Spring and fall blood concentrations of adrenocorticotropic hormone, insulin, and thyroxine in healthy horses in Saskatchewan

Abstract

Objective

To characterize concentrations of adrenocorticotropic hormone (ACTH), insulin, and thyroxine (T4) at 2 times of the year in healthy horses in Saskatchewan.

Animals and procedure

A prospective, observational study was carried out in 52 healthy, client-owned horses. Inclusion criteria were no recent travel outside of Saskatchewan, normal physical examination findings, and no history or evidence of ongoing illness. Blood concentrations of ACTH, insulin and T4 were determined by chemiluminescence. Samples were collected in spring and fall and compared using the paired Student’s t-test or Wilcoxon signed-rank test. Generalized estimating equations were used to assess the associations between ACTH, T4, and insulin concentrations and gender, age, season, body condition score, glucose concentration, and breed.

Results

There were increases in both ACTH and insulin concentrations in the fall compared to spring (P < 0.001 and P = 0.001, respectively). Other than season, insulin concentration was associated with breed, whereas ACTH concentration was associated with age. Finally, T4 concentration was associated with breed and glucose concentration, but not with season.

Conclusion and clinical relevance

These results highlighted the differences between spring and fall concentrations of both ACTH and insulin in healthy horses residing in the Canadian prairie provinces, which are known for extreme differences in summer and winter temperatures as well as day length. Geographically adjusted reference values are necessary to account for these variations, to improve diagnostic accuracy. This is the first published Canadian study evaluating these factors and their associations with hormone concentrations in clinically healthy animals.

Introduction

Equine endocrine diseases, including pituitary pars intermedia dysfunction (PPID) and equine metabolic syndrome (EMS), are chronic, sometimes concurrent conditions affecting middle-aged to older horses (1). The prevalence of PPID is ~20% in animals 15 y or older (2). Clinical signs include lethargy, abnormal hair coat shedding, laminitis, loss of muscle mass, polydipsia, and polyuria (3). In addition, EMS is a group of clinical signs and pathologic conditions associated with an increased risk of endocrinopathic laminitis (4). One of its key components is insulin dysregulation, including hyperinsulinemia and insulin resistance. Hyperinsulinemia, but not hyperglycemia, has been associated with development of laminitis (4). Although increased generalized or regional adiposity are common, EMS can also affect lean animals (5).

Diagnosis is based on signalment, history, clinical signs, and diagnostic testing, with measurement of plasma adrenocorticotropic hormone (ACTH) and serum insulin concentrations for PPID and EMS, respectively. The ACTH concentration has seasonal and geographical variations in normal horses (6–8), with increased concentrations in fall months. Seasonality in insulin concentration is well-documented in humans, and both cold weather and reduced daylight length have been associated with decreased insulin sensitivity (9,10). Seasonality is also suspected to influence normal insulin concentration in horses (11,12).

Hypothyroidism has been reported in the Canadian prairie provinces as a congenital endocrinopathy affecting newborn foals, associated with inadequate nutrition of the pregnant mare (13). In adult horses, concentrations of thyroid hormones, including triiodothyronine (T3) and thyroxine (T4), fluctuate with season and time of day, with higher values in summer (14).

Improving diagnostic accuracy of equine endocrinopathies could lead to earlier detection and treatment of affected animals. However, geographical variation can make available information difficult to interpret in areas with extreme conditions such as the Canadian prairie provinces, where weather, day length, and other factors could affect hormone dynamics of both healthy and diseased horses. Our objective was to describe variations in basal ACTH, insulin, and T4 concentrations in spring and fall for clinically healthy horses residing in Saskatchewan. Characterizing fluctuations in hormone concentrations in Saskatchewan is important for improving diagnostic accuracy in this geographic region.

Materials and methods

Study design

The study was conducted in accordance with the animal care standards of the University Animal Care Committee, University of Saskatchewan (Saskatoon, Saskatchewan) (Animal Use Protocol 20200089, approved September 30, 2020). A prospective observational study was designed by recruiting client-owned horses. All owners gave informed consent. Inclusion criteria were healthy animals between 1 and 20 y old with no history of having lived outside the province for > 1 wk at a time in the year before the beginning of the study and throughout the study period. Animals were considered healthy in the absence of history or physical examination findings suggestive of systemic illness, including PPID and EMS. Exclusion criteria included a previous diagnosis of PPID or EMS or a history suggestive of 1 or both conditions, including laminitis, recurrent infections, or abnormal shedding. Animals suffering from conditions associated with chronic pain or receiving long-term treatment with steroidal or nonsteroidal anti-inflammatory drugs were excluded.

Animals were sampled in spring (April 1 to April 23, 2021) and fall (October 4 to October 14, 2021). Animals were not fasted before sampling and were receiving their usual diet. Owners were instructed to avoid feeding grain in the 12 h before sampling. Blood samples were collected by jugular venipuncture into plastic EDTA (ACTH and glucose) and plain (T4 and insulin) Vacutainer tubes (BD, Franklin Lakes, New Jersey, USA) and stored at 4°C until processed. A physical examination was done at the time of sampling. Body condition score (BCS) was assessed with the Henneke scale (15). Level of exercise was subjectively classified, following the 2007 National Research Council categorization (16), as light, moderate, or heavy.

Sample processing and analysis

Samples were processed within 4 h after collection. Centrifugation (10 min at 2000 × g) and separation of plasma or serum was completed, and samples were stored at −72°C until analyzed. The interval between collection and analysis was 9 mo for samples collected in the spring and 1 mo for samples collected in the fall. Plasma ACTH and T4 concentrations and serum insulin concentrations were quantified using chemiluminescence (Immulite 1000 Immunoassay System and Immulite/Immulite 1000 assays and control modules for ACTH, insulin, and T4; Siemens, Malvern, Pennsylvania, USA). The kits were validated for use with equine plasma and serum at the Endocrine Laboratory, Prairie Diagnostic Services (Saskatoon, Saskatchewan). Quality control materials provided by the manufacturer were used daily. For ACTH, the coefficients of variation for lower- (21.82 pg/mL) and upper- (326.82 pg/mL) level quality control materials were 9.2 and 3.3%, respectively. For insulin, the coefficients of variation for lower- (69.9 pmol/L) and upper- (365.7 pmol/L) level quality control materials were 8.5 and 3.9%, respectively. For T4, the coefficients of variation for lower- (11.1 nmol/L) and upper- (109.5 nmol/L) level quality control materials were 13 and 4.6%, respectively. Reference intervals for T4 in adult horses were obtained from the Animal Health Diagnostic Center Laboratory at Cornell University (Ithaca, New York, USA), whereas reference intervals for ACTH and insulin in adult horses were obtained from the Animal Health Laboratory at the University of Guelph (Guelph, Ontario). Blood glucose was measured with the enzymatic endpoint method using the hexokinase reaction (Roche, Mannheim, Germany).

Data analyses

Descriptive statistics were used to summarize the baseline characteristics of the animals. Data are expressed as mean ± SD, median and interquartile range, or numbers and percentages. Statistical analysis was carried out with Stata/SE 17.0 (StataCorp LLC, College Station, Texas, USA) and R (17) software. A P-value < 0.05 was considered statistically significant. Adrenocorticotropic hormone, insulin, T4 and glucose concentrations were compared between seasons using the paired Student’s t-test with Welch correction (for normally distributed data) or Wilcoxon signed-rank test (for non-normally distributed data). Univariable regression was used to select variables with unconditional associations with the outcomes, using a P-value < 0.2 as a threshold for inclusion in the model. Insulin, T4, and ACTH concentrations were included as outcome variables. Gender, age, season, BCS, glucose concentration, farm, and breed were included in the analysis as explanatory variables. To account for the presence of clustered data and repeated measures, generalized estimating equations (GEE) were used to estimate marginal models with unstructured correlation for the effect of the explanatory variables on each outcome (insulin, T4, and ACTH concentrations). For this part of the analysis, age was categorized into 5-year intervals. To assess its association with T4 and ACTH concentrations, breed was grouped into 6 categories: thoroughbred (TB) and crosses, quarter horse (QH) and crosses, Arabian and crosses, pony and crosses, warmblood (WB) and crosses, and Clydesdale. For the outcome insulin only, breed was transformed into a dichotomous variable (pony and crosses versus all other breeds) based on exploratory data analysis results. Wald tests were carried out to estimate the significance of each variable, and nonsignificant ones (P ≥ 0.05) were removed from the model. Excluded variables were evaluated for their potential confounding effects by assessing whether their removal resulted in ≥ 20% changes in the coefficients of significant variables in the model. Interactions between variables were also assessed.

Results

Fifty-two horses were sampled in the spring, and 43/52 in the fall, with the remaining 9 lost to follow-up. Data and comparisons between spring and fall are shown in Table 1. The horses resided on 9 premises within a radius of 50 km of the city of Saskatoon (Saskatchewan). Horses were housed in paddocks and received additional roughage with or without grain, depending on their age and level of exercise. All horses were on a light exercise regime, except for the horses < 4 y old, which were not exercised, and 1 gelding (19 y) that was retired from exercise between the first and second samplings. The BCS did not change between spring and fall, except for 1 gelding whose BCS decreased from 6 to 5/9.

Table 1.

Clinical characteristics and basal hormone concentrations of horses enrolled in each season.

Variable	Spring (n = 52)	Fall (n = 43)	P-value
Age (y)	9.89 ± 5.47	10.39 ± 5.47	—
Sex	F: 46.2% (24/52) G: 5% (26/52) M: 3.8% (2/52)	F: 44.1% (19/43) G: 51.2% (22/43) M: 4.6% (2/43)	P = 0.85
Breed	15 TB, 7 WB, 7 AP, 5 QH, 3 Clydesdale, 5 pony, 7 mixed-breed, 1 Friesian, 1 Lusitano, 1 Andalusian	13 TB, 5 WB, 7 AP, 4 QH, 3 Clydesdale, 3 pony, 7 mixed-breed, 1 Friesian	—
BCS	5 [1]	5 [1]	—
ACTH (pg/mL)	20.77 [10.84]	31.32 [13.23]	P < 0.001
Insulin (pmol/L)	27.05 [31.05]	34.60 [55.5]	P = 0.001
T4 (nmol/L)	23.45 [9.6]	21.9 [6.5]	P = 0.43
Glucose (mmol/L)	5.3 [0.5]	5 [0.85]	P = 0.002

Data are expressed as mean ± SD, median [interquartile range], or percentage.

Statistical significance (defined as P < 0.05) was assessed using a paired Student’s t-test with Welch correction for normally distributed data, or a Wilcoxon signed-rank test for non-normally distributed data.

ACTH — Adrenocorticotropic hormone; AP — American paint; BCS — Body condition score; F — Female; G — Gelding; M — Intact male (stallion); QH — Quarter horse; TB — Thoroughbred; T4 — Thyroxine; WB — Warmblood.

Concentrations of ACTH, T4, insulin, and glucose are shown in Figure 1. Glucose concentration (Figure 1 A) decreased between spring and fall (P = 0.002). Serum insulin concentration was higher in fall compared to spring (Figure 1 B). Serum T4 concentration (Figure 1 C) did not change significantly between spring and fall. Plasma ACTH concentration (Figure 1 D) increased in fall compared to spring. In 6 horses, fall concentration was lower than spring concentration. No common factors (age, sex, breed, BCS, level of exercise, or premises) were identified in these horses. Concentrations of insulin and ACTH by age and season are further illustrated in Figures 2 and 3, respectively.

Figure 1.

Spring (n = 52) and fall (n = 43) levels of glucose (A), insulin (B), T4 (C), and ACTH (D) in clinically healthy horses in Saskatchewan.

ACTH — Adrenocorticotropic hormone; T4 — Thyroxine.

Figure 2.

Spring (grey squares, n = 52) and fall (red circles, n = 43) levels of insulin as a function of age in clinically healthy horses in Saskatchewan.

Figure 3.

Spring (grey squares, n = 52) and fall (red circles, n = 43) levels of ACTH as a function of age in clinically healthy horses in Saskatchewan.

ACTH — Adrenocorticotropic hormone.

Analyses of GEE parameter estimation are shown in Table 2. Insulin concentration was significantly higher in fall compared to spring (Figure 2). There was also a breed effect, with the category including ponies and their crosses exhibiting a higher insulin concentration than all other breeds, when analyzed both separately (overall P = 0.01, result not shown) and as a single, “non-pony” category. Thyroxine concentration was significantly associated with glucose concentration and breed (overall P = 0.001), with TB and their crosses exhibiting a lower hormone concentration than the remaining breed categories. For both insulin and T4, interaction effects were not significant. In contrast, analysis using GEE revealed significant influences on ACTH concentration by season, age category, and their interaction (overall P = 0.001). Animals in age categories 1 (1 to 5 y old), 3 (11 to 15 y old), and 4 (16 to 20 y old) had significantly higher blood ACTH concentrations in fall compared to spring. This was not the case for animals in age category 2 (6 to 10 y old). There were no effects of gender or BCS on ACTH, T4, or insulin concentrations. The level of exercise also did not exhibit any significant associations with the outcome variables analyzed in this study.

Table 2.

Estimated coefficients, 95% CI, and P-values in multivariable marginal models using generalized estimating equations for insulin, T4, and ACTH. Only the factors retained in the final models have been included.

Outcome: Insulin	Coefficient	95% CI	P-value
Season (reference: spring)
Fall	17.08	7.05 to 27.11	0.001
Breed (reference: non-pony)
Pony and crosses	84.75	58.87 to 110.64	< 0.001
Intercept	27.41	18.00 to 36.81	< 0.001
Outcome: T4
Breed (reference: TB and crosses)
QH and crosses	6.53	2.93 to 10.12	< 0.001
Arabian and crosses	11.04	5.31 to 16.78	< 0.001
Pony and crosses	7.59	2.46 to 12.73	0.004
WB and crosses	7.5	3.08 to 11.93	0.001
Clydesdale	6.38	0.64 to 12.11	0.029
Glucose (mmol/L)	1.77	0.06 to 3.47	0.043
Intercept	9.53	0.64 to 12.11	0.03
Outcome: ACTH
Season (reference: spring)
Fall	15.33	10.45 to 20.21	< 0.001
Age (reference: 1 to 5 y old)
6 to 10 y old	−4.16	−10.3 to 1.97	0.184
11 to 15 y old	3.21	−2.76 to 9.19	0.292
16 to 20 y old	9.12	2.57 to 15.68	0.006
Season × age
Fall vs spring, 1 to 5 y old	10.65	2.94 to 18.36	0.006
Fall vs spring, 6 to 10 y old	8.33	−0.47 to 17.12	0.058
Fall vs spring, 11 to 15 y old	15.62	7.24 to 24.00	< 0.001
Fall vs spring, 16 to 20 y old	31.31	21.48 to 41.14	< 0.001
Intercept	13.5	7.35 to 19.65	< 0.001

ACTH — Adrenocorticotropic hormone; CI — Confidence interval; QH — Quarter horse; TB — Thoroughbred; T4 — Thyroxine; WB — Warmblood.

Discussion

Diagnosing endocrine conditions in horses is essential to mitigate their effects on quality of life and performance. However, fluctuations in concentrations of measured biomarkers can interfere with interpretation of test results and lead to misdiagnosis (18). In the present study, an increase in ACTH concentration was observed in the fall compared to spring in most horses. However, 6 animals had an opposite trend, with a decline in ACTH concentration; this could be attributed to individual variation or to factors not accounted for in the study, such as increased circulating cortisol concentrations. The general trend was similar to that in a previous report (19). Adaptability, and thus, variability in hormone secretion has been reported to decrease with age but not season (19–21). In studies of hormone secretion in humans (20) and circadian and seasonal rhythmicity in horses (19,21), hormone secretion patterns changed with age. Mean cortisol concentrations are increased in elderly humans, with both cortisol and ACTH exhibiting decreases in rhythm amplitude (20). These age-related changes have been attributed to a reduction in adaptability against various stressors. In the present study, the association between age and ACTH concentration was evident in the fall, but not in the spring. The fact that 1 age category (horses between 6 and 10 y old) did not follow this pattern might stem from an insufficient number of horses in the study in general and in that age group in particular. A larger study population would be necessary to better characterize this trend.

There was an increase in fall insulin concentration in this study that was consistent across all ages. Seasonal fluctuations in basal insulin concentration have been previously reported, including increases in fall (12,22) or spring months (23), but have not been identified in other works (24,25). Seasonal changes in feed nutritional content, especially in soluble carbohydrates and starch, have been proposed as a possible reason for these discrepancies. Temperature and day length affect insulin sensitivity in dogs (26), and these factors might also have a role in horses. An increase in fall insulin concentration in Finnhorse mares grazing cultivated, high-yielding pasture versus semi-natural grassland has been described (27). However, the BCS of those animals had increased since the beginning of the study in the spring. Therefore, changes in metabolic activity, including decreased insulin sensitivity or tissue responsiveness leading to a higher compensatory insulin secretion, could explain their seasonal differences. In our study, animals had a similar BCS at the first (spring) and second (fall) sampling, and thus, associated changes in insulin sensitivity between sampling times were unlikely. Since all samples were taken in the morning in both seasons, circadian variation would also not account for this finding. Further studies in healthy horses are underway to better characterize seasonal insulin variations.

Insulin sensitivity has been reported to decrease with age due to a diminished response to insulin signaling (28,29). A higher concentration of circulating insulin in response to intravenous glucose and after a meal has been detected in aged horses (28). In the present study, only the basal concentration of insulin was measured. A mild, nonsignificant increase in basal insulin concentration was detected in association with age and might reflect these age-related changes in insulin sensitivity. A larger number of animals is likely needed to override individual variation and detect clear differences in insulin concentration.

In line with previous studies (22,30), no differences were observed between spring and fall in T4 concentration. Some authors (31) reported a weak but statistically significant negative correlation between insulin and T4 concentrations. The relationship between thyroid status and insulin sensitivity has been studied in humans, with conflicting results. Clinical hypothyroidism has been associated with hyperinsulinemia in children and adults (32,33), and there is evidence that subclinical hypothyroidism can present with insulin resistance due to impaired glucose disposal in peripheral tissues (34). A proposed underlying mechanism is decreased perfusion to peripheral tissues, mediated by impaired vasodilation and decreased sympathetic response (35). In the present study, there was no association between age and T4 concentrations. In contrast, Breuhaus (30) reported a negative correlation between age and total T4 concentration in healthy horses. In the absence of other markers of thyroid function, including free and total T3 and T4, as well as thyrotropin-releasing hormone, interpreting this discrepancy is challenging and warrants further investigation. The same applies for any breed association with T4 concentration. Although TB and their crosses had consistently lower T4 concentrations than the other breeds included, this apparent breed-specific difference should be explored using larger numbers of animals.

No seasonal variation in glucose concentration in horses was reported in a previous study (22). Although a significant decrease in blood glucose concentration was observed in the present study, its absolute magnitude (0.3 mmol/L) may not be clinically relevant. Despite no associations between exercise and ACTH, insulin, and T4 concentrations, the enrolled horses performed at a low level of exercise or were not being exercised at the time of the study. Further research involving animals performing at higher levels in a variety of disciplines may uncover an effect of physical activity on these biomarkers. However, other conditions associated with exercise should be accounted for, to avoid confounding.

This study had several limitations. Basal concentrations of ACTH and insulin were determined, but dynamic testing was not done. The latter can increase sensitivity and therefore prevent some animals in early endocrine disease stages from being erroneously classified as healthy. For example, dynamic testing for PPID with a thyrotropin-releasing hormone stimulation test improves diagnostic accuracy compared to basal ACTH concentration when reference intervals are used for interpretation (36). Health status was assessed before inclusion based on history and clinical signs, and the likelihood of missing affected animals was thus reduced. Breed effect was accounted for in the statistical analysis. However, given the low sample sizes of some of the breeds included, it is possible that further breed effects might be present, with a larger number of animals needed to assess this further. Although age-related differences in hormone concentrations could be detected, the relatively lower number of aged animals (15 to 20 y old) compared to middle-aged horses (5 to 14 y old) might be masking a more pronounced effect of age on basal insulin and, especially, ACTH concentrations. Finally, effects of confounding variables, such as feed composition and total feed intake, have not been explored and might explain some of the differences observed. In this regard, Williams et al (12) reported higher concentrations of water- and ethanol-soluble carbohydrates in pastures in spring compared to fall. Although these differences in feed composition did not affect glucose or insulin concentrations in horses in their study, Frank et al (37) reported a positive correlation between insulin concentration and the amount of ethanol-soluble carbohydrates in pasture.

In line with previous works, this study supports the need for geographic-specific data for markers of endocrine function in healthy horses. Circannual fluctuations in ACTH concentration vary depending on geographic location (7,8), and this, in turn, might affect interpretation of diagnostic test results. Certain biomarkers, including ACTH and potentially insulin, additionally require the use of season-specific reference intervals, and possibly age-specific intervals, to account for normal fluctuations related to these factors. The circannual variation of ACTH concentration in equine plasma means that establishing discrete categories will probably be inaccurate, as concentrations fluctuate in a continuum throughout the year. Therefore, studies aimed at establishing monthly or even biweekly changes in ACTH and other biomarkers would provide a more accurate picture and aid in establishing regional reference intervals. Results that differ, and sometimes contrast, between studies highlight the importance of generating geographic-specific and seasonally adjusted data for markers of endocrine function in equids. Further studies including large numbers of animals of various breeds, ages, and levels of exercise are needed to improve assessment of an individual animal’s health status.

Funding Statement

This work was supported by Prairie Diagnostic Services, Inc., who conducted the laboratory analyses.