Glucagon therapy in canines with an insulinoma: A retrospective descriptive study of 11 dogs

Abstract

Canine insulinomas are uncommon neoplasms, which often result in refractory hypoglycemia. Glucagon is one readily available treatment for insulin-induced hypoglycemia. The primary objective of this study was to evaluate blood glucose trends and outcome (survival to discharge versus death or euthanasia) for dogs with insulinoma that were treated with glucagon. Secondary objectives included the description and influence of other variables such as abnormalities on diagnostic tests, physical examination abnormalities, concurrent administration of dextrose and/or glucocorticoids, and seizures. The median glucagon constant rate infusion dose was significantly higher for the non-survivors than for survivors. No other correlation was found between any of the independent variables evaluated when comparing blood glucose trends, length of hospitalization, and outcome. The main conclusion of the study is that glucagon therapy in insulinomas is an effective treatment to manage hypoglycemia.

Introduction

Insulinoma is a beta cell neoplasm of the pancreas that results in insulin-induced hypoglycemia (1,2). The function of the endocrine pancreas is to secrete hormones directly into the circulation. The hormones originate from the Islets of Langerhans, which contain 3 cell types: alpha, beta, and delta cells (1). Alpha cells account for 25% of total islet cells and secrete glucagon (1). Beta cells account for 60% of the islet cells and secrete insulin (1).

In healthy dogs, insulin secretion is inhibited once the blood glucose levels reach 4.44 mmol/L (3). Insulinomas secrete insulin independently from the circulating blood glucose levels and are relatively uncommon in dogs (1,2). Most canine insulinomas are malignant, commonly spreading to regional lymph nodes and the liver. Metastasis to the lung is rare (1,3).

The average age in dogs with insulinomas is 9 y (1). Insulinomas tend to occur in large breed dogs with golden retrievers, Labrador retrievers, boxers, German shepherds, Irish setters, standard poodles, and mixed breed dogs being overrepresented (1,4). Clinical signs are variable, but include weakness, collapse, ataxia, disorientation, and seizures (1). Clinical signs are associated with neuroglycopenia, as the central nervous system uses glucose as its sole energy source (1).

Suspicion of insulinoma is raised when a patient has fasting hypoglycemia (blood glucose < 3.33 mmol/L) with concurrent hyperinsulinemia (> 138.9 nmol/L) (5). Complete blood (cell) count (CBC), serum chemistry, and radiographic imaging are largely non-specific (1). The CBC is typically normal (6). In addition to hypoglycemia, serum chemistry may show elevations in alkaline phosphatase (ALP) and alanine aminotransferase (ALT) (6). Abdominal and thoracic radiographs often show no significant abnormalities in the absence of metastasis (6). Abdominal ultrasound has been shown to identify pancreatic masses in less than 50% of cases (6). Other diagnostic tools include advanced imaging such as computed tomography (CT), or magnetic resonance imaging (MRI) to identify a pancreatic nodule (3).

Medical management of insulinomas is aimed at increasing blood glucose levels and include the use of glucagon, glucocorticoids, and dextrose supplementation (2,4). Other medical interventions include treatment with diazoxide and streptozotocin (4). In an acute hypoglycemic crisis dextrose can be administered (1,6,7). However, a complication of administering dextrose is rebound release of insulin from the pancreas causing worsening hypoglycemia (6).

Prognosis varies and depends on the clinical stage of the disease, with higher stages of metastatic disease having a poorer prognosis compared to patients without signs of metastatic changes (6). Medical therapy alone is reported to result in a median survival time of 196 d (8). Almost 50% of dogs with metastatic changes typically succumb to the disease within 6 mo (6).

The primary objective of this study was to evaluate blood glucose trends and outcome (survival to discharge versus death or euthanasia) for insulinoma patients that were treated with glucagon. Secondary objectives included the description and influence of other variables such as concurrent administration of dextrose and/or glucocorticoids, physical examination abnormalities, seizures, and abnormalities on diagnostic tests. We hypothesized that administration of glucagon continuous rate infusion (CRI) would improve systemic blood glucose concentration in the current patient population.

Materials and methods

Electronic medical record databases of 2 veterinary referral specialty hospitals were used to search for medical records between December 2011 and September 2017 which contained the keywords glucagon, insulin:glucose (I:G) pair, and insulinoma. Inclusion criteria were dogs with hypoglycemia (defined as a blood glucose < 3.33 mmol/L), a recorded I:G pair consistent with an insulinoma (defined as blood glucose < 3.33 mmol/L with concurrent hyperinsulinemia > 138.9 nmol/L) and treatment with glucagon. The following data were collected from medical records for patients that met the inclusion criteria:

• signalment,

• presenting complaint(s),

• presence of seizures,

• examination findings,

• results of biochemical parameters,

• I:G pair,

• blood glucose concentrations (Accu-check glucometer; Roche Diagnostics, Indianapolis, Indiana, USA) prior to and during treatment,

• doses of glucagon CRI,

• additional therapies such as glucocorticoids and/or dextrose, histopathology or cytology (if available),

• length of hospitalization and outcome.

Descriptive statistics were calculated for all clinical and laboratory data. Statistical analyses were conducted using SPSS v.24 software (IBM Statistics for Windows Version 24.0; IBM Corp, Armonk, New York, USA). Binary logistic regression was used to assess the relationship of initial blood sugar, duration of time on glucagon, highest glucagon CRI dose, elevations in ALP and ALT, and the presence of seizures on the likelihood of survival to discharge. Pearson’s product-moment correlation coefficient was used to test for an association between initial blood glucose and time on a glucagon CRI as well as elevations in ALP and ALT and length of time on a glucagon CRI. Fisher’s exact test was used to assess correlation between ALP and ALT and metastasis to the liver. A least squares linear regression was used to assess for a relationship between highest blood glucose concentration and dose of glucagon. A Mann-Whitney U-test compared time to euglycemia between patients that received dextrose and glucagon versus patients that just received glucagon. A P < 0.05 was considered statistically significant.

Results

A total of 13 dogs met the inclusion requirements. Two patients were excluded due to incomplete records. Of the remaining 11 patients, 6 were spayed females and 5 were castrated males. The average age was 9.9 y (range: 7 to 13 y). Breeds represented included mixed breed dogs (n = 3), American pitbull (n = 2), and 1 each of bullmastiff, Boston terrier, Kerry blue terrier, bichon frise, Labrador retriever, old English bulldog, and golden retriever. There was no gender or breed predilection.

The most common presenting complaints were seizures (n = 7), weakness (n = 3) and referral for confirmed hypoglycemia (n = 2). Seizure activity was reported in 8/11 patients. All seizure activity had started within 72 h of presentation. A total of 7/11 patients had at least 1 seizure during treatment (range: 0 to 9 seizures). Seizures did not correlate with patient outcome (P > 0.21).

All patients had normal heart rate, temperature, and respiratory rate on arrival (defined as heart rate 80 to140 beats/min, respiration rate 24 to 46 breaths/min, and rectal temperature 37.6°C to 38.8°C). Abnormalities found on examination were weakness (n = 5), a mammary mass (n = 1), and airway stertor (n = 1). Examination was reported as normal in 3/11 patients. On presentation, no patients had active seizures nor were they in the postictal period.

All patients had an I:G pair which was consistent with an insulinoma (Table 1). The average serum insulin level was 638.04 pmol/L (reference range: 52.09 to 138.9 pmol/L; range: 354.2 to 1024.39 pmol/L). The average glucose measurement during submission of the I:G pair was 1.81 mmol/L (range: 1.22 to 2.16 mmol/L) (Table 1).

Table 1.

Summary of blood glucose (BG_{t = 0}) values on presentation, serum insulin values and insulin glucose pair results.

Patient	BGt = 0 (mmol/L)	Insulin values (pmol/L)	Glucagon CRI starting dose (ng/kg BW/min)	BGmax (mmol/L)	BGmin (mmol/L)	Duration of administration of glucagon CRI (hours)	Outcome
1	1.77	424.3	15	6.33	1.33	47	D
2	1.83	563.2	10	12.34	1.77	40	D
3	1.88	354.1	4.4	5.71	1.83	150	E
4	1.94	708.3	6.0	5.44	1.61	60	D
5	1.94	379.8	15	10.06	1.77	154	E
6	2.66	645.2	15	3.33	1.51	8	E
7	1.28	375.0	10.4	11.83	2.27	82	D
8	2.22	711.8	25	12.22	1.78	36	D
9	1.66	729.9	13.8	7.94	1.44	46	E
10	2.00	1024.3	10	8.83	2.28	64	D
11	1.55	972.3	9.4	9.16	1.44	84	D

BW — body weight; BG_max — highest blood glucose concentration recorded during glucagon administration; BG_min — lowest blood glucose concentration recorded during glucagon administration; CRI — constant rate infusion; E — euthanasia; D — survival to discharge.

Initial CBC and serum chemistry parameters were available for all patients. Complete blood (cell) count was within reference ranges for all patients. The most common abnormality on serum chemistry was hypoglycemia. The median presenting blood glucose concentration was 1.82 mmol/L (range: 1.27 to 2.67 mmol/L) detected by handheld glucometer. No significant association was found between the presenting blood glucose reading and the likelihood of survival (P > 0.21).

A total of 3/11 patients had elevated ALP (> 140 IU/L) (median: 131 IU/L; range: 226 to 365 IU/L). There was no correlation between elevated ALP and number of hours receiving glucagon (P = 0.312). All 3 patients with elevated ALP had liver histopathology performed. No correlation was found between the ALP and metastasis to the liver. Elevation in ALP did not correlate with survival to discharge (P = 0.9).

A total of 3/11 patients also had an elevated ALT (> 120 IU/L) on initial serum chemistry (median: 105.5I U/L; range: 121 to 336 IU/L). There was no correlation between elevated ALT and number of hours receiving glucagon (P = 0.837). All 3 patients with elevated ALT had liver histopathology performed. Of those patients, 2/3 had evidence of metastatic changes to the liver on histopathology. There was no correlation between elevated ALT and metastasis to the liver (P > 0.99). The ALT value also did not correlate with survival to discharge (P = 0.63).

Thoracic radiographs were taken in 9/11 patients. Of those 9, 1 had an enlarged sternal lymph node and another had a cranial thoracic mass. An abdominal ultrasound examination was completed in 9/11 patients. Of those, 3 identified the presence of a pancreatic nodule. Other abnormalities seen on ultrasound included enlarged lymph node, hepatomegaly, and cystoliths (n = 1 each).

Six patients underwent an exploratory laparotomy with subsequent histopathology. All of the patients had pancreatic biopsies submitted and 4/6 included liver samples. One patient had a fine-needle aspirate of the liver. All of the histopathology or cytology results from liver and pancreatic samples were compatible with islet cell carcinoma (insulinoma).

All patients were placed on a glucagon CRI. The average blood glucose concentration at the time of glucagon administration was 2.36 mmol/L (range: 1.3 to 4.88 mmol/L). No correlation was identified between initial blood glucose reading and duration of glucagon CRI (P = 0.716). The lowest blood glucose reading after initiation of the glucagon CRI was 1.85 mmol/L (range: 1.33 to 2.77 mmol/L). All patients had at least 1 episode of hypoglycemia (defined as blood glucose concentration < 3.33 mmol/L), while 10/11 patients experienced hyperglycemia (defined as a blood glucose concentration > 6.67 mmol/L) after initiation of glucagon CRI (average: 8.95 mmol/L, range: 5.44 to 12.3 mmol/L). There was no significant linear association between highest recorded blood glucose reading and corresponding glucagon CRI dose (P = 0.353). The mean length of time from initiation to euglycemia and highest blood glucose from initiation of the glucagon CRI was < 1 and 22 h, respectively. The average frequency of blood glucose monitoring was 1.8 h (range: 1 to 6 h).

The starting glucagon CRI average dose was 11.8 ng/kg body weight (BW) per minute (range: 4.4 to 25 ng/kg BW per minute) and this dose did not correlate with survival to discharge (P = 0.953). The median maximum rate of infusion of the glucagon CRI was 10 ng/kg BW per minute (range: 6 to 34 ng/kg BW per minute and was significantly higher for non-survivors (P = 0.042). The mean length of time on the glucagon CRI was 70.18 h (range: 9 to 150 h). There was no correlation between duration of time on the glucagon CRI and outcome (P = 0.21)

A total of 4/11 patients had dextrose supplementation (range: 2.5% to 5.0% solution) in addition to glucagon CRI. The average blood glucose concentration of the patients which had a dextrose CRI at any time during their glucagon infusion was 4.16 mmol/L, while for patients which never received dextrose during their glucagon infusion this value was 3.95 mmol/L. There was no difference in the time to euglycemia (P > 0.99) or duration of glucagon CRI (P = 0.573) between those patients receiving glucagon alone versus those receiving glucagon and dextrose.

Prednisone was administered to 7/11 patients. The average dose of prednisone administered was 0.91 mg/kg BW per day (0.3 to 2.0 mg/kg BW per day). There was no significant difference between use of prednisone when comparing time to euglycemia (P = 0.601) or dose of glucagon CRI (P = 0.573). There was no significant difference between prednisone use and duration of glucagon infusion for groups which received glucagon and dextrose supplementation (P = 0.601).

The average number of days in hospital for all patients was 5.5. A total of 4/11 patients were euthanized. The reasons for euthanasia included post-surgical complications (1 with diabetes, 1 with aspiration pneumonia), lack of response to treatment (refractory hypoglycemia, n = 1), and identification of more than 1 type of neoplasia (mammary mass, n = 1). No patients died of natural causes. No adverse reactions pertaining to glucagon administration were noted for any patients. The average hospitalization time for survival versus non-survival groups was 4.71 d (range: 3 to 10 d) and 3.5 d (range: 1 to 6 d), respectively. Hospitalization time between survivors and non-survivors did not differ significantly (P = 0.222).

Discussion

Glucagon infusion was successful in raising blood glucose concentrations in patients diagnosed with insulinoma. Consistent with other publications, the average age of patients herein was 9.9 y and there was no breed or gender predilection (1,3,6,9).

The duration of clinical signs prior to diagnosis has been documented to be variable with ranges between 1 d and 3 y (3,6). In our study, clinical signs were reported to be more acute (typically within 72 h), possibly because some clinical signs of insulinomas are mild and gradual. Owners may rationalize initial symptoms as age-related. The most common presenting complaint in this study was the presence of new seizures, which has been documented to occur in 63% of patients (1). Owners may perceive the sudden onset of seizures as the only clinical sign, and therefore report a more acute onset of symptoms. Weakness was the second most common presenting complaint which has also been noted to occur in up to 40% of patients (10). This exemplifies the effects of hypoglycemia including neuroglycopenia and peripheral polyneuropathy (1,3,6). Interstitial brain glucose concentrations are 20% to 30% less than that of plasma, but the brain has 3 times the metabolic rate of peripheral tissues, making the central nervous system prone to hypoglycemia-induced cellular damage (11). Initial clinical signs include weakness and tremors (11). As hypoglycemia progresses to 1.0 mmol/L severe signs such as seizure activity, brain damage, and even death occur (11). Compensatory pathways to increase peripheral blood glucose levels occur by increasing endogenous levels of glucagon, catecholamines, and cortisol with suppression of endogenous insulin (11). Intracranial compensation arises from changes in cerebral blood flow and glucose delivery by upregulation of glucose transporters (11).

The peripheral polyneuropathy is possibly due to the effects of hyperinsulinemia on the peripheral nerves (1,10). Elevated insulin levels prevent the nerves from utilizing fatty acids and amino acids as energy and make them reliant on glucose as the sole energy source (12). However, the extent of the peripheral neuropathy and duration or severity of the hypoglycemia are not correlated (1,9,11).

Diagnosis of an insulinoma can be established by an I:G pair that shows hypoglycemia with inappropriately elevated insulin levels. During hypoglycemia, endogenous pancreatic insulin concentrations should be low or non-detectable due to negative feedback (1). A previous study has documented that an insulin level > 417 pmol/L was sufficient to confirm an insulinoma in dogs (13). Laboratory interpretation guidelines state normal (reference range: 7 to 139 pmol/L) or elevated insulin concentration in the presence of hypoglycemia is supportive of an insulinoma. All patients in the current study had an insulin level > 139 pmol/L.

In this study, elevations in ALT and/or ALP (1,5,6) did not correlate with survival to discharge or length of time of glucagon CRI. There are no studies with which to compare this finding. Unlike in the present study, a correlation between elevations in ALP and ALT with metastasis to the liver has been reported in 1 study (5). The most common areas of metastasis include the regional lymph nodes and liver, with 46% to 75% having signs of metastatic changes seen on exploratory laparotomy (9,14). Metastatic spread can hinder gluconeogenesis and glycogenolysis (10,15). However, the duration and dose of glucagon CRI can be altered to maintain euglycemia. This was not found herein possibly due to the small sample size.

Radiographs are of limited diagnostic value for diagnosis of insulinomas (1,10). Beta cell secreting tumors typically are < 3 cm in diameter at the time of diagnosis, do not produce a mass effect within the cranial abdomen, and rarely will cause pulmonary metastatic changes until much later in the disease course (1,10). No patients in this study had evidence of metastatic spread on thoracic radiographs but the sensitivity of radiographic detection of metastatic changes to visceral organs is only 18% (9). Therefore, patients in this study may have had metastatic spread that was not evident on radiographs. No patients underwent advanced imaging such as CT or MRI.

In this patient population abdominal ultrasound may visualize a mass(es) within the pancreas and also determine if there are metastatic changes in other organs. Three of nine pancreatic nodules were identified on abdominal ultrasound. Ultrasonography for detection of insulinomas has reported low specificity, detecting insulinomas in 46% to 75% of cases (9). Detection of pancreatic nodules was reported in only 5/13 and 12/16 dogs in 2 prior studies (16,17). Limitations of transabdominal ultrasound include user experience, patient body condition score, patient cooperation, and size of the tumor (1,5). Abdominal ultrasound is useful and a recommended diagnostic tool to rule out other causes of hypoglycemia and can be used to determine if there is metastatic spread (1,10).

Glucagon is an insulin antagonistic hormone that causes an increase in blood glucose by increasing glycogenolysis, promoting gluconeogenesis, increasing ketogenesis, increasing lipolysis, and decreasing glycogenesis (2,5,18). Once in circulation, the peak plasma concentration of glucagon occurs within 20 min of administration (19). Previous studies show that maximum glucose levels occur within 40 to 80 min after glucagon administration (1,6). In the current study, euglycemia was achieved in < 1 h after initiation of the glucagon CRI.

Side effects of glucagon infusion in humans include dizziness and nausea. In a study in dogs, subcutaneous administration of glucose caused mild sedation (19). Other side effects of glucagon administration include diarrhea, ataxia, and tachypnea (19,20,27). The current study did not show any side effects of glucagon administration.

In the current study, the average glucagon CRI dose was higher in non-survivors. High serum insulin levels at the time of initiation of glucagon and a delay in glucagon administration have been shown to decrease the efficacy of glucagon (21,22). In addition, prolonged hyperinsulinemia can lead to a depletion of glycogen storage in the liver, inhibiting glycogenolysis (15,21,23). Furthermore, dogs with normal or elevated glucose concentrations have a significantly better prognosis compared with dogs that had persistent hypoglycemia (24).

Our data showed that a much lower glucagon CRI dosing range was needed to maintain euglycemia compared with that in a previous study (25). This led the clinician to consider starting the glucagon CRI at a lower dose. Also, refractory hyperglycemia that can be seen with higher glucagon CRI doses may be avoided (2).

Dextrose can be administered for short-term use and initial stabilization if the patient is showing signs of hypoglycemia such as seizures, trembling, weakness, or abnormal mentation (1,9,16,24). Dextrose can stimulate further insulin secretion resulting in rebound hypoglycemia and should therefore be used with caution (1,2,9,24). In the current study a lack of refractory hyperglycemia could have been due to the concurrent administration of glucagon.

Most patients in the current study were administered glucocorticoids concurrently. Glucocorticoids increase blood glucose levels by antagonizing insulin at a cellular level, increasing hepatic gluconeogenesis, elevating glucose 6-phosphate activity, decreasing tissue uptake of glucose, and stimulating endogenous glucagon release (24). In a previous retrospective study, there was no significant increase in blood glucose levels in patients receiving prednisone until a glucagon CRI was instituted, which is consistent with the current findings (25).

All non-survivors herein were euthanized at owners’ request due to other comorbidities including post-surgical complications, lack of response to treatment, and identification of more than one type of neoplasia. Long-term outcome could not be assessed due to lack of follow-up, as only the medical records were reviewed. Most patients were seen several years before data collection and long-term communication with the owners or referring veterinarians was not documented within the medical records, limiting long-term evaluations.

Due to the retrospective descriptive nature of this study many limitations are noted. Keyword search of the medical record database may have missed patients that were coded incorrectly. Many patients had to be excluded due to incomplete records or lack of inclusion criteria. These exclusions can lead to type II error, making it difficult to draw any statistical conclusions. Furthermore, insulinomas are an uncommon neoplasm making acquisition of a large population of patients difficult. The blood glucose monitoring devices also varied among hospitals and hematocrit values were not taken into consideration when measuring spot blood glucose readings (26). Abnormally low or high hematocrit readings can affect blood glucose values with anemic patients having a falsely higher blood glucose reading and patients with hemoconcentration reading falsely low glucose values (26). Another limitation to the study is that nearly half of the patients were euthanized. Treatment regimens were variable and based on doctor discretion. Limitations related to treatments include the use and timing of adjunctive therapies (glucocorticoids, dextrose, and chemotherapeutic agents), starting dose, and time to initiation of the glucagon CRI.

In conclusion, the current retrospective observational study shows that glucagon CRI is an effective medication to stabilize blood glucose concentrations in patients with an insulinoma. Due to the small sample size, further conclusions could not be made pertaining to the more effective dosing of glucagon, concurrent prednisone administration, dextrose administration, and outcome. Further studies are warranted with a larger patient population and standardization of the glucagon therapy to make definitive recommendations. CVJ