The preliminary findings of this study were presented in abstract form as a Free Communication at the European College of Veterinary Internal Medicine Annual Forum, Munich, September 2002.
Anti-Insulin Antibodies in Diabetic Dogs Before and After Treatment with Different Insulin Preparations
Version of Record online: 3 OCT 2008
Copyright © 2008 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 22, Issue 6, pages 1317–1325, November–December 2008
How to Cite
Davison, L.J., Walding, B., Herrtage, M.E. and Catchpole, B. (2008), Anti-Insulin Antibodies in Diabetic Dogs Before and After Treatment with Different Insulin Preparations. Journal of Veterinary Internal Medicine, 22: 1317–1325. doi: 10.1111/j.1939-1676.2008.0194.x
- Issue online: 30 OCT 2008
- Version of Record online: 3 OCT 2008
- Submitted March 14, 2008; Revised June 2, 2008; Accepted July 30, 2008.
- Bovine insulin;
- Glycemic control;
- Porcine insulin
Background: Anti-insulin antibodies (AIA) occur in diabetic dogs after insulin therapy, although their clinical significance is unclear.
Hypothesis: Treatment of diabetic dogs with heterologous insulin is more likely to stimulate production of AIA than is treatment with homologous insulin.
Animals: Diabetic dogs sampled before insulin therapy (n = 40), diabetic dogs sampled following treatment with porcine (homologous) insulin (n = 100), bovine (heterologous) lente insulin (n = 100), or bovine protamine zinc (PZI) insulin (n = 20), and nondiabetic control dogs (n = 120).
Methods: Prospective observational study. Sera were analyzed by ELISA for antibodies against porcine insulin, bovine insulin, insulin A, B, or C peptides, and control antigens; canine distemper virus (CDV) and canine thyroglobulin (TG). Canine isotype-specific antibodies were used to determine total and anti-insulin IgG1 : IgG2 ratios.
Results: There was no difference in CDV or TG reactivity among the groups. AIA were detected in 5 of 40 newly diagnosed (untreated) diabetic dogs. There was no significant difference in AIA (ELISA optical density reactivity) comparing control and porcine insulin-treated diabetic dogs (P > .05). Anti-insulin reactivity was most prevalent in bovine PZI insulin-treated dogs (90%; P < .01), and bovine lente insulin-treated dogs (56%; P < .01). AIA induced by treatment were enriched for the IgG1 isotype.
Conclusions and Clinical Importance: This study indicates that bovine insulin is more immunogenic than porcine insulin when used for treatment of diabetic dogs.
The majority of diabetic dogs have insulin deficiency and therefore require insulin therapy to maintain glycemic control.1 The underlying pathogenesis of insulin deficiency diabetes in dogs remains to be established, although β cell loss associated with exocrine pancreatic disease,2 immune-mediated attack,3 or both has been suggested to be involved. Insulin autoantibodies can be present at disease onset and before treatment in human type 1 diabetic patients, reflecting an autoimmune process against β cell antigens. Despite the presence of anti-islet reactivity in some diabetic dogs, insulin autoantibodies are not a common feature of the disease.4 The majority of studies have failed to find significant insulin autoreactivity,a,b although there is a report of the presence of autoantibodies in 2 diabetic dogs.c
There are currently 2 licensed insulin products for treatment of canine diabetes in the United Kingdom. Insuvet (Schering Plough Animal Health) contains bovine insulin and is available in soluble (short acting), lente (intermediate acting), or protamine zinc (PZI; long acting) formulations. Caninsulin (also marketed in North America as Vetsulin; Intervet) is a mixed soluble and lente form of porcine insulin. In addition to their use in the United Kingdom, bovine and porcine insulin products have been used to treat diabetic dogs throughout Europe, North America, and Australia, although bovine PZI and lente preparations are no longer widely available in the United States. Because most diabetic dogs are treated by injecting insulin on a daily basis, it is perhaps not surprising that some dogs develop anti-insulin antibodies (AIA) as a result of treatment.a,b,c,d It is not easy to distinguish between insulin autoantibodies and AIA once insulin therapy has been initiated. In human beings, the former are usually present at lower concentrations, are more likely to recognize linear rather than conformational epitopes, and are more species specific.5
Initial studies have suggested that approximately 60% of diabetic dogs treated with bovine insulin show serological evidence of anti-insulin reactivity.6 However, it is not clear why some dogs develop AIA whereas others do not. Many factors are believed to contribute to the immunogenicity of different types of insulin preparation including species of origin, formulation, and method of injection.7 Previous studies in human diabetic patients have suggested that some form of adjuvant is required to stimulate an immune response to insulin.8 Such adjuvants could include impurities in insulin preparations9 or substances designed to prolong the activity of insulin (protamine), which itself is capable of inducing antibody responses in some individuals.10
The insulin molecule is similar across species, which is why insulin, purified from bovine or porcine pancreas tissue, could be used to treat human diabetes before the advent of recombinant protein technology. Although insulin of animal origin has a similar biological activity to the recombinant form in humans, heterologous insulin is more likely to stimulate an antibody response.11 Currently, there is no recombinant canine insulin available for treatment of diabetic dogs and use of bovine or porcine insulin is still widespread. The B chains of canine, bovine, and porcine insulin are identical. Canine and porcine A chains have identical amino acid sequences, whereas there are 2 A chain amino acid differences (positions 8 and 10) comparing canine and bovine insulin. Thus, in immunological terms, porcine insulin can be considered to be a self-antigen in dogs, whereas bovine insulin is a heterologous protein. Although it is only the A chain that differs, comparing canine and bovine insulin, a previous study suggested that AIA recognize conformational rather than linear epitopes (6), although little is known of the precise epitopes of the insulin molecule involved in AIA binding.
Evaluation of antibody isotype has proven to be a useful tool in evaluating the anti-insulin response in humans. Antibody isotype is determined by the cytokine environment in which the B cell is activated and the nature of T-cell help provided. IgG2a antibodies in mice are associated with “T-helper type 1” (TH-1) responses and IgG1 antibodies with “T-helper type 2” (TH-2) responses.12 Although pancreatic β cell destruction is thought to occur in association with a predominant TH-1 cytokine environment,13 insulin autoantibodies of both IgG1 and IgG2 isotypes have been reported in experimental rodent models of diabetes14 and in diabetic children.15 It has been suggested that AIA induced by insulin therapy in mice are more likely to be IgG1, associated with TH-2 responses, following subcutaneous injection of insulin.16 The isotype of AIA has yet to be investigated in dogs and the association between isotype and T-cell cytokine responses has not been fully elucidated in this species.
The clinical significance of AIA is controversial. It has been postulated that the presence of AIA might be associated with poor or erratic glycemic control, because of their potential to bind and neutralize insulin.17 Furthermore, there is the potential for immune complex formation in the pancreatic islets, possibly exacerbating the pathological process. Conversely, antibody binding to insulin and subsequent dissociation might be beneficial by prolonging the duration of action.
The aim of the current study was to investigate the nature of the AIA response in a large group of diabetic dogs. The objectives were to determine the prevalence of insulin autoantibodies in untreated diabetic dogs, to test the hypothesis that heterologous insulin is more immunogenic than homologous insulin, and to determine the subunit specificity and isotype of AIA in insulin-treated diabetic dogs.
Materials and Methods
Case Recruitment and Sample Collection
Dogs with naturally occurring diabetes (n = 260) were recruited from 1st opinion veterinary practices and referral centers primarily in the United Kingdom, with some additional samples from France, Germany, and Holland. Diabetes mellitus was diagnosed based on a combination of consistent clinical signs, persistent hyperglycemia, and glucosuria. Blood was collected by sterile venepuncture and serum was submitted to the Royal Veterinary College as part of the Canine Diabetes Database and Archive initiative. Practitioners provided information for each animal with respect to age, breed, sex, weight, date of diagnosis, and concurrent disease as well as dose, type, frequency, and duration of insulin treatment. Some diabetic dogs were sampled before initiation of insulin therapy (n = 40) and repeat samples were provided from some of these dogs (n = 9) after the 1st 3 months of insulin treatment. The treated diabetic dog groups consisted of animals sampled while receiving treatment with either Caninsulin (n = 100), Insuvet lente (n = 100), or Insuvet PZI insulin (n = 20). All dogs receiving insulin had been treated for at least 1 month and had not received any other type of insulin. After completion of diagnostic testing, excess serum was stored at −70 °C until analysis.
The control group consisted of 45 healthy dogs and 75 dogs that had been referred to the Queen's Veterinary School Hospital, University of Cambridge, for reasons other than diabetes mellitus. Dogs with any other endocrine or immune-mediated disease or receiving corticosteroid treatment were excluded from the control group. Control dogs were confirmed as normoglycemic (range 3.3–5.6 mmol/L) either as part of a routine biochemistry panel or with a commercial glucometer.e
Measurement of serum antibodies to canine distemper virus (CDV) or canine thyroglobulin (TG) was determined using commercially available antigens.f,g In addition, flat bottomed 96-well microtiter platesh were coated with 50-μL insulin antigen (bovine insulin, porcine insulin, bovine insulin A chain peptide, bovine insulin B chain peptide,i or canine insulin C-peptidej at 10 μg/mL in 0.05 M carbonate/bicarbonate buffer [pH = 9.6] overnight at 4 °C. For measurement of total serum IgG1 and IgG2 concentrations, individual serum samples diluted 1 : 10,000 in 50 μL carbonate/bicarbonate buffer were directly bound to ELISA plate wells. To allow quantification of total and AIA, standard curves were prepared using a reference serumk (total IgG = 31 mg/mL, IgG1 = 17 mg/mL, IgG2 = 14 mg/mL).
Antigen-coated ELISA plates were washed twice with 0.15 M phosphate-buffered saline supplemented with 0.1% Tween 20 (PBST) and blocked by incubating with 100 μL PBST supplemented with 2% skimmed milk powderl and 10% sheep serumj (PTMS) at room temperature for 1 hour. After 2 further washes, 50 μL patient serum, diluted in PTMS (1 : 100 for TG and CDV serology or 1 : 10 for anti-insulin serology) was added to duplicate wells. Positive and negative control sera for TG, CDV, and AIA were used where indicated. Plates were washed with PBST 4 times and antibody binding detected using 1 : 20,000 dilution in PTMS of either sheep anti-canine IgG horse-radish peroxidise (HRP) conjugate, goat anti-canine IgG1-HRP, or sheep anti-canine IgG2-HRP.k Alternatively, IgG isotypes were detected using tissue culture supernatants containing monoclonal antibodies from the B6 hybridoma (mouse anti-canine IgG1) or E5 hybridoma (mouse anti-canine IgG2), generously provided by Prof MJ Day, University of Bristol.18
After 4 final washes, plates were developed by the addition of 50 μL substrate (3,3′ 5,5′ tetramethyl benzidinej), incubated for 10 minutes and the reaction stopped using 100 μL 2 M sulfuric acid. The optical density (OD) of each well was measured at 450 nm using a Titertek MultiScan ELISA plate reader.m
The mean absorbance OD value for the negative control sample (PTMS diluent only, without serum, performed in duplicate) was subtracted from test serum sample mean OD values on each plate. A canine serum sample known to be positive for AIA at a 1 : 100 dilution was selected following pilot experiments and was used as a positive control for each anti-insulin ELISA. This was also used to normalize OD values for test serum samples, to compensate for interassay variability between plates. Quantification of AIA was achieved by measuring samples against a standard curve, generated with the canine reference serum.k This was used in all isotype experiments to allow quantification of total serum IgG1 and IgG2 in each sample as well as anti-insulin IgG1 and IgG2. Isotype results were then calculated as a ratio rather than absolute values where
Statistical analyses were performed using a commercial software package (SPSS v11.5 for Windows). An ELISA OD value of greater than the mean plus 1.96 × SD of the control group (95% confidence interval) was considered positive. Antibody reactivity in control and diabetic groups was compared using the Mann-Whitney U-test with the Bonferroni correction. Correlation between anti-bovine insulin and anti-porcine insulin antibodies was determined using the Spearman rank correlation coefficient. P values < .05 were considered significant.
In the diabetic group, the median duration of treatment was 6.3 months in the Caninsulin-treated dogs (range 1–48 months), 7 months in Insuvet lente-treated dogs (range 1–93 months), and 12 months in the Insuvet PZI-treated dogs (range 1–60 months) (Table 1). Seventy-two dogs in the Caninsulin-treated group, 70 dogs in the Insuvet lente-treated group, and all 20 PZI-treated dogs were receiving insulin once daily, with all other diabetic dogs receiving twice daily injections. The median dose of Caninsulin was 1.0 IU/kg/injection, Insuvet lente was 1.2 IU/kg/injection, and Insuvet PZI was 1.2 IU/kg/injection.
|Group (Number of Dogs)||Age (Years) Median [Range]||Sex||Weight (kg) Median [Range]||Duration of Insulin Therapy (Months) Median [Range]||Frequency of Insulin Injection||Insulin Dose (IU/ kg/injection) Median [Range]||Reactivity (OD) to CDV Antigen Median [Range]||Reactivity (OD) to Canine TG Median [Range]||Anti-Bovine Insulin IgG (μg/mL) Median [Range]||Anti-Porcine Insulin IgG (μg/mL) Median [Range]|
|Healthy control dogs (n = 45)||>1||19M (11N) 26F (19N)||NA||NA||NA||NA||1.397 [0.150–2.301]||0.244 [0.093–0.614]||<0.3 [<0.3–0.5]||<0.3 [<0.3–0.5]|
|Nondiabetic control patients (n = 75)||8 [0.5–14]||44M (15N) 31F (22N)||NA||NA||NA||NA||1.203 [0.023–2.524]||0.228 [0.048–1.377]||<0.3 [<0.3–0.5]||<0.3 [<0.3–0.5]|
|Newly diagnosed diabetics (n = 40)||10 [0.4–11]||17M (9N) 23F (22N)||NA||0||NA||NA||1.045 [0.042–2.153]||0.214 [0.045–1.021]||0.4 [<0.3–8.1]||0.4 [<0.3–8.4]|
|Caninsulin-treated diabetics (n = 100)||10 [2.5–16.5]||46M (30N) 54F (50N)||14.4 [3.2–73.5]||6.3 [1–48]||72 q24h 28 q12h||1.0 [0.2–3.4]||1.033 [0.061–2.326]||0.277 [0.057–1.329]||<0.3 [<0.3–2.7]||<0.3 [<0.3–2.4]|
|Insuvet lente-treated diabetics (n = 100)||10 [1.9–17]||42M (27N) 58F (56N)||14.9 [3.9–54]||7.0 [1–93]||70 q24h 30 q12h||1.2 [0.2–3.6]||1.049 [0.125–2.357]||0.224 [0.053–0.871]||0.8 [<0.3–25.6]||0.8 [<0.3–33.8]|
|Insuvet PZI-treated diabetics (n = 20)||11.3 [3.0–15]||12M (8N) 8F (6N)||17.4 [4.1–28.6]||12 [1–60]||16 q24h 4 q12h||1.2 [0.7–2.4]||1.042 [0.132–2.239]||0.175 [0.077–0.463]||1.9[<0.3–>51.2]||1.9 [0.3–>51.2]|
Five of 40 newly diagnosed diabetic dogs had antibody reactivity to insulin (Fig 1). Repeated testing was performed on 9 dogs several months after initiation of insulin therapy. At this point, 3 of 6 dogs treated with Insuvet lente had seroconverted, whereas all 3 Caninsulin-treated dogs that were retested remained AIA negative (Fig 1).
Most dogs were reported to be vaccinated against CDV, although vaccination status was not reported in a proportion of dogs. There was wide variation in CDV reactivity in each of the groups (Fig 2). The majority of dogs were positive for CDV antibodies and there was no significant difference comparing groups (P>.05). A small number of dogs in each group were positive for anti-TG antibodies, but there was no significant difference in anti-TG reactivity comparing diabetic and control groups (P>.05) and, in general, reactivity was low compared with a TGAA-positive control serum sample.
There was no significant difference in anti-insulin reactivity comparing Caninsulin-treated dogs with either of the 2 control groups (P > .05) (Fig 2c and d). However, there was significantly higher insulin reactivity in the Insuvet lente and the Insuvet PZI-treated dogs compared with controls (P < .01), although there was a wide range of AIA within these treatment groups. In AIA-positive dogs, there was a significant association between anti-bovine and anti-porcine insulin reactivity (r2= 0.96, P < .01).
Using the mean plus 1.96 × SD of the combined control groups as the 95% confidence limit, 12% of dogs treated with Caninsulin had AIA reactivity, compared with 56% in the Insuvet lente-treated group. The Insuvet PZI-treated group contained the highest proportion of dogs demonstrating insulin reactivity, with 17 of 20 dogs being AIA positive. This group also contained the dogs with the highest concentration of antibodies, when anti-insulin IgG was quantified (Table 1).
AIA Subunit Specificity and Isotype Analysis
The insulin subunit specificity of the AIA response was investigated in selected Insuvet lente-treated diabetic dogs. Samples (n = 45) were selected to represent a range of anti-insulin reactivity with the majority being positive for AIA. Canine insulin C-peptide was used as a negative control antigen, because this is not present in the Insuvet lente preparation, and no significant difference in reactivity was found between diabetic dogs and controls (Fig 3d). Forty-one of the 45 treated diabetic dogs were positive for AIA against the native molecule, 6 of which were positive for insulin A chain reactivity and 10 of which were positive for insulin B chain reactivity.
Similar trends were seen using either the Bethyl reagents or the monoclonal antibodies, although the correlation between values obtained by each method was poor (r2= 0.14 for IgG1 and 0.26 for IgG2, P >.05). By either method, there was no significant difference in total serum IgG1 or IgG2 concentration comparing diabetic dogs and controls (P > .05). Isotype analysis of AIA demonstrated a bias in favor of the IgG1 isotype (Fig 4).
This study investigated the prevalence and nature of AIA in diabetic dogs at diagnosis and after treatment with insulin. A small proportion of dogs demonstrated AIA before diagnosis, consistent with an autoimmune response. Antibody development following treatment was most predominant in dogs treated with (heterologous) bovine insulin preparations, appeared to be enriched for IgG1 and in some cases was directed against individual insulin A chain or B chain peptides.
Five of 40 newly diagnosed diabetic dogs had anti-insulin reactivity in the present study. Because blood samples had been taken before initiation of insulin therapy, this suggests that these are insulin autoantibodies. However, it remains to be established as to whether these are indicative of an immune-mediated pathogenesis, or a consequence of immune reactivity to β cell antigens, released following cellular destruction by some other disease process.
In a previous study evaluating anti-islet autoreactivity in the sera of newly diagnosed canine diabetic patients, 12 of 23 dogs were found to be positive for islet cell autoantibodies by immmunofluoresence.19 The larger proportion of cases with autoreactivity compared with the present study might relate to differences in sensitivity of the techniques used, but it is also possible that autoantibody recognition of other islet autoantigens apart from insulin led to islet immunofluoresence. Such antigens might include glutamic acid decarboxylase 65 or insulinoma antigen-2, which are important in human type 1 diabetes, with preliminary veterinary studies also suggesting a potential role in the disease in dogs.3
In human beings, ELISA is not considered to be useful for detection of insulin autoantibodies and a more sensitive liquid phase radioimmunoassay is performed. Selected canine serum samples (n = 30) were tested against human insulin using this technique and revealed a large amount of nonspecific reactivity in both control and diabetic samples (data not shown). This made interpretation of results difficult, with no significant difference seen between controls and diabetic dogs. Because autoantibodies can be more species specific than antibodies induced by injection of an exogenous protein,20 a liquid phase assay using radio-labelled porcine insulin rather than human insulin might be preferable for detection of such autoantibodies in dogs.
Insulin autoantibodies in humans can precede clinical signs of diabetes by several years, although not all antibody-positive patients will progress to overt disease.21 In the nonobese diabetic mouse, a rodent model of type 1 diabetes, high insulin autoantibody concentrations can be correlated with an increased risk of progression to disease.22 Using the 95% confidence limit, anti-insulin reactivity was detected in 4 of 120 control patients. So far, these dogs have not progressed to developing diabetes, although these might simply represent false positives of the assay.
In the small number of newly diagnosed diabetic dogs that were retested following initiation of insulin therapy, seroconversion occurred in 3 of 6 (50%) of dogs treated with Insuvet lente but none of the Caninsulin-treated dogs. AIA in human beings are usually produced within 2–4 months of initiating insulin therapy, reaching a plateau at around 6 months but persisting as long as the same type of insulin is administered.23 A similar pattern of reactivity was seen in the current study. Those dogs with no evidence of insulin reactivity at 3 months did not make AIA even at later time points (Fig 1). Because these data suggested that Insuvet lente is more immunogenic than Caninsulin in dogs, it was decided to evaluate a larger group of treated diabetic patients.
The majority of diabetic and control dogs demonstrated reactivity to the positive control CDV antigen, because most had been vaccinated. Four of 220 diabetic and 5 of 120 control dogs demonstrated reactivity to canine TG, used as an autoantigen control. None of the animals recruited into the study had been reported as suffering from thyroid disease at the time of sampling. However, serum thyroxine and thyroid-stimulating hormone concentrations were not measured. Thyroglobulin autoantibodies in canine serum have been associated with an increased risk of thyroid dysfunction,24 but anti-TG autoantibodies have also been documented in dogs with nonthyroidal illness.25,26 Diabetic dogs with concurrent hypothyroidism could be considered to have disease equivalent to human autoimmune polyendocrine syndrome type 2.27 However, confirmed hypothyroidism is rare in the diabetic dog population currently enrolled in the RVC Canine Diabetes Database and Archive (3 out of 900), and results from the current study suggest that TG autoantibodies are no more prevalent in the diabetic population than in nondiabetic animals.
Diabetic dogs that had received Caninsulin demonstrated less serological response to insulin than those that had received Insuvet lente. When anti-insulin IgG was quantified, the concentrations of AIA measured in the Caninsulin-treated dogs were also lower than those treated with Insuvet lente. This supports the hypothesis that porcine insulin is less immunogenic than bovine insulin in canine diabetic patients. However, other components of the commercial insulin preparations might also differ between the 2 products, which might influence the immunogenicity of the insulin protein. It is tempting to speculate that the anti-porcine insulin antibodies seen in Caninsulin-treated dogs are more indicative of pancreatic autoimmunity rather than a response to therapy, because these are, in effect, autoantibodies.
Insuvet PZI was found to be the most immunogenic of the insulin preparations evaluated and induced the highest concentrations of AIA. The difference between the 2 bovine insulin preparations (lente versus PZI) lies in the formulation in which the insulin is suspended. Insuvet PZI contains zinc and protamine, to promote slow release of insulin from a depot, prolonging the duration of insulin action. Protamines are cationic fish chromosomal proteins that retard absorption of insulin from the injection site.28 Insuvet lente is a shorter acting insulin zinc preparation, lacking protamine. Protamine itself is known to be immunogenic in humans,29 and it would be revealing to measure anti-protamine antibodies in Insuvet PZI-treated dogs. As well as acting as an adjuvant, chemical linkage of protamine to insulin has the potential to allow protamine-reactive T cells to drive insulin-reactive B-cell antibody responses.
Not all diabetic dogs treated with bovine insulin developed AIA. Results from the longitudinal study indicated that some dogs completely fail to respond serologically to administration of Insuvet lente, suggesting that there is a state of immunological tolerance to insulin that is not broken by exposure to xenogeneic antigen. It is also possible that some dogs become AIA positive but then peripheral tolerance mechanisms act to regulate immune reactivity and antibody levels fall. Because the AIA detected were of the IgG isotype(s), this infers that there was interaction between B cells and CD4+T-helper cells. Antigen presentation of insulin would be required for B cells to recruit T-cell help and class-switch to IgG. It is possible that dog leukocyte antigen genes, encoding MHC Class II molecules that are responsible for presentation of peptide epitopes, play a role in determining whether or not an AIA response is made.
The insulin subunit specificity of AIA was also investigated in this study. Most AIA-positive diabetic samples showed no reactivity to the individual bovine insulin B or A subunits, so it is likely that these antibodies were directed against conformational rather than linear epitopes. This is consistent with the specificity of human AIA.30 There were more anti-B chain than anti-A chain reactors, even though it is the A chain that differs between bovine and canine insulin. Insulin self-reactive B cells are clearly present, but will only become activated following processing and presentation of insulin-derived peptides. In the case of porcine insulin, only self-peptides can be generated and there is likely to be T-cell tolerance to these. In contrast, following processing of bovine insulin, foreign A-chain peptides can be presented, resulting in T-cell activation. Such activated autoreactive T cells can drive the production of antibodies by B cells that could recognize any part of the insulin molecule. Thus, it is not necessarily B cells that discriminate between bovine and porcine insulin, but rather T cells, reactive against bovine A-chain peptide(s), that could be responsible for driving AIA responses.
Previous studies of canine IgG isotypes in various diseases have revealed conflicting results.18,31,32 Much seems to depend upon the secondary antibodies used for detecting canine IgG1 or IgG2, with results obtained using polyclonal antisera not necessarily consistent with those where monoclonal antibodies have been used.33 Thus, in the current study, both methods were used to evaluate serum AIA isotypes. Furthermore, to avoid quantification errors associated with differences in the affinities of the secondary antibodies, AIA (IgG1 and IgG2) were calculated as a ratio in relation to the total serum antibody for each isotype.
Both isotype ELISA techniques indicated that AIA were enriched for IgG1, which is similar to that seen in human beings, where antibodies induced by insulin therapy are predominantly IgG1 whereas insulin autoantibodies are predominantly IgG2.30 Although the general conclusions from the isotype analysis of AIA are similar, it is a concern that there are such discrepancies between the results obtained using the 2 methods, with poor correlation between the data sets. Further investigation is required to ensure validity of results generated using either the commercial or in-house assays.
The clinical significance of AIA in dogs remains to be determined. In human diabetic patients, AIA can compromise glycemic control by inducing insulin resistance,34,35 with a change of insulin type leading to a reduction in antibody concentration and improvement of glycemic control.11 Glycemic instability was reported in 1 canine study in association with the presence of AIA,b although the current data suggest that there are many stable canine diabetic patients with circulating AIA. This study was not designed to test the relationship between AIA and glycemic control; however, there was no association between AIA concentration and the dose of insulin required or between anti-insulin reactivity and fructosamine concentrations (data not shown). These findings suggest that the relationship between AIA and glycemic control requires further investigation and might not be straightforward or predictable.
The current study has demonstrated that insulin autoantibodies can be detected in a small number of diabetic dogs, although the significance of this finding in terms of the pathogenesis of β cell dysfunction/deficiency remains to be established. Treatment of diabetic dogs with Insuvet lente or PZI is more likely to stimulate AIA than treatment with Caninsulin, although these do not seem to be clinically important in the majority of dogs. Studying immune responses to exogenously administered insulin in dogs might be useful for investigating the immunogenetics of tolerance and immune reactivity to a defined peptide antigen.
aHarb-Hauser, M, Nelson RW, Gershwin L, et al. Prevalence of anti-insulin antibodies in diabetic dogs. Proceedings of the American College of Veterinary Internal Medicine forum 1998, Vol. A61: 213 (abstract)
bFerasi, K, Fuimicelli M, Bernardini D, et al. Detection of anti-insulin antibodies in diabetic dogs after insulin treatment using a solid phase enzyme immunoassay. British Small Animal Veterinary Association Congress 1999 Proceedings: 258 (abstract)
cCatchpole B, Ristic, J, Herrtage ME. Cross-reactivity of antibodies to bovine and porcine insulin in diabetic dogs. British Small Animal Veterinary Association 2001 Proceedings: 550 (abstract)
dFeldman EC, Nelson RW, Karam JH. Reduced immunogenicity of pork insulin in dogs with insulin dependent diabetes mellitus. Diabetes 1983; 32 (Suppl 1): 153A, (abstract)
eGlucometer Espirit, Bayer, Newbury, UK
fCDV Antigen, Churchill Applied Biotechnology, Huntingdon, UK
gCanine thyroglobulin autoantibody immunoassay kit, Oxford Laboratories, Oxford, MI
hMaxi-sorp, Nunc, Paisley, UK
iSigma, Poole, UK
jAmerican Peptide Company, Sunnyvale, CA
kBethyl Laboratories, Montgomery, TX
lMarvel, Premier Beverages, Stafford, UK
mTitertek, Huntsville, AL
This study was funded in part by Intervet Pharma R&D (Europe). LJD was supported by a Royal Veterinary College Clinical Research Fellowship.
The authors are grateful to Dr Ezio Bonifacio, University of Milan, for testing selected samples by a liquid phase radioimmunoassay to human insulin.
The authors would also like to thank the veterinary surgeons who submitted samples for this study.
- 1Diabetes mellitus. In: FeldmanEC, NelsonRW, eds. Canine and Feline Endocrinology and Reproduction. Philadelphia, PA: WB Saunders; 1996:339–391.,