A Pilot Study Comparing the Diabetogenic Effects of Dexamethasone and Prednisolone in Cats
Fourteen cats received either daily prednisolone (4.4 mg/kg per os [PO]) or dexamethasone (0.55 mg/kg PO) for 56 days. These doses were clinically equipotent. Serum fructosamine and urine glucose were measured on days 0, 28, and 56. Insulin sensitivity, glucose tolerance, and peak insulin secretion were measured in each group prior to and at the end of the courses of glucocorticoid administration. On day 56, the prevalence of glucosuria was significantly greater (P=0.027), and a trend was seen toward greater fructosamine concentrations (P=0.083) in dexamethasone-treated cats compared to prednisolone-treated cats. The results of this pilot study also showed a trend toward a greater decrease in insulin sensitivity (P=0.061) and a significantly lower compensatory increase in insulin secretion (P=0.081) in the dexamethasone-treated cats than in cats administered prednisolone. These preliminary data suggest that dexamethasone exhibits greater diabetogenic effects in cats than equipotent doses of prednisolone. Further study is justified to support this hypothesis.
Introduction
Diabetes mellitus is one of the most common endocrine disorders of the cat, with an estimated incidence of between one in 50 to one in 400 cats.1,2 Feline diabetes mellitus has been likened to type II diabetes in humans, a condition in which hyperglycemia results from resistance of the peripheral tissues to the effects of insulin. Reduced sensitivity to insulin may initially be compensated for by increased insulin secretion from the pancreatic β-cells; however, with time, β-cell exhaustion and a resultant insulin-dependent state may arise.2,3
Glucocorticoids are one of the most commonly prescribed classes of drugs in veterinary medicine, but they carry the risk of significant side effects.4 A wide variety of synthetic glucocorticoids are available for use, yet little is known about the safety profiles of these drugs in cats. In comparison with other species, cats are considered relatively resistant to many of the deleterious effects of glucocorticoids, and other investigators have claimed that cats require higher doses of glucocorticoids than dogs to achieve equivalent effects.4,5 This may in part be related to the lower number of glucocorticoid receptors that have been documented in the skin and liver of cats.6 Nonetheless, glucocorticoid side effects, such as a decreased tolerance to glucose, do occur in cats.7 Some authors believe that cats are particularly predisposed to this glucocorticoid-induced hyperglycemia.4 Decreased glucose tolerance can unmask or exacerbate a preexisting diabetic state in susceptible cats. This is supported by the fact that 80% of cats with hyperadrenocorticism develop diabetes mellitus.8
Insulin sensitivity, defined as the capacity of insulin to mediate the disposal of glucose, is a marker of insulin resistance, which is a feature of both human and feline diabetes mellitus.9 Decreased insulin sensitivity values have also been observed in diabetic cats, and documentation of these decreases can be used to detect preclinical diabetes mellitus.10 The two most widely accepted methods for measurement of insulin sensitivity are the euglycemic clamp and minimal model analysis of the frequently sampled intravenous glucose tolerance test (FSIVGTT).10–12 The FSIVGTT differs from the standard intravenous glucose tolerance test (IVGTT) in that more frequent glucose and insulin measurements are made through the test. The euglycemic clamp is a labor-intensive protocol wherein blood glucose is held constant by a variable-rate infusion of glucose, while a constant-rate infusion of insulin is provided.9 Under these conditions, the rate of uptake of glucose by the body is equal to the glucose infusion rate. Minimal model analysis provides an alternative protocol that relies on the mathematical modeling of glucose and insulin concentrations during a FSIVGTT to arrive at a calculated insulin sensitivity value. In cats, insulin sensitivity estimates derived from minimal model analysis were shown to correlate highly with those obtained from the euglycemic clamp (correlation coefficient [r] = 0.91).9 A recent study suggested that basal insulin concentrations can also be of use in comparing insulin sensitivity between groups of cats.11
Glucose tolerance is the ability to dispose of an oral or intravenous (IV) load of glucose; both the sensitivity of tissues to insulin and the ability of the pancreas to secrete insulin are taken into account. Glucose tolerance has been determined in cats based on glucose half-life (t½), the disappearance of glucose (Kglucose), and absolute glucose concentrations at various time points throughout an IVGTT.13–15 In addition to their effects on insulin resistance, glucocorticoids have been shown to target the pancreatic β-cell and inhibit insulin secretion directly in rodent models.16 Arginine is a more potent insulin secretagogue than glucose in cats, and it may be more appropriate for assessing β-cell function in this species.17
Prednisolone and dexamethasone are two of the most commonly used glucocorticoids in small animal practice, and some clinicians believe that adverse events are more common with dexamethasone.18 Because of the high incidence of diabetes mellitus in the feline population and the widespread usage of glucocorticoids in cats, it is important to determine which drug, if any, is more diabetogenic. The goal of this study was to determine whether a significant difference exists in the diabetogenic potential of dexamethasone versus prednisolone in cats when the drugs are given at the doses commonly recommended in practice and considered equipotent in terms of their antiinflammatory effect. We hypothesized that dexamethasone would cause significantly greater changes consistent with a prediabetic or diabetic state than an equipotent dose of prednisolone.
Materials and Methods
Animals
This study was reviewed and approved by the Institutional Animal Care and Use Committee. Fourteen healthy, intact male, purpose-bred domestic shorthair cats were obtained from a research facility.a Animals were cared for according to the principles outlined in the National Institutes of Health Guide for Care and Use of Laboratory Animals. All cats were between 9 and 14 months of age. The mean body weight of the cats was 4.7 kg (range 4.4 to 5.2 kg), and body condition ranged from three to four using a five-point body condition scoring system.
Experimental Protocol
All cats were acclimated to their environment and diet for 2 weeks prior to any testing. All cats were assessed as healthy based on physical examination, complete blood cell count (CBC), serum biochemical analysis, and urinalysis. The cats were housed individually, and dry commercial cat foodb was fed once daily. Water was available ad libitum. The 14 cats were randomly assigned to two groups of seven. No significant differences in age, weight, or body condition were seen between the two groups. For accurate dosing, prednisolone and dexamethasone were prepared at 50 mg/mL and 5 mg/mL, respectively, in an oral fish-flavored suspension. Cats in the prednisolone group received 4.4 mg/kg per os [PO] q 24 hours, while cats in the dexamethasone group were administered 0.55 mg/kg PO q 24 hours. Both corticosteroids were given for 56 days. Prior to drug administration (day 0), urine glucose and serum fructosamine concentrations were assessed. On day -2 the IVGTT was performed, and on day 0 the arginine stimulation test was performed. Urine glucose and fructosamine concentrations were again assessed at day 28 and day 56, while the IVGTT and arginine stimulation test were repeated on days 56 and 58, respectively. Urine was obtained as a free-flow sample using nonabsorbent litter.c Serum glucose concentrations and CBCs were also repeated on day 56.
Intravenous Glucose Tolerance Test
All cats were anesthetized with ketamine and Valium 2 days prior to testing, and jugular catheters were placed for blood sampling and drug administration. All cats were fasted for 24 hours prior to testing. A bolus of 0.5 g/kg of glucose was given over 30 seconds into the jugular catheter and then flushed with 2 mL of saline. Blood samples were collected for glucose and insulin measurement prior to glucose administration and at 5, 10, 15, 30, 45, 60, 90, and 120 minutes. Glucose was measured using a hand-held glucose meterd on whole blood, after which plasma was separated and frozen for subsequent insulin assay. Hand-held portable blood glucose meters have been used and evaluated extensively in veterinary cases.19–22 The blood glucose values measured by the glucose meter used in this study agreed closely with those values obtained using the standard hexokinase method (data not shown). Multiple repeated measurements on various feline blood samples were also performed, and the coefficients of variation were found to vary from 0.9% to 3.9%, demonstrating a high level of precision (data not shown). Insulin was measured using an insulin radioimmunoassay kite previously validated for use in the cat.23
Arginine Stimulation Test
Two days after the IVGTT (on days 0 and 58 of the study), an arginine stimulation test was performed following a 24-hour fast. Jugular catheters were kept patent between tests by flushing them with heparinized saline once daily. On the day of testing, 0.1 g/kg of arginine was injected through the jugular catheter, which was then flushed with 2 mL of saline. Blood samples for glucose and insulin determination were collected (as previously described) prior to injection of arginine and at 2, 4, 6, 8, 10, 20, and 30 minutes following arginine injection. Peak insulin concentration was defined as the highest concentration of insulin detected during the arginine stimulation test.
Insulin Sensitivity
Two indices of insulin sensitivity were assessed. The first was determined using the MINMOD Millennium computer programf and minimal model analysis of the IVGTT. The second was the basal insulin value that was calculated as the average insulin value, prior to glucose or arginine injection, during the IVGTT and arginine stimulation test, respectively.
Glucose Tolerance
Glucose tolerance was determined using the t½ of glucose, Kglucose, and whole blood glucose concentrations at 0, 60, 90, and 120 minutes during the IVGTT. Linear regression analysis of the semilogarithmic plot of glucose concentration versus time during the IVGTT was used to calculate the t½. The disappearance coefficient of glucose (K) value was calculated as K = (0.693/t½)100.14,15
Statistical Analysis
Using a multivariate repeated measures analysis of variance (ANOVA),g mean fructosamine concentrations on days 0, 28, and 56, as well as changes from days 0 to 28, 28 to 56, and 0 to 56 were compared between the treatment groups. The following were also compared between the treatment groups, using a multivariate repeated measures ANOVAg: insulin sensitivity from minimal model analysis; basal insulin concentrations; glucose t½; Kglucose; glucose concentrations during the IVGTT on days -2 and 56; peak insulin concentrations on days 0 and 58; and changes from days -2 or 0 to days 56 to 58 in these parameters (depending on the test). Using an exact test of the difference between two binomial proportions,h the prevalence of glucosuria was compared between groups. Statistical significance was accepted at P<0.05 for all comparisons.
Results
The results of the fructosamine measurements are given in Table 1. No significant difference in fructosamine concentrations was seen between the prednisolone- or dexamethasone-treated cats on day 0 (P=0.90) or day 28 (P=0.38). By day 56, a trend toward higher mean fructosamine concentrations was observed in the dexamethasone-treated cats (P=0.083). A significant increase in the fructosamine concentration occurred in all cats over time (P<0.001), and the magnitude of the increase in fructosamine concentration was significantly greater in the dexamethasone-treated cats compared to the prednisolone-treated cats (P=0.047).
The results of the urine glucose measurements are given in Figure 1. None of the cats in either group displayed glucosuria at day 0. On day 28, 29% of cats administered prednisolone and 71% of cats administered dexamethasone had glucosuria; however, this difference was not significant (P=0.090). On day 56, the prevalence of glucosuria was significantly greater in the dexamethasone-treated cats (P=0.027) than in the cats administered prednisolone.
Table 2 shows the results of the insulin sensitivity measurements. When estimated from minimal model analysis, insulin sensitivity was significantly lower in the prednisolone-treated cats than in the dexamethasone-treated cats on day 0 (P=0.013). On day 56, no significant difference in insulin sensitivity was present between the two groups (P=0.88). In addition, a significant decrease in insulin sensitivity was observed among all cats from day 0 to day 56 (P=0.0011), and a trend toward a greater decrease in insulin sensitivity occurred in the dexamethasone-treated cats (P=0.061).
Table 3 shows the results of the basal insulin measurements. The mean basal insulin concentration was lower in the dexamethasone-treated cats than in the prednisone-treated cats on day 0, although this difference was not statistically significant (P=0.052). No difference in mean basal insulin values was seen at day 56 (P=0.54). In all cats, the mean basal insulin concentration increased from day 0 to day 56 (P<0.001), and no difference was seen between the groups in the way the mean basal insulin concentration changed over time (P=0.66).
The results of the IVGTT are given in Tables 4 and 5. No significant difference was seen between the two treatment groups in any of the parameters used to assess glucose tolerance at either time point. Mean glucose t½ increased significantly (P=0.0050), and mean Kglucose decreased significantly (P=0.0035) from day -2 to day 56, but no difference existed between groups in terms of the way these parameters changed (P=0.46 and 0.42, respectively). Mean glucose concentrations increased significantly in all cats from day 0 to day 56 (P<0.001), and no significant difference existed between groups in terms of the way mean glucose concentrations changed from day -2 to day 56 (P=0.61).
During the arginine stimulation test, the mean peak insulin concentration in the dexamethasone-treated cats was 52.2 μU/mL, while that of the prednisolone-treated cats was 92.7 μU/mL on day 0. This difference was statistically significant (P=0.0030). On day 58, the mean peak insulin concentration in the dexamethasone-treated cats was 106.2 μU/mL, while that of the prednisolone-treated cats was 191.0 μU/mL. This difference was also statistically significant (P=0.0082). The mean peak insulin concentration increased from day 0 to day 58 in both groups of cats (P<0.001), and a lower increase in peak insulin concentration was seen in the dexamethasone-treated cats—although this difference did not reach statistical significance (P=0.081).
Discussion
The most striking differences between the two treatment groups in our study were in the parameters most commonly used in clinical practice: fructosamine concentrations and glucosuria. Serum fructosamine concentrations indicate glycemic control over the previous 2 to 3 weeks.24 A trend toward higher fructosamine concentrations was seen in the dexamethasone-treated cats on day 56 compared to the prednisolone-treated cats. Similarly, on day 56, a significantly greater number of cats in the dexamethasone-treated group displayed glucosuria. Worth noting is that although glucose was present in the urine of eight cats at day 56, the serum glucose concentrations (as measured by both the blood glucose meter and the laboratory using the standard hexokinase method) did not exceed the published renal threshold of 290 mg/dL in any cat.13 Although glucocorticoid-associated glucosuria is often attributed to hyperglycemia, evidence has shown that in other species urinary excretion of glucose is often higher than would be expected from serum glucose levels in patients treated with glucocorticoids, and these drugs may selectively influence the glucose reabsorption system in the kidney.25 The differences in the prevalence of glucosuria between the two groups in this study may therefore reflect not only differences in glucose tolerance, but also the degree to which these steroids affect renal glucose transport systems. This effect is important to keep in mind in order to interpret urinalyses correctly while monitoring cats treated with glucocorticoids.
Compared to cats in the prednisolone group, the dexamethasone-treated cats had significantly greater insulin sensitivity estimates, as well as lower basal and peak insulin concentrations prior to any drug administration (day 0). The reason for these differences prior to treatment was not obvious. Cats had been randomly assigned by arbitrary selection to their respective treatment groups, and no pretreatment difference in age, weight, body condition, or other parameter was known to be associated with impaired insulin sensitivity or secretion. Considerable inter-day variation has been shown in measurements of feline insulin sensitivity, and the pretreatment differences were likely due to random variation and also to the well-known imprecision of nonlinear modeling.26 Because of this unexpected pretreatment difference, the amount of change in each group between day 0 and day 56 was also compared. Dexamethasone caused a greater decrease in insulin sensitivity, as estimated from minimal model analysis, and a lesser increase in peak insulin secretion—though neither of these differences achieved statistical significance. In fact, given that the fructosamine concentrations were greater in the dexamethasone-treated cats on day 56, one would expect a greater increase in insulin secretion in order to compensate; yet, in reality, the opposite was seen. Despite greater fructosamine concentrations, the dexamethasone-treated cats were unable to mount greater compensatory increases in insulin secretion. It is difficult, however, to determine the level of influence that treatment and the observed difference in baseline values had on the magnitude of the changes.
Minimal model analysis has been evaluated by several investigators as a means of determining insulin sensitivity in cats; however, the values obtained are usually generated from a FSIVGTT, or insulin-modified FSIVGTT, rather than the standard IVGTT used here.9,10,26,27 During a FSIVGTT, blood samples for insulin and glucose evaluation are taken much more frequently following the glucose bolus than was done in this study.9,10 Nonetheless, MINMOD was able to model successfully the abbreviated data provided in this study and to generate insulin sensitivity values similar to those reported in other studies. While the use of abbreviated data could increase the variance of the estimates of the parameters, the estimates would not be systematically increased (or decreased) in one group. Additionally, minimal model analysis of IVGTTs has been used in human models, and the results have correlated well with results obtained using FSIVGTTs.28
Although many indices of glucose tolerance were assessed, the only parameter that approached a significant difference between the two groups was the glucose concentration at time 0 on day 56. This value was higher in the dexamethasone-treated cats. On day 56, the glucose concentration at time 0 was above previously reported reference ranges, indicating impaired glucose tolerance in the dexamethasone-treated cats but not in those cats treated with prednisolone.13,14 Additionally, the mean serum glucose concentration, as determined by the laboratory hexokinase method, was greater in the dexamethasone-treated cats on day 56 (data not shown). With increased serum glucose concentrations, a compensatory increase in insulin secretion is expected. Under normal conditions, one would therefore expect a similar or greater peak insulin value in the dexamethasone-treated cats at day 56, given that fructosamine and basal glucose levels were highest in this group. Pancreatic β-cells, however, have been shown to be directly inhibited by glucocorticoids in vivo, and in this study, dexamethasone-treated cats had both a lower mean peak insulin response on day 56 and a lesser compensatory increase in mean peak insulin concentration from day 0 to day 56.16 These results suggest a greater suppression of β-cell function in the dexamethasone-treated group.
The doses of glucocorticoids used in this study are considered immunosuppressive in cats, and they were chosen to maximize the chances of detectable differences in a healthy feline population.4,29,30 Whether similar differences would exist between prednisolone and dexamethasone at lower doses is unknown. Possibly at the high doses used here, the diabetogenic effect was so potent in both groups that the ability to detect a difference was impaired. Similar studies using antiinflammatory doses may provide further insight into this matter.
The differences detected in this study might not reflect a greater diabetogenic effect of dexamethasone, but perhaps the doses of dexamethasone and prednisolone selected were not equipotent in terms of the glucocorticoid effects. Dexamethasone has been suggested to be from 6.25 to 10 times more potent than prednisolone, depending on the species.4,5,29–32 In humans, the affinity of dexamethasone for the glucocorticoid receptor has been shown to be 6.25 times greater than that of prednisolone, and this relationship between the potencies of the two drugs has been verified by more recent pharmacokinetic and pharmacodynamic studies. 33–35 No similar information has been published on this topic in cats, and doses are largely extrapolated from clinical experience. In humans, the relative potency of these drugs has also been shown to vary with the target tissue.31 Since the diabetogenic effect of a glucocorticoid is an unwanted side effect, the doses chosen for use in practice are typically based upon the drug’s antiinflammatory potential. Since no previous studies have been performed to document what constitutes an equipotent dose of dexamethasone and prednisolone in cats, we chose in this study to base “equipotency” on the relative antiinflammatory nature of dexamethasone and prednisolone. Dexamethasone was therefore assumed to be eight times more potent than prednisolone, which is in agreement with many previously published estimates and, most importantly, reflects the comparison commonly used in practice. 4,5,29,30,32 However, because this is the first study to evaluate the relative potency of these two drugs against any body system in cats, an additional interpretation of these data could be that dexamethasone is more than eight times more potent than prednisolone. Further studies on the relative antiinflammatory potential of the two drugs would be required to verify or disprove this hypothesis.
An additional concern is that the duration of action for these glucocorticoids also differs. In a recent study, the elimination t½ of serum prednisolone concentrations in cats was shown to be 0.66 hours, while to our knowledge the serum t½ of dexamethasone in cats has not been determined. 36 After binding to their receptors, glucocorticoids induce many of their effects through the modification of gene transcription; because of this, the biological t½ is more important than the plasma t½. The biological t½ correlates with the t½ of induced proteins and allows categorization of these drugs into short-, intermediate-, and long-acting compounds. 37
Based upon work done in humans, dexamethasone is considered a long-acting glucocorticoid, with a duration of action between 35 and 72 hours. Prednisolone is considered an intermediate-acting glucocorticoid, with a duration of action between 12 and 36 hours.30,38 In cats, no studies on the exact duration of action for prednisolone or dexamethasone have been performed; however, in one study, dexamethasone was shown to suppress cortisol to less than detectable levels for 32 hours.39 The duration for which a glucocorticoid suppresses the hypothalamic-pituitary-adrenal axis parallels the duration of its antiinflammatory activity. 37,40 This information, along with clinical experience, suggests that dexamethasone and prednisolone might also be considered long- and intermediate-acting glucocorticoids, respectively, in cats. It is therefore possible that some of the differences in insulin and glucose homeostasis detected in this study, where dexamethasone and prednisolone were both given once daily, reflect differences in the duration of activity for these drugs. The main goal of this study was to determine whether differences occur at commonly used doses, and both of these drugs are commonly given once daily for immunosuppressive doses;37,41 therefore, once-daily administration was deemed an appropriate initial dosing regimen to study.
While the absorption and bioavailability of oral prednisolone have been studied in cats, few studies exist on the pharmacokinetics of oral dexamethasone in this species.36 In one previous study, the absorption of oral dexamethasone was shown to be highly variable between cats, although some of this variation was attributed to expectoration of the dose (a problem that was not seen in the current study).42 Unfortunately, the pharmacokinetics of the two drug formulations administered in the current study were not directly assessed. While the doses given may reflect how these drugs are administered clinically, future studies should take into account possible variations in absorption and pharmacokinetics that may contribute to observed differences.
A P value of <0.05 is often used as the cutoff for determining statistical significance; however, this value is not appropriate for all cases and should not arbitrarily be used without regard for the design and type of experiment performed. 43 The chances of committing a type II error exist in the current study, because a small number of cats were used, and parameters shown to be associated with a high degree of variability were measured. While statistical adjustments for multiple comparisons can be useful, their cost is the increase in frequency of incorrect statements that assert no relation between two factors.44 As this study was preliminary in nature, adjustments for multiple comparisons were not used because of the potential to convert associations that may have justified further investigation to those less worthy of attention.44 Though many of the differences detected in the current study did not achieve statistical significance, several important trends were observed that may have biological significance. These trends justify further studies using larger numbers of cats to help strengthen the findings of this pilot study.
Although the inclusion of a negative control group would have been ideal to assess the effects of any handling stress on the cats, we believe this would have had minimal impact on the end results. The treatments administered were eagerly accepted by the cats and not seen as an additional source of stress. Previous work has shown that significant changes in most of the parameters assessed in this study do not occur in untreated cats when tested under similar conditions.27 Additionally, any stresses induced by the housing, testing, or treatments used in this study would be equivalent between the two treatment groups. Since the primary goal of this study was to compare the effects of two treatments to each other, we believed a comparison of solely prednisolone to dexamethasone would yield meaningful results.
Conclusion
The results of this study support the hypothesis that dexamethasone, when given at doses reported to be equipotent to prednisolone, may be a more potent inducer of a prediabetic or diabetic state than an equivalent dose of prednisolone. Our data suggest that dexamethasone treatment, in comparison to prednisolone, results in the following: greater fructosamine concentrations; greater decreases in insulin sensitivity; a lesser degree of insulin secretion from pancreatic β-cells in the face of higher glucose and fructosamine concentrations; and a greater prevalence of glucosuria. Further work is needed to verify these findings in a larger and more widely varied population of cats. Knowledge of this difference should aid the clinician in selecting glucocorticoids for use in the cat.
Liberty Research Inc., Waverly, NY 14892
Harlan Cat Diet; Harlan Co., Madison, WI 53713
Nosorb; Catco Inc., Cape Coral, FL 33990
OneTouch Ultra; Johnson & Johnson, Milpitas, CA
Coat-A-Count Insulin RIA; DPC, Los Angeles, CA 90045
MINMOD Millenium; Minmod Inc., Los Angeles, CA
Systat v 11.01; Systat Inc., Point Richmond, CA 94804
StatXact v. 7; Cytel, Cambridge, MA 02139
Acknowledgments
The authors thank Sandra Grable, CVT; Dr. Kenneth Keppel; Dr. Erin Lecky; Alyssa Galligan, CVT; and Tara Maggio, CVT, for their technical assistance.
Citation: Journal of the American Animal Hospital Association 45, 5; 10.5326/0450215
Prevalence of glucosuria.
