• Open Access

Anti-Erythrocyte Antibodies and Disease Associations in Anemic and Nonanemic Dogs


Corresponding author: Dr Steven Dow, DVM, PhD, DACVIM, Department of Clinical Sciences, Colorado State University, Fort Collins, CO 80523; e-mail: sdow@colostate.edu.


Background: Flow cytometry has been used to detect anti-red blood cell (RBC) antibodies in dogs with immune-mediated hemolytic anemia (IMHA), but the prevalence of anti-RBC antibodies in anemic and nonanemic dogs with a variety of different diseases has not been assessed previously.

Hypothesis: We hypothesized that anti-RBC antibodies would be more common in anemic dogs and in dogs with immune-mediated disorders and cancer.

Animals: Blood samples from 292 dogs were analyzed prospectively by flow cytometry for anti-RBC antibodies.

Methods: Blood samples from 147 anemic and 145 nonanemic dogs were evaluated by flow cytometry to detect surface-bound immunoglobulin (Ig) G and IgM antibodies on RBC. Disease associations with RBC antibodies were determined, as was the correlation between disease status and the percentage of Ig+ RBC. The specificity and sensitivity of flow cytometry and clinical variables for the diagnosis of IMHA were compared by Bayesian analysis.

Results: Anemic dogs were significantly more likely to be positive for anti-RBC antibodies (IgG, IgM, or both) than nonanemic dogs. Anemic dogs also had significantly higher percentages of Ig+ RBC than nonanemic dogs, whereas dogs with IMHA had significantly higher percentages of Ig+ RBC than dogs with all other diseases. Dogs with IMHA, infectious diseases, and immune-mediated thrombocytopenia were significantly more likely to have anti-RBC antibodies than dogs with other medical or surgical diseases.

Conclusions: Anemic dogs with immune-mediated diseases and infectious diseases were at the highest risk for the development of anti-RBC antibodies, and flow cytometry for the detection of IgG on RBC was highly sensitive and specific for the diagnosis of IMHA.

Immune-mediated hemolytic anemia (IMHA) is the most common disease associated with production of antired blood cell (RBC) antibodies in dogs.1–4 Most cases of IMHA in dogs develop spontaneously, although the development of IMHA and anti-RBC antibodies has also been reported after vaccination, after treatment with certain antibiotics, and after certain infections.5–7 In some dogs with IMHA, the target antigens for RBC autoantibodies have been identified, although in most dogs with IMHA the antigens are unknown.8 Previous studies have found that both immunoglobulin (Ig) G and IgM antibodies can be detected on the surface of RBC.7,9–13 The primary consequence of antibody binding to the RBC surface is premature removal of RBC from circulation by macrophages in the spleen and liver.2–4

Currently, the direct antiglobulin test (DAT; Coomb's test) is the primary test used clinically to screen for the presence of anti-RBC antibodies. However, the DAT test is an indirect assay, is relatively insensitive, and at best provides only semiquantitative information on the degree of RBC antibody binding.12,13 Flow cytometry can also be used to detect anti-RBC antibodies and is considered a more sensitive and quantitative assay for the detection of anti-RBC antibodies than the Coombs test.11–13 The use of flow cytometry to detect anti-RBC antibodies was reported previously in 2 studies of a small numbers of dogs with IMHA.12,13 The majority of dogs in those studies had IgG antibodies, whereas a few animals had only IgM antibodies. In a more recent study, flow cytometry was also compared with the DAT test, using 2 different secondary antibody reagents.11 In that study of anemic dogs, flow cytometry detected RBC antibodies in more than 50% of patients that were negative by the DAT test.

Antibodies of the IgG class are thought to be the most pathogenic autoantibodies and can result in the rapid clearance of RBC by splenic and hepatic macrophages.14 Dogs with IMHA often experience dramatic and rapid RBC destruction, and the disease is associated with high case fatality rates, ranging from 50 to 70%.3,4,15,16 At present, the pathogenesis of IMHA in dogs is poorly understood. In addition, little is known regarding the prevalence of RBC autoantibodies in dogs in general, including nonanemic as well as anemic dogs and dogs with nonimmune disorders.

Regarding the development of new diagnostic tests, an appropriate evaluation of diagnostic tests is critical for understanding how these tests should be applied to the clinical setting and to aid in appropriate interpretation of test results. In the past, the most common method for evaluating diagnostic tests was to identify and select 1 “best” diagnostic method to serve as the reference test, sometimes called the “gold standard” test, and to then compare results from the new test under investigation to this reference test.17 This criterion-based method of evaluating tests assumes that the reference test is perfect and that there is therefore no misclassification of the results. However, this assumption inherently creates bias in the resulting characterizations because there is always some degree of misclassification with any diagnostic testing. Therefore, alternative statistical methods have become available recently, which allow characterization of diagnostic tests without assuming that 1 test serves as a “gold standard” reference.18,19 In addition to providing theoretically unbiased estimates of test sensitivity and specificity, these approaches also have the advantage of providing unbiased estimates of disease prevalence in study populations.

Therefore, we conducted a study by flow cytometry to determine the overall prevalence of Ig+ RBC in healthy and sick dogs and to determine disease associations. In addition, we used advanced statistical analyses to evaluate the sensitivity and specificity of flow cytometry for the diagnosis of IMHA compared with diagnosis of IMHA based on clinical criteria.

Materials and Methods

Study Overview

Dogs with anemia (n = 145) and without anemia (n = 147) were enrolled prospectively in a study designed to estimate the prevalence of antibody binding among patients with a variety of different conditions. Blood samples from dogs were evaluated by flow cytometry to determine the percentage of RBCs with surface-bound antibodies of the IgG or IgM subclasses. This information was also used in the evaluation of the sensitivity and specificity of flow cytometry for diagnosis of IMHA. These studies were approved by the Institutional Animal Care and Use Committee at Colorado State University.

Patient Population and Disease Categories

Dogs were enrolled in the study from dogs treated at the Colorado State University Veterinary Teaching Hospital over a 12-month period based only on knowledge of the animal's hematocrit, without prior knowledge of the disease status. Anemic and nonanemic dogs were enrolled in an approximately 1:1 proportion. After enrollment, patients were classified into 7 different categories based on the diagnoses made at the time of sample collection (IMHA, thrombocytopenia, cancer, noninfectious medical conditions, infectious diseases, surgical or other conditions, and dogs with no abnormal findings), as noted in Table 1. Independent of this first set of classifications, dogs were also classified as having IMHA, thrombocytopenia, being nonanemic and clinically healthy, anemic and clinically ill, or anemic (without IMHA) but clinically healthy (Table 2). For this study, the diagnosis of IMHA was based on the presence of 3 criteria: anemia, evidence of regeneration (reticulocytosis), and evidence of RBC destruction (spherocytosis, hemoglobinemia, and/or hemoglobinuria). Anemia was defined as hematocrit <43% and reticulocytosis was defined as >60,000 reticulocytes/μL blood.

Table 1.   Clinical diagnosis for 292 dogs evaluated for RBC immunoglobulin (Ig) by flow cytometry.
Disease categoryAnemicNonanemicTotal
  1. IMHA, immune-mediated hemolytic anemia.

Surgical and miscellaneous19625
Table 2.   Classification of anemic and nonanemic dogs based on immunoglobulin (Ig)-positive or -negative red blood cells (RBCs).
Disease categoryIgG+IgM+IgG+/IgM+IgTotal
  1. IMHA, immune-mediated hemolytic anemia.


Flow Cytometry

Blood samples were collected in ethylene diamine tetraacetic acid (EDTA) tubesa and stored at 4°C for <24 hours before analysis. Flow cytometric assessment of anti-RBC antibodies was carried out with commercially available fluorescein isothiocyanate (FITC)-conjugated goat anti-dog IgGb and goat anti-dog IgMc reagents. These reagents were titrated before the study to determine a concentration that resulted in <1% binding to RBC from clinically normal dogs (data not shown). Additional negative controls were evaluated at the beginning of the study to assure specificity of secondary antibody binding. These included incubation of dog RBC with an irrelevant FITC-conjugated goat antibody to c-mycd (data not shown). In addition, EDTA blood samples from dogs known to have positive Ig binding to RBC were stored at 4°C and evaluated at multiple time points over a 24-hour period to ensure that storage conditions did not decrease antibody bound to the surface of dog RBC (data not shown).

An aliquot of RBC from study patients was placed in a well of a 96-well, round bottom platee and washed twice in FACS buffer (phosphate-buffered saline plus 2% fetal bovine serum plus 0.1% sodium azide), then diluted to a final concentration of 1 × 107 RBC/mL in 100 μL FACS buffer with anti-IgG or anti-IgM reagents. Erythrocytes were incubated with appropriate dilutions of the anti-IgG and -IgM reagents for 30 minutes at 4 °C, then washed twice and resuspended in FACS buffer before analysis. RBC fluorescence was assessed with a Cyan ADP flow cytometer.f Red blood cells were analyzed by gating on typical forward- and side-scatter characteristics, with exclusion of white blood cells, as reported previously.13 For each blood sample, an unstained sample of RBC was analyzed together with an immunostained sample of RBC. Analysis gates for determining the percentage of RBC with surface-bound Ig were set based on the unstained control sample for each patient.


A complete blood count was also performed on all samples. Hematologic analyses included reticulocyte counts, which were performed with an automated analyzerg as well as manual analysis of RBC morphology. Anemia was defined as hematocrit <43% and reticulocytosis was defined as >60,000 reticulocytes/μL blood.

Data and Statistical Analyses

Using the percentage of Ig-positive RBC estimated from flow cytometry, samples were classified as being positive or negative for Ig-binding of RBCs using a cut-off threshold of 5% (>5% versus ≤5%). Classifications for IgG and IgM binding were made independently. This cut-off value was chosen because it was >2 standard deviations higher than the mean anti-Ig staining percentage among samples collected from clinically healthy control dogs. Diagnosis classifications for study subjects were made by information gathered from reviewing patient records. Differences between diagnosis categories in the percent of Ig-bound RBCs were analyzed by the Kruskal–Wallis test for overall differences, and then pair-wise comparisons were made by the Wilcoxon ranked-sum test. IgM- and IgG-binding status (positive versus negative) was compared with anemia status (anemic versus nonanemic) and disease classifications by logistic regression.h Odds ratios (OR) and 95% confidence intervals (95% CI) were estimated from these models. A critical α of 0.05 was used for all statistical analyses.

Bayesian analysis methods18 were used to characterize the sensitivity and specificity of IMHA diagnoses made by traditional methods (clinical diagnosis), diagnoses made by IgG- and IgM-binding (positive versus negative), and diagnoses made by IgG and IgM results interpreted in parallel (if either test was positive then the dog was classified as having IMHA). Briefly, frequency cross-tabulations of results for the 4 tests (clinical diagnosis, IgG only, IgM only, and IgG and IgM) were prepared for anemic and nonanemic dogs. These data were included in the Markov Chain Monte Carlo simulation modeling by Gibbs sampling.18,19

Two models were used to investigate performance of these tests. In the first model, clinical diagnosis, flow cytometry for IgG, and flow cytometry for IgM were evaluated simultaneously, and in the second model, clinical diagnosis and the parallel interpretation for IgG or IgM were evaluated simultaneously. Prior assumptions about test sensitivities and specificities and disease prevalence were obtained from expert opinion of the senior investigator (S.D.), and β distributions regarding these assumptions were obtained using available software (BetaBuster 1.0 software, available free at http://www.epi.ucdavis.edu/diagnostictests). Specifically, assumptions used in this Bayesian modeling were: that the most likely true prevalence of IMHA in anemic dogs was assumed to be 25% with 95% confidence that it was >10%; the most likely true prevalence value for IMHA among nonanemic dogs was assumed to be 2% with 95% confidence that it was <5%; the modal sensitivity for traditional diagnosis was assumed to be 90% with 99% confidence that it was >75%; the modal specificity for traditional diagnosis was assumed to be 90% with 95% confidence that it was >75%; the modal sensitivity for flow cytometry detection of IgG was assumed to be 90% with 99% confidence that it was >80%; the most likely value for specificity of the IgG assay was assumed to be 85% with 99% confidence that it was >0.80; the most likely value for sensitivity of the IgM assay was assumed to be 10% with 99% confidence that it was <20%; the modal specificity for the flow cytometry detection of IgM was assumed to be 95% with 99% confidence that it was >90%.

Bayesian models were created for 2 populations assuming that IgG and IgM flow cytometry assays were conditionally dependent whereas clinical diagnosis was conditionally independent (WinBUGS 1.4.2, freely available at: http://www.mrc-bsu.cam.ac.uk/bugs/welcome.shtml). Covariance variables for test dependency were modeled by a uniform distribution. The code for these models was adapted from previously published information (http://www.epi.ucdavis.edu/diagnostictests/AB3tests1popn.html). The mean and 95% probability intervals (95% PI) were estimated from posterior distributions for the test sensitivities and specificities, along with estimates for the true prevalence of IMHA in anemic and nonanemic dogs. For each analysis, an initial burn-in of 10,000 iterations was discarded and node estimates were based on the subsequent 100,000 iterations. Convergence for each model was assessed by simultaneously running 5 chains with widely differing starting values. A sensitivity analysis was run for each model by changing the prior β distributions for all test variables to ensure that the results were repeatable. Positive and negative predictive values were calculated across a range of true prevalence values and then plotted.17


Detection of RBC Antibodies by Flow Cytometry

A typical histogram indicates flow cytometric analysis of IgG bound to RBCs obtained from an IMHA patient and a healthy patient (Fig 1). Strong binding of IgG to the RBC surface was detected in the IMHA patient (solid fill), with negligible binding noted in the healthy control patient (cross-hatched). Analysis of IgM binding to RBC gave very similar flow cytometric results (data not shown).

Figure 1.

 Flow cytometric detection of surface immunoglobulin (Ig) G in a dog with immune-mediated hemolytic anemia (IMHA) and in a healthy control dog. Red blood cells from a healthy control dog (cross-hatched histogram) and a dog with IMHA (solid filled histogram) were immunostained with fluorescein isothiocyanate (FITC)-conjugated anti-goat anti-dog IgG reagent, as described in Materials and Methods, and then analyzed by flow cytometry.

Patient Population Characteristics

Blood samples from 292 patients (147 anemic, 145 nonanemic) were analyzed in the study (Table 1). This included samples from 22 dogs diagnosed clinically with IMHA, 17 dogs with thrombocytopenia, 11 dogs with infectious diseases, 28 dogs with noninfectious medical disorders (excluding IMHA, thrombocytopenia, and cancer), 157 dogs with various types of cancer, 25 dogs with surgical or other miscellaneous diseases, and 32 clinically normal dogs.

RBC Antibodies in Nonanemic Dogs

Dogs were also reclassified according to their anemic versus nonanemic status and their clinical status (Table 2). Thus, these dogs were also classified as having IMHA, thrombocytopenia, being nonanemic and clinically healthy, anemic (without IMHA) and clinically ill, or anemic (without IMHA) but clinically healthy. Among the 145 nonanemic dogs, there were 12 dogs (8.3%) with RBC antibodies, and positive test results appeared to be more common in dogs with infectious diseases and cancer. Six dogs had IgG antibodies, 4 dogs had IgM antibodies, and 2 dogs had both IgG and IgM antibodies. Cancer was the most common diagnosis for nonanemic dogs with RBC antibodies (5 of 81 dogs), followed by infectious disease (3 of 5 dogs). Four of the 8 RBC antibody-positive dogs with cancer had lymphoma, although it should also be noted that lymphoma was overrepresented among the cancer study population (36% of the total cancer cases). Two of the dogs in the infectious disease category, which were positive for IgM antibodies, had chronic ehrlichiosis, as determined by serology.

Anti-RBC Antibodies in Anemic Dogs

Among the 147 anemic dogs included in this study, 26 (17.7%) had anti-RBC antibodies (Table 2). Within this population of 26 Ig-positive dogs, 8 dogs had IgG antibodies and 18 dogs had both IgG and IgM antibodies, whereas none had only IgM antibodies. Seventeen of 22 dogs (77%) with IMHA had anti-RBC antibodies, whereas 5 of 14 dogs (36%) with thrombocytopenia and 3 of 71 dogs (4%) with cancer had anti-RBC antibodies. None of the anemic dogs had antibodies to only IgM, whereas mixed antibody responses (IgG + IgM) overall were more common than IgG-only responses. All of the 26 anemic dogs that were positive for anti-RBC antibodies had evidence of regenerative anemia, as reflected by increased numbers of reticulocytes and macrocytosis (data not shown). In contrast, none of the 12 nonanemic dogs with anti-RBC antibodies had evidence of RBC regeneration (data not shown).

Antibody Binding and Disease Status

One of the advantages of flow cytometry over DAT for analysis of RBC Ig is that flow cytometry provides quantitative data on the percentage of RBC that are positive for surface-bound Ig. In contrast, DAT testing generally provides yes/no answers, or at best only semiquantitative estimates of Ig binding. Therefore, we evaluated the correlation between the percentage of Ig+ RBC and disease status. For the 38 dogs with IgG+ RBC, the mean percent IgG+ RBC was 27.9%, with a median of 15.2% positive (25th percentile = 10.3%, 75th percentile = 40.0%). For this analysis, dogs were classified as anemic and sick (111 dogs), nonanemic and well (32 dogs), nonanemic and sick (110 dogs), IMHA (22 dogs), or thrombocytopenia (17 dogs). Overall, there were statistically detectable differences in the percentage of RBCs with antibody binding between the groups (P < .001). Investigation of pair-wise comparisons indicated that anemic dogs had significantly greater proportions of antibody-bound erythrocytes than did nonanemic and well dogs (P < .0001), but there were no detectable differences between anemic and nonanemic sick dogs (P = .51). Nonanemic dogs that were sick also had significantly higher proportions of IgG-bound RBCs than did nonanemic dogs that were well (P < .0001). Dogs with IMHA had significantly higher RBC IgG binding than any of the other groups of dogs. Thus, analysis of the relative amount of surface-bound IgG, as reflected by the percentage of IgG+ RBC, indicated that this parameter was significantly associated with disease state. The highest RBC IgG levels were found in anemic animals and in animals with IMHA.

Comparison of RBC Antibody Status and Disease Status

Anemic dogs were about 2 times more likely to have Ig-positive RBC samples (either IgG- or IgM positive) than were nonanemic dogs (OR = 2.3, 95% CI = 1.12–4.78, P= .02). This association with anemia status was even stronger when IgG results were considered alone (OR = 3.6, 95% CI = 1.56–8.19, P= .002) and also when IgG and IgM results were interpreted in series (ie, samples were both IgG- and IgM-positive; OR = 9.7, 95% CI = 2.22–43.50, P= .0003). There were significant differences in the likelihood of dogs having Ig-positive RBCs (by parallel interpretation, ie, either IgG- or IgM-positive) among the different clinical disease categories of the study population (IMHA, thrombocytopenia, infectious diseases, cancer, other noninfectious medical diseases, surgical and other conditions, or clinically normal; Table 3). Compared with dogs with cancer, dogs with IMHA were 63 times more likely to have positive flow cytometry results (OR = 63.3; 95% CI = 20.0–238.5). Similarly, dogs with thrombocytopenia or infectious disease were both about 7 times more likely to have positive flow cytometry results (OR for idiopathic thrombocytopenic purpura [ITP] = 7.8, 95% CI = 2.08–27.28; OR for infectious diseases = 6.98, 95% CI = 1.34–30.04). However, there were no detectable differences in the likelihood of detecting Ig-binding between dogs with cancer and those with other medical conditions, those with surgical or other miscellaneous conditions, or those that were clinically normal (see Table 2).

Table 3.   Logistic regression analysis of the likelihood of immunoglobulin (Ig+) red blood cells (RBCs) based on disease category.
DiagnosisOR95% CIP-value
  1. Logistic regression analysis was carried out on a population of 292 dogs that had blood samples analyzed for the presence of anti-RBC antibodies, as described in Materials and Methods.

  2. OR, odds ratio; 95% CI, 95% confidence intervals; IMHA, immune-mediated hemolytic anemia.

Infectious disease7.01.34–30.04.01
Other noninfectious medical conditions1.20.18–5.26.79
Surgical and miscellaneous0.80.04–4.51.81
Clinically normal1.20.18–5.26.79

Bayesian Analysis of Flow Cytometry for Diagnosis of IMHA

Previous studies have established that flow cytometry is more sensitive than DAT for detection of RBC antibodies, especially in dogs with IMHA.11,12 However, the sensitivity and specificity of flow cytometry for the diagnosis of IMHA have not been determined previously in a large study. Estimates of the sensitivity and specificity obtained from Bayesian analyses suggested that flow cytometry detection of IgG bound to erythrocytes was superior to the other methods of testing for IMHA, including traditional diagnostic methods (Table 4). Among the diagnostic methods evaluated, IgM detection with flow cytometry had the lowest sensitivity and traditional diagnostic methods had the lowest specificity. The Bayesian estimate of true IMHA prevalence among anemic dogs was 29.7% (95% PI = 7.6–59.1%), and among nonanemic dogs was 4.1% (95% PI = 0.5–11.3%). At the estimated true prevalence of IMHA among anemic dogs in this study population (about 30%), the positive predictive value for IMHA diagnosis by detection of IgG–erythrocyte binding was 69% and the negative-predictive value was 94% (Fig 2). These estimates suggest that in a population with this true prevalence of disease (about 30%), a positive test result would accurately predict true IMHA status about 70% of the time, and a negative test result would accurately predict true IMHA status about 95% of the time.

Table 4.   Results of Bayesian analysis of the sensitivity and specificity of standard clinical criteria or flow cytometric analysis of red blood cell (RBC) immunoglobulin (Ig) for the diagnosis of immune-mediated hemolytic anemia (IMHA) in dogs.
Diagnostic testMean sensitivity
(95% PI)
Mean specificity
(95% PI)
  • a

    Regenerative anemia with evidence of RBC destruction.

  • b

    b Test results interpreted in parallel.

  • 95% PI, 95% prediction interval.

Traditional diagnosisa83.8% (60.2–97.7%)50.0% (29.6–70.2%)
Flow cytometic detection of IgG87.6% (74.6–96.3%)83.1% (69.7–96.3%)
Flow cytometic detection of IgM12.4% (3.7–25.2%)91.1% (76.9–98.9%)
Flow cytometic detection of either IgG or IgMb87.7% (78.3–94.7%)74.2% (56.9–88.2%)
Figure 2.

 Predictive values for immune-mediated hemolytic anemia (IMHA) diagnosis by flow cytometry to detect immunoglobulin G on the surface of red blood cells. The dashed vertical line represents the estimated true prevalence of IMHA in anemic dogs included in this study population.


In this large, prospective study, we assessed the prevalence and disease associations of anti-RBC antibodies in both healthy and sick dogs. Flow cytometry was used to screen blood samples because this technique is much more sensitive than DAT testing for the detection of anti-RBC antibodies.11,12 One important finding from the study was that the prevalence of anti-RBC antibodies in both anemic (17.7%) and nonanemic dogs (8.3%) was higher than determined previously. Anemic dogs were significantly more likely to have anti-RBC antibodies (including IgG, IgM, or both) than nonanemic dogs. In addition, the study found that the sensitivity and specificity of flow cytometry for the diagnosis of IMHA in dogs were greater than those of standard clinical diagnostic criteria, including regenerative anemia and spherocytosis.

These findings suggest that more widespread use of flow cytometric screening may be indicated in the workup of patients with unexplained anemia. This would be particularly true for dogs that this study identified as at high risk of developing anti-RBC antibodies, including anemic dogs with infectious diseases and ITP as well as dogs with suspected IMHA. Early identification of patients with Ig+ RBC would prompt additional diagnostic tests to identify a cause of RBC autoantibodies plus quicker medical intervention in the form of immunosuppressive drugs to stop ongoing RBC destruction.

Rigorous criteria were established for diagnosis of a positive RBC sample for this study. For example, we specified that >5% of RBC had to have surface-bound IgG or IgM in order to be considered positive. This cut-off was determined to be at least 2 standard deviations higher than background Ig positivity observed in healthy animals, based on analysis of blood samples from at least 20 healthy control dogs and research Beagles (data not shown). The 5% cut-off was established to exclude low-level Ig+ binding to RBC, which is normally present and represents a mechanism by which senescent RBC are normally targeted for elimination.20

The quantitative data generated by flow cytometric analysis of anti-RBC antibodies also allowed us to correlate the percentage of Ig+ RBC with the diagnosis and clinical status of the patients. For all animals with RBC antibodies (IgG, IgM, or both), those animals that were anemic had significantly higher percentages of RBC with surface-bound Ig than did nonanemic animals. This finding provides further support for the idea that RBC antibodies in anemic dogs are likely to play an important role in the pathogenesis of anemia. The pathogenic role of RBC antibodies is well established for dogs with IMHA, but RBC antibodies may also play a role in the anemia observed in dogs with ITP and infectious diseases, especially rickettsial diseases.

The ability to quantitate levels of surface-bound Ig could also be used to monitor the effectiveness of immunosuppressive therapy, particularly in dogs with IMHA. For example, we have observed rapid decreases in the percentage of Ig+ RBC after initiation of immunosuppressive therapy in dogs with IMHA, before changes in hematocrit or reticulocytosis became apparent (S.D.; unpublished data).

The clinical relevance of detecting RBC antibodies in nonanemic dogs is unclear at present. None of the 12 nonanemic dogs with RBC antibodies had evidence of RBC regeneration (data not shown). Thus, if RBC antibodies in these animals were responsible for increased RBC destruction, their overall effect on RBC turnover was relatively modest. However, it is also possible that over time even a modest but sustained increase in RBC destruction attributable to surface-bound Ig might lead to chronic anemia. It is also possible that the RBC antibodies detected in nonanemic dogs may represent antibodies directed against antigens expressed by senescent RBC.20–24 However, if this were the case, one would also expect a higher prevalence of anti-RBC antibodies in the healthy control animals. Because the cut-off for detection of positive RBC antibodies in the present study was set relatively high, it is likely that the threshold of antibody detection was sufficiently high to screen out naturally occurring RBC antibodies.24

Flow cytometry has been used previously to detect anti-RBC antibodies in dogs with IMHA.11–13 The consensus from these studies is that flow cytometry is more sensitive than DAT for detection of RBC antibodies, and equivalent to DAT in terms of specificity. This analysis confirmed the high sensitivity (88%) of screening samples for RBC IgG by flow cytometry. Using different criteria for defining Ig+ RBC samples, it was determined that the sensitivity of flow cytometry was 92% for diagnosis of IMHA in 1 study, whereas in another study the sensitivity of flow cytometry was 100%.12,13 It should be noted that the methods used for estimating the sensitivity and specificity in this study were different from those used in previous studies. Thus, the lower value for sensitivity that we obtained probably reflects the fact that this analysis accounted for the potential bias created by more traditional diagnostic methods for IMHA. In addition to diagnostic bias created by use of imperfect tests, this difference between apparent prevalence and estimated prevalence may be partially attributable to the severity of disease. In addition, in a previous study, the cut-off criteria for defining Ig+ RBC samples were lower than in the present study.13

In the present study, which included 22 dogs diagnosed with IMHA by traditional methods, RBC antibodies were detected in 77% of dogs with IMHA. Using conventional clinical diagnostic criteria, we found that the apparent prevalence of IMHA among anemic dogs was 15% (22 of 147). However, we also calculated that the theoretically unbiased estimate of true prevalence of IMHA among anemic dogs was 29.7%. This finding highlights the importance of using analytical methods for characterizing diagnostic tests, which can account for diagnostic biases inherent in imperfect testing methods. Of all the diagnostic methods evaluated for the diagnosis of IMHA in this study, we found that flow cytometric detection of IgG bound to RBCs had the highest combined sensitivity and specificity (see Table 4).

Once a test is run, clinicians are typically more interested in the predictive values for test results. Whereas sensitivity and specificity are fixed characteristics for a given test, predictive values vary dramatically in populations depending on the true prevalence of disease. Predictive values obtained by the true prevalence of IMHA estimated from this study population of anemic dogs indicated that there was excellent diagnostic value in using flow cytometry to detect IgG binding to RBC in anemic dogs. Clinicians evaluating dog populations with a different true prevalence of IMHA would obtain different predictive values (see Fig 2). However, the prevalence of IMHA seen in this population of anemic dogs should be similar to other patient populations from large referral hospitals as there was no attempt to subselect within anemic dogs at the time enrollment.

The fact that 4 of the 8 dogs with cancer and RBC antibodies had lymphoma is noteworthy, although lymphoma was also the most common cancer diagnosis (36% of the total) in this study. For example, a possible link between immune-mediated disease and lymphoma in dogs has been noted previously.25 An association between cancer (soft tissue sarcoma) and IMHA has also been reported previously in a case report.26 However, the overall prevalence of RBC antibodies in the 157 dogs with cancer in this study was actually low (5%), compared with the prevalence of RBC antibodies in other diseases, including thrombocytopenia (29%) and infectious disease (27%). Moreover, dogs with cancer and RBC antibodies had significantly lower percentages of Ig+ RBC (mean = 1.5% Ig+) than did dogs with immune diseases, including IMHA (mean = 30.8% Ig+) and thrombocytopenia (mean = 7.4% Ig+). In addition, IgM antibodies, which are generally considered to be less pathogenic, were present only in nonanemic dogs and not at all in anemic dogs. Thus, it appears unlikely that RBC antibodies play an important role in the pathogenesis of anemia in dogs with cancer, regardless of whether the dogs with cancer had regenerative or nonregenerative anemia.

The results of this study have several important implications for the clinical management of dogs with anemia. The greater sensitivity of flow cytometry, coupled with the greater availability today of relatively inexpensive flow cytometers, suggests that greater use of flow cytometric screening of dogs for the presence of anti-RBC antibodies may be warranted. Detection of RBC antibodies can be particularly useful in assessing patients with regenerative anemias and selecting appropriate courses of treatment. For example, regenerative anemia caused by RBC destruction secondary to cell membrane changes (eg, hypophosphatemia, Zn intoxication) or infection with certain RBC parasites (eg, mycoplasma infection) would generally not be associated with production of anti-RBC antibodies. Likewise, nonregenerative anemias such as aplastic anemia would also typically not be associated with anti-RBC antibodies. For example, none of the 5 dogs with aplastic anemia in the current study had RBC antibodies. In contrast, immune-mediated anemias would be associated with detection of high levels of anti-RBC antibodies. Therefore, flow cytometric screening may be considered an important tool in the management of anemia in dogs.


aVacutainer, Becton-Dickinson, Franklin Lakes, NJ

bBethyl Laboratories, Montgomery, TX

cBethyl Laboratories

dSanta Cruz Biotech, San Diego, CA

eCorning Life Sciences, Lowell, MA

fDakoCytomation, Fort Collins, CO

gAdvia Hematology System, Siemens Medical Solutions Diagnostics, Tarrytown, NY

hRroc Genmod, SAS v 9.2, Carey, NC


The authors wish to acknowledge the assistance of Ms Briana Harris with this study.

This study was supported by a grant from the Morris Animal Foundation.