Materials and Methods

Horses—Twenty-four adult mares from the University of Tennessee teaching and research herd were included in the study. The study was performed during 5 months from September 2007 through January 2008. Horses were admitted to the University of Tennessee Veterinary Teaching Hospital in groups of 4 and remained there for 22 days. At the conclusion of the study, horses were returned to the research herd. To eliminate differences attributable to sex, only mares were used. Mean ± SD weight of horses upon admission was 519.4 ± 36.3 kg. The age of horses ranged from 5 to 16 years (mean ± SD, 10.4 ± 3.6 years); breeds included Quarter Horse (n = 11), Quarter Horse–Tennessee Walking Horse crossbred (10), Tennessee Walking Horse (1), Standardbred (1), and Thoroughbred (1). Body condition scores²⁰ on a scale of 1 to 9 ranged from 4 to 6. The study protocol was approved by the University of Tennessee Institutional Animal Care and Use Committee.

Experimental design—By use of a completely randomized design, horses were assigned to 1 of the 3 groups, each containing 8 horses. A levothyroxine group received 48 mg of levothyroxine sodium powder^a mixed with water and 200 g of oats for 15 days, a dexamethasone group received 20 mg of dexamethasone^b (2 mg/mL of injectable solution) orally via syringe and 200 g of oats for 15 days, and a control group received only oats. All horses underwent an endotoxin challenge following pretreatment. Horses entered the study in groups of 4, and the order was control (n = 4), levothyroxine (4), dexamethasone (4), levothyroxine (4), control (4), and dexamethasone (4). As horses entered the study, they were weighed and physical examinations were performed on day 1. Each horse was then housed separately in a 3.7 × 3.7-m stall in the veterinary teaching hospital. Grass hay was provided to horses in amounts equivalent to 2% of body weight, and horses had continual access to water. Horses were acclimated to their new environment for approximately 72 hours before procedures were initiated. On day 4, a 14-gauge IV catheter^c was inserted into the left jugular vein and left in place for approximately 24 hours. A baseline FSIGTT was performed on day 5 starting at 9:00 AM. The pretreatment period began the next day, which was day 6 of the study, and continued for 15 days. Treatments were administered between 7:00 AM and 9:00 AM. On day 20, horses were weighed, and IV catheters were inserted and left in place for approximately 48 hours. The posttreatment FSIGTT procedure was performed on day 21 starting at 9:00 AM. On that day, treatments were administered at 12:00 PM, after completion of the posttreatment FSIGTT. An LPS bolus infusion was administered IV at 1:00 PM on the same day. The post-LPS FSIGTT procedure was performed the following day starting at 9:00 AM, which was day 22 of the study and 20 hours after LPS was administered. Horses received tapering doses of their respective treatments after the conclusion of the study. The dexamethasone dosage was lowered to 8 mg administered once daily for 2 days, then 4 mg administered once daily for 2 days, and finally 2 mg administered 48 hours apart for 2 doses. Levothyroxine was administered at a dosage of 24 mg once daily for 3 days and then 12 mg once daily for 4 days.

LPS administration—Escherichia coli O55:B5 LPS^d was mixed with 30 mL of sterile saline (0.9% NaCl) solution under a fume hood with gloves, and a respirator was worn to minimize exposure. The LPS solution (20 ng/kg) was infused via the IV catheter during a 15-minute period.

CBC analysis—To evaluate individual responses to endotoxin, CBCs were performed in each horse. Blood was collected from the indwelling jugular catheter into tubes containing EDTA before the LPS infusion was initiated and 2 hours later (ie, 1 hour and 45 minutes after completing the LPS infusion). Samples were immediately transported to the clinical pathology laboratory for analysis. Leukopenia was defined by the laboratory reference range as WBC count < 5.4 × 10³ cells/μL.

FSIGTT procedure—A 14-gauge polypropylene catheter was inserted into the horse's left jugular vein the day before each FSIGTT was performed. During tests, horses were allowed access to grass hay and water. Patency of the IV catheter was maintained before the test by injection of 5 mL of saline solution containing heparin (4 U/mL) into the catheter every 6 hours. For the FSIGTT procedure, an injection cap and infusion set^e (length, 30 cm; internal diameter, 0.014 cm) were attached to the catheter. The FSIGTT procedure first described by Hoffman et al⁴ was modified so that 100 mg of glucose/kg was infused as a bolus within 1 minute by use of 50% (wt/vol) dextrose solution,^f followed by 20 mL of heparinized saline solution. Blood samples were then collected via the catheter 10, 5, and 1 minute before and 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, and 19 minutes after infusion of dextrose. At 20 minutes, regular insulin^g (20 mU/kg, IV) was administered followed by 20 mL of heparinized saline solution. Blood samples were subsequently collected via the catheter at 22, 23, 24, 25, 27, 30, 35, 40, 50, 60, 70, 80, 90, 100, 120, 150, and 180 minutes relative to the dextrose infusion. At each time point, 3 mL of blood was withdrawn from the infusion line and discarded. A 6-mL blood sample was then collected, followed by infusion of 5 mL of heparinized saline solution. Half the volume of blood was transferred to a tube containing sodium heparin, which was immediately cooled on ice before plasma was harvested via centrifugation. The remaining blood was transferred to a tube containing no anticoagulant, and the samples were allowed to clot at 22°C for 1 hour before serum was harvested. Plasma and serum were harvested via low-speed (1,000 × g) centrifugation and stored at −20°C until further analyzed.

Plasma glucose and serum insulin concentrations—Plasma glucose concentrations were measured by use of a colorimetric assay^h on an automated discrete analyzer.ⁱ Serum insulin concentrations were determined by use of a radioimmunoassay^j that has been validated for use in horses.²¹ Each sample was assayed in duplicate, and intra-assay coefficients of variation < 5% or < 10% were required for acceptance of glucose and insulin assay results, respectively.

Interpretation of FSIGTT data by use of the minimal model—Values for SI, Sg, AIRg, and disposition index were calculated for each FSIGTT in accordance with the minimal model²² by use of commercially available software^k,l and described methods.^4,23 Disposition index was calculated via multiplication of AIRg by SI.

Statistical analysis—Physical examination and CBC values were compared among groups by use of mixed-model ANOVA. In each horse, AUCg and AUCi were calculated for all 3 FSIGTT procedures (baseline, after treatment, and after LPS administration) from the glucose and insulin concentrations measured during the FSIGTT, by use of the trapezoidal method, with commercially available computer software.^m Mixed-model ANOVA was used to determine effects of treatment and time (baseline, after treatment, and after LPS administration) on Sg, SI, AIRg, disposition index, AUCi, and AUCg. Period could not be included in the model as a fixed effect because of the confounding effect of treatment. However, each treatment group consisted of 2 subgroups containing 4 horses, and these subgroups were evaluated at different times of the year. No significant differences were detected between earlier and later subgroups in the same treatment group. If data were nonnormally distributed or unequal variance was encountered, either square root or logarithmic transformation was applied. When a significant treatment × time effect was detected, the protected least significant difference mean separation method was used. Significance was defined at a value of P < 0.05.

Results

Mean ± SD body weight of horses receiving levothyroxine (n = 8; 510.6 ± 27.6 kg) decreased significantly (P < 0.001) by a mean of 18.3 kg relative to baseline value (528.9 ± 25.3 kg) by the end of the 15-day pretreatment period, whereas mean body weight remained unchanged in the dexamethasone and control groups. Mean baseline SI values did not differ significantly (P = 0.942) among groups (Table 1), but 1 horse in the levothyroxine group had a low baseline SI value (0.59 × 10⁻⁴ L•min⁻¹•mU⁻¹) at the beginning of the study that was judged to be an outlier (having a residual value of 2.14 × SD), so data from that horse were excluded from the analysis of treatment and time effects on glucose and insulin dynamics. Insulin sensitivity significantly (P < 0.001) decreased over time in all groups across the 15-day pretreatment period. Mean SI values decreased by 47%, 52%, and 78% relative to baseline value for control, levothyroxine, and dexamethasone groups, respectively. A significant (P = 0.021) treatment-by-time effect was detected; mean SI for the dexamethasone group was significantly (P = 0.032) lower than that of the control group by the end of the pretreatment period. Significant (P < 0.001) treatment-by-time effects were also detected for AUCg and AUCi. Mean AUCg significantly (P < 0.001) increased over time in the dexamethasone group, but did not change in the control (P = 0.626) and levothyroxine (P = 0.878) groups, relative to baseline value. Mean AUCi significantly (P < 0.001) increased over this 15-day period in control and dexamethasone groups, but remained unchanged in the levothyroxine group (P = 0.153). No significant changes in Sg, AIRg, or disposition index were detected.

All horses responded to administration of LPS with colic signs, and leukopenia was detected 2 hours after LPS administration (Table 2). Heart rate, respiratory rate, and rectal temperature significantly (P < 0.001) increased and WBC count significantly (P < 0.001) decreased in all treatment groups following LPS administration. Horses in the dexamethasone group had significantly higher maximal heart rates after LPS administration than horses in the levothyroxine group (treatment-by-time; P < 0.001) and control group (treatment-by-time; P = 0.006), and maximal respiratory rates were higher in the dexamethasone group, compared with the control group, during the same period (treatment-by-time; P = 0.006).

Insulin sensitivity significantly decreased after LPS administration in control (P = 0.002) and dexamethasone (P = 0.015) groups, but remained unaffected (P = 0.695) in the levothyroxine group (Table 1). Additionally, AUCi and AUCg were significantly (P < 0.001) higher in the dexamethasone and control groups after LPS administration, compared with values obtained beforehand (Figures 1 and 2). Mean AUCg increased significantly (P = 0.016) over time in the levothyroxine group, but AUCi remained unchanged (P = 0.329) relative to values obtained before LPS administration. Mean SI measured 20 hours after LPS administration was significantly (P = 0.002) higher in the levothyroxine group than the dexamethasone group, whereas the difference between the levothyroxine and control groups did not reach significance (P = 0.075).

View Full Size

Mean ± SE plasma glucose concentrations during FSIGTTs performed in horses (n = 8 horses/group) undergoing 15 days of control (panel A), levothyroxine sodium (panel B), or dexamethasone (panel C) pretreatment, followed by IV administration of 20 ng of LPS/kg. Testing was performed on day 5 at baseline (black squares), day 20 after pretreatment and before LPS was administered (white circles), and day 21 after LPS infusion (black triangles).

Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.60

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Mean ± SE serum insulin concentrations during FSIGTTs in the same horses as in Figure 1. See Figure 1 for key.

Citation: American Journal of Veterinary Research 71, 1; 10.2460/ajvr.71.1.60

• Download Figure
• Download figure as PowerPoint slide

No significant treatment-by-time effects were detected for Sg, AIRg, or disposition index.

Discussion

Insulin sensitivity decreased significantly over time in all treatment groups across the 15-day pretreatment period, with the lowest mean value detected in the dexamethasone group. Administration of LPS caused an additional decrease in SI in the dexamethasone and control groups, but the magnitude of this response did not differ between those 2 groups, so our hypothesis that resting insulin sensitivity determines the degree of insulin resistance induced by endotoxemia was not supported. Insulin sensitivity markedly decreased when LPS was administered to horses pretreated with dexamethasone, which suggested that glucocorticoid excess and endotoxemia combined to increase the severity of insulin resistance in affected horses. Results also revealed that pretreatment with levothyroxine prevented the reduction in SI associated with endotoxemia, indicating that this treatment should be considered for the prevention of endotoxin-induced insulin resistance in horses.

Dexamethasone and levothyroxine alter glucose and insulin dynamics in horses, so these treatments were selected for this study with the aim of decreasing and increasing resting SI, respectively.^6,7 In the present study, SI significantly decreased in horses receiving dexamethasone, and a mean ± SD value of 0.73 ± 0.63 × 10⁻⁴ L•min⁻¹•mU⁻¹ was detected at the end of the 15-day pretreatment period. This result compared favorably with the mean ± SE value of 0.53 ± 0.13 × 10⁻⁴ L•min⁻¹ •mU⁻¹ reported for 6 Standardbreds after dexamethasone was administered IV at 0.08 mg/kg every other day for 21 days.⁶ Dexamethasone induces insulin resistance by altering insulin signaling pathways in adipose,²⁴ skeletal muscle,²⁵ and liver²⁶ tissues. Phosphorylation of phosphoinositol-3-kinase, a central molecule in the insulin signal transduction pathway, substantially decreases in rat liver following dexamethasone treatment.²⁶ In skeletal muscle, glucocorticoid-induced insulin resistance is accompanied by reduced insulin-mediated phosphorylation of protein kinase B, which causes glycogen synthase kinase-3 to remain in its unphosphorylated, active form.²⁵ This enzyme phosphorylates and thereby inactivates glycogen synthase, which reduces glycogen synthesis and contributes to the development of insulin resistance.²⁵ In horses, treatment with dexamethasone for 21 days impaired insulin-mediated phosphorylation of glycogen synthase kinase-3 in skeletal muscle, but did not alter either the expression of GLUT4 and protein kinase B or insulin-stimulated phosphorylation of protein kinase B.²⁷

Mean SI decreased by 47% and 52% relative to baseline value across the pretreatment period in the levothyroxine and control groups, respectively, and reached values of approximately 1.7 × 10⁻⁴ L•min⁻¹•mU⁻¹. A cutoff value for insulin resistance has not been defined in horses, but Treiber et al²⁸ reported SI values ranging from 0.14 × 10⁻⁴ L•min⁻¹•mU⁻¹ to 0.78 × 10⁻⁴ L•min⁻¹•mU⁻¹ in the lowest quintile of a population of 46 horses. If 0.78 × 10⁻⁴ L•min⁻¹•mU⁻¹ were used as a cutoff value in the present study, horses in the levothyroxine and control groups would have had a reduction in SI but not developed insulin resistance during the pretreatment period. Potential explanations for this reduction in SI include alterations in diet, exercise, estrous cycle, or housing conditions. Before enrollment in the study, horses were kept on pasture with only round bales of hay provided as additional feed. Horses were then transported to the teaching hospital and kept in stalls. They were unable to exercise, and the diet was altered to grass hay fed in amounts equivalent to 2% of body weight divided between 2 meals and 200 g of oats fed once daily. Exercise increases SI in obese and lean horses,²⁹ and skeletal muscle GLUT4 content increases by a factor of 2 to 3 following 6 weeks of exercise training.³⁰ Sedentary behavior is also implicated in the development of insulin resistance in humans³¹ and provides a potential explanation for the decrease in SI over time during the pretreatment period. Alterations in SI have been associated with changes in the estrous cycle in mares.ⁿ Mean ± SE insulin sensitivity significantly decreased in mares from 5.0 ± 0.6 × 10⁻⁴ L•min⁻¹•mU⁻¹ measured during the follicular phase to 3.1 ± 0.6 × 10⁻⁴ L•min⁻¹•mU⁻¹ detected during the luteal phase.ⁿ The estrous cycle was not monitored in mares in the present study, so its influence on results could not be assessed. However, it is likely that mares were in winter anestrus³² during the latter part of the study, and no differences were detected among subgroups when they were compared within treatment groups.

Stress associated with hospitalization may have affected glucose and insulin dynamics in the present study because horses were confined to stalls in the teaching hospital. Effects of stall confinement on SI have not been measured, but results of previous studies^33–36 suggest that horses are stressed when confined to stalls. Agitation and increased vocalization were observed when horses were placed in stalls for the first time in a study³⁶ of 2-year-old mares. Stress-related behavior was recorded for the first 24 hours of stall confinement, and a higher mean score was detected for mares placed in stalls for the first time, compared with mares kept in a paddock. In 1 report,³⁷ cortisol concentrations were more than 2 times as great in 3 of 5 horses removed from a herd and placed in an enclosed barn, compared with control horses that remained with the herd. Conversely, other studies^35,36 have failed to detect that blood or saliva cortisol concentrations or ACTH concentrations in pituitary venous blood³³ are higher in confined horses. Additionally, loss of the normal diurnal blood cortisol rhythm has been described in horses in association with stall confinement.³⁸ Irvine and Alexander³⁸ reported that blood cortisol concentration did not decrease during the day in confined horses. Loss of the cortisol diurnal rhythm may therefore provide evidence of stress and could affect insulin sensitivity in horses confined to stalls. It can also be speculated that artificial lighting affects diurnal patterns of hormone secretion in confined horses. In the study reported here, horses were moved into the teaching hospital for the pretreatment period and subjected to almost continuous lighting. Further studies are required to examine the effects of housing environment on SI in horses, but the reduction in SI detected in this study suggested that hospitalization adversely affected glucose and insulin dynamics in the horses of the present study. Results indicated that both endotoxemia and hospitalization are important risk factors for laminitis in horses¹⁸ and may also explain why hyperadrenocorticism and laminitis are associated.³⁹

Insulin sensitivity increases in response to levothyroxine administration in healthy horses, and this alteration is accompanied by weight loss.⁸ Insulin sensitivity increased from 1.8 ± 1.0 × 10⁻⁴ L•min⁻¹•mU⁻¹ to a peak value of 4.4 ± 2.0 × 10⁻⁴ L•min⁻¹•mU⁻¹ when levothyroxine was administered to 6 healthy mares at a dosage of 48 mg administered daily for 48 weeks, and body weight significantly decreased in treated horses.⁷ However, in the study reported here, mean SI decreased across the 15-day pretreatment period in all 3 treatment groups, and levothyroxine did not ameliorate this response. One explanation for this finding is that levothyroxine failed to prevent the reduction in SI induced by factors associated with hospitalization. This question has not been addressed in the authors' previous studies because study subjects were kept in stalls with adjoining paddocks or on pasture.^7,8 It is also possible that the length of levothyroxine pretreatment was too short to affect SI. Levothyroxine has been administered to horses at a dosage of 48 mg/d for 14 days, but glucose and insulin dynamics were only measured at the beginning and end of an 8-week treatment period after the dosage was incrementally raised to a maximum of 96 mg/d.⁸ Although levothyroxine failed to ameliorate the reduction in SI associated with hospitalization, mean body weight significantly decreased by 18.3 kg in horses that received levothyroxine in the present study. This suggested that the effects of levothyroxine on SI may have been influenced by factors associated with hospitalization, but effects on body weight were preserved.

All horses included in the study had clinical signs of endotoxemia after LPS administration, including leukopenia and increased rectal temperature, heart rate, and respiratory rate. These findings were consistent with results of other studies^40,41 of experimentally induced endotoxemia in horses.^40,41 However, mean rectal temperature, heart rate, and respiratory rate values were higher in the dexamethasone group, and this was an unexpected finding. Administration of dexamethasone to ponies 5 minutes, 3 hours, 9 hours, and 24 hours after LPS infusion fails to prevent the decrease in WBC count induced by endotoxemia.⁴² However, dexamethasone pretreatment suppresses LPS-induced interleukin-1β and interleukin-6 responses in rats.⁴³ These mediators play a central role in the inflammatory reaction induced by LPS.¹⁷ Nevertheless, the same study⁴³ also established that exposure to an acute stressor preceding dexamethasone administration eliminates the inhibitory effect of this drug on proinflammatory cytokine production and release. Results of the study reported here suggested that dexamethasone exacerbated the clinical response to endotoxemia.

Endotoxin-induced insulin resistance was more pronounced in horses with lower SI resulting from dexamethasone administration. This suggests that insulin resistance will be exacerbated in horses or ponies with preexisting insulin resistance that develop endotoxemia. If carbohydrate overload occurs in horses grazing on pasture and endotoxemia develops as it does in experimental models,¹⁴ these intestinal events could exacerbate insulin resistance in insulin-resistant horses and predispose them to laminitis. Although the single endotoxin dose used in the study reported here failed to induce laminitis, naturally occurring endotoxemia is likely to be more sustained. Further studies are required to examine relationships among endotoxemia, SI, and laminitis in horses.

Insulin sensitivity significantly decreased after LPS administration in the dexamethasone and control groups. This effect of endotoxemia on glucose and insulin dynamics was established by use of FSIGTT and euglycemic hyperinsulinemic clamp methods of measuring SI.^2,3 Humans have a biphasic response to LPS, with an initial increase in glucose metabolism followed by a progressive decrease in SI.⁴⁴ Endotoxin-mediated insulin resistance affects the liver, skeletal muscle, and adipose tissues.^45,46 In cultured 3T3-L1 adipocytes, LPS-induced activation of Toll-like receptor 4 stimulates the expression of inflammatory cytokines and provokes insulin resistance.⁴⁶ In rats, sustained endotoxemia adversely affects the early steps of the insulin signaling pathway by inhibiting tyrosine phosphorylation of the insulin receptor substrate-1 molecule in skeletal muscle and liver tissues.⁴⁵

Both SI and AUCi remained unchanged following LPS infusion in horses that were pretreated with levothyroxine in the present study, and this may be explained by the effects of thyroid hormone on glucose transport at hepatic and extrahepatic sites.^47–49 In 3T3-L1 adipocytes, treatment with triiodothyronine increases GLUT1 and GLUT4 content and favors partitioning to the plasma membrane, which facilitates transmembrane glucose transport into cells.⁴⁸ Glucose transport has been enhanced in the presence and in the absence of insulin.⁴⁸ The same effect of thyroid hormone on glucose transport was detected in an in vivo study⁵⁰ of rats treated with levothyroxine. Increased insulin-stimulated glucose transport was detected along with higher GLUT1 and GLUT4 content and increased functional activity. Thyroid supplementation increases basal and insulin-stimulated glucose uptake in skeletal muscle, with the increase proportional to the increase in GLUT4 abundance.⁴⁹ In adult rat liver-15 cells, thyroid hormone stimulates glucose transport by increasing GLUT1 content and enhancing its partition to the cell surface.⁴⁷

In the present study, resting SI did not determine the percentage reduction in SI induced by endotoxin, but results suggested that glucocorticoid excess caused by stress or disease exacerbated endotoxin-induced insulin resistance. Horses in all groups had a reduction in SI across the 15-day pretreatment period, indicating that hospitalization was a risk factor for insulin resistance. Levothyroxine did not prevent the reduction in SI associated with hospitalization, but did protect against endotoxin-induced insulin resistance when administered for 15 days beforehand.