Portions of this study were presented at the American College of Veterinary Internal Medicine Forum, Anaheim, CA, USA, June 9–12, 2010.a,b
Corresponding author: M. Kjelgaard-Hansen, Department of Small Animal Clinical Sciences, Faculty of Life Sciences, University of Copenhagen, Gronnegaardsvej 3, DK-1870 Frederiksberg C, Denmark; e-mail: email@example.com.
Background: The cytokine response in immune-mediated hemolytic anemia (IMHA) is poorly characterized and correlation with outcome is unknown.
Hypothesis/Objectives: To determine if cytokine activity is correlated with outcome in dogs with IMHA.
Animals: Twenty dogs with primary IMHA and 6 control dogs.
Methods: Prospective study on dogs with IMHA with blood sampling at admission. Serum activity of interleukin-2 (IL-2), IL-4, IL-6, IL-7, IL-8, IL-10, IL-15, IL-18, monocyte chemoattractant protein-1 (MCP-1), granulocyte-macrophage colony stimulating factor (GM-CSF), interferon-inducible protein-10, interferon-gamma, and keratinocyte chemoattractant (KC) was assessed.
Results: Thirty-day case fatality rate was 25% (5/20 dogs). Increased concentrations (median [range]) of IL-2 (45.5 ng/L [0;830] versus 0 ng/L [0;46.8]), IL-10 (8.2 ng/L [0;60.6] versus 0 ng/L [0;88.2]), KC (1.7 μg/L [0.3;4.7] versus 0.5 μg/L [0.2;1.1]), and MCP-1 (162 ng/L [97.6;438] versus 124 ng/L [90.2;168]) were observed in dogs with IMHA compared with controls. The cytokine profile was indicative of a mixture of pro- and anti-inflammatory cytokines of various cellular origins. Cytokines/chemokines strongly associated with macrophage/monocyte activation and recruitment were significantly increased in nonsurvivors compared with survivors; IL-15 (179 ng/L [48.0;570] versus 21.3 ng/L [0;193]), IL-18 (199 ng/L [58.7;915] versus 37.4 ng/L [0;128]), GM-CSF (134 ng/L [70.0;863] versus 57.6 ng/L [0;164]), and MCP-1 (219 ng/L [135;438] versus 159 ng/L [97.6;274]), respectively. Logistic regression suggested increased IL-18 and MCP-1 concentrations were independently associated with mortality in this population (P<.05, Wald's type 3).
Conclusions and Clinical Importance: A mixed cytokine response is present in dogs with IMHA and mediators of macrophage activation and recruitment might serve as prognostic indicators.
Primary immune-mediated hemolytic anemia (IMHA) in dogs is an idiopathic disorder characterized by antierythrocyte antibody production and immune system dysregulation.1,2 Increased serum concentrations of IL-2, IL-4, and TNF-α and an imbalance of IL-10 and IL-12 concentrations have been identified in human IMHA patients.3 Since cytokines including IL-4, IL-6, and IL-10 are known to promote antibody formation, it is suggested that type-2 T-helper (Th2) cells mediate development of autoimmune diseases such as IMHA.4 Studies of cytokine production by peripheral mononuclear cells in people with IMHA show that in basal conditions the T-cell proliferative response is twice that of controls suggesting hyperactivation.5
Antierythrocyte antibody binding can lead to intra- or extravascular hemolysis. Extravascular hemolysis occurs when erythrocytes opsonized by autoantibodies are removed by cells of the mononuclear phagocytic system following interaction of immunoglobulin constant fragments(Fc) with cell-surface Fc receptors.6 The spleen is the primary site for this process although hepatic Kupffer cells are also involved. Mononuclear cells are mobilized from bone marrow and recruited to sites of inflammation by monocyte chemoattractant protein-1 (MCP-1), a cytokine derived from both endothelial and mononuclear cells. Consistent with the key role played by the reticuloendothelial system in dogs with IMHA, increased concentrations of MCP-1 have recently been reported in dogs with IMHA.7
IMHA in dogs is associated with considerable case fatality rate, which appears to be related with the development of thromboembolic complications. Thrombosis is frequently identified at necropsy and it is well established that IMHA in dogs is associated with a prothrombotic state.2,8,9 Cross-talk between systemic inflammation and the coagulation system is implicated in the genesis of this prothrombotic state,10 which may involve erythrocyte phosphatidylserine exposure,11 procoagulant microparticle release,12 and cytokine-induced TF expression on monocytes and endothelial cells.13
Thus, current knowledge suggests IMHA in dogs is a proinflammatory process partially driven by increased cytokine activity. To date the systemic cytokine profiles of dogs with primary IMHA have not been studied. Increased activity of specific groups of cytokines could identify cell-populations critical to the pathophysiology and outcome of IMHA in dogs. Therefore, the purpose of this study was to determine if excess pro- or anti-inflammatory cytokine activity in primary IMHA in dogs was related to specific cell populations and whether certain cytokine activities correlated with outcome.
Materials and Methods
The study population comprised client-owned dogs admitted to the Queen Mother Hospital for Animals, at the Royal Veterinary College between October 2008 and October 2009. The study was approved by the RVC Ethics and Welfare Committee. Any dog with a diagnosis of primary IMHA was considered eligible. The diagnosis of primary IMHA required: documented anemia (PCV < 37%) with either a positive in-saline agglutination test, a positive Coombs' test or the observation of moderate to marked spherocytosis on a blood smear by a board certified clinical pathologist with no evidence of any underlying disease processes. Dogs were considered ineligible if their platelet count at study entry was <100,000/μL or if they had been pretreated with corticosteroids for >3 days, consistent with a recent publication on acute phase proteins in IMHA where most dogs received <3 doses of corticosteroids and all received <6 doses.14 Dogs that received antithrombotic medications including aspirin, unfractionated or low-molecular weight heparin, warfarin, or clopidogrel before enrolment were also excluded. To maximize recruitment, all dogs with a provisional diagnosis of IMHA were enrolled pending completion of diagnostic testing. Signalment, medical history, physical examination findings, and initial clinicopathologic data were recorded at hospital admission. In hospital, 30-day, and 6-month survival were recorded.
Blood samples were collected at admission by jugular venipuncture with a 21 G needle and a 10 mL syringe. Blood samples were immediately aliquoted into 1.1 or 1.3 mL nonvacuum, polypropylene tubes.c CBC samples were collected into potassium EDTA. Samples for serum biochemistry, C-reactive protein (CRP) and cytokine concentrations were collected into gel separator tubes. Serum samples for CRP measurement and cytokine analyses were centrifuged, separated, and frozen at −80°C. Samples were then batch analyzed.
CBCs and serum biochemical analyses were performed with automated analyzers,d,e and crosschecked by a board certified clinical pathologist. CRP was measured at the University of Copenhagen by an automated human turbidimetric assay validated for canine use15 calibrated with purified canine CRP.f Serum cytokine activity was assessed by a canine-specific multiplex assayg including internal quality control material with an automated analyzerh for interleukin-2 (IL-2), IL-4, IL-6, IL-7, IL-8, IL-10, IL-15, IL-18, MCP-1, granulocyte-macrophage colony stimulating factor (GM-CSF), interferon-inducible protein-10 (IP-10), interferon-gamma (IFN-γ), and keratinocyte chemoattractant (KC).
Six healthy dogs were enrolled from the hospital blood donation program to act as controls for the cytokine analyses. These dogs were declared healthy on the basis of history, absence of travel outside the United Kingdom, and normal full physical examination, CBC and serum biochemistry profile. These dogs were of various breeds and could not be age and sex matched to the study population.
Before test selection, data were assessed for normality. Although some variables were parametric, most were not and thus nonparametric tests were used throughout. Mann-Whitney U-tests were used to compare CRP and cytokine concentrations between healthy controls and dogs with primary IMHA and to compare 30-day survivors with nonsurvivors. Mann-Whitney U-tests were also used to compare concentrations of cytokines and of CRP between survivors and nonsurvivors, as well as BUN, bilirubin, albumin, and monocytes since these latter markers have previously been suggested to have prognostic significance.16–18 To test the independency of cytokines as prognostic factors for 30-day survival multiple logistic regression analysis with backwards exclusion followed by forward inclusion was performed as described previously.19 Variables were included in the multivariate analysis if their concentrations differed significantly between survivors and nonsurvivors based on univariate analyses. Correlation of cytokine concentrations to leukocyte count and duration of clinical signs before admission was assessed by Spearman's correlation coefficient. All analyses were conducted by commercial statistical software (Normality, Mann-Whitney U-test, and Spearman's correlation testi and logistic regressionj). Alpha was set at 0.05 for all tests.
Study Population Characteristics
Twenty dogs were enrolled between October 2008 and October 2009. The mean age was 7.2 ± 2.9 years. Eleven breeds were represented, including 6 English Springer Spaniels, 3 Cocker Spaniels, and 2 Labrador Retrievers. There was 1 each of Bedlington Terrier, Border Terrier, Dachshund, Jack Russell Terrier, Miniature Schnauzer, Shih Tzu, Staffordshire Bull Terrier, and Weimeraner. There was 1 crossbred dog. There was a high proportion of female dogs in the study population (17/20, 85%); there were 15 spayed females, 2 entire females, 2 entire males, and 1 neutered male. The median duration of clinical signs before presentation was 3 days (1–7). Nine dogs received no treatment from the referring veterinarian before presentation. Six dogs received corticosteroids before presentation: prednisolone (n = 4), dexamethasone (n = 2). No dog received corticosteroids for more than 3 days before presentation; the median duration of prior corticosteroid therapy was 1 day (1–3). Other therapies included sucralfate (n = 3), clavulanate-potentiated amoxicillin, metronidazole, vitamin K, ranitidine, azathioprine (all n = 2), enrofloxacin (n = 1), and opioids (n = 1). Two dogs received fresh whole blood before presentation.
On presentation, mean heart rate was 127 ± 19 bpm, median respiratory rate was 34 bpm (20–68), and mean rectal temperature was 38.6 ± 0.5°C. Eighteen dogs had mucous membrane pallor and 10 dogs were clinically icteric. Eight dogs had audible systolic cardiac murmurs, all grade ≤3/6. All dogs had a positive in-saline agglutination test. A blood type was established in 19 dogs, 10 were dog erythrocyte antigen (DEA) 1.1 positive, 9 were DEA 1.1 negative. Initial clinicopathologic data are summarized in Table 1. Eight out of the 20 dogs had systemic inflammatory response syndrome.20
Table 1. Summary hospital admission clinicopathologic data from 20 dogs with primary IMHA (median [range]).
n = 19, due to missing data on 1 nonsurvivor. No statistical significant differences were found between survivors and nonsurvivors for analytes previously suggested as prognosticators; albumin, bilirubin, and BUN (P > .05, Mann-Whitney U-test).
All dogs underwent diagnostic evaluation to exclude a primary cause of IMHA, including abdominal ultrasonography (n = 20), thoracic radiography (n = 19), urine culture (n = 18), urinalysis (n = 17), and polymerase chain reaction screening for tick-borne disease (n = 4). No underlying causes were found.
All 20 dogs received corticosteroids. Sixteen dogs received dexamethasone, median dose 0.3 mg/kg (0.25–0.5) q24h; 17 dogs received prednisolone, median dose 2 mg/kg (1.67–2.2) q12h. Seven dogs received azathioprine, median dose 2 mg/kg (1.44–2.2) q24h. Five dogs received mycophenolate mofetil, median dose 16.6 mg/kg (10–20 mg/kg) all but 1 administered q24h. One dog received cyclosporine 10 mg/kg q24h. Two dogs received human immunoglobulin, 1 received 0.49 g/kg, and another received 0.69 g/kg. Sixteen dogs received packed red blood cells, median total volume 25.3 mL/kg (6.6–46.7). Two dogs each received 450 mL of fresh whole blood and 2 dogs were administered a total of 500 mL of a hemoglobin-based oxygen carrying solution.k One dog received fresh frozen plasma during therapeutic plasma exchange. Nineteen dogs received aspirin, all 0.5 mg/kg q24h; 2 dogs received dalteparin both 150 U/kg q8h and 1 dog received clopidogrel 1.1 mg/kg q24h. Eleven dogs received maropitant, all 1 mg/kg and 8 dogs received omeprazole, all 1 mg/kg.
CRP, Clinicopathologic, and Cytokine Analyses
CRP, IL-2, IL-10, KC, and MCP-1 concentrations at admission were significantly increased in dogs with primary IMHA when compared with healthy controls (Table 2). Concentrations of IL-15, IL-18, GM-CSF, and MCP-1 were significantly higher in nonsurvivors versus survivors (Table 2 and Fig 1). No significant difference in BUN, bilirubin, and albumin concentrations or monocyte counts were observed between survivors and nonsurvivors (P= .55, .40, .92, and .66, respectively). Multiple logistic regression analysis revealed that concentrations measured at admission of IL-18 and MCP-1 (P= .0001 and .0015 [Wald's type 3], respectively) were independently predictive of outcome at 30-days. No correlation was observed between leukocyte count (monocyte, lymphocyte, or neutrophil count) and concentrations of IL-15, IL-18, GM-CSF, and MCP-1, respectively (P > .05, Spearman's correlation analysis). No relationship between duration of clinical signs and cytokine concentrations were identified (P > .3, Spearman's correlation analysis).
Table 2. Canine serum C-reactive protein (CRP) and cytokine concentrations in dogs with primary immune-mediated hemolytic anemia (IMHA) and healthy controls (median [range]).
Analytes significantly different in concentration between group of IMHA and healthy controls, and between survivors and nonsurvivors marked with superscript a and b, respectively (P < .05, Mann-Whitney U-test).
n = 4, due to missing data. Concentrations below assay detection limit were set to zero.
Sixteen dogs were discharged alive. Three dogs died while hospitalized, 1 due to uncontrollable hemolysis and an inability to identify cross-match compatible units, 1 due to secondary acute kidney injury, and 1 due to suspected pulmonary thromboembolism (PTE). Necropsy failed to confirm the clinical diagnosis of PTE but did identify hepatic sinusoidal thrombi, suggesting this dog died of thromboembolic complications. One dog was euthanized during initial hospital stay, due to acute deterioration. The dog did not undergo necropsy. Of the 16 dogs that were discharged from the hospital, 15 were alive at 30 days postadmission and 11 dogs were alive at 6 months postadmission, equivalent to a 55% 6-month survival rate. Of the 4 dogs that died between 30 days and 6 months, 2 dogs were euthanized, 1 due to systemic fungal disease and 1 with melena and nonregenerative anemia. Two dogs died suddenly at home after the 30-day time point. No premonitory signs were identified in 1 dog, while the other reportedly developed respiratory distress immediately before death. Necropsies were not performed in any of these 4 cases.
IMHA in dogs is a proinflammatory disease process involving loss of immunologic tolerance, autoantibody formation, and active removal of opsonized erythrocytes by cells of the mononuclear phagocytic system.6 Comparable to previous reports, dogs in this study had excessive proinflammatory activity with significantly increased concentrations of CRP and fibrinogen (Tables 1 and 2) and markedly increased peripheral leukocyte counts.14,21
The cytokine profiles of dogs with IMHA in this study were significantly different to those of healthy controls suggesting that multiple inflammatory cell types are involved in the pathogenesis of this disease. Concentrations of 4 cytokines were significantly higher in IMHA dogs compared with controls. IL-2 is a Th (CD4+) related cytokine with key immunoregulatory functions and both pro- and anti-inflammatory effects.22 IL-10 is the archetypical anti-inflammatory cytokine produced by several types of leukocytes which down-regulates inflammatory activity, although it may also have immunostimulatory activity.23,24 MCP-1 is a potent chemotactic factor expressed by vascular endothelium and mononuclear phagocytic cells, which regulates monocyte recruitment and trafficking.25–27 KC is a chemokine whose expression is induced by proinflammatory activity of monocytes and endothelium. KC is involved in chemotaxis and activation of inflammatory cells, thought to mediate some of the effects of MCP-1.28
Although the cytokine profiles of dogs with IMHA suggest contributions from multiple inflammatory cell types, only cytokines mainly related to monocyte/macrophage recruitment were predictive of outcome at 30 days postadmission. IL-15, IL-18, GM-CSF, and MCP-1 concentrations which were all increased in nonsurvivors compared with survivors are all mainly secreted by cells of the monocyte/macrophage lineage and principally target, recruit, and activate these cells.25–29 Furthermore, both IL-18 and MCP-1 were identified as independent prognostic factors which may indicate a critical role for the monocyte/macrophage cell line in determining outcome. Activated monocytes and macrophages are the main source of IL-18, which was first identified as a factor produced by IFN-γ stimulated macrophages. The effects of IL-18 on cells of both the innate and adaptive immune systems are dependent of the concurrent concentration of the endogenous antagonist IL-18-binding-protein induced in a negative feedback loop.28 To the authors' knowledge the role of IL-18 has not been studied in dogs with naturally occurring disease. Similarly, only limited data exist documenting MCP-1 concentrations in naturally occurring or experimentally induced disease in dogs.7,27 Further analysis of outcome at 6 months was not conducted because of the limitations of statistical power due to the reduction in case numbers at this time interval.
The possibility of ex vivo preanalytical release of these cytokines from leukocytes in the blood sample exists, however, a correlation between the observed cytokine concentrations and the concentration of the various leukocyte types would then be expected, which was not identified. Additionally, monocyte count itself was not found to be prognostic.
With the proinflammatory nature of primary IMHA in dogs, increased concentrations of key proinflammatory cytokines, such as IL-6 and IFN-γ, would be expected when compared with controls. This is particularly true of IL-6, which is the principal inducer of the CRP response. As such, the low concentrations of IL-4, IL-6, IP-10, and IFN-γ (Table 2) in the present study may suggest compromised analytical sensitivity rather than truly low concentrations. No obvious cause for this was detected, however. Internal quality control measurements of these cytokines were within acceptable ranges and careful sample handling (eg, no freeze-thaw cycles, sample analysis within 12 months) suggests that preanalytical variation should be limited.30 The canine-specific cytokine assay used has previously demonstrated acceptable analytical performance for these cytokines, although admittedly, in very high concentrations in an experimental setting.31 Alternative explanations include rapid, transient increases in IL-6 and IFN-γ that evaded detection; however, this is improbable during an active inflammatory disease process such as IMHA. We had expected a consistently high concentration of IL-6, in at least a fraction of the dogs, as is observed in human IMHA.32 Thus, despite a lack of obvious reasons for compromised analytical sensitivity, any conclusions on the apparent lack of prognostic ability of IL-4, IL-6, and IFN-γ in the present study should be made with caution. Further evaluation of IL-4, IL-6, and IFN-γ in dogs with IMHA is warranted given their important role in other autoimmune and inflammatory conditions.
Single protein concentration measurements in complex inflammatory conditions such as IMHA in dogs limit the information that can be obtained from each sample. The trend is to evaluate such systems with multiplex assays which can be run on very small volumes.33,34 In the present study, evaluation of the inflammatory status of dogs with IMHA was enabled by multiplex technology which allowed 13 cytokines to be measured using only 50 μL serum. Regardless of technique, the chance of identifying cytokine activity derived from cell populations critical to outcome is clearly dependent on the panel of cytokines evaluated. We were limited to those cytokines available in canine-specific kits for the multiplex methodology. The panel used included cyto- and chemokines representing several cell populations and both pro- and anti-inflammatory actions and thus represented a wide variety of analytes. However, studying additional panels of cyto-/chemokines in primary IMHA in dogs might reveal additional cytokine activities critical to outcome.
As in other fields, much attention has been focused on identification of prognostic biomarkers for IMHA in dogs. Analytes demonstrated to be prognostic in other studies include platelet, erythrocyte and band neutrophil counts, clotting times, fibrinogen degradation products, albumin, BUN, bilirubin,16–18 and thromboelastography derived parameters.35 However, the majority of these markers are downstream markers, which are indicative of the effect of the disease process. In the current study, noncytokine analytes (eg, serum albumin, bilirubin, and BUN) were not statistically different between survivors and nonsurvivors and did not have prognostic value. In contrast, the cytokines identified in this study suggest that excessive monocyte/macrophage activation can be prognostic and may be viewed as upstream markers of the pathophysiology itself rather than its effects. As such, this may suggest a target for future development of novel treatment strategies focusing on excessive monocyte/macrophage activity. Targeting of IL-18 and MCP-1 in in vivo models of human inflammatory diseases has proven beneficial.28,36 Since the spleen is the principal site of erythrophagocytosis by cells of the mononuclear phagocytic system in dogs with IMHA therapies which focus on this organ might be particularly beneficial to dogs with high concentrations of monocyte/macrophage derived cytokines. Two such therapies are already available. Splenectomy has been reported previously for IMHA in dogs37 and a recent small case series suggested early surgical intervention may be beneficial.38 Experimental work suggests that liposome encapsulated clodronate (dichloromethylene diphosphonate) which is phagocytosed by macrophages, inducing apoptosis thereby depleting macrophage numbers and blocking clearance of opsonized cells, is safe to use in the dog and that low doses are capable of blocking RBC clearance in naturally occurring IMHA in dogs.39 A prospective clinical trial of liposomal clodronate in dogs with IMHA is ongoing,40 but clearly, further work will be necessary to determine whether subsets of dogs with IMHA might particularly benefit such targeted interventions.
In conclusion, the present study confirms the inflammatory nature of primary IMHA in dogs but suggests that its pathophysiology is the result of a complex set of pro- and anti-inflammatory processes under the direction of multiple inflammatory and immune cell populations. We have identified for the 1st time that the activation and recruitment of the monocyte/macrophage cell line as indicated by IL-18 and MCP-1 concentrations is prognostic in primary IMHA in dogs, suggesting the potential for novel targeted therapeutic interventions to be developed in the future.
a Goggs R, Wiinberg B, Kjelgaard-Hansen M, Chan DL. Serial analysis of coagulation parameters in dogs with immune-mediated hemolytic anemia (IMHA) using thromboelastography (TEG). J Vet Intern Med 2010;24:680 (abstract)
b Kjelgaard-Hansen M, Goggs R, Wiinberg B, Chan DL. Increased concentrations of cytokines related to macrophage/monocyte activity and recruitment are risk factors for mortality in primary canine immune-mediated hemolytic anemia. J Vet Intern Med 2010;24:680 (abstract)
c Pediatric Tube, International Scientific Supplies, Bradford, UK
d Cell-Dyn 2500, Abbott Laboratories, Abbott Park, IL
e ILAB 600, Instrumentation Laboratory, Warrington, UK
f Canine C-reactive Protein, LifeDiagnostics, West Chester, PA
g CCYTO-90K, Millipore, Billerica, MA
h Luminex 200, Luminex Corporation, Austin, TX
i GraphPad Prism 5.01 for Windows, GraphPad Software, San Diego, CA