Canine idiopathic immune-mediated haemolytic anaemia: a review with recommendations for future research

Abstract

Idiopathic immune-mediated haemolytic anaemia (IMHA) is one of the most common immune-mediated diseases of dogs. The aim of this article is to review current knowledge of canine IMHA, its etiology, clinical presentation, diagnosis, complications, and treatment, in an attempt to establish why its outcome is still so poor. Clinical signs of anaemia develop within 3 days and dogs present with a median haematocrit of 13%, leucocytosis, a left shift, and reticulocytosis. Coagulation test results support the presence of disseminated intravascular coagulation. About 50% of dogs die in the first 2 weeks after presentation, and analysis of risk factors suggests that mortality is associated with hypercoagulability, inflammatory response, and liver and kidney failure. A positive direct agglutination test, spherocytosis, and true autoagglutination are widely accepted tests to demonstrate anti-erythrocyte antibodies, but are not yet standardized. To date, there is no evidence to support the efficacy of immunomodulators in addition to corticosteroids in the treatment of IMHA. Despite numerous investigations, the prognosis of IMHA remains dismal. There is an urgent need to validate and standardize diagnostic tests and criteria, and clinical trials might benefit from stratifying dogs by mortality risk. Analysis of samples from well-defined cases of canine IMHA might provide insight into the aetiology and pathophysiology of IMHA.

Keywords:

• dog
• canine
• anaemia
• haemolysis
• idiopathic
• immune-mediated

1. Background

Idiopathic immune-mediated haemolytic anaemia (IMHA) is one of the most common immune-mediated diseases of dogs (Giger 2005; Balch and Mackin 2007). The incidence was estimated as 0.2% in a case-load study of a general veterinary university hospital (Keller 1992). The first case series of 19 dogs with IMHA was described in the 1960s (Lewis et al. 1965). Six of the 19 dogs died during the initial haemolytic episode, and a further five during recurrences (Lewis et al. 1965). Despite numerous studies since then, the mortality of IMHA remains high (Reimer et al. 1999; Burgess et al. 2000; Grundy and Barton 2001; Weinkle et al. 2005; Piek et al. 2008, 2011).

The aim of this article is to review current knowledge of canine IMHA, its diagnosis and treatment, with a view to understanding why the prognosis remains so poor. The review concludes with recommendations to optimize future research.

2. Methods

A Pubmed search (http://www.ncbi.nlm.nih.gov/sites/entrez) using the search terms ‘canine IMHA’ was performed. The focus was on aetiology, clinical signs, diagnosis, risk factors, and evaluation of treatment in canine IMHA. Only studies published in peer-reviewed journals were used.

3. Results

3.1. Etiology

3.1.1. Breed and sex predisposition

Breed predisposition and familial occurrence suggest that a genetic component contributes to the susceptibility for IMHA (Table 1; Day and Penhale 1992; Day 1996). Moreover, some susceptible breeds may have a higher incidence of more than one immune-mediated disease (Day and Penhale 1992; Day 1996; Wilbe et al. 2010). The recognition of foreign proteins or self-proteins by the major histocompatibility complex (MHC) proteins is one of the key events in the development of immune-mediated disease (McDevitt 2000). Canine IMHA is associated with both susceptible and protective dog leucocyte antigen (DLA) haplotypes which, interestingly, are associated with different effects in specific breeds (Kennedy et al. 2006; Table 1). The presence of autoreactive T-cells in dogs with IMHA supports the view that MHC molecules are candidate susceptibility genes for IMHA (Corato et al. 1997).

Table 1. Breed predispositions, DLA-haplotypes, and sex in canine IMHA.

Breed	OR	HR	95% CI	Reference
Airdale terrier	45.3	–	5.5–∞	(McAlees 2010)
Bichon frise	5.3	–	1.2–22.5	(Miller et al. 2004)
Cocker spaniel	–	–	–	(Klag et al. 1993; McManus and Craig 2001; Carr et al. 2002; Mason et al. 2003; Ishihara et al. 2010)
	12.2	–	4.5	(Miller et al. 2004)
	–	3.3		(Burgess et al. 2000)
	–	5.9		(McManus and Craig 2001)
	–	5		(Weinkle et al. 2005)
	6.7	–		(Reimer et al. 1999)
English Springer spaniel	10	–	1.3–74.7	(McAlees 2010)
	32.4	–	–	(Reimer et al. 1999)
Finnish spitz	69.9	–	2.1–2287	(Miller et al. 2004)
Hungarian Viszla	10	–	1.3–74.7	(McAlees 2010)
Maltese	2.8	–	1.5–4.9	(McAlees 2010)
Miniature pincher	7.4	–	1.2–47.1	(Miller et al. 2004)
Miniature schnauzer	–	4	–	(Weinkle et al. 2005)
Old English sheepdog	–	–	–	(Day and Penhale 1992)
Rough-coated Collie	22.2	–	–	(Klag et al. 1993; Miller et al. 2004)
Shih Tzu	–	–	–	(Ishihara et al. 2010)
Welsh corgi	–	–	–	(Ishihara et al. 2010)
DLA-haplotype
DRB1*00601/DQA1*005011/DQB1*00701	1.8	–	1.2–3	(Kennedy et al. 2006)
DRB1*015/DQA1*00601/DQB1*00301	2.6	–	1.1–5.8	(Kennedy et al. 2006)
DRB1*001/DQA1*00101/DQB1*00201	0.4	–	0.2–0.9	(Kennedy et al. 2006)
Sex
Females	2.1		1.1–4.1	(Miller et al. 2004)
Females	–	–	–	(Carr et al. 2002; Mason et al. 2002; Weinkle et al. 2005)
Neutered females	–	–	–	(Carr et al. 2002; Weinkle et al. 2005)
Neutered dogs	–	–	–	(Carr et al. 2002; Weinkle et al. 2005)
Notes: Risk factors for IMHA were obtained in these studies by comparison of breed, haplotype, and sex incidence in a population with IMHA dogs versus the general hospital population. In case of Bichon frise, Finnish spitz, Miniature Pinscher, and the Rough-coated Collie, the author acknowledged that the numbers of dogs with this breed with IMHA may have been too low in study to accurately assess risk on IMHA. The risks associated with a specific DLA haplotype differed between breeds (Kennedy et al. 2006).
OR, Odds Ratio; HR, Hazard Ratio; and –, not reported.

The distribution and frequency of both DLA alleles and haplotypes vary substantially between breeds as a result of selective breeding (Angles et al. 2005; Kennedy et al. 2007a, 2007b, 2007c). Odds ratios or hazard rates attributed to breed are higher than those conferred by DLA haplotype suggesting that the risk for IMHA cannot be explained solely by the DLA haplotype (Klag et al. 1993; Miller et al. 2004; Kennedy et al. 2006; McAlees 2010). Several other genes are involved in the immune response cluster in the MHC region and show strong linkage disequilibrium with MHC genes (Horton et al. 2004). Therefore, susceptibility to IMHA might be linked to another gene within the MHC region. An increased incidence of IMHA has been observed in female (Carr et al. 2002; Mason et al. 2003; Miller et al. 2004; Weinkle et al. 2005) and neutered (Carr et al. 2002; Weinkle et al. 2005) dogs and an association has been reported with oestrus and whelping (Dodds 1977).

3.2. Clinical presentation

3.2.1. Clinical signs

IMHA can occur at any age. Although most reports describe an onset after the first year (Klag et al. 1993; Reimer et al. 1999; Burgess et al. 2000; Scott-Moncrieff 2001; Balch and Mackin 2007), idiopathic IMHA developed in eight of 222 dogs before 1 year of age in another study (Piek et al. 2011). The mean age of onset of IMHA is 6 years (combined result of n = 340; Klag et al. 1993; Duval and Giger 1996; Reimer et al. 1999; Burgess et al. 2000; McAlees 2010).

Most dogs with IMHA develop anaemia rapidly, possibly over a period as short as 3 days (Table 2). Non-specific signs, such as lethargy and loss of appetite, occur in most dogs and are accompanied by vomiting and diarrhoea in 15–30% of cases (Mason et al. 2003; Piek et al. 2008, 2011). Signs more specific for haemolysis are yellow to orange discolouration of the faeces and red urine (Reimer et al. 1999; Burgess et al. 2000; Mason et al. 2003; Piek et al. 2008, 2011).

Table 2. Duration and incidence of clinical signs at presentation in canine idiopathic IMHA.

	n = 60 (Burgess et al. 2000)^a	n = 42 (Klag et al. 1993)	n = 18 (Mason et al. 2003)	n = 70 (Reimer et al. 1999)	n = 149 (Piek et al. 2008)^b	n = 73 (Piek et al. 2011)^c,d
Duration (days)	–	90% <3	83% <3	–	median 6	median 3
Lethargy (%)	93	86	94	99	98	–
Pallor (%)	77	76	–	97	98	99
Anorexia (%)	–	90	83	99	80	76
Vomiting (%)	–	–	30	–	30	37
Diarrhoea (%)	–	–	17	–	15	23
Icterus (%)	45	50	67	51	37	37
Red urine (%)	32	–	44	13^e	44	24
Heart murmur (%)	27	–	–	47	–	–
Organomegaly^f (%)	38	25	–	43	34	–
Notes: In this table, the results of six studies on canine idiopathic IMHA have been summarized. In three studies, however, 1–5% of dogs had petechiation that may have been due to concurrent ITP (Burgess et al. 2000; Scott-Moncrieff et al. 2001; Piek et al. 2008; Piek et al. 2011).
^aIn one dog (2%) petechiae were noted.
^bIn eight dogs (5%) petechiae were noted.
^cFever in 46% of dogs.
^dPetechiae in 1% of dogs.
^eHaematuria
^fCranial abdominal organomegaly established during physical exam.

Table 3. CBC in dogs with IMHA.

	Mean	Median	Range	References
Ht (%)	14.8	12–14.5	4–32	(Klag et al. 1993; Reimer et al. 1999; Burgess et al. 2000; Scott-Moncrieff 2001; Carr et al. 2002; Mason et al. 2003; Piek et al. 2008; McAlees 2010; Piek et al. 2011)
	n = 353	n = 412	n = 444
Corrected reticulocytes (%)	1.3	0.9–2.7	0–19.2%	(Klag et al. 1993; Burgess et al. 2000; Carr et al. 2002; Piek et al. 2008; Piek et al. 2011)
	n = 60	n = 287	n = 282
WBC (×10^9/L)	34.4	21.7–38.7	2.1–130	(Reimer et al. 1999; Burgess et al. 2000; Scott-Moncrieff 2001; Carr et al. 2002; Mason et al. 2003; Piek et al. 2008; McAlees 2010; Piek et al. 2011)
	n = 312	n = 554	n = 444
Band neutrophils (×10^9/L)	2.1 ± 2.6	1–1.4	0–22.1	(Scott-Moncrieff et al. 2001; Carr et al. 2002; Mason et al. 2003; Piek et al. 2008; Piek et al. 2011)
	n = 72	n = 294	n = 294
Notes: In this table, the results of the CBC at presentation in a referral institution are summarized for 614 dogs based on data from nine different studies (Klag et al. 1993; Reimer et al. 1999; Burgess et al. 2000; Scott-Moncrieff 2001; Carr et al. 2002; Mason et al. 2003; Piek et al. 2008; McAlees2010; Piek et al. 2011). The means for Ht, corrected reticulocytes, WBC, and band neutrophils have been calculated from the mean and the number of dogs as reported in the studies referred to in this table. In case of the median, the range of the medians that were reported are given. The range given in the table encompasses all the ranges reported in the respective studies. The inclusion criteria for these studies consisted of different combinations of the following criteria: a restriction on the Ht (<30–35%), positive Coombs’ test, autoagglutination, evidence of hemolysis (hyperbilirubinemia, bilirubinuria, haemoglobinemia, haemoglobinuria, and/or spherocytosis), and regenerative erythroid response. Exclusion criteria focused on the exclusion of possible underlying disorders such as vaccination, medication, neoplasia, and infectious disorders, pre-treatment with cytostatics or longstanding immunosuppressive therapy.
WBC, white blood cell count and Ht, haematocrit.

The physical examination reveals clinical signs caused by anaemia, such as, tachycardia, tachypnoea, steep pulse, pale mucous membranes, and systolic murmur (Reimer et al. 1999; Burgess et al. 2000). Fever is a common clinical sign (Mellett et al. 2010), occurring in 46% of dogs (Piek et al. 2008, 2011). Petechiation as a result of concurrent severe thrombocytopenia is reported incidentally (2–5% of cases) (Burgess et al. 2000; Piek et al. 2008) and may be due to concurrent immune-mediated thrombocytopenia (ITP). Cranial abdominal organomegaly, due to splenomegaly and hepatomegaly, is found in up to 40% of cases (Klag et al. 1993; Reimer et al. 1999; Burgess et al. 2000; Piek et al. 2008).

3.2.2. Complete blood count

The complete blood count (CBC) results are remarkably similar in different populations of dogs with idiopathic IMHA despite differences in study inclusion and exclusion criteria (Table 3). At the time of presentation, most dogs have severe anaemia with a haematocrit of 12–14%, but some dogs may show a more chronic disease course and have a higher haematocrit (Day 1999). Tissue oxygenation is severely impaired at a haematocrit below 10% (Cain 1977) and can result in severe exercise intolerance, tachypnoea, and tachycardia, which may be the main reasons for referral since dogs with these symptoms require transfusion.

Pronounced leucocytosis with a left shift is a common laboratory feature at presentation, or leucocytosis may develop in dogs with leucopenia or normoleukaemia at presentation (Mitchell et al. 2009). Monocytosis is present in about 50% of cases (Thompson et al. 2004).

3.2.3. Coagulation

The prothrombin time is increased in up to 50% of dogs, and the activated partial thromboplastin time (APTT) is increased in 50–60% of dogs with idiopathic IMHA (Burgess et al. 2000; Scott-Moncrieff et al. 2001; Carr et al. 2002; Piek et al. 2008). Thrombocyte counts are below 50 × 10⁹/L in about 20% of dogs (Klag et al. 1993; Burgess et al. 2000; Scott-Moncrieff et al. 2001; Carr et al. 2002; Piek et al. 2008). These changes suggest the presence of diffuse intravascular coagulation in many dogs, which is supported by the finding of a low fibrinogen concentration in 20% of dogs and increased concentrations of D-dimer and fibrin degradation products and a decreased antithrombin activity (Scott-Moncrieff et al. 2001; Piek et al. 2008). However, another study found fibrinogen concentrations to be raised in 30–90% of dogs, which might be because fibrinogen is secreted during the acute-phase response (Murata et al. 2004).

3.3. Diagnosis

3.3.1. Diagnostic testing

The laboratory diagnosis of IMHA rests upon the demonstration of an immune-mediated mechanism for the haemolysis. The direct Coombs’ test is still the main method used to demonstrate anti-erythrocyte antibodies, alternative methods, however, include flow cytometry (Morley et al. 2008), enzyme-linked immunosorbent assay (Jones et al. 1987), and a gel test (Lapierre et al. 1990), although these tests are not routinely available in veterinary practice. The diagnostic performance of the direct Coombs’ test depends on many factors, such as use of polyspecific instead of monospecific Coombs reagents, a one-dilution only tube test instead of multititre plates, and incubation at 37°C only versus additional incubation at 4°C (Wardrop 2005). The differences in the set-up of the Coombs’ test in different laboratories may explain the reported large range in sensitivity, 50–80% (Jones et al. 1990; Overmann et al. 2007).

Anti-erythrocyte antibodies are characteristic of both secondary and idiopathic IMHA. The pattern of antibody reactivity in the direct Coombs’ test may be of diagnostic significance. IgM reactivity is more common in dogs with underlying disorders, and in one study was associated with less severe anaemia (Slappendel 1979; Wardrop 2005; Warman et al. 2008). Low Coombs’ titres may be seen in sick dogs without obvious signs of haemolysis (Jacobs et al. 1984; Overmann et al. 2007). The development of a commercially available gel test to detect antigen–antibody reactions suggests that it may be possible to standardize anti-erythrocyte antibody testing (Lapierre et al. 1990).

Currently, the use of multiple tests is advocated to overcome the problem of the low sensitivity of the direct Coombs’ test for diagnosing IMHA. Spherocytosis can be the result of immune-mediated erythrophagocytosis, but may also be found in some hereditary spherocytic disorders, such as spectrin deficiency (Slappendel et al. 2005). These diseases are rare, and spherocytosis is generally accepted as pathognomonic for IMHA. Spherocytes are quantified microscopically, usually as counts per high-power field (Weiss 1984; Griebsch et al. 2009, 2010). Potential disadvantages of this method are a high interobserver variability and the fact that the number of spherocytes per high-power field depends on the haematocrit.

Autoagglutination is generally accepted as a diagnostic criterion for IMHA (Giger 2005; Day 2008; Couto 2009). To avoid a false-positive outcome, erythrocytes that aggregate due to the presence of anti-erythrocyte antibodies, the so-called true autoagglutination, must be distinguished from rouleaux, which are conglomerates of erythrocytes that develop for non-immunological reasons, especially at lower temperatures. The basis of the in-saline slide agglutination test used for this purpose is the fact that saline may break up rouleaux formation but not erythrocyte aggregates. The ratio of saline to erythrocytes used differs between protocols from a 1:1 to 10:1 ratio (Giger 2005; Day 2008; Couto 2009). Agglutination can be viewed macroscopically on a slide, but most investigators prefer to confirm the absence of aggregates by microscopic examination. The disappearance of erythrocyte aggregates in the in-saline slide agglutination test does not exclude IMHA, because repeated washing of erythrocytes may break up erythrocyte aggregates (Day 2008).

3.3.2. Secondary IMHA

It is clinically important to distinguish between idiopathic IMHA and secondary IMHA because therapy and prognosis may be different. The incidence of secondary IMHA in dogs is 20–25% (Jackson and Kruth 1985; Piek et al. 2008).

Risk factors for secondary IMHA can be grouped into medications, vaccinations, neoplasia, infections, and systemic immune-mediated disorders (Giger 2005). Most of these risk factors have been taken from the human literature (Sokol et al. 1992, 1994), and relatively few have been confirmed in dogs. Sulphonamides, cephalosporin, and carprofen have been linked with IMHA (Bloom et al. 1988; Trepanier et al. 2003; Trepanier 2004; Mellor et al. 2005; Lavergne et al. 2006). The association of vaccination in one study was not confirmed in two other studies (Duval and Giger 1996; Astrup et al. 1998; Carr et al. 2002). In contrast with the situation in humans, where the occurrence of non-Hodgkin lymphoma is associated with IMHA (Ekstrom Smedby et al. 2008), in dogs, ITP has been found to be associated with lymphoma, but no such association has been confirmed for IMHA (Keller 1992). IMHA may be diagnosed as a part of a systemic immune-mediated disease such as systemic lupus erythematosus (Jones 1993), or as part of Evans syndrome (Evans and Duane 1949).

Many studies have reported on the seasonality of the onset of IMHA (Jackson and Kruth 1985; Klag et al. 1993; Duval and Giger 1996; Burgess et al. 2000; Weinkle et al. 2005; McAlees 2010) with increased exposure to infectious agents in certain seasons being suggested to trigger the disease. The occurrence of IMHA secondary to vector-borne diseases affects dogs that live or have visited the geographic area where the vector is endemic (Morita et al. 1995; Frank and Breitschwerdt 1999; Shaw et al. 2001). Inconsistencies in disease incidence between countries might be explained by differences in exposure risk in different geographic locations. For example, despite cases that may suggest the opposite (Bundza et al. 1976; Kenny et al. 2004), Haemoplasma canis is not an important trigger for IMHA in dogs in the UK (Warman et al. 2010).

3.3.3. Differential diagnosis

The generally accepted diagnostic features of idiopathic IMHA include immune-mediated haemolysis and the exclusion of underlying triggering disorders (Giger 2005). Knowledge of the known triggers for IMHA will optimize the identification of secondary IMHA. Many cases of secondary IMHA are relatively straightforward to diagnose because the dogs may display signs and symptoms due to the underlying disorder and different from those expected in idiopathic IMHA. Two clinical entities, Evans syndrome and pure red cell aplasia (PRCA), merit special attention since their presentation is similar to that of idiopathic IMHA.

3.3.4. IMHA and ITP

Evans syndrome, defined as the combined or sequential occurrence of IMHA and ITP, was first reported by Evans et al. in 1951 and is a rare disease in humans (Evans et al. 1951). The prevalence of Evans syndrome in dogs has been estimated at 0.01% (Orcutt et al. 2010). In addition to signs due to IMHA, dogs with Evans syndrome develop signs of haemorrhagic diathesis, such as petechiae or ecchymosis (in 13 of 21 dogs; Goggs et al. 2008), melena (in 4 of 12 dogs; Orcutt et al. 2010), oral bleeding (in 4 of 21 dogs and 3 of 12 dogs; Goggs et al. 2008; Orcutt et al. 2010, respectively), and haematemesis (in 1 of 12 dogs; Orcutt et al. 2010). The diagnosis of ITP should be made based upon exclusion of other causes of thrombocytopenia, preferably in combination with the demonstration of antiplatelet antibodies (Campbell et al. 1984; McVey and Shuman 1989; Kristensen et al. 1994a, 1994b). However, antiplatelet antibody testing is not routinely available in veterinary practice. The thrombocyte count cut-off to distinguish between ITP and disseminated intravascular coagulation (DIC) in dogs has been reported as 15 × 10⁹/L (Rozanski et al. 2002), although another study reported median thrombocyte counts of 32 × 10⁹/L in dogs with ITP and 55 × 10⁹/L in dogs with DIC (Botsch et al. 2009).

The prognosis of Evans syndrome is reported to be worse than that of the individual constituent diseases (Jackson and Kruth 1985; Goggs et al. 2008), although another study reported the prognosis to be similar to that of the individual constituent diseases (Orcutt et al. 2010).

3.3.5. Non-regenerative anaemia: non-regenerative IMHA or PRCA

Up to 33% of dogs with idiopathic IMHA present with non-regenerative anaemia. As reticulocytosis takes about 4–5 days to develop and most dogs develop clinical signs about 3 days before presentation, regenerative anaemia should develop a few days after first presentation (Giger 2005). If reticulocytosis fails to develop in this period, non-regenerative IMHA and PRCA should be considered in the differential diagnosis (Jonas et al. 1987; Stokol et al. 2000; Weiss, 2002, 2008). PRCA is characterized by total erythroid aplasia, in contrast with non-regenerative IMHA, where there is bone marrow erythroid hyperplasia or erythroid maturation arrest (Stokol et al. 2000; Weiss 2008).

Dogs with non-regenerative IMHA characterized by erythroid hyperplasia were found to have a lower 60-day survival than dogs with non-regenerative IMHA characterized by erythroid maturation or PRCA (Weiss 2008). Besides erythroid hyperplasia, these dogs had multiple pathological events, such as dysmyelopoiesis, myelonecrosis and myelofibrosis, interstitial oedema, haemorrhage, acute inflammation, and haemophagocytic syndrome that might have aggravated ineffective erythropoiesis (Weiss 2008). In contrast, in dogs with PRCA erythroid colony formation is suppressed by serum and immunoglobulin G (Weiss 1986).

3.4. Mortality

3.4.1. Mortality and thromboembolism

The death rate of canine IMHA may be as high as 80% (Reimer et al. 1999), and most deaths occur in the first 2 weeks after diagnosis (Piek et al. 2008, 2011). As many as half of these deaths are due to thromboembolism in the lung, liver, spleen, or in multiple organs, which was found in 50–80% of cases in three post-mortem studies (McManus and Craig 2001; Scott-Moncrieff et al. 2001; Carr et al. 2002). Risk factors inherent to IMHA are thrombocytopenia, increases in serum bilirubin concentrations, and hypoalbuminaemia, but treatment-related factors, such as glucocorticoid therapy, blood transfusion, and intravenous catheterization, may contribute as well (Klein et al. 1989; McManus et al. 2001; Carr et al. 2002). Many dogs with IMHA can be considered hypercoagulable, based on thromboelastography recordings (Sinnott and Otto 2009; Kol and Borjesson 2010; Fenty et al. 2011). P-selectin expression is considered to be a marker of activated platelets and is increased in dogs with IMHA (Weiss and Brazzell 2006; Ridyard et al. 2010).

An inflammatory reaction is present in most dogs with IMHA (Table 3). As a matter of fact, thrombogenesis and inflammation are interrelated. In inflammation, cell adhesion molecules called selectins are up-regulated and facilitate leucocyte transmigration. Additionally, interaction of P-selectin with its receptor, PSGL-1, stimulates the production of thrombogenic microparticles from leucocytes (mainly monocytes), which together with platelets and endothelial cells play an important role in thrombogenesis (Ramacciotti et al. 2009; Wakefield et al. 2009). The presence of an inflammatory response in dogs with IMHA is also evidenced by the activity of chemotaxins, such as monocyte chemoattractant protein-1 and granulocyte macrophage colony-stimulating factor, and acute-phase proteins (Tecles et al. 2005; Griebsch et al. 2009; Mitchell et al. 2009; Duffy et al. 2010; Yuki et al. 2010).

Table 4. Variables influencing the mortality risk in canine idiopathic IMHA.

Risk factors	HR	95% CI	Reference
Clinical characteristics
Male	↑		(Ishihara et al. 2010)
Body weight	↑		(Ishihara et al. 2010)
Warm season, Japan	↑		(Ishihara et al. 2010)
Vaccination	↑		(Duval and Giger 1996)
DEA 7 in Cocker Spaniel	↓		(Miller et al. 2004)
Icterus	2	1.07–3.77	(Duval and Giger 1996; Piek et al. 2008; Holahan et al. 2010)
Haematological laboratory characteristics
Autoagglutination	↑		(Duval and Giger 1996; McAlees 2010)
Spherocytosis	0.48	0.25–0.92	(Piek et al. 2008)
Decreased PCV	↑		(Klag et al. 1993)
PCV < 20%	↑		(Ishihara et al. 2010)
Reticulocytopenia	↑		(Klag et al. 1993)
WBC > 60 × 10^9/L	↑		(Burgess et al. 2000)
Bands > 3 × 10^9/l	↑		(Weinkle et al. 2005)
Thrombocytopenia	↑		(Carr et al. 2002)
<150 × 10^9/L	↑		(Weinkle et al. 2005)
<200 × 10^9/L	↑		(Ishihara et al. 2010)
Per 50 × 10^9/L increase	0.90	0.82–0.99	(Piek et al. 2010)
PT > 10 s	↑		(Burgess et al. 2000)
PT > 9 s	↑		(Ishihara et al. 2010)
PT per second increase	1.44	1.14–1.82	(Piek et al. 2008)
PT per second increase	1.46	1.05–2.02	(Whelan et al. 2009)
APTT (s)	1.03	1.01–1.06	(Piek et al. 2008)
FDP ≥ 10 mg/l	↑		(Ishihara et al. 2010)
Increase Fibrinogen (g/L)	0.77	0.65–0.9	(Piek et al. 2008)
Biochemistry laboratory characteristics
Total protein < 60 g/L	↑		(Ishihara et al. 2010)
Hypoalbuminemia	↑		(Carr et al. 2002)
<30 g/L	↑		(Weinkle et al. 2005)
<26 g/L	↑		(Ishihara et al. 2010)
Increased bilirubin	↑		(Duval and Giger 1996; Holahan et al. 2010)
	1.03	1–1.07	(Whelan et al. 2009)
>2 g/L	↑		(Ishihara et al. 2010)
>5 g/L	↑		(Carr et al. 2002)
>10 g/L	↑		(Klag et al. 1993)
Increased AP	↑		(Holahan et al. 2010)
Sodium <140 mmol/L	↑		(Ishihara et al. 2010)
Potassium <3.5 meq/L	↑		(Weinkle et al. 2005)
Serum CK >250 U/L	↑		(Weinkle et al. 2005)
Increased urea	↑		(Holahan et al. 2010)
>25 g/L	↑		(Ishihara et al. 2010)
Per 20 mmol/L increase	2.11	1.14–3.15	(Piek et al. 2008)
Creatinine per 50 µmol/L increase	1.27	1.12–1.43	(Piek et al. 2008)
Decreased HCO3^−	↑		(Holahan et al. 2010)
Decreased BE	↑		(Holahan et al. 2010)
Increased lactime	↑		(Holahan et al. 2010)
Notes: The factors that were associated with increased (↑) or decreased (↓) mortality risk are summarized in this table. These risk factors were identified either by univariate Cox's proportional hazards analysis or by contingency table analysis.
Lactime was defined as the duration of increased plasma lactate concentration.
AP, alkaline phosphatase and BE, base excess.

Table 5. Effects of change in pre-test probability for two different Coombs’ test sensitivities on the posterior probability of IMHA.

3.4.2. Risk factors for mortality

Several studies have identified risk factors for death (Table 4), the most important being variables related to DIC, inflammation, and liver and kidney failure. Tissue hypoxia due to the often severe anaemia that develops during IMHA has been suggested to play a central role in the development of pathology (McManus et al. 2001; Piek et al. 2008), yet only a few investigators have identified anaemia as a risk factor (Klag et al. 1993; Ishihara et al. 2010). It may be that a significant association with anaemia has been overlooked in other studies because the haematocrit was evaluated as a continuous rather than a dichotomous variable. Studies have shown that tissue oxygenation is severely impaired when the haematocrit is below 10% (Cain 1977; Cain and Chapler 1978), which suggests that tissue hypoxia only develops in the severest cases of anaemia. The duration of severe anaemia may be another factor to consider. A study of dogs with IMHA demonstrated that the duration of hyperlactaemia was associated with a poor survival, which provides indirect evidence for the impact of anaemia since lactate concentrations are inversely correlated with the haematocrit (Holahan et al. 2010). The effect of hypoxia may be difficult to identify since blood transfusions are a standard component of palliative therapy in IMHA.

Multivariate models suggest that liver and renal failure, DIC, and inflammation contribute independently to the risk of death (Piek et al. 2008, 2011). Hypoxia may cause tissue damage (McManus et al. 2001), and chemotaxins released during the ensuing acute-phase response then orchestrate an inflammatory response. Local tissue necrosis, endothelial cell activation, and the production of thrombogenic particles by leucocytes may activate coagulation (Ramacciotti et al. 2009; Wakefield et al. 2009). The occurrence of thromboembolism may disturb tissue perfusion and further aggravate hypoxic tissue necrosis. The cumulative effect of multiple independent risk factors can be assessed by multiplication of the independent risks. The observation that these risk factors may potentiate each other explains the severely increased mortality risk if they are simultaneously present.

3.5. Treatment

3.5.1. Immunomodulation

Immunomodulation is the mainstay of treatment of IMHA, with the aim of decreasing erythrophagocytosis and suppressing immunoglobulin production, and can be combined with whole blood or packed red cell transfusions and anticoagulation (Balch and Mackin 2007). Although the efficacy of glucocorticoids in IMHA has never been investigated in a clinical trial, it is generally assumed that they are effective and their side-effects are taken for granted (Whitley and Day 2011). In contrast with recommendations for lifelong immunosuppression, the results of two studies suggest that immunosuppression for approximately 3 months is sufficient (Piek et al. 2008, 2011). However, at least 10% of dogs had a recurrence of IMHA up to 4 years after the first haemolytic crisis (Piek et al. 2008, 2011). Available evidence suggests that the combination of azathioprine plus prednisolone is not more effective than prednisolone alone in the treatment of canine IMHA (Piek et al. 2011) despite reports suggesting a positive effect of the addition of azathioprine. However, the treatment arms were not randomized in these studies, the duration of azathioprine therapy was suboptimal, and disease duration and severity may have influenced treatment choice, thereby confounding the outcome (Reimer et al. 1999; Burgess et al. 2000; Piek et al. 2011). There is also no evidence supporting the use of cyclophosphamide in canine IMHA (Mason et al. 2003), whereas there is evidence that cyclophosphamide may be associated with increased mortality (Burgess et al. 2000; Grundy and and Barton 2001). Human intravenous immunoglobulin has not been found to have a beneficial effect (Kellerman and Bruyette 1997; Gerber et al. 2002; Whelan et al. 2009). Splenectomy has been investigated in two small retrospective cohort studies of dogs with IMHA (Feldman et al. 1985; Horgan et al. 2009), but because the dogs also received medical treatment and blood transfusions and neither study included a control group, definite conclusions could not be drawn about the efficacy of splenectomy in IMHA. Liposomal clodronate improved survival in dogs with IMHA compared with that of a historical control group matched for disease severity (Mathes et al. 2006).

3.5.2. Anticoagulation

A retrospective cohort study reported superior survival in dogs with IMHA treated with ultra low-dose aspirin (Weinkle et al. 2005), and in a controlled randomized trial ultra-low dose aspirin was as effective as clopidogrel, another antiplatelet drug (Mellett et al. 2010). There is no evidence that heparin is effective, even though it is commonly used as adjunct therapy (Mathes et al. 2006; Whelan et al. 2009). It may be difficult to establish the dose of heparin to be used because of differences in the pharmacokinetics of heparin in healthy and diseased dogs (Breuhl et al. 2009). Pharmacokinetic studies suggest that levels of unfractionated heparin and low-molecular weight heparin should be monitored on the basis of anti-Xa activity rather than on the basis of coagulation tests such as APTT (Mischke and Grebe 2000; Mischke et al. 2001; Mischke and Jacobs 2001b). A recent randomized clinical trial in which the heparin dose was adjusted based on plasma anti-Xa activity demonstrated superior survival in the heparin treatment arm (Helmond et al. 2010).

3.5.3. Blood transfusion

It is not possible to evaluate the true effect of blood transfusion on survival because it is part of supportive therapy in both retrospective studies and randomized clinical trials. There seems to be no evidence that transfusions are contraindicated. Several retrospective studies failed to find transfusion to be associated with survival (Burgess et al. 2000; Carr et al. 2002; Weinkle et al. 2005; McAlees 2010), and only one study found transfusion to adversely affect survival (Piek et al. 2008), possibly because transfusions were given to the most severely ill patients. Bovine oxyglobin transfusion may be an alternative to blood transfusions, but no randomized controlled trials have been reported. One retrospective study reported an increased relative mortality risk in dogs with IMHA that had been treated with bovine haemoglobin solution (Grundy and and Barton 2001).

4. Discussion

Despite numerous preclinical and clinical studies, the prognosis of canine IMHA remains poor. There is no evidence that immunomodulators in addition to glucocorticoids have a convincing positive effect on the outcome of IMHA. This may be due to a true lack of effect of this additional therapy or may be the reflection of a suboptimal study design. To optimize the ability of a trial to detect treatment-related differences in outcome, it is essential that factors that have an impact on outcome, such as leucocytosis, left shift, bilirubinaemia, low thrombocyte counts, and increased coagulation times, are equally distributed between treatment arms. This is especially true because risk factors for death appear to be multiplicative such that dogs with icterus (hazard ratio (HR) = 2), increased creatinine by 20 µmol/L (HR = 1.7), increased PT by 5 s (HR = 1.4), increased APTT by 10 s (HR = 1.1), and thrombocyte counts of 100 × 10⁹/L (HR = 1.1) (Table 4) have a 30-fold higher risk of death than dogs without these risk factors. A recent study showed that dogs with IMHA can be stratified by risk of death according to the haematocrit (<20%), thrombocyte count < 200 × 10⁹/L, total protein concentration (<60 g/L), sex, and season (Ishihara et al. 2010).However, other investigators have not identified sex and season as mortality risk factors (Table 4). A more solid scoring system that is universally valid should be developed using mortality risk factors identified in different populations, such as those given in Table 4. The cut-offs for such a scoring system are discretionary, but could, for example, identify IMHA dogs with low (<25%), intermediate (25–75%), and high (>80%) risk of death.

The risk of death is highest during the first 2 weeks after diagnosis (Piek et al. 2008, 2011), and the main factors that contribute to this risk are coagulation disturbances, and liver and kidney failure, which are probably due to anaemia-induced tissue hypoxia (McManus et al. 2001; Piek et al. 2008, 2011; Holahan et al. 2010). It may be unrealistic to expect a single therapy to be effective in these virtually moribund patients. Dogs at intermediate risk of dying may be a better population in which to study therapies to ameliorate the major risk factors within this 2-week time frame, such as anticoagulant therapy, immunomodulators to stop antibody-mediated erythrophagocytosis, or therapies that reverse anaemia and hypoxia. The subset of dogs with IMHA with a low mortality risk score is perfectly suited to the study of immunomodulating therapies that have an expected onset of action after 2 weeks, such as azathioprine and cyclosporine. The duration, dosing, and efficacy of immunosuppression should be evaluated in this subset of IMHA dogs. Given the relatively low incidence of IMHA, multicentre studies are the only way to recruit large enough groups of patients to study the effects of drugs (Figure 1). However, before this can happen, inclusion and exclusion criteria and methodological aspects need to be standardized.

Figure 1. The output of PS power and sample size calculations. PS is an interactive program for performing power and sample size calculations by Dupont W.D. and Plummer W.D. (http://biostat.mc.vanderbilt.edu/PowerSampleSize; accessed 27 March 2011) that is freely available on the web. PS was used to calculate sample size for a hypothetical randomized controlled trial of a new treatment that would lead to 50% less mortality in dogs with IMHA with a mortality risk of 50%.

Idiopathic IMHA is a diagnosis made largely by hypothetico-deductive reasoning. Each step in the deductive process increases the prior probability of IMHA. Evidence for immune-mediated erythrocyte destruction is in most cases obtained with the direct Coombs’ test. However, investigators are worried about the low sensitivity of the test and the lack of standardized methodology (Jones et al. 1990; Wardrop 2005; Overmann et al. 2007). Although this is certainly true, it should be realized that not only sensitivity but also disease prevalence determines the positive predictive value of a diagnostic test. An essential step in the work-up of a patient with idiopathic IMHA is the use of exclusion criteria in order to differentiate primary IMHA from secondary ones. The diagnostic procedures in the work-up of IMHA patients may raise the pre-test probability of IMHA to such a level that the positive predictive value of the direct Coombs’ test is minimally influenced by its sensitivity (Table 5).

Analysis of studies of the prognosis and treatment of dogs with idiopathic IMHA suggests that collaborative efforts may be the only way to improve the currently dismal prognosis of canine idiopathic IMHA. This will require the standardization of laboratory testing and the development of a scoring system to stratify dogs according to their mortality risk. Studies of the etiological and pathophysiological aspects of canine idiopathic IMHA will benefit from the improved access to samples from well-defined canine IMHA cases.