytopenia lends circumstantial evidence to the hypothesis that platelet activation plays a role in the pathogenesis of TE. However, two of the dogs in this subset (platelet counts of 5 and 10×109/l, respectively), which survived to discharge, had no risk factors for TE or additional markers suggestive of TE or DIC and may have had concurrent IMT. Heightened platelet activation in these dogs may be a consequence of concurrent IMT, irrespective of IMHA or TE. In support of this hypothesis, heightened P-selectin expression has also been reported in people with autoimmune thrombocytopenia (Panzer and others 2006), although interestingly, in these patients, platelets exhibiting PAC-1 binding (an alternative to fibrinogen binding for detecting alterations in GpIIb/IIIa) were rare.

Increased circulating PLAs in several conditions associated with thrombosis (Barnard and others 2005) suggest that they have procoagulant properties, via expression of tissue factor (CD142) (Michelson and others 2001). However, in the present study there was a significant decrease in the number of platelet-granulocyte aggregates in IMHA. This may have been due to the strong correlation between platelet-granulocyte aggregates and platelet count (Pearson’s product moment correlation coefficient P=0·816, P<0·001; data not shown); a similar but less robust pattern was observed for platelet-monocyte aggregates and platelet count (P=0·419, P=0·030). Only when dogs with D-dimer more than 1000 ng/ml were compared with the rest of the IMHA group, was a relative increase in platelet-monocyte aggregates observed. Although the small numbers in this group (n=4) make it difficult to draw any meaningful conclusions, the combination of increased D-dimer concentration and platelet-monocyte aggregates is worth noting. Evaluation of tissue factor expression by PLAs, rather than enumeration, is likely to give a better reflection of their procoagulant role.

One of the major limitations of this study was the difficulty in accurately diagnosing TE. Measurement of D-dimers has shown promise as a non-invasive laboratory marker of TE (Nelson and Andreasen 2003); a D-dimer concentration more than 1000 ng/ml, in the absence of haemoabdomen, was reported to have a specificity of 100%. In using this stringent cut-off we may have underestimated the number of cases with TE, as the reported sensitivity of the test at this cut-off was reported to be only 80%. Interestingly, if we adopt the lower cut-off of 500 ng/ml, with a reported sensitivity of 100% (specificity 70%), 9 of 10 of the cases would have had values consistent with TE, even though only one of these dogs was suspected clinically of having TE, perhaps reflecting the subtle symptomatic presentation of TE. Further evaluation is clearly required.

Overall, our study has shown convincing evidence of enhanced platelet activation in a subset of patients with IMHA, namely those dogs with concurrent severe thrombocytopenia. Further studies are required to further investigate the role of platelets, PMPs and PEAs in IMHA and TE, particularly to determine if early therapeutic intervention aimed at reducing platelet activation may prevent thromboembolic complications.


The authors gratefully acknowledge BSAVA Petsavers for funding this work. They also like to thank the clinicians at the University of Edinburgh Hospital for Small Animals and referring practitioners who assisted with these cases, the staff in the University of Edinburgh Veterinary Pathology Unit and Sue Cade of the Animal Health Trust for assistance with laboratory analyses and Dr Niall McHugh, Ms Alison Burrells, and Professor Ivan Morrison of The Roslin Institute of the University of Edinburgh for assistance with flow cytometry. Dr Paul Harrison of the Oxford Haemophilia Centre and Thrombosis Unit, Churchill Hospital, Oxford, is thanked for his enthusiasm and guidance.