All work performed at the University of Tennessee, Knoxville, TN. Previously presented in part at the 2009 American College of Veterinary Internal Medicine Forum, Montreal, Canada and published in the proceedings of that meeting.
Corresponding author: Dr Nicholas Frank, Department of Large Animal Clinical Sciences, College of Veterinary Medicine, University of Tennessee, 2407 River Drive, Knoxville, TN 37996; e-mail: firstname.lastname@example.org.
Background: Pituitary pars intermedia dysfunction (PPID) is a risk factor for pasture-associated laminitis, which follows a seasonal pattern.
Hypothesis: Hormonal responses to season differ between PPID and unaffected horses.
Animals: Seventeen horses aged 8–30 years (14 horses ≥ 20 years of age).
Methods: Longitudinal observational study. Blood was collected monthly from August 2007 until July 2008 after pasture grazing and again after overnight stall confinement. Blood hormone and metabolite concentrations were measured and pasture grass samples were analyzed to determine carbohydrate content. Analysis of variance analysis for repeated measures was performed.
Results: Mean ACTH concentrations varied significantly over time (P < .001), with higher concentrations detected in August, September, and October compared with November–April. Pasture × time effects were detected for glucose and insulin concentrations, with peaks observed in September. Horses were retrospectively allocated to PPID (n = 8) and control (n = 9) groups on the basis of plasma ACTH concentrations. Changes in insulin concentrations over time differed in the PPID group when compared with the control group. Insulin concentrations were positively correlated with grass carbohydrate composition.
Conclusions and Clinical Importance: PPID did not affect the timing or duration of the seasonal increase in ACTH concentrations, but higher values were detected in affected horses. Insulin concentrations differed between groups, but hyperinsulinemia was rarely detected. Glucose and insulin concentrations peaked in September when horses were grazing on pasture, which could be relevant to the seasonal pattern of laminitis.
Pituitary pars intermedia dysfunction (PPID), which is also known as Equine Cushing's Disease, has been associated with laminitis in horses,1–4 but the mechanisms responsible for this association have not been fully elucidated. One potential explanation for this association is that horses with PPID are insulin resistant.5 Insulin resistance (IR) is an important predisposing factor for pasture-associated laminitis in ponies, cortisol antagonizes the action of insulin within tissues,6,7 and some PPID-affected horses have reduced insulin sensitivity.8,9 Hyperinsulinemia is detected in some, but not all, horses and ponies with PPID, and hyperglycemia occurs in a smaller number of animals.10–12
The incidence of pasture-associated laminitis follows a seasonal pattern that might be relevant to the association between PPID and laminitis. Laminitis develops between March and May in ponies in Virginia,6,13 and between September and May in a group of 40 horses that included 28 animals with suspected PPID.3 This increase in laminitis incidence in September is of interest because it coincides with seasonal upregulation of the hypothalamic-pituitary-adrenal axis in equids.14–16 Plasma concentrations of ACTH and α-melanocyte-stimulating hormone are higher in September, and false positive dexamethasone suppression test results occur more frequently at this time of the year.14,15
Seasonal alterations in hormone concentrations warrant further examination because they appear to coincide with a higher incidence of laminitis in the autumn.3 Furthermore, it must be determined whether these seasonal alterations are more profound in horses with PPID, which could explain the association between this disorder and laminitis. We therefore hypothesized that hormonal responses to season would differ between PPID and unaffected horses. It was also hypothesized that changes in pasture grass composition would induce seasonal alterations in glucose and insulin concentrations and PPID would affect these responses. Blood hormone and metabolite concentrations and responses to pasture grazing were therefore examined across a 12-month period.
Materials and Methods
Seventeen adult light breed horses (9 mares; 8 geldings) ranging in age from 8 to 30 years (14 horses aged ≥ 20 years) were included in the study. None of the horses included in this study were receiving medical treatment for PPID.
A longitudinal study was performed across a 12-month period extending from August 2007 until July 2008. Horses from a facility located in Kingston, Tennessee, in the southeastern region of the United States were included in the study. Evaluations were performed during the first week of every month and consisted of visits to the farm on 2 consecutive days. Blood samples were collected via jugular venipuncture between 8:00 a.m. and 10:00 a.m. on both days. On day 1, blood samples were collected from horses after they were brought in from pasture and housed in stalls. Physical measurements and grass samples were also obtained on day 1 after all blood samples had been collected. Horses were subsequently returned to pasture until 6:00–7:00 p.m., when they were brought back to their stalls for the night. Two flakes of hay were given to each horse, but no grain or additional hay was provided until blood samples had been collected the following morning (day 2). The study protocol was approved by the University of Tennessee Institutional Animal Care and Use Committee.
Feeding and Management Practices
Horses were routinely housed on pasture, except for a 30-minute to 2-hour period between 7:00 a.m. and 9:00 a.m. when they were brought into stalls for feeding. A 12% protein sweet feed or a complete pelleted feed for senior horses was fed in the morning, with amounts varying according to the individual horse and time of year. Hay was fed during the winter months. Feed amounts were recorded.
Body weight was measured by weight tape and the body condition score (BCS) was assessed with the 1–9 scale.17 Neck measurements were obtained as described previously.18 Any abnormalities of the haircoat, including dullness, longer hair length, and curling of the hair, were recorded at this time. These observations were subjective and made by different investigators throughout the year.
Blood was collected into tubes containing potassium EDTA, sodium heparin, or no anticoagulant. Tubes were chilled on ice (plasma) and then placed in racks within coolers containing ice packs or left at ambient temperature to clot for 1 hour (serum) before being transferred to a cooler for transportation. Plasma and serum was harvested by low-speed (1,000 ×g) centrifugation within 2 hours of collection and then stored at −80°C until further analysis.
Serum insulin concentrations were measured with a radioimmunoassay kit (Coat-A-Count insulin radioimmunoassay)a validated previously for equine sera19 and revalidated by our laboratory within 6 months of samples being analyzed.
Plasma glucose, triglyceride, and cholesterol concentrations were measured by colorimetric assaysb,c and an automated discrete analyzer.d Nonesterified fatty acid (NEFA) concentrations were measured with an enzymatic colorimetric test kitc and microtiter plate reader.e For all assays performed on site, measurements were performed in duplicate with all samples analyzed on the same day, and intra-assay coefficients of variation of <5% were required for acceptance of results, with the exception of insulin, which had a cut-off value of 10%.
Frozen plasma samples were packaged with ice packs and sent via overnight mail to the Animal Health Diagnostic Center at Cornell Universityf for measurement of plasma ACTH concentrations. A chemiluminescent ACTH immunoassayg validated previously4 for use with equine plasma was used, with samples analyzed in duplicate. A reference range of 9–35 pg/mL was provided by the laboratory.
Pasture Grass Analysis
Wire exclusion cages were maintained on pastures. One grass sample was collected from each pasture between 9:00 a.m. and 10:00 a.m. on day 1 with electric shears, with the stems cut approximately 1 cm above the ground. Samples were placed in plastic bags and then immediately transferred to a cooler that contained ice packs, which remained closed at all other times. Samples were transported to the laboratory within 2 hours of collection and stored at −20°C. Carbohydrate analysis was performed by the Dairy One Forage Laboratory.h Carbohydrate composition was determined by wet chemistry analysis and amounts of ethanol-soluble carbohydrates, water-soluble carbohydrates, and starches were measured. Depending upon the pastures being used at different times, data from 2 to 7 samples were pooled for each month.
Normality was assessed by examining plotted results and performing Shapiro-Wilk tests. Adrenocorticotropin hormone and insulin data required logarithmic transformation to fit a normal distribution before statistical tests were performed. Geometric mean values with 95% confidence intervals are displayed for these variables. Mean (SD) values are reported for glucose and NEFA concentrations. Mixed-model ANOVA for repeated measures was performed by use of statistical softwarei to determine the effects of time (month), and subsequently group (PPID versus control), on measured variables. Effects of pasture grazing were also included in the same model for all variables, with the exception of ACTH because this variable was only measured on day 1. When a significant effect was detected, the Bonferroni test for multiple comparisons was used to identify significant differences among least squares means. Effects of sex, initial body weight, and the amount of feed provided were also examined, but were subsequently removed from the model because they did not affect results. Pearson correlation coefficients were calculated for mean blood variable concentrations and mean pasture grass carbohydrate percentages. Significance was defined at a value of P < .05.
All horses remained healthy throughout the study, with the exception of 1 horse that required tissue debridement and application of a foot cast because of recurrent sole abscesses. Glucose, insulin, and lipid data from this horse were excluded from the analysis because marked hyperinsulinemia was detected, with a peak insulin concentration of 955 μU/mL observed in April. Another horse suffered from chronic degenerative joint disease of both carpi and received phenylbutazone intermittently during the study.
Time effects were significant for body weight (P < .001), neck circumference (P < .001), and BCS (P < .001; Table 1). Mean body weight (via weight tape) was highest in December and mean midneck circumference was lowest in June. There was no discernable pastern for BCS.
Table 1. Physical examination measurements for 17 horses across a 12-month sampling period.
Mean ± SD (n = 17)
Body Weight (kg)
Midneck Circumference (cm)
Within a column, values with different superscript letters differ significantly (P < .05).
BCS, body condition score.
472 ± 72cd
5.5 ± 1.0bc
90.9 ± 8.7ab
481 ± 75abcd
6.0 ± 1.5ab
90.7 ± 7.6ab
488 ± 67ab
6.5 ± 1.5a
92.9 ± 7.6a
488 ± 60a
5.0 ± 1.5c
90.7 ± 6.7ab
493 ± 68a
6.5 ± 1.5ab
92.0 ± 8.0ab
484 ± 72abc
6.0 ± 1.5abc
92.6 ± 8.4ab
490 ± 68a
6.5 ± 1.5ab
89.4 ± 6.8ab
480 ± 68abcd
6.0 ± 1.5abc
89.3 ± 6.7b
480 ± 67abcd
6.0 ± 1.0abc
91.0 ± 6.9ab
479 ± 68abc
6.0 ± 1.5abc
90.9 ± 6.6ab
466 ± 64d
6.0 ± 1.5abc
85.6 ± 7.8c
472 ± 66bcd
6.0 ± 2.0abc
90.2 ± 7.9ab
P < .001
P < .001
P < .001
Plasma ACTH concentrations were significantly (P < .001) affected by time, with higher mean values detected in August, September, and October compared with the November–April period (Fig 1). Effects of pasture grazing on plasma ACTH concentrations were not assessed because this hormone was measured once each month.
Time (P < .001), pasture (P < .001), and pasture × time (P < .001) effects were significant for glucose concentrations, with a peak in September when horses were grazing on pasture (Fig 2). Insulin concentrations also peaked in September when samples were collected after grazing and time (P < .001), pasture (P < .001), and pasture × time (P < .001) effects were significant for this variable (Fig 3). A positive correlation (r= 0.22; P= .002) existed between mean ethanol-soluble carbohydrate content of the grass reported on a dry matter basis and mean insulin concentrations measured in grazing horses. Monthly mean ethanol-soluble carbohydrate, water-soluble carbohydrate, and starch content (dry matter basis) within pasture grass ranged from 2.0 to 9.1, 1.6 to 12.7, and 1.0 to 2.0% across the 12-month sampling period. Pasture, time, and pasture × time effects were also significant for triglyceride and NEFA concentrations, with higher NEFA concentrations detected after stall confinement. Total cholesterol concentrations were affected by pasture and time, but did not follow a recognizable seasonal pattern.
Horses were subsequently allocated to PPID (n = 8) and control (n = 9) groups on the basis of plasma ACTH results. A presumptive diagnosis of PPID was made when plasma ACTH concentrations exceeded 35 pg/mL on ≥3 occasions between December and June. Five of 8 horses in the PPID group had persistently increased plasma ACTH concentrations throughout this 7-month time period. Horses in the PPID group ranged in age from 18 (estimated) to 30 years (median; 28.5 years) compared with 8–26 years (median; 21 years) for the control group. Both groups contained 7 horses that were ≥20 years of age. Five mares and 3 geldings were included in the PPID group and the control group contained 4 mares and 5 geldings. Breeds represented in the PPID group included Arabian (n = 1), Arabian/Quarter Horse (1), Quarter Horse (2), Saddlebred (1), mixed breed (1), and Thoroughbred (1), whereas the control group contained Appaloosa (1), Quarter Horse (3), Thoroughbred (4), and Tennessee Walking Horse (1) horses. Initial mean ± SD body weight was 449 ± 88 kg for the PPID group and 493 ± 51 kg for the control group. The amount of feed provided to each horse varied by individual animal and over time, but mean values did not differ significantly between groups (2.4 ± 0.6 versus 2.4 ± 0.5 lb/d for PPID and control groups, respectively; P= .963). Abnormalities of the haircoat were noted in 6 of 8 horses from the PPID group and 4 of 9 horses in the control group. Three horses aged 27, 29, and 30 years in the PPID group exhibited a long curly haircoat consistent with hirsutism and fat redistribution. One of these horses was the animal that developed sole abscesses. Haircoat abnormalities recorded for other horses included subjective observations of longer hair length and dullness. There were no reports of laminitis, polyuria, or polydipsia and the owner did not raise concerns about accelerated weight loss in any of the horses.
Physical examination variables did not differ significantly between groups. Plasma ACTH concentrations were higher in the PPID group (P < .001), but the group × time effect was not significant (P= .847; Fig 4). Plasma ACTH concentrations >35 pg/mL were detected at 1 (n = 1), 2 (n = 3), or 3 (n = 4) of the August, September, and October time points in 8 of 9 horses from the control group, but did not exceed this cut-off value in the remaining horse. In contrast, plasma ACTH concentrations were persistently increased between August and October in the PPID group. Maximum ACTH values were 1,250 and 105 pg/mL for PPID and control groups, respectively.
Overall mean insulin concentrations did not differ significantly between PPID and control groups (P= .185), but a group × time effect (P= .037) was detected (Fig 5). The group × pasture × time interaction (P= .962) was not significant. A significant group × pasture effect (P= .004) was also detected for NEFA concentrations (Fig 6).
Horses in this study exhibited upregulation of the hypothalamic-pituitary-adrenal axis in the late summer and autumn, as evidenced by significantly higher plasma ACTH concentrations. Horses grazing on pasture had higher glucose and insulin concentrations in September and insulin concentrations were positively correlated with carbohydrate composition when horses were grazing on pasture. PPID did not alter the timing or duration of this seasonal change in ACTH concentrations, although higher concentrations were detected in affected animals. Variation in insulin concentrations over time differed between groups, but hyperinsulinemia was rarely detected when horses were sampled after stall confinement.
A presumptive diagnosis of PPID was made in this study on the basis of plasma ACTH concentrations exceeding 35 pg/mL on 3 or more occasions outside of the late summer and autumn period. This method was selected because the owner would not permit other diagnostic tests to be performed because of concerns about inducing laminitis. However, 3 horses in the PPID group had clinical signs consistent with the disorder, including overt hirsutism and fat redistribution. Other results from the haircoat examinations must be evaluated within the context of the methods used because observations were subjective and some signs, such dullness of the haircoat are nonspecific. Objective criteria for diagnosing haircoat abnormalities attributable to PPID should be used in future studies to address this deficiency. Hirsutism has been used as a gold standard for PPID in 1 previous study,4 which supports the allocation of 3 horses to the PPID group. However, all other horses in the PPID group were allocated on the basis of ACTH concentrations alone, so results should be interpreted accordingly.
Horses with presumed PPID did not differ with respect to the timing or duration of the seasonal alterations in plasma ACTH concentrations, but higher concentrations were detected in this group, and this was most apparent in August, September, and October. Diagnostic testing has been avoided during this time period since Donaldson et al14 published their finding of increased ACTH concentrations in September, but warrants reevaluation in light of our findings. In the future, it might be possible to test horses in the late summer/autumn season if season-specific reference ranges for ACTH are established. Three of 8 horses in the PPID group also had ACTH concentrations that were within reference range on 1 or more occasion between December and July, which suggests that test accuracy could be improved by collecting more than 1 blood sample throughout the year.
Adrenocorticotropin hormone was measured by a chemiluminescent immunoassay that has been validated previously for the measurement of ACTH in equine plasma.4 A cut-off value of 35 pg/mL has been adopted in previous studies,3,14,16 but upper limits of 45,20,21 50,2 or 70 pg/mL12 have also been used for diagnosis of PPID. A seasonal rise in plasma ACTH concentrations was detected in this study when values from August, September, and October were compared with those from December to April, consistent with previous reports.14,16 Resting ACTH concentrations were significantly higher in horses during the autumn, compared with the winter and spring. Results of these previous studies and the one reported here indicate that horses undergo upregulation of the hypothalamic-pituitary-adrenal axis during the late summer and autumn.
Seasonal alterations in ACTH concentrations are likely to be linked to changes in photoperiod, with the reduction in daylight hours triggering alterations in the hypothalamic-pituitary-adrenal axis. Melatonin is thought to play an important role in this process because it has a circadian rhythm of low concentrations during the day followed by higher concentrations at night, and this pattern changes with season as days get shorter.22,23 Horses also gain body fat in response to decreasing photoperiod duration, presumably in preparation for winter.22 Seasonal weight gain was observed in this study, although weight tape and neck circumference measurements may have been confounded by the growth of winter haircoats.
Insulin concentrations differed between groups, but hyperinsulinemia (>20 μU/mL; reference range for laboratory) was rarely observed, except in response to pasture grazing. There was little evidence of reduced insulin sensitivity in this study, but insulin concentrations are affected by alterations in pancreatic output as well as tissue insulin sensitivity, so this situation requires further investigation. Insulin and glucose data from 1 mare were excluded from statistical analyses because of severe lameness and marked hyperinsulinemia. Interestingly, this mare had a history of obesity, regional adiposity, and laminitis before losing weight and developing hirsutism in recent years. This suggests that she suffered from equine metabolic syndrome24 before the development of PPID, and therefore makes it difficult to determine the cause of hyperinsulinemia in this animal. Results of this study differ from those of previous reports because PPID has been associated previously with hyperinsulinemia.8,10,12,21 It should be noted, however, that only horses were examined in the study reported here, whereas PPID groups contained ponies in the previous studies. Blood samples were also collected under fed conditions in the aforementioned studies. These points could be relevant because ponies have lower insulin sensitivity when compared with horses,25,26 and results of this study indicate that insulin concentrations are affected by feeding conditions.
Higher glucose and insulin concentrations were detected in horses after pasture grazing and were affected by changes in season, with peaks detected in September. Insulin concentrations also increased again in April, but this peak was not statistically significant. A weak correlation existed between mean insulin concentrations and mean ethanol-soluble content of the pasture grass when postgrazing results were examined. Ethanol-soluble carbohydrates include simple sugars,27 so higher glucose and insulin concentrations appeared to correspond with increases in sugar intake on pasture. The ethanol-soluble carbohydrate content of the pasture grass may also reflect its growth, so these may have been times of the year when the grass was more abundant because of increased rainfall and sunlight.28 Unfortunately, grass intake could not be measured in this study.
There were several statistically significant alterations in blood lipid variables, but only NEFA concentrations followed a recognizable pattern. Higher NEFA concentrations were detected when blood samples were collected on day 2 after horses experienced a period of fasting. This is a normal physiological response to reduced feed intake because negative energy balance stimulates hormone-sensitive lipase and stored triglycerides are hydrolyzed to yield glycerol and fatty acids, which are used for energy.29,30 A significant group × pasture effect was also detected for this variable, with stall confinement having a greater effect on NEFA concentrations in control horses. The lower response detected in the PPID group was unexpected because hyperadrenocorticism and IR would be expected to increase hormone-sensitive lipase activity.31 One explanation for this finding may be that PPID-affected horses remain calmer when deprived of feed, although no differences in demeanor were recognized in this study.
Retrospective studies provide evidence of an association between PPID and laminitis in horses and ponies,1–4 with 1 report demonstrating that laminitis occurred more commonly in September and May.3 Laminitis was not detected in the study reported here, but glucose and insulin concentrations peaked in September when plasma ACTH concentrations were increased. Although not statistically significant, a second insulin peak was also observed in the spring when a higher incidence of pasture-associated laminitis has been reported in ponies.6,13 Interestingly, the insulin peak detected in September was higher than the one observed in April, despite similar increases in pasture grass ethanol-soluble carbohydrate content. This suggests that upregulation of the hypothalamic-pituitary-adrenal axis accentuated the insulinemic response, although further research is required to examine this relationship. Differences in total grass intake might also explain this discrepancy.
In conclusion, our hypothesis that horses with PPID would respond differently to changes in season was partially supported because changes in insulin concentrations over time were more pronounced in affected horses. However, seasonal changes in ACTH concentrations did not differ significantly in timing or duration between groups. Further studies are required to establish season-specific reference ranges for ACTH and determine whether the magnitude of the seasonal increase in hormone concentrations can serve as a diagnostic test for PPID in horses. Our results also demonstrate that pasture grazing raises glucose and insulin concentrations at specific times of the year in horses. Glucose and insulin concentrations peaked in September at the same time that ACTH concentrations were increased, and this convergence of risk factors may be relevant to the development of laminitis.
aSiemens Medical Solutions Diagnostics, Los Angeles, CA
bGlucose, Roche Diagnostic Systems Inc, Somerville, NJ
cWako Chemicals USA, Richmond, VA
dCobas Mira, Roche Diagnostic Systems Inc
eELx800 Microplate Reader, Bio-Tek, Winooski, VT
fAnimal Health Diagnostic Center, Cornell University, Ithaca, NY
gSiemens Medical Solutions Diagnostics
hDairy One Forage Laboratory, Ithaca, NY
iPROC MIXED, SAS, version 9.1, SAS Institute Inc, Cary, NC
The work was supported by a grant from the American College of Veterinary Internal Medicine Foundation. The authors thank Ms Sharon Lloyd and Dr Eric Etheridge for their assistance with this study.