Article Text


Glucagon stimulation test for estimating endogenous insulin secretion in dogs
  1. T. Fall, DVM1,
  2. B. Holm, DVM2,
  3. Å. Karlsson, BSc1,
  4. K. M. Ahlgren, MSc3,
  5. O. Kämpe, MD, PhD3 and
  6. H. von Euler, DVM, PhD1
  1. 1 Department of Clinical Sciences, Swedish University of Agricultural Sciences (SLU), SE-750 07 Uppsala, Sweden
  2. 2 Regiondjursjukhuset Blå Stjärnan, Gjutjärnsg 4, SE-417 07 Göteborg, Sweden
  3. 3 Department of Medical Sciences, University Hospital, Uppsala University, SE-751 85 Uppsala, Sweden


Fifty-one dogs (27 diabetic dogs, four that had recovered from diabetes and 20 healthy control dogs) were given 0·5 or 1·0 mg glucagon intravenously. Blood samples were taken before the injection and 10 and 20 minutes after it. Samples were analysed to determine C-peptide, insulin and glucose concentrations, and one sample from each dog was analysed for fructosamine. The median (interquartile range) concentrations of C-peptide in the samples taken at 10 minutes were 0·5 (0·3 to 0·8) nmol/l in the control dogs, 0·1 (0 to 0·2) nmol/l in the diabetic dogs, and 0·3 (0·2 to 0·4) nmol/l in the dogs that had recovered from diabetes. Seven of the 51 dogs showed mild adverse reactions after the injection of glucagon.

Statistics from

DIABETES mellitus is a common disease in dogs but its underlying causes are not fully understood. Its aetiology is considered to be multifactorial and may be broadly divided into insulin deficiency and insulin resistance (Catchpole and others 2005). The dog has recently emerged as an ideal species for the mapping of disease genes in complex disorders owing to the unique structure of its genome (Sutter and others 2004, Lindblad-Toh and others 2005) and the existence of more than 300 breeds, each of which can be regarded as a well defined genetic isolate. The incidence of diabetes in the different breeds differs widely, suggesting that there may be important genetic factors in its aetiology (Catchpole and others 2005, Fall and others 2007). As a better classification is essential for genetic studies aimed at identifying the genes responsible for the disease, the role of the glucagon stimulation test in the subclassification of diabetes in dogs was investigated.

Diabetes due to insulin deficiency is considered to be caused by the autoimmune destruction of insulin-producing β-cells, pancreatitis or islet hypoplasia (Kramer 1981, Hoenig and Dawe 1992, Cook and others 1993). Insulin-resistant diabetes may occur as a result of hormonal disturbances; for example, hyperadrenocorticism and progesterone-induced acromegaly (Eigenmann and others 1983, Peterson and others 1984). If the cause of the insulin resistance is treated, the dog may go into remission. However, untreated insulin-resistant diabetes can lead to insulin deficiency because of chronic hyperglycaemia (Catchpole and others 2005) that can induce permanent β-cell dysfunction (Imamura and others 1988).

In human beings, several different types of diabetes have been described, and the capacity for the secretion of insulin varies with the type. The immunological destruction of β-cells is the crucial factor leading to type 1 diabetes (Bennet and Knowler 2005) and the resulting failure of insulin secretion. In patients with type 2 diabetes some residual capacity for the secretion of insulin is common, and its extent can be fairly accurately estimated by the measurement of C-peptide. Neither type 2 diabetes nor the much rarer maturity-onset diabetes of the young, a monogenic form of the disease caused by the inheritance of mutations in any one of six known genes (Olek 2006), has yet been observed in dogs. In human beings diabetes can also develop as a result of pancreatitis or pancreatic surgery, so-called secondary diabetes.

C-peptide is released in equimolar concentrations with insulin when proinsulin is cleaved in β-cells (Kaneko 1989), but because it has a negligible hepatic extraction and a much longer half-life than insulin (Polonsky and others 1984) it is a better marker of endogenous insulin secretion than insulin. Moreover, exogenous insulin does not interfere with the analysis, because it does not contain any C-peptide. However, exogenous insulin could reduce glucose levels and change the ratio of insulin to C-peptide indirectly. In most instances, the stimulated secretion of insulin and C-peptide provides more information than basal secretion (Madsbad and others 1981).

The glucagon stimulation test for evaluating residual β-cell function was first proposed by Faber and Binder (1977) and has been widely used to assess endogenous insulin secretion for clinical and research purposes in human medicine (Scheen and others 1996). The C-peptide response to an intravenous bolus injection of 1 mg glucagon correlates well with the β-cell response to mixed meals and other stimuli commonly used to characterise endogenous insulin secretion (such as oral or intravenous glucose or arginine), and has the advantages of having a shorter duration than other stimuli and being easily standardised (Faber and Binder 1977, Scheen and others 1996).

However, the glucagon stimulation test has not been used extensively in veterinary medicine. Montgomery and others (1996) studied 42 diabetic dogs stimulated with glucagon and Watson and Herrtage (2004) have used the test to assess β-cell function in dogs with chronic pancreatitis. In first line veterinary practice, no assessment of insulin secretion is made during the diagnosis of diabetes mellitus in dogs.

The purpose of this prospective study was to investigate whether the glucagon stimulation test could be an effective and safe method for estimating insulin secretion in healthy and diabetic dogs, and be useful in clinical work and genetic studies.


Between January and October 2006, an intravenous glucagon stimulation test was applied to 51 dogs. The dogs were privately owned and brought to the five participating clinics by their owners on the day of the test.

The 51 dogs were divided into three groups. Twenty-seven of them were diabetic; their mean age was 8·8 years, with a range from 3·1 to 12·2 years, and 23 of them had been treated with intermediate acting porcine insulin (Caninsulin; Intervet) or recombinant human nph insulin (Insulatard; Novo Nordisk) for between three weeks and 2·4 years before entering the study. Two of the diabetic dogs also had hyperadrenocorticism and three were being treated for hypothyroidism. The other four dogs had not been given insulin treatment. The diagnosis of diabetes in each dog was based on its clinical signs, including polyuria and polydipsia, a persistent fasting hyperglycaemia of more than 8 mmol/l and glycosuria. A second group consisted of four females, with a mean age of 11·8 years (range 9·4 to 13·9 years), that had undergone remission from diabetes after ovariohysterectomy three weeks to 3·7 years before the study; three of them had been treated with insulin. The third group consisted of 20 healthy control dogs with a mean age of 3·7 years (range 0·8 to 9·6 years). The inclusion criteria for these control dogs were that they were clinically normal, had received no medication during the previous three months and had normal fasting serum glucose concentrations. The distribution of breeds in the three groups is shown in Table 1.

Glucagon stimulation test

Food and insulin were withdrawn for at least 10 hours before the test was carried out. A bolus of biosynthetic glucagon (Glucagon; Novo Nordisk) was administered within a period of two minutes via the cephalic vein. Dogs weighing less than 10 kg received a dose of 0·5 mg glucagon and dogs weighing 10 kg or more received a dose of 1 mg glucagon. Blood samples were taken into serum tubes before the glucagon was administered and 10 and 20 minutes after it had been administered. Within an hour after the first sample had been taken the samples were centrifuged and the serum was separated and divided into two tubes. One tube was sent unfrozen by overnight mail to the Laboratory of Clinical Pathology, slu, and the other tube was stored at −20°C and transported to the lab on dry ice at the end of the study. The veterinarian performing the test was instructed to record any adverse effects on the dogs.

The stability of C-peptide, with or without the addition of the protease inhibitor aprotinin was determined in a small scale experiment in which three fresh samples were each divided into four tubes. The concentration of C-peptide was measured in the fresh samples and in samples stored for 24 hours and 48 hours at room temperature.

The treatments and sampling of the dogs were approved by the owners and conformed to the decision of the Swedish Animal Ethical Committee (C267/5) and the Swedish Animal Welfare Agency (2005-2038).

Analytical methods

Serum glucose concentrations were determined by a glucose hexokinase method (Thermo Clinical Labsystem) in the samples transported on dry ice. Serum C-peptide concentrations were measured in the samples sent by post and in the samples transported on dry ice, with a commercial dog-specific radioimmunoassay, which does not cross-react with insulin, glucagon or proinsulin (Canine C-Peptide ria; Linco Research). The manufacturer's instructions were followed accurately, except that the protease inhibitor aprotinin (Trasylol; Bayer) was added just before the analysis instead of immediately after sampling.

The detection limit of C-peptide in serum was 0·05 nmol/l and samples below the detection limit were assumed to have a concentration of 0·025 nmol/l for the statistical analysis. The intra-assay repeatability was assessed with two pools of sera from dogs with low (0·42 nmol/l) and high (1·76 nmol/l) C-peptide concentrations; the respective intra-assay coefficients of variation for two extracted samples of each pool were 7·6 per cent and 4·1 per cent respectively. In the 14 assay runs, the interassay coefficient of variation of the serum pools of the dogs was 7·0 per cent. Fructosamine was measured in one sample from each dog by a colorimetric assay, based on the ability of ketoamines to reduce nitrotetrazolium blue formazans in an alkaline medium (reference range 200 to 400 μmol/l) (Pentra Fructosamine; abx Diagnostics). Serum insulin was measured with a sandwich elisa with monoclonal human antibodies (Insulin elisa; Merdocia). The validity of the assay was assessed in terms of its linearity with samples of different concentrations. A serum sample from a dog with a high concentration of insulin (560 pmol/l) was diluted with the manufacturer's blank sample down to 8·75 pmol/l (1/64). The absorbance results from the diluted sera were plotted on a curve and compared with the calibration curve established with the samples of known insulin concentration supplied by the manufacturer. The curves were parallel up to the concentration of the test sample (560 pmol/l).

Data analysis

For all the statistical analyses, the results from the samples transported on dry ice were used, except for the comparison of the samples sent by mail and the samples transported on dry ice. Descriptive statistics (median and interquartile ranges) were calculated for the measurements of C-peptide, glucose and insulin. Values of P<0·05 were considered statistically significant. The Levene test (Levene 1960) showed that the variances of the C-peptide concentrations of the different groups were not equal (P<0·001), and non-parametric statistics were therefore used. Box plots for the concentrations of C-peptide and glucose at the three time points were constructed. A Kruskall-Wallis test was used to assess whether the median values of the measurements of C-peptide and glucose at 0 and 10 minutes differed significantly between the three groups. Pair-wise comparisons with Wilcoxon's rank sum test were made if the Kruskall-Wallis test revealed significant differences (Conover 1999). A Wilcoxon rank sum test was also used to evaluate the differences in the concentration of C-peptide at 10 minutes between male and female dogs with diabetes and also between dogs that had been diagnosed with the disease less than three months earlier and dogs that had had the disease for longer periods. Spearman's correlations were used to evaluate the correlations between the concentrations of insulin and C-peptide in the dogs in the study that had not received insulin treatment and those that had. For this analysis, only data from the 10 minute samples were used, resulting in 26 samples from dogs not previously treated with insulin and 21 from insulin-treated dogs (four values were missing because there was too little serum for the insulin elisa).

To assess whether transport by mail at room temperature or on dry ice influenced the results, samples were compared by the Wilcoxon signed rank test, for which 57 samples from 19 dogs were available. The statistical calculations were made using jmp software, v 5.1 (sas Institute).


The median and interquartile ranges of the serum concentrations of C-peptide, glucose, fructosamine and insulin for the three groups of dogs are shown in Table 2. Boxplots of the concentrations of C-peptide and glucose are shown in Figs 1 and 2. There were significant differences between the groups in the concentrations of C-peptide (P<0·0001) and glucose (P=0·007) before (0 minutes) and 10 minutes after the administration of glucagon. Pair-wise comparisons indicated that there were significant differences at 0 minutes in glucose and C-peptide between the diabetic and control dogs (C-peptide P<0·001, glucose P<0·0001), and in glucose between the recovered and diabetic dogs (P=0·02) but not in C-peptide. There were no significant differences in C-peptide or glucose concentrations between the control dogs and the recovered dogs. At 10 minutes there were significant differences in C-peptide and glucose between the diabetic and control dogs (C-peptide P<0·001, glucose P<0·0001), between the recovered and diabetic dogs (C-peptide P=0·020, glucose P=0·019), but not between the control dogs and the recovered dogs.

At 10 minutes the diabetic dogs that had been diagnosed in the previous three months had a median C-peptide concentration of 0·13 nmol/l (interquartile range 0·03 to 0·35) compared with 0·06 nmol/l (0·03 to 0·10) for the dogs that had been diagnosed more than three months earlier; the difference was not significant. The male diabetic dogs had a median C-peptide concentration of 0·07 (0·03 to 0·10) and female dogs had a median concentration of 0·10 (0·07 to 0·20); the difference was not significant.

Spearman's correlation between the 10-minute insulin and C-peptide concentrations in 21 untreated dogs (16 control dogs and five diabetic dogs) was 0·91 (P<0·0001), and in 26 treated dogs it was 0·41 (P=0·04) (Fig 3).


Box plots of the serum concentration of C-peptide before (0 minutes) and 10 and 20 minutes after the intravenous administration of a bolus dose of glucagon to 20 healthy control dogs, 27 diabetic dogs and four dogs that had recovered from diabetes


Box plots of the serum concentration of glucose before (0 minutes) and 10 and 20 minutes after the intravenous administration of a bolus dose of glucagon to 20 healthy control dogs, 27 diabetic dogs and four dogs that had recovered from diabetes


Relationships between the serum concentrations of insulin and C-peptide 10 minutes after the intravenous administration of a bolus dose of glucagon to (a) 21 dogs that had not been treated with insulin and (b) 26 dogs that had been treated with insulin

The mean C-peptide concentration of the 54 blood samples (from 19 dogs) that were sent by mail was significantly higher (P=0·001) than that of the aliquots sent on dry ice, but the difference was only 0·05 nmol/l, with a range from 0 to 0·16 nmol/l.

Seven of the 51 dogs showed adverse side effects after the administration of glucagon: four of them were diabetic dogs, two had recovered from diabetes and one was a control dog. Three of the dogs (a basenji, a giant schnauzer and an Irish setter) had one episode of diarrhoea each, minutes after the glucagon injection; two (a poodle and a dachshund) developed ataxia, one (a dachshund) showed signs of fatigue and one dog had an increased respiratory rate. In each case these clinical signs disappeared within a few minutes, without medical intervention.

There were no substantial differences between the concentrations of C-peptide in the fresh samples of serum and in the samples stored with or without aprotinin after 24 and 48 hours at room temperature. Aprotinin was added immediately before the analysis if it had not been added before the samples were stored.


A better classification of diabetes mellitus in dogs is important in clinical practice and for future genetic studies of the genes responsible for the disease. The results of this study show that the glucagon stimulation test is a safe and simple method for estimating residual insulin production.

On the basis of the results of earlier studies, blood samples were taken 10 and 20 minutes after the administration of glucagon. In dogs the concentrations of C-peptide and insulin peak at approximately 10 minutes and glucose peaks at about 20 minutes after the administration of glucagon (Montgomery and others 1996). These authors showed that the measurement of plasma C-peptide concentration could differentiate healthy dogs, dogs with impaired β-cell function, that is, with diabetes mellitus, and dogs with increased β-cell responsiveness to glucagon, that is, insulin resistance. The results of the present study support these findings and indicate that most dogs diagnosed with diabetes mellitus are insulin deficient.

Faber and Binder (1977) tested a group of healthy people and found a mean C-peptide concentration of 1·28 nmol/l (range 0·9 to 1·9) after stimulation with glucagon. Montgomery and others (1996) tested 24 healthy dogs; their mean concentration of C-peptide increased from 0·17 nmol/l to 0·57 nmol/l. In the present study the 20 control dogs had a median C-peptide concentration of 0·51 nmol/l after stimulation. These results indicate that healthy dogs may have a lower peripheral concentration of C-peptide after stimulation with glucagon than people, and a lower cut-off concentration should therefore be considered for diagnosing insulinopenia in dogs.

In the study by Montgomery and others (1996), a few diabetic animals had a low residual C-peptide secretion. In the present study the owners of the two female diabetic dogs with the highest C-peptide concentrations at 10 minutes (0·43 and 0·73 nmol/l) were contacted at the end of the study and the dogs were found to be in a non-insulin-dependent state. Both dogs had had an ovariohysterectomy before the test and both had had diabetes for a short period. It would be of interest to study the insulin secretion of intact female dogs when they are diagnosed with diabetes to determine whether a high C-peptide concentration could predict remission.

There was a statistically significant difference between the initial median concentrations of C-peptide in the three groups. However, there was a big overlap between the individual concentrations of the dogs in the groups. This overlap was much smaller in the samples taken 10 minutes after the administration of glucagon, indicating that it is necessary to perform the test to evaluate insulin secretion in individual dogs. The basal secretion of insulin/C-peptide is not sufficient to assess β-cell function in individual dogs.

In the four dogs with transient diabetes that were no longer insulin-dependent before the study, the 10-minute concentration of C-peptide was significantly greater than in the other diabetic dogs. However, studies of more dogs would be needed to draw any conclusions.

It has been suggested that in human beings glucagon induced insulin secretion might be markedly affected by blood glucose levels. Acute hypoglycaemia inhibits and acute hyperglycaemia potentiates post-glucagon secretion of insulin (Samols and others 1966, Oakley and others 1972). In cases of chronic hyperglycaemia, it has been suggested that glucotoxicity reduces the secretory function of β-cells (Unger and Grundy 1985). In this study, the dogs were fasting and their insulin treatment had been withdrawn for at least 10 hours before the glucagon stimulation test. The median glucose concentration of the diabetic dogs before the test was 17·9 mmol/l (interquartile range 8·4 to 21·5) and that of the control dogs was 4·7 mmol/l (4·5 to 5·1). The median fructosamine concentration of the diabetic dogs was moderately high, 539 μmol/l, compared with a reference range of 200 to 400 μmol/l. The high value may indicate an inadequate control of blood glucose (Nelson 2005).

The hyperglycaemia was probably partly chronic and partly acute in character and may have changed the C-peptide response of the diabetic dogs. It might have been preferable to have performed the test with the dogs in a normoglycaemic state, induced by treating them with exogenous insulin.

In the dogs that had not been treated with insulin, the concentrations of C-peptide and insulin at 10 minutes correlated well. The correlation was poor in the dogs already treated with insulin. This implies that measurements of insulin can only be used instead of C-peptide in the glucagon stimulation test when testing dogs not already on insulin replacement therapy.

In human beings, there are adverse side effects after the administration of glucagon in less than one in 10,000 cases, nausea and vomiting being the most common (Davis 2006). In dogs it has been suggested that doses as low as 0·01 mg/kg can cause transient tachycardia (Chernow and others 1986). Mochiki and others (1998) found that in dogs glucagon affects duodenal and jejunal contractions, which may be responsible for its gastrointestinal side effects. Seven of the 51 dogs suffered mild adverse side effects, diarrhoea being the most common, probably because of these gastrointestinal effects. The signs of ataxia and fatigue suffered by a few of the dogs were possibly caused by tachycardia or nausea.

A small-scale stability study had shown that the concentration of C-peptide did not change significantly during storage whether or not the protease inhibitor aprotinin was added to the samples. It was therefore decided not to add aprotinin to the samples before they were stored.

The samples sent by post that had been stored at room temperature for one to two days, had a slightly, but statistically significant, greater concentration of C-peptide than the samples stored at −20&deg;C. The samples transported on dry ice had been stored at −20&deg;C for between one and 10 months before their transport and the C-peptide might therefore have been affected. The differences in C-peptide concentration were small and for clinical use of negligible importance, showing that it is possible to send samples by normal overnight mail, thus extending the clinical usefulness of the glucagon stimulation test.

Most of the dogs diagnosed with diabetes were insulin deficient. One exception appeared to be newly diagnosed intact females, and further studies are required to investigate whether high C-peptide concentrations in a newly diagnosed female are indicative of transient diabetes. In genetic studies and in clinical work, the glucagon stimulation test may be valuable for distinguishing subtypes of diabetes. The measurement of C-peptide instead of insulin is particularly useful in dogs that have been treated with insulin.


This study was supported by grants from the Foundation for Research, Agria Insurance and the Swedish Research Council. The authors thank the participating veterinary clinics of Nordvärmlands smådjurspraktik, Regiondjursjukhuset Strömsholm, Regiondjursjukhuset Bagarmossen, Landskrona Smådjurspraktik, Regiondjursjukhuset Blå Stjärnan (Göteborg) and the University Animal Hospital, Uppsala, Sweden. They also thank Inger Lilliehök, Nils Fall and Agneta Egenvall, slu, for their intellectual help.


View Abstract

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.