Coombs’, haemoplasma and retrovirus testing in feline anaemia
Objective: To investigate the associations between Coombs’ testing, haemoplasma and retroviral infections, and feline anaemia.
Methods: Haematology, Coombs’ testing (including assessment of persistent autoagglutination) and selected infection testing (haemoplasma, feline leukaemia virus/feline immunodeficiency virus provirus) were performed in blood samples collected from 60 anaemic and 60 non-anaemic cats.
Results: No association between infection and anaemia or Coombs’ positivity existed. Anaemic cats (21.7%) were significantly more likely than non-anaemic cats (0%) to have cold autoagglutination (P<0.0001), but significance (set at ≤0.0025 due to multiple testing) was not quite reached when Coombs’ positivity was compared between anaemic (40.4% and 21.7% positive at 4°C and 37°C, respectively) and non-anaemic (20% and 3.3% positive, P=0.021 and P=0.004, at 4°C and 37°C, respectively) cats. Cats with immune-mediated haemolytic anaemia were significantly more likely to have persistent cold autoagglutination (P<0.0001) and be Coombs’ positive at 37°C with polyvalent (P<0.0001), immunoglobulin (Ig)G (P<0.0001) or any antiserum (P<0.0001). Haemoplasmas and retroviruses were uncommonly detected.
Clinical Significance: Cats suspected of having immune-mediated haemolytic anaemia should be evaluated for persistent autoagglutination at 4°C as well as performing Coombs’ testing at 37°C, but positive results may occur in with other forms of anaemia. Testing for erythrocyte-bound antibodies should always be interpreted in parallel with documentation of haemolysis in anaemic cats.
Anaemia is commonly encountered in feline practice and arises due to many different causes (Weingart and others 2004). One cause of anaemia is haemolysis following the binding of antibody and/or complement molecules to the surface of erythrocytes (Switzer and Jain 1981). In cats, such immune-mediated haemolytic anaemia (IMHA) is often secondary to an underlying cause such as haemoplasma infection, feline leukaemia virus (FeLV) infection, feline immunodeficiency virus (FIV) infection, feline infectious peritonitis (FIP), neoplasia (e.g. lymphoproliferative or myeloproliferative disorders), administration of drugs such as propylthiouracil or methimazole, systemic lupus erythematosus, glomerulonephritis, neonatal isoerythrolysis or incompatible blood transfusions (Scott and others 1973, Peterson and others 1984, Werner and Gorman 1984, Zulty and Kociba 1990, Day 1996, Person and others 1997, Bucheler 1999, Gunn-Moore and others 1999, Piek and others 2003, Weingart and others 2004, Lenard and others 2007, Tasker and others 2009). However anaemia in cats with infectious disease need not have an immune-mediated pathogenesis since cats with haemoplasmosis may be anaemic but not have autoagglutination or a positive Coombs’ test (Willi and others 2006a), whereas cats with FeLV infection may have aplastic anaemia, pure red cell aplasia or myelodysplastic syndrome (Hisasue and others 2000, 2001, Weiss 2006a,b). FIV-associated anaemia is less well understood, but anaemia is common in FIV-infected cats (Arjona and others 2000, Fujino and others 2009).
Primary IMHA, where no underlying aetiology is identified, was thought to be uncommon in the cat compared with the dog, but recent reports suggest otherwise (Husbands and others 2002, Kohn and others 2006). Diagnosis of IMHA is usually at least partly based upon the detection of agglutination due to the presence of erythrocyte-bound antibodies or complement in a direct Coombs’ test (Wardrop 2005), although false negative and false positive Coombs’ results have been reported in patients with IMHA (Honeckman and others 1996, Wardrop 2005, Morley and others 2008). Persistent autoagglutination of erythrocytes following washing in phosphate-buffered saline (PBS), which generally precludes completion of full Coombs’ testing, also indicates the presence of erythrocyte-bound antibodies (Warman and others 2008) and has been used as an additional diagnostic criterion for IMHA in both cats (Kohn and others 2006) and dogs (Klag and others 1993, Burgess and others 2000, Wilkerson and others 2000, Grundy and Barton 2001, Carr and others 2002, Mason and others 2003, Warman and others 2008). Persistent autoagglutination after washing is a more stringent form of the simple “in-saline agglutination test” that is widely used in practice to distinguish between true autoagglutination and rouleaux formation.
Inclusion criteria for the diagnosis of IMHA in published studies have varied but generally include multiple factors such as the presence of anaemia, positive Coombs’ test, autoagglutination, erythroid regeneration and/or evidence of haemolysis (Slappendel 1979, Klag and others 1993, Reimer and others 1999, Burgess and others 2000, Wilkerson and others 2000, Grundy and Barton 2001, Carr and others 2002, Mason and others 2003, Weinkle and others 2005, Overmann and others 2007, Warman and others 2008). Studies of feline IMHA are limited. In one investigation (Kohn and others 2006) either a positive Coombs’ test or persistent autoagglutination of erythrocytes was used to support a diagnosis of IMHA. All studies of canine and feline IMHA are hampered by lack of a “gold standard” method of diagnosis (Morley and others 2008).
Only one previous prospective study has specifically evaluated Coombs’ testing in cats (Dunn and others 1984) but the nature of the erythrocyte-bound antibodies in this study of 40 cats was only partly characterised since only limited Coombs’ reagents were utilised. A more recent study (Kohn and others 2006) performed Coombs’ testing using a panel of Coombs’ reagents but only a limited number of non-anaemic cats (14) were tested.
The aim of the present prospective study was to perform full Coombs’ testing and to evaluate persistent autoagglutination in a large group of anaemic and non-anaemic cats. Associations between anaemia, Coombs’ test positivity, persistent autoagglutination, retrovirus and haemoplasma infection status and underlying disease were investigated in 60 anaemic and 60 non-anaemic cats.
Materials and methods
Surplus ethylenediaminetetraacetic acid (EDTA) blood available from cats from which blood had been submitted for haematological examination was used. For each sample, haematological analysis was performed using a Cell Dyn 3700 analyser (Abbott, IL, USA). A Wright’s-stained blood smear was used for differential white blood cell count and to detect any changes in red blood cell (RBC) morphology, including the presence of spherocytes, and to confirm an automated platelet count as being low, normal or high. Blood samples from 60 anaemic (defined as a haemoglobin [Hb] concentration of < 8 g/dl; reference range 8 to 15 g/dl) and 60 non-anaemic (Hb ≥ 8 g/dl) cats were collected. For cats less than one year of age, anaemia was defined where Hb < 6 g/dl according to published reference ranges for kittens (Jacobs and others 2000).
Coombs’ testing was performed within 24 hours of collection of blood samples. EDTA blood was suspended in an excess volume (10×) of cold PBS (pH 7.4, 0.1M) and centrifuged (3500 rpm, 3 min). The supernatant and buffy coat were removed before resuspending the pelleted RBCs in PBS and repeating centrifugation. The RBCs were washed three times in cold (4°C) PBS. A 2.5% suspension of the packed, washed RBCs was made in PBS. Serial dilutions (from 1:5 to 1:10,240) in PBS of polyvalent feline Coombs’ reagent (ICN Flow, Basingstoke, UK), anti-feline immunoglobulin (Ig)G Fc, anti-feline IgM Fc and anti-feline IgA Fc antiserum (Nordic Laboratories, Tilberg, The Netherlands) were made (25 μl volumes) in two U-bottom 96 well plates. An anti-cat complement C3b antibody could not be sourced for this study, as the European distributors of the reagent used previously (Kohn and others 2006) no longer marketed this product. Four control wells with PBS alone were included as negative controls. To each well, 25 μl of the 2.5% RBC suspension was added and mixed by tapping the plate. The two plates were then incubated for 1 hour at either 4°C or 37°C. The plates were examined for the presence of agglutination and the titre, thermal activity and deduced antibody class were recorded. If agglutination was present in the PBS control wells, the Coombs’ test was deemed uninterpretable at that temperature, and these cases were defined as having persistent autoagglutination.
Signalment and clinical data
Copies of case records from each of the cats were obtained from the attending veterinary surgeon. Age, breed and sex were recorded in addition to the final diagnosis when this was determined with confidence. A diagnosis of IMHA required (1) the presence of anaemia based on haemoglobin concentration, (2) evidence of haemolysis (haemoglobinaemia or haemoglobinuria and/or bilirubinaemia in the absence of cholestasis), (3) lack of evidence of other causes of erythrocyte destruction and blood loss, and, when given, (4) evidence of response to immunosuppressive treatment. Where lack of information in case records precluded a final diagnosis, the diagnosis was classified as undetermined. Each case in which a final diagnosis was made was categorised into one of the following disease groups, similar to those described previously (Morley and others 2008): neoplasia, gastrointestinal disease, pancreatitis, infectious disease, non-neoplastic bone marrow diseases, IMHA, other miscellaneous medical diseases, and surgical diseases. IMHA was classified as primary if underlying causes (retrovirus or haemoplasma infection, other infections, neoplasia, previous blood transfusions, administration of medications known to be association with IMHA etc.) were rigorously ruled-out, as described previously (Kohn and others 2006). If an underlying cause was identified, IMHA was classified as secondary.
Haemoplasma and retrovirus testing
Remaining surplus EDTA blood (minimum 0.1 ml) was stored at −20°C. DNA was extracted from the samples using a commercially available kit (DNeasy kit, Qiagen, Crawley, UK) and subjected to polymerase chain reaction (PCR) assays for all three feline haemoplasma species: Mycoplasma haemofelis, “Candidatus Mycoplasma haemominutum” and “Candidatus Mycoplasma turicensis” as described previously (Peters and others 2008). Additionally, samples were subjected to PCR assays for FeLV and FIV provirus (Pinches and others 2006). All samples were also subjected to feline 28S rDNA PCR as an internal control (Pinches and others 2006) to confirm the presence of amplifiable DNA in samples. Appropriate positive (from known infected cats) and negative (from known non-infected cats) controls for each infectious agent were included, both in the DNA extraction process and PCR runs to screen for any extraction/PCR problems or contamination.
Data were entered into a database (Excel 2002, Microsoft Ltd, Reading, UK) and exported into statistics software (Statistical Package for Social Scientists, Woking, UK). Descriptive data pertaining to Coombs’ test results and diagnosis were evaluated for all 120 study cats. The chi-squared testing was used to compare categorical variables, with exact statistics used when necessary. Persistent autoagglutination, Coombs’ test results, haemoplasma infection, and FeLV and FIV provirus positivity were compared between anaemic and non-anaemic cats. Associations between Coombs’ reactivity at different temperatures and/or persistent autoagglutination were investigated. Disease category, haemoplasma infection, and FeLV and FIV provirus positivity were compared between cats that showed persistent autoagglutination and those that did not, and between Coombs’ positive and negative cats.
When analyses of Coombs’ test results at 4°C were performed, cats with persistent autoagglutination at 4°C were excluded from analysis as Coombs’ titres and reagent specificity could not be confirmed in these cases. Coombs’ test results were analysed by taking a positive result to be at any titre at 4°C and/or 37°C, at 4°C only and at 37°C only, each with any antiserum. Titres ≥ 160 were separately evaluated.
The Bonferroni correction was applied to account for multiple testing. Statistical significance was set at P≤0.0025 in the comparison of anaemic and non-anaemic cats, and P≤0.0009 in the comparison of cases with and without persistent autoagglutination at 4°C and those that were Coombs’ test positive and negative.
Disease categories, Coombs’, haemoplasma and retroviral provirus testing
A diagnosis was not determined in 14 of the 120 (11.7%) cases and 1 case was deemed to be clinically well at the time of sample collection. The number of cases in each of the disease categories and their distribution between the anaemic and non-anaemic groups is shown in Table 1.
Table 1. Disease categories classified according to anaemic status
|Miscellaneous medical diseases||42 of 120 (35%)||14 of 42 (33.3%)||28 of 42 (66.7%)|
|Infectious diseases||13 of 120 (10.8%)||9 of 13 (69.2%)||4 of 13 (30.8%)|
|Neoplasia||11 of 120 (9.2%)||8 of 11 (72.7%)||3 of 11 (27.3%)|
|IMHA||11 of 120 (9.2%)||11 of 11 (100%)||0 of 11 (0%)|
|Gastrointestinal diseases||8 of 120 (6.7%)||1 of 8 (12.5%)||7 of 8 (87.5%)|
|Pancreatitis||7 of 120 (5.8%)||1 of 7 (14.3%)||6 of 7 (85.7%)|
|Surgical diseases||7 of 120 (5.8%)||1 of 7 (14.3%)||6 of 7 (85.7%)|
|Non-neoplastic bone marrow diseases||6 of 120 (5%)||6 of 6 (100%)||0 of 6 (0%)|
|Diagnosis undetermined||14 of 120 (11.7%)||9 of 14 (64.3%)||5 of 14 (35.7%)|
|Clinically well||1 of 120 (0.8%)||0 of 1 (0%)||1 of 1 (100%)|
Of the 120 cats, 13 (10.8%) had persistent autoagglutination in PBS control wells at 4°C which precluded determination of Coombs’ test titres at 4°C since agglutination was present in all test wells. Autoagglutination in the PBS control wells at 37°C was not seen in any cat.
In Coombs’ testing with polyvalent antiserum at 4°C, 96 cats (80%) were negative and 11 cats (9.2%) had positive reactions with titres of 10 (n=4), 20 (n=3), 160 (n=2) and 320 (n=2). With IgG antiserum at 4°C, 103 cats (85.8%) were negative and 4 cats (3.34%) were positive with titres of 640 (n=2), 1280 (n=1) and 2560 (n=1). With IgM antiserum at 4°C, 79 cats (65.8%) were negative, whereas 28 cats (23.3%) were positive with titres of 10 (n=7), 20 (n=12), 40 (n=3), 80 (n=4), 320 (n=1) and 5120 (n=1). IgA antiserum was only incorporated into the Coombs’ test of 99 cats in which there was adequate sample volume. With IgA antiserum at 4°C, 84 cats (84.8%) were negative and 7 cats (7.07%) were positive with titres of 10 (n=4), 20 (n=2) and 40 (n=1). Eight cats (8.1%) could not have IgA titres determined due to agglutination in the PBS controls.
In Coombs’ testing with polyvalent antiserum at 37°C, 105 cats (87.5%) were negative and 15 cats (12.5%) were positive with titres of 5 (n=1), 10 (n=1), 20 (n=2), 40 (n=4), 80 (n=4), 160 (n=2) and 320 (n=1). With IgG antiserum at 37°C, 106 cats (88.3%) were negative and 14 cats (11.6%) were positive with titres of 20 (n=2), 80 (n=2), 160 (n=2), 320 (n=2), 640 (n=1) and 1280 (n=3). With IgM antiserum at 37°C, 119 cats (99.2%) were negative and only 1 cat (0.8%) was positive (titre 160). None of the 99 cats tested with IgA antiserum at 37°C was positive.
PCR testing for haemoplasmas and retroviral provirus was performed on all 120 samples. The internal control, 28S rDNA, was amplified from all 120 samples. Thirteen samples were positive for “Candidatus M. haemominutum” but none were positive for M. haemofelis or “Candidatus M. turicensis”. Two samples were positive for FIV provirus and eight samples positive for FeLV provirus.
Comparison of anaemic and non-anaemic cats
Comparison of anaemic and non-anaemic cats found no significant difference in breed (77.6% non-pedigree and 22.4% pedigree versus 58.6% non-pedigree and 41.4% pedigree), sex (56.1% males and 43.9% females versus 35.1% males and 64.9% females) or age (median [range] of 5 [0.3 to 17] years versus 6 [0.2 to 16] years] (full data not shown). The median [range] Hb value was 4.9 [1.5 to 7.8] g/dl in the anaemic group and 11.5 [8 to 15] g/dl in the non-anaemic group. A significant difference in the presence of persistent autoagglutination at 4°C was found between anaemic (13 of 60; 21.7% positive) and non-anaemic (0 of 60 positive) cats (P<0.0001). Table 2 presents the analyses of Coombs’ test result comparisons between anaemic and non-anaemic cats, when any agglutination at any titre was regarded as positive. Anaemic cats tended to be more likely to be Coombs’ positive at 37°C (P=0.004) and to have positive reactions with polyvalent antiserum at 37°C (P=0.004) but significance was not reached. Raising the threshold considered positive to a titre of ≥ 160 raised P values (data not shown) for each reaction pattern.
Table 2. Comparison of Coombs’ test positivity between anaemic and non-anaemic cats
|Any antiserum at either 4°C and/or 37°C||19 of 47 (40.4%)||12 of 60 (20%)||P=0.021|
|Any antiserum at 4°C only||19 of 47 (40.4%)||12 of 60 (20%)||P=0.021|
|Any antiserum at 37°C only||13 of 60 (21.7%)||2 of 60 (3.3%)||P=0.004|
|Polyvalent antiserum at 4°C only||9 of 47 (19.1%)||2 of 60 (3.3%)||P=0.010|
|IgG antiserum at 4°C only||2 of 47 (4.3%)||2 of 60 (3.3%)||P=0.100|
|IgM antiserum at 4°C only||17 of 47 (36.2%)||11 of 60 (18.3%)||P=0.037|
|IgA antiserum at 4°C only||4 of 32 (12.5%)||3 of 59 (5.1%)||P=0.236|
|Polyvalent antiserum at 37°C only||13 of 60 (21.7%)||2 of 60 (3.3%)||P=0.004|
|IgG antiserum at 37°C only||12 of 60 (20%)||2 of 60 (3.3%)||P=0.008|
|IgM antiserum at 37°C only ||1 of 60 (1.7%)||0 of 60 (0%)||P=0.100|
|IgA antiserum at 37°C only||0 of 40 (0%)||0 of 59 (0%)|| |
Of the 13 samples positive for “Candidatus M. haemominutum”, 10 were from non-anaemic cats and 3 were from anaemic animals, with no significant difference between the two groups (P=0.075). The real-time PCR threshold cycle values for the “Candidatus M. haemominutum” infected cats that were anaemic were not significantly different from those that were non-anaemic (mean±sd: 32.6±4.1 versus 27.9±4.6, data not shown). Two samples were positive for FIV provirus, one from an anaemic cat and one from a non-anaemic cat. Of the eight samples positive for FeLV provirus, two were from anaemic cats and six from non-anaemic cats, with no significant difference present between the two groups (P=0.272). The real-time PCR threshold cycle values for the FeLV provirus infected cats that were anaemic were not significantly different from those that were non-anaemic (25.7±10.8 versus 31.9±7.5, data not shown).
Further description of cats with a diagnosis of IMHA
Eleven cats were diagnosed with IMHA. The age (median [range]) of these animals was 2.5 (0.5 to 14) years (data from n=10). Seven cats (63.6%) were male (four neutered) and three (27.3%) were female (two neutered) with gender not recorded in one animal. Eight cats (72.7%) were non-pedigree, one (9.1%) was a British Shorthair and breed was unrecorded in one case. Nine cats had primary IMHA and two had secondary disease attributed to infection with “Candidatus M. haemominutum” and concurrent gastrointestinal neoplasia respectively. All cats with IMHA had positive tests at 4°C: nine showed persistent autoagglutination and two cases had positive Coombs’ reactions with polyvalent antiserum (titres of 20 or 160), IgM antiserum (titres of 20 or 40) and IgG antiserum (titre of 640, one cat only). Nine of the eleven cats with IMHA had positive Coombs’ test results at 37°C with positive reactions with polyvalent antiserum (titres of 20 [two cats], 40 [three cats], 80 [two cats] and 160 [two cats]), IgG antiserum (titres of 80 [two cats], 160 [two cats], 320 [three cats] and 1280 [one cat]) and IgM antiserum (titre of 160, one cat only).
Association between Coombs’ test positivity at different temperatures and persistent autoagglutination
Cats that showed persistent agglutination at 4°C were more likely to be Coombs’ positive (any antiserum at any titre) at 37°C (9 of 13; 69.2%) than negative (4 of 105; 3.8%) (P<0.0001). Omitting the cats that had persistent autoagglutination at 4°C, cats that were Coombs’ positive at 4°C were also more likely to be Coombs’ positive at 37°C (6 of 6; 100%) than negative (25 of 101; 24.8%) (P<0.0001). Significance was still reached when titres ≥ 160 were considered (data not shown).
Comparison of cats with and without persistent autoagglutination
A significant difference in the number of cats showing persistent autoagglutination at 4°C was found between the different disease categories (P<0.0001) (Table 3). Evaluation of the contribution of each of the disease categories to the chi-square statistic showed that the IMHA disease category was responsible for the significant difference (P<0.0001), with 9 of the 11 (81.8%) cats with IMHA having persistent autoagglutination at 4°C.
Table 3. Disease category comparisons showing significant differences in persistent autoagglutination and Coombs’ test reaction patterns
|Miscellaneous medical diseases||0||42||0||42||0||42||0||42|
|Non-neoplastic bone marrow diseases||0||6||0||6||0||6||0||6|
No association between persistent autoagglutination at 4°C and PCR positive status for “Candidatus M. haemominutum”, FeLV or FIV provirus was found (all P=1.000).
Comparison of cats with positive and negative Coombs’ test results
A significant difference in the number of cats between the different disease categories was found for the following Coombs’ test results at 37°C: positive with any antiserum at any titre (P<0.0001), positive with any antiserum with a titre of ≥ 160 (P<0.0001), positive with polyvalent antiserum at any titre (P<0.0001), positive with IgG antiserum at any titre (P<0.0001) and positive with IgG antiserum with a titre of greater than or equal to 160 (P<0.0001) (Table 3). Evaluation of the contribution of each of the disease categories to the chi-square statistic again showed that the IMHA disease category (Table 3) was responsible for the significant difference (P<0.0001) in each of these cases. A trend towards a significant difference (P=0.002) in the number of cats between the different disease categories with positive Coombs’ tests with polyvalent antiserum at 4°C at any titre was found but none of the other statistical comparisons of Coombs’ reactive patterns and disease categories gave significant P values (0.008 to 0.476, data not shown).
No association between Coombs’ test positivity, either at any titre or with a titre of ≥ 160, and PCR positive status for “Candidatus M. haemominutum”, FeLV or FIV provirus was found (P values ranging from 0.034 to 1.000).
The present study represents the first prospective description of full Coombs’ testing in a large group of anaemic and non-anaemic cats.
Only anaemic cats (13 of 60) showed persistent autoagglutination of erythrocytes at 4°C resulting in a significant association between the presence of anaemia and persistent cold autoagglutination. The detection of such autoagglutination is performed as a routine part of the Coombs’ test and is reported separately to reactions with specific antisera (Dunn and others 1984, Kohn and others 2006). Cold autoagglutination in these 13 cats precluded further reading of the Coombs’ test (for specific antisera) at that temperature, necessitating their exclusion from statistical analyses of Coombs’ test results at 4°C. This may have contributed to the failure to detect significant associations between anaemia and positive Coombs’ tests at 4°C only, and may also have impacted on the analysis of the number of cats with positive Coombs’ tests at 4°C and/or 37°C in the anaemic and non-anaemic groups, since cats with persistent cold autoagglutination were also excluded from this analysis despite some of these cases having a positive Coombs’ test at 37°C.
Cold autoagglutination is recognised to represent the presence of haemagglutinating IgM antibodies that elute from the surface of erythrocytes at higher temperatures, and this would appear to be the case in the 13 anaemic cats in the current study since parallel testing at 37°C failed to reveal persistent autoagglutination. Cold agglutinins, by definition, react more strongly at 4°C than at higher temperatures (Gehrs and Friedberg 2002). The clinical significance of such cold autoagglutination is widely debated since this in vitro phenomenon is not observed at physiological temperatures. Indeed, no cats in the current study showed signs of cold agglutinin disease (e.g. ear or tail tip necrosis) questioning their role in the pathogenesis of anaemia. Human beings with cold agglutinin syndrome have pathological cold agglutinins that are thermally active up to at least 30°C (Rosse and Adams 1980, Gehrs and Friedberg 2002, Petz 2008). Although not performed as part of this investigation, a more complete study of the thermal reactivity of feline cold agglutinins would be of interest. However, a recent study evaluating Coombs’ testing in M. haemofelis infected cats (Tasker and others 2009) reported that the cold agglutinins detected in those animals were not active at room temperature (18°C to 20°C), suggesting a narrow thermal activity range for those cold agglutinins. Rather than assigning a pathogenic role in the induction of anaemia to such cold agglutinins, an alternative hypothesis could be that they arise as a result of the development of anaemia, maybe due to changes in erythrocyte membranes resulting from erythrocyte damage or haemolysis.
Although not reaching significance due to the necessary adjustment to the critical P value to allow for multiple testing, anaemic cats also tended to have more positive Coombs’ tests with any antiserum and with polyvalent antiserum (both P=0.004) at 37°C. Such warm antibodies or complement could be involved in the pathogenesis of anaemia as they are active at physiological temperatures.
Despite these associations some non-anaemic cats also had a positive Coombs’ test. In the current study, 20% of non-anaemic cats were Coombs’ positive for any antiserum at 4°C; mostly (in 10 of the 12) due to low titred (≤80) IgM (±IgA) reactions, but two cats had high titred IgG (±polyvalent) reactions. Only two (3.3%) non-anaemic cats were Coombs’ positive at 37°C and these animals both had high titred reactions with polyvalent antiserum and anti-IgG. These were the same two cats with high titred reactions at 4°C, and both were diagnosed with pancreatitis (see below). The significance of the positive Coombs’ test or the presence of persistent autoagglutination in non-anaemic cats is not known. Low-titred cold agglutinins were described in 20% of healthy non-anaemic cats in one previous study (Dunn and others 1984). However, no evidence of cold or warm agglutinins in five healthy or nine sick non-anaemic cats was reported recently by Kohn and others (2006). Our discovery of a significant number of Coombs’ positive non-anaemic cats, including some with high titres, could be due to the larger number of non-anaemic cats that we were able to sample, or reflect the fact that all but one of our 60 non-anaemic cats were ill, in contrast to the other studies that included healthy cats. It should also be emphasised that the data presented here pertain to the reagents and the procedure used for Coombs’ testing in our laboratory and that variations in reagents and procedure across different laboratories mean that direct extrapolation of the results of our study to results generated by other laboratories may not be possible. Our Coombs’ test, for example, may have had greater sensitivity resulting in a larger number of (or higher titred) positive reactions in the non-anaemic cats.
Although no statistical association was demonstrable, it is to be noted that five of seven cats in the present study with pancreatitis had a positive Coombs’ test; three cats had low-titred cold agglutinins (one of which was anaemic with an Hb of 5.98 g/dl), and the other two were the only non-anaemic cats to have warm agglutinins (which were high-titred and present in association with high-titred cold agglutinins). An association between pancreatitis and a positive Coombs’ test has not previously been reported in cats, and this warrants further study although the animals with pancreatitis in the current study had no signs of haemolysis. Pancreatitis has been associated with haemolytic anaemia in man (Druml and others 1991) and in experimental rodent models (Saruc and others 2007), and has been described in Babesia-induced haemolysis in dogs (Mohr and others 2000).
The associations detected between anaemia and the pattern of Coombs’ test reactivity described above were found to be interrelated, in that cats with Coombs’ positivity at 37°C were more likely to show persistent autoagglutination at 4°C and to have a positive Coombs’ test at 4°C. Kohn and others (2006) reported that all Coombs’ positive cats in their study (which were all diagnosed with IMHA) were positive at both 4°C and 37°C. Although this could be interpreted as supportive evidence for the performance of Coombs’ testing at one temperature only, we found a number of cats to have reactions at 4°C only, including two cats with IMHA (one with persistent autoagglutination and one positive for IgM antiserum). Both of these cats had signs of haemolysis without evidence of cholestasis or blood loss, and both responded to prednisolone treatment. Of importance are the results of a recent study of 65 dogs with IMHA in which 10 cases had Coombs’ test reactivity at 4°C only (Warman and others 2008), suggesting that evaluation at 4°C can be important in cases of suspected IMHA.
Previous reports of feline IMHA have not consistently described the class and titre of erythrocyte-associated antibodies and statistical analyses have not been performed. One article reported the presence of IgG, IgM or both molecules on the surface of erythrocytes from 12 cats with IMHA (Werner and Gorman 1984), whereas another demonstrated IgG antibody in five cats with primary IMHA, although an anti-cat IgM reagent was not used (Person and others 1997). Kohn and others (2006) reported cats with IMHA to be mostly positive for IgG, or both IgG and IgM, with just 1 of 18 cases positive for IgM only. In our study cats with IMHA were more likely to have persistent cold autoagglutination and show positive Coombs’ test results at 37°C with any antiserum, with polyvalent antiserum, or with IgG antiserum alone, supporting a recommendation that cats suspected of having IMHA should be evaluated for persistent autoagglutination at 4°C as well as performing Coombs’ testing at 37°C. Ours is the first study to investigate the presence of RBC-bound IgA in cats. IgA antibody has been reported to be associated with erythrocytes in most dogs with IMHA using a sensitive ELISA-based assay (Barker and others 1993) and can also be found in some people with IMHA (Reusser and others 1987, Dubarry and others 1993, Bardill and others 2003). However, reactions with this antiserum were uncommon and none of the cats with IMHA were positive. This could be due to the Coombs’ test in the current study having poor sensitivity for the detection of IgA antibody or indicates that IgA is infrequently associated with IMHA in cats.
Piek and others (2003) evaluated various laboratory tests, including measurement of titres of IgG and IgM erythrocyte-bound antibody, for their ability to differentiate between idiopathic (primary) or secondary feline IMHA but none were found to be useful. Very few cases of IMHA were identified in the current study to determine whether any features of the Coombs’ test correlated with primary or secondary disease, as reported in dogs (Warman and others 2008).
One of the two cats with IMHA secondary to underlying disease had infection with “Candidatus M. haemominutum”. This organism is regarded as being poorly pathogenic (Foley and Pedersen 2001) and is found in 20.3% of ill cats in UK (Tasker and others 2003). This cat had both persistent cold autoagglutination and a positive Coombs’ test at 37°C with polyvalent and IgG antisera. Experimental infection studies of the pathogenic M. haemofelis have shown an association between this pattern of Coombs’ test reactivity (that is persistent cold autoagglutination and a warm-reactive IgG antibody), anaemia and parasitaemia as assessed by quantitative PCR, although such an association was not demonstrated with “Candidatus M. haemominutum” (Tasker and others 2009). Nevertheless, “Candidatus M. haemominutum” can be associated with anaemia (George and others 2002, De Lorimier and Messick 2004) and the case in the current study responded to appropriate treatment for haemoplasma-induced IMHA. Therefore, it seems wise to recommend that any cat with haemolytic anaemia be tested for haemoplasma infection.
Few cats in this study had haemoplasma or retrovirus infection, despite the use of sensitive PCR methodology and the fact that these agents are recognised causes of feline anaemia, including IMHA (Piek and others 2003, Kohn and others 2006). No cats infected with “Candidatus M. turicensis” or M. haemofelis were identified, yet previous reports have confirmed the presence of these organisms in the UK (Tasker and others 2003, Willi and others 2006b), albeit at low prevalence (Peters and others 2008).
FeLV and FIV infection was not associated with anaemia, despite the fact that both agents can cause anaemia (Shelton and Linenberger 1995). Serological diagnosis of FeLV and FIV infection was not performed in this study, as matched serum samples were not available from the majority of cats. Retrospective evaluation of the case records did provide some serological data; however, these serological tests were not necessarily performed contemporaneously with PCR so interpretation of any discrepant results is difficult. Using the information in the case records no cat had positive serology but negative PCR. Both cats with FIV provirus had FIV antibody. Two of eight cats with FeLV provirus were seropositive but five other animals tested were seronegative. Consistent with the study of Pinches and others (2006), the two FeLV seropositive cats had a higher provirus relative copy number (Ct values of 18.1 and 18.8 versus 26.6 to 36.9, data not shown). FeLV provirus positive but seronegative cats have likely been previously exposed to FeLV but have developed an immune response allowing suppression of viremia and antigen production (Torres and others 2005, Pinches and others 2006). Even when all FeLV provirus positive cats were included in analysis, there was no association between past FeLV exposure and anaemia.
Limitations of the current study include the fact that follow-up data were not objectively collected to allow accurate descriptions of response to treatment. This would have been particularly useful in the IMHA cases as an additional aid to help differentiate primary (likely to respond to immunosuppressive treatment only) from secondary IMHA, although in the current study other causes of secondary IMHA were rigorously excluded in the cats diagnosed with primary IMHA. Additionally, current therapy was not always recorded at the time of submission of samples for Coombs’ testing and it is possible that concurrent immunosuppressive therapy could have influence Coombs’ test results in the cats. It would be useful to consider such limitations in the design of future studies on feline Coombs’ testing and IMHA.
In conclusion, this study has documented associations between IMHA and persistent cold autoagglutination, and between IMHA and certain positive Coombs’ test reactions (with any antiserum, with polyvalent antiserum and with IgG antiserum) at 37°C supporting a recommendation that cats suspected of having IMHA should be evaluated for the presence of persistent autoagglutination at 4°C as well as undertaking a Coombs’ test at 37°C. Negative results for these tests would make IMHA very unlikely. However, the positive reactions described were not specific to IMHA as persistent autoagglutination at 4°C also occurred in some cats with other forms of anaemia, and positive Coombs’ tests at 37°C also occurred in cats with other forms of anaemia as well as two non-anaemic cats. Thus, testing for the presence of erythrocyte-bound antibodies should always be interpreted in parallel with documentation of haemolysis when suggesting that anaemia is immune-mediated. The reason for the presence of erythrocyte-bound antibodies in cats with other forms of anaemia warrants further investigation. Haemoplasmas and retroviruses were uncommonly detected and were not a major cause of anaemia in this sample of cats.
The work in this study was funded by a grant awarded by BSAVA Petsavers and their support is gratefully acknowledged. The authors thank E. Crawford, S. Cue and K. Papasouliotis for the haematological analyses and K. Egan for obtaining some of the samples. The assistance of colleagues is also gratefully acknowledged, particularly those from the University of Bristol Veterinary School (A. Harvey, I. Battersby, R. Dean, R. Giles, M. Goodfellow, S. Warman, F. Rizzo, S. Rudd and S. Wills). S. Warman and K. Papasouliotis are thanked for helpful discussions. J. Murray holds a position funded by Cats’ Protection and their support is gratefully acknowledged.