Corresponding author: Barbara Skelly, Department of Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge CB3 0ES, UK; E-mail: firstname.lastname@example.org
Despite being the most prevalent autoimmune disease of dogs, there is considerable variation between individuals and institutions in the treatment regimens that are employed for the management of immune-mediated hemolytic anemia. The aim of this review was to evaluate evidence relating to the treatment of the disease systematically and to use this evidence to draw conclusions that are applicable in wider veterinary practice. Search tools were employed to identify relevant articles and these were assessed according to stated criteria. The overall quality of published evidence was poor, with many articles failing to provide details of the enrollment, treatment, monitoring, and assessment stages of the study process. In view of this, firm conclusions cannot be drawn regarding the treatment of this disease and further research of a higher quality is required.
Immune-mediated hemolytic anemia (IMHA) is caused by an autoimmune response directed at endogenous antigens expressed on the surface of erythrocytes. Targeted erythrocytes are lysed intra- or extravascularly, resulting in severe acute anemia.
Treating animals with IMHA represents a considerable diagnostic and therapeutic challenge in veterinary practice and the disease is commonly associated with a mortality rate of 50–70%.[2-7] Although the presenting clinical signs of the disease are related to the development of anemia, it has become increasingly clear that other pathophysiologic processes, such as concurrent thromboembolic disease, contribute greatly to overall morbidity and mortality.[8-10]
Despite many previous endeavors, there is still great uncertainty as to the optimal immunosuppressive regimen that should be implemented when an animal is diagnosed with IMHA. Numerous immunosuppressive drugs have been employed and a variety of adjunctive therapies have also been trialed in the last two decades in an attempt to reduce the incidence of thromboembolism. In spite of this, clinical choices are still largely guided by experience, anecdote, and personal preference rather than by objective evidence.
The aim of this study was to systemically evaluate the current evidence concerning the use of immunosuppressive and antithrombotic drugs in the management of IMHA in dogs.
Materials and Methods
Identification of Studies
The review was conducted after those of Olivry and others and according to the principles of the QUORUM statement. The Medline, CABI abstract, and ISI Web of Knowledge databases were searched from 1980 using the criteria ‘(IMHA OR AIHA OR hemolytic anemia OR immune-mediated hemolytic anemia OR autoimmune hemolytic anemia) AND (dog OR canine)’. The search was conducted on January 17, 2012. The abstracts of each article identified were scanned by the primary author to identify those of relevance and the full text of these studies was obtained.
Selection of Studies
Studies were selected if they were presented as full reports in peer-reviewed journals and reported on the outcome of therapeutic regimens directed at the management of primary idiopathic IMHA in dogs. Primary idiopathic IMHA was defined as anemia (packed cell volume <35%) associated with the presence of antibodies directed at endogenous erythrocyte antigens (as indicated by in-saline agglutination, a titer of 1 : 16 or greater in a Coombs test or significant spherocytosis) for which no underlying cause could be detected through biochemical and hematologic analysis of blood samples or by imaging of the chest and abdomen or for which no underlying cause was subsequently detected at postmortem examination.
Trials were also required to present at least 1 outcome measure that related to the survival of individual treatment groups rather than assessing changes in clinical or biochemical parameters during the management of the disease. Studies were not considered if the report contained fewer than 5 cases and if survival data were not presented separately for each treatment group described.
Selected reports were reviewed independently by the 2 authors and assessed according to the criteria described below. Data were extracted in tabular form and, where differences were identified between the authors, these were resolved by consensus.
Assessment of Enrollment Criteria
Studies were assigned to 1 of 3 groups according to the description of the enrollment criteria for subjects:
Clearly defined and complete: Diagnosis based on the presence of in-saline agglutination or a significant positive titer in a Coombs test or the presence of significant spherocytosis. Biochemical and hematologic analysis of blood samples and imaging conducted in all cases.
Clearly defined, but incomplete: Diagnosis based on the presence of in-saline agglutination or a significant positive titer in a Coombs test or the presence of significant spherocytosis. Analysis of blood samples and imaging not conducted in every case or nature of additional tests performed not reported or unclear.
Unclear: Diagnosis not based on tests suggesting the presence of antibodies specific for endogenous erythrocyte antigens nor spherocytosis or diagnostic criteria unclear.
Assessment of Therapeutic Regimens
Studies were assigned to 1 of 3 groups according to the description of the therapeutic regimens that were implemented:
Clearly defined and complete: Name and dose range provided for all immunosuppressive and antithrombotic products administered to each animal. Duration of treatment recorded.
Clearly defined, but incomplete: Name and dose range provided for all immunosuppressive and antithrombotic products administered to each animal. Duration of treatment not recorded.
Unclear: Name and dosage not supplied for all immuno-suppressive and antithrombotic drugs administered to each animal.
Evaluation of Prospective Studies
Prospective studies were evaluated according to the following parameters:
Randomization: (method of generation and allocation to treatment groups)
Masking: (blinding of the observers and participants to the allocation of therapeutic regimens)
Intention to treat (ITT) analysis: (statistical analysis based on number of enrolled subjects rather than those remaining at the close of the study period).
Each study was assigned the grade ‘adequate’, ‘unclear’, or ‘inadequate’ for these parameters as described previously. The term ‘not applicable’ was applied for ITT analysis if all cases were accounted for.
Grading of Evidence Quality
Each study was assigned to an evidence grade after assessment of the criteria listed above and based on the design of the study. The evidence grades were as follows:
A (Blinded randomized controlled trial comparing 2 treatment interventions)
B (Controlled trial lacking either blinding or randomization)
C (Prospective cohort study)
D (Prospective case–control study)
E (Retrospective cohort or case–control study)
F (Prospective study with single intervention)
G (Retrospective case series with single intervention)
1. (>50 cases per group)
2. (21–50 cases per group)
3. (10–49 cases per group)
4. (<10 cases per group)
Assessment of Outcome Measures
The primary outcome variables assessed for each study were mean and median survival times and percentage mortality at stated time points after diagnosis. Additional outcome indicators included relapse rate and incidence of adverse events, including thromboembolic events.
Reporting of Results
Variation in the presentation of outcome data across studies precluded quantitative synthesis. Details concerning study design, patient enrollment, treatment, and survival measures were extracted into tabular format and major conclusions were described.
Evidence and Quality
Nineteen articles met the inclusion criteria for the review (Fig 1, Tables 1 and 2; Supplemental Tables 1 and 2), and these collectively reported data from 843 cases of IMHA. Of the studies, 4 (21%) were controlled trials comparing 2 or more interventions (evidence grade A or B), 10 (53%) were retrospective studies comparing 2 or more interventions (grade E), 2 (11%) were prospective studies that described the use of single interventions in groups of patients (grade F), and 3 (16%) were retrospective case series that described the use of single interventions (grade G). The years of publication ranged from 1981 to 2011 and the review included articles based on populations in the United States (10, 74%), UK (2, 11%), Netherlands (2, 11%), and Germany (1, 5%).
Table 1. Summary of articles describing 2 or more antithrombotic regimens
No significant difference between groups in survival. TE confirmed at necropsy of 3 dogs (13%)
TE events suspected or confirmed in 7 dogs (50%), of which 29% occurred in the IAD group and 71% in CD group. Percentage mortality at 6 months and incidence of TE significantly different between groups
Aspirin group had significantly lower mortality than the group that only received azathioprine, whereas heparin group had significantly higher mortality. Aspirin and heparin group had significantly lower mortality than heparin group.
No significant differences between groups. TE events confirmed at necropsy in 9 dogs (35%).
Table 2. Summary of article describing single antithrombotic regimen
Heparin (UFH) 300 IU/kg SC q6h, then adjusted after 40 hours according to an antifactor Xa assay
Number of cases
Length of treatment
Median for UFH: 14 days (range 1–23), not stated for immunosuppressive drugs
Prednisone 2 mg/kg PO q12h (n = 14) or dexamethasone 0.28 mg/kg IV q12h (n = 1).
Azathioprine (2 mg/kg PO q24h) (n = 3) or cyclophosphamide (50 mg/m2 IV q24h for 4 days, then 3 days off) (n = 1) if no increase in HCT by 14 days after start of treatment or 2 blood transfusions needed
Within 24 hours of diagnosis (n = 15), within 2–4 days (n = 3)
Median survival time (days)
Mean survival time (days)
17 at discharge, 28 at 1 month, 39 at 1 year
Other outcome measures
TE disease detected in 3 dogs (17%) at necropsy
Of the 4 controlled trials, 3 were assigned to the grade ‘adequate’ for randomization, but this process was considered to be ‘unclear’ in the remaining study as no details were provided. In 3 of the 4 trials, none of the enrolled subjects was lost to follow-up (rendering ITT analysis inapplicable), but an animal was censored in the 4th because its owners withdrew it from the trial and the study was therefore deemed ‘inadequate’ with regard to ITT. Masking was ‘adequate’ in 1 trial, remained ‘unclear’ in 1 trial, was not implemented in 1 trial, and was lifted during the final trial, and the latter two were therefore considered to be ‘inadequate’.
All of the studies reported some criteria for the inclusion of animals, but in 4 (21%) of these, the diagnosis of IMHA could have been based on the presence of anemia with clinical evidence of hemolysis and enrollment criteria were therefore assessed as ‘unclear’. In the remaining studies, the diagnosis was based on tests that detect the presence of antibodies directed at erythrocytes and, of these, 8 (42%) also described the use of diagnostic imaging in every case of IMHA to exclude underlying causes of the disease (graded as ‘clearly defined and complete'). The remainder stated that such procedures were conducted in some cases only or did not provide any information and were therefore graded as ‘clearly defined but incomplete’.
The use of immunosuppressive drugs in the management of IMHA was described in all of the studies with 5 (26%) presenting data primarily relating to the use of antithrombotic agents. In 9 (47%) of the studies, stated details regarding treatment regimens were graded as ‘clearly defined and complete’ while a further 8 (42%) were graded as ‘clearly defined but incomplete’ because they failed to state the duration of treatment. The treatment regimens of the remaining 2 (11%) studies were considered to be ‘unclear’ because the dose rates of key drugs were not stated.
Of the immunosuppressive drugs, corticosteroids were used in every animal with IMHA, although the product, formulation, and dose varied between reports. Across studies, the dose ranges of the 5 most widely used drugs were prednisone/prednisolone 1–8 mg/kg/day, dexamethasone 0.2–1 mg/kg/day, azathioprine 1–2.7 mg/kg/day, cyclosporine 3–8 mg/kg/day, and cyclophosphamide 1.1–3.3 mg/kg/day or 50 mg/m2/day for 4 days or an initial dose of 200 mg/m2 followed by 50 mg/m2/day for 3 days. Antithrombotic drugs were used in 58% studies according to set dose rates or according to the results of serial APTT or factor Xa assays.
Percentage mortality figures for each treatment group are shown in Tables 1 and 2 and Supplemental Tables 1 and 2. The reported prevalence of thromboembolic complications varied from 3 to 50% across studies, although these events were diagnosed both post and antemortem using a variety of different tests. Relapse rate was reported in 3 studies with a range of 6–13%, although the timing of relapses differed between reports.
The review shows that there are few pieces of high-quality evidence available to inform clinicians managing cases of IMHA in dogs. The absence of higher quality material likely relates to the relatively low incidence of the disease across the whole canine population and the concomitant difficulty in recruiting large numbers of cases to prospective trials based at single centers. Diseases with such dynamics are often amenable to retrospective analysis, but this facility comes at the price of uncertainty regarding enrollment criteria that are applied in hindsight, inconsistency in the precise nature of the intervention performed within and between treatment groups, and the risk that the conclusions will be out of date when they are produced.
In the context of IMHA, retrospective studies may be especially prone to bias if clinicians feel compelled to introduce more aggressive regimens in cases that are initially unresponsive or because some clinicians prefer to alter the treatment protocol in animals that start to show adverse effects when treated with high doses of glucocorticoids. Data obtained from retrospective analyses may therefore reflect the severity of the presentation and the preferences of the treating veterinarian rather than the effect of the intervention under investigation. At least 2 studies included in this review described an intervention that was used only in patients that had not responded to other forms of treatment or were not expected to do so,[16, 17] and several others described the introduction of additional immunosuppressive drugs in patients that had shown no response to an earlier protocol after a defined period.
While prospective studies occupy a higher level of evidence quality according to the accepted hierarchy of evidence-based medicine, the ability to draw representative conclusions from the RCTs assessed here was limited by concerns over study design and observation of multiple deviations from the CONSORT reporting template. In addition to variations in conducting and reporting randomization, blinding and ITT analysis, only one of the RCTs included a description of preliminary power calculations and concluded that the sample size obtained was adequate to detect a 50% difference in mortality between groups. It is likely that this and similar studies suffered from type 2 errors (ie, a genuine difference existed, but the study was unable to detect it) because of their small sample sizes.
Many studies used a follow-up period that is unlikely to have captured the majority of the anticipated data events (ie, deaths due to IMHA). While mortality data calculated between 1 and 6 months after diagnosis were often reported, several studies presented data only for the 1st week of treatment or until discharge, ie, periods that may not represent the true course of a disease that often requires treatment over 3–6 months.
All of the cases described in the review were presented to referral hospitals, which are likely to offer a higher standard of critical care than the majority of veterinary practices. Blood transfusions, plasma transfusions, and intensive monitoring were employed in a large proportion of cases, and the resultant mortality figures are unlikely to be transferable to centers where these procedures are unavailable or infrequently performed. Animals treated at tertiary institutions may also represent those cases with a lower likelihood of survival as it is more likely that animals will be referred if they do not respond to the treatment offered at a first-opinion center. The origin of the study population for this review also explains the overrepresentation of therapeutic products that are rarely encountered in veterinary general practice, particularly human gamma globulin and injectable cyclophosphamide.
Publication dates for the studies considered in this review span a period of 30 years, with 3 studies using historical control populations. It is likely that clinical management of cases will have changed considerably over this period, particularly in the provision of supportive care and this means that conclusions drawn from comparing populations widely separated in time may not be reliable.
Enrollment Criteria and Treatment Regimens
IMHA results from a type 2 (antibody-mediated) autoimmune response and the detection of antierythrocyte antibodies must underpin a reliable diagnosis. In several cases, stated inclusion criteria were such that a diagnosis could be based on evidence of hemolysis without a positive Coombs test, evidence of autoagglutination, or detection of spherocytes on a blood smear. Even where the diagnosis of IMHA was based on the detection of autoantibodies, there may have been variations in the exact method of the Coombs test used and in the dilution of saline and washing of erythrocytes when performing in-saline agglutination. This inconsistency in methods could be reflected in the characteristics of the populations recruited to different studies and this should be considered when directly comparing their results.
Detection of spherocytes was considered acceptable for the diagnosis of IMHA in anemic patients and, although a value of 5 spherocytes per high-power field has been considered to be supportive of a diagnosis of IMHA in previous reports,[10, 26, 37] it is not clear if this definition was used in many of the studies considered here. Spherocyte counts determined by microscopy are likely to vary with the severity of anemia, and published data indicate that counts can be very variable between cases with IMHA.[10, 20]
Very few studies gave a full account of the investigations that were performed to exclude underlying causes of the disease, and this affected the reliability of the stated diagnosis, of primary IMHA. The authors consider diagnostic imaging of the thorax and abdomen to be essential in excluding macroscopic neoplasia or other forms of systemic disease that can induce IMHA,[1, 21] but reports of these procedures were presented only infrequently. Several reports described the use of serologic or PCR-based tests to detect infectious agents that may cause secondary IMHA. These tests were not applied consistently within or between studies and this probably reflects variations in levels of clinical suspicion for individual patients and in geographic prevalence of relevant pathogens. Although all animals would ideally also undergo postmortem examination to eliminate underlying causes, the uptake for this procedure was low across all of the studies where it was offered and this is always likely to be the case in populations of client-owned animals.
There was some variability in the exclusion criteria applied, particularly with respect to platelet counts. In some cases, animals with low platelet counts at diagnosis were excluded, whereas most others included these animals or did not state specific criteria. IMHA with severe thrombocytopenia has traditionally been considered as a separate entity to “uncomplicated” IMHA because it has been associated with a poorer outcome in several reports.[4, 6, 22, 23] More recent evidence suggests that this subset may have percentage mortality figures similar to cases of IMHA when animals with evidence of consumptive coagulopathy (such as disseminated intravascular coagulation, DIC) are excluded, although there are other factors which could explain this difference. This evidence presented by Orcutt and colleagues highlights the fact that there is no reliable way to determine whether thrombocytopenia results from an immune-mediated process, bystander killing or the onset of DIC and the authors do not feel that the term ‘Evan's syndrome’ is applicable in veterinary medicine until the various causes of thrombocytopenia can be separated reliably.
This area of discussion also highlights a point made recently by Piek regarding the heterogeneity of the population of animals presenting with IMHA with regard to clinical and biochemical parameters that can act as prognostic factors.[2, 4-6, 26] Conclusions drawn from composite samples may be biased by the presence of particularly mild or severe cases within the analysis or may not be relevant to the whole population if one type of case predominates. Studies of the type conducted by Goggs and others and Orcutt and others will therefore be useful to investigate subsets of the IMHA population and scoring systems, such as that described by Ishihara and others may be used from the outset in prospective studies in future to categorize cases according to severity.
Several studies failed to provide important information about the treatment protocols employed with only 47% reporting the duration over which immunosuppressive treatment was administered and only one giving a clear account of how immunosuppressive doses were tapered as animals entered remission. In several cases, a “typical” dose rate was reported for immunosuppressive drugs without an average value or range and dose rates for adjunctive treatment were frequently withheld. While major treatment groups were often defined in the retrospective studies, small numbers of animals in these groups had often received other drugs before or during the treatment period under examination and this is a possible source of confounding in the outcome measures presented.
Outcome or endpoint data took such widely different forms among reports that it is difficult to compare the results of studies that had ostensibly similar designs. The most prevalent outcome measure was percentage mortality, but values for this parameter were frequently presented without any indication as to the time point after recruitment when they were calculated. As the uptake of postmortem examination was so low, mortality figures presented often refer to all-cause mortality, which is likely to overestimate the mortality because of IMHA in studies without a control or comparison group.
Relapse is a constant threat with IMHA, but the rate of this occurrence was reported infrequently across the studies examined here, perhaps in part attributable to uncertainty as to how this is defined. Similarly, the protocol for dealing with relapses was reported in only 1 study. It is likely that relapses greatly influence the overall success of individual cases, but to the authors knowledge, this association has not been examined.
Several studies that examined the use of antithrombotic drugs suffered from the lack of a reliable indicator of thrombotic status in animals that survived the event and it is likely that future studies will make increasing use of thromboelastography and factor Xa assays to assess this.
The evidence considered here suggests that use of glucocorticoids alone will result in a successful outcome in a high proportion of cases. A wide range of glucocorticoid dosages and regimens is described, but the effect of dosage on outcome is difficult to assess across studies without complete information regarding duration of treatment, method of tapering, and use of additional therapeutic products.
There is no consensus on the duration of time over which animals should receive treatment, on the way in which the dose should be tapered over time, or on the appropriate response to relapses. Only the study produced by Piek and others gave a full account of a 3-month tapering treatment protocol with a return to the starting dose in animals that suffered relapses.
The incidence of adverse effects associated with the use of glucocorticoids and the impact of these on patient quality of life, client satisfaction, and decisions regarding euthanasia has received relatively little attention so far. In the authors experience, doses of prednisolone above 2 mg/kg BID are likely to result in unacceptable adverse effects without apparent improvements in short- or long-term outcome. In nonresponsive cases or in situations where the clinician would prefer to use a lower corticosteroid dose, a 2nd drug may be introduced.
“Second Line” Treatment
Azathioprine and cyclosporine are widely used as “second line” drugs and these can be introduced immediately or added later. Both drugs cause additional immunosuppression, but their major benefit may lie in the fact that the glucocorticoid dose can often be reduced when these drugs are used concurrently.
The use of cyclosporine has been reported infrequently in the management of IMHA and these data are largely of poor quality. One retrospective study suggested that it may be associated with a poorer outcome than prednisolone alone or prednisolone with azathioprine, but there were insufficient cases to prove this association. An incomplete report of a small prospective trial indicated that the use of cyclosporine may exert no beneficial effect over the use of glucocorticoids alone,1 but this was not presented as a complete article and was therefore not considered in this review.
Multiple studies (including 13 of those considered here) have described the use of azathioprine in cases with IMHA. Three retrospective reports suggested that it may be associated with a more favorable outcome,[2, 26, 28] but comparison groups differed widely between studies and the authors do not consider their conclusions to be directly comparable. A recent large retrospective study apparently showed that azathioprine conferred no benefit over the use of prednisolone alone, but the 2 populations were separated in time and differed significantly according to platelet count, a parameter which the same workers identified as a negative prognostic indicator.
There is, therefore, no published clinical evidence available to guide a choice between azathioprine, cyclosporine, or no “second line” drug in the treatment of IMHA.
There is some retrospective evidence to suggest that human gamma globulin may improve hematologic parameters in refractory cases,[16, 30] but these studies and a RCT showed that it exerted no benefit in terms of survival.
A small RCT showed that the use of cyclophosphamide with prednisolone conferred no benefit over prednisolone alone and 2 further retrospective studies suggested that the use of this drug resulted in an increased mortality compared with other interventions.[2, 3] Based on this evidence, the authors do not consider the use of cyclophosphamide to be justified when alternative products are available.
The use of individually adjusted doses of heparin and of low-dose aspirin was each associated with improved survival in cases of IMHA. The evidence for the use of heparin was of higher quality as it derived from a RCT, but this intervention is unlikely to be adopted widely because of the limited availability of the factor Xa assay, which was used as a monitoring tool. In contrast, although the use of aspirin may represent a cost-effective intervention, the evidence supporting its use was obtained from a retrospective study, which provided incomplete information regarding enrollment criteria and treatment regimens.
The use of the platelet inhibitor clopidogrel has been investigated in a small RCT, but it is likely that this was subject to a type 2 error and the authors believe that this area warrants further investigation with studies of a higher statistical power before treatment recommendations can be made.
Recommendations for Future Research
The review highlights the importance of prospective research of a high quality in guiding practical decisions in the management of cases of IMHA. Future research efforts should take care to apply robust and consistent enrollment criteria, which do not leave the diagnosis in doubt and to design and report consistent treatment protocols, which take account of long-term treatment, tapering, and responses to relapses. As discussed above, consideration should also be given to the heterogeneity of cases in terms of clinical presentation as it is likely that particularly mild or severe cases will have a good or poor prognosis regardless of the treatment regimen employed. The low incidence of the disease also emphasises the importance of multi-center co-operation to maximize the enrollment of suitable cases.
The quality of evidence available to guide clinical decisions in the treatment of IMHA is generally poor and further research of a higher quality will be required to investigate existing and novel therapeutic regimens for the disease.
The authors are grateful to Drs F. Schiborra, A. Kovacevic, and B. Gerber. The study was not supported by a grant. The study was not presented at a meeting.
Conflict of Interest: Authors disclose no conflict of interest.
Husbands B, Polzin D, Armstrong PJ, et al. Prednisone and cyclosporine versus prednisone alone for treatment of canine immune mediated hemolytic anemia (IMHA). Abstract presented at 22nd annual ACVIM forum, Minneapolis, MN 2004