Abnormalities of insulin metabolism include hyperinsulinaemia and insulin resistance, and these problems are collectively referred to as insulin dysregulation in this review. Insulin dysregulation is a key component of equine metabolic syndrome: a collection of endocrine and metabolic abnormalities associated with the development of laminitis in horses, ponies and donkeys. Insulin dysregulation can also accompany prematurity and systemic illness in foals. Causes of insulin resistance are discussed, including pathological conditions of obesity, systemic inflammation and pituitary pars intermedia dysfunction, as well as the physiological responses to stress and pregnancy. Most of the discussion of insulin dysregulation to date has focused on insulin resistance, but there is increasing interest in hyperinsulinaemia itself and insulin responses to feeding. An oral sugar test or in-feed oral glucose tolerance test can be performed to assess insulin responses to dietary carbohydrates, and these tests are now recommended for use in clinical practice. Incretin hormones are likely to play an important role in postprandial hyperinsulinaemia and are the subject of current research. Insulin resistance exacerbates hyperinsulinaemia, and insulin sensitivity can be measured by performing a combined glucose-insulin test or i.v. insulin tolerance test. In both of these tests, exogenous insulin is administered and the rate of glucose uptake into tissues measured. Diagnosis and management of hyperinsulinaemia is recommended to reduce the risk of laminitis. The term insulin dysregulation is introduced here to refer collectively to excessive insulin responses to sugars, fasting hyperinsulinaemia and insulin resistance, which are all components of equine metabolic syndrome.
Equine metabolic syndrome (EMS) is a collection of endocrine and metabolic abnormalities associated with the development of laminitis in horses, ponies and donkeys . Insulin dysregulation, hyperleptinaemia, increased adiposity and hypertriglyceridaemia are components of EMS, and are risk factors for laminitis . Hyperinsulinaemia is a particular concern because laminitis has been experimentally induced in healthy ponies and horses by infusing exogenous insulin with glucose i.v. [3, 4]. Generalised obesity and/or regional adiposity accompany insulin dysregulation in many equids with EMS. Horses and ponies with EMS should also be evaluated for evidence of laminitis, including divergent hoof growth rings (founder lines) or third phalanx rotation. Mild laminar pathology does not always cause overt lameness and can go unrecognised by horse owners. Increased adiposity is evident before laminitis in most animals, but a lean EMS phenotype also exists, with insulin dysregulation and laminitis occurring in horses and ponies with normal appearance. Other terms used in the past to describe EMS include prelaminitic metabolic syndrome, syndrome X, insulin resistance syndrome, obesity-associated insulin resistance (IR) and prediabetes mellitus [5, 6].
The discussion of hyperinsulinaemia in equids has been complicated by the different approaches used in the past to assess glucose and insulin dynamics. Insulin concentrations have been measured in the fed state in several studies [2, 7-10]. In others, horses have been fasted [11-13], and fasting prior to sample collection was recommended in the 2010 American College of Veterinary Internal Medicine consensus statement .
Fasting hyperinsulinaemia results from persistent stimulation of pancreatic β cells, and in obese human individuals increased insulin secretion during the interprandial period is attributed to elevated fatty acid concentrations . Hyperinsulinaemia has also been associated with obesity in equids and higher fatty acid concentrations were detected in obese insulin-resistant horses in one study, although hay was available at the time of sampling . Beta-cell hyperplasia is also a potential cause of fasting hyperinsulinaemia. Several hormones and nutrients act as insulin secretagogues and increase insulin concentrations, including incretin hormones , glucagon, glucose, fatty acids and arginine. Insulinomas can be accompanied by hypoglycaemia and an increased insulin:glucose ratio, and are a rare cause of hyperinsulinaemia in other animals [16, 17]. Only one case of insulinoma has been reported in equids, in a 12-year-old Shetland pony that developed seizures . Hypoglycaemia and hyperinsulinaemia were detected during seizures and increased insulin secretion was demonstrated during an oral glucose tolerance test and an i.v. glucagon tolerance test.
Postprandial hyperinsulinaemia is a concern in equids because feeding occurs on an almost continuous basis when horses and ponies are grazing on pasture. Even when fasted, digestion continues for many hours within the equine caecum and large intestine, with production of the gluconeogenic substrate propionic acid and amino acids . Testing horses in the fed state allows for better assessment of insulin dysregulation, but feeds vary in composition and this makes it difficult to establish reference ranges. Standardised tests are therefore recommended to diagnose hyperinsulinaemia in horses. The first option is an oral sugar test that is gaining acceptance in the USA, where corn syrup is readily available for purchase , and the second is an in-feed oral glucose test that is used in the UK, performed by mixing glucose powder into a chaff-based feed in an amount equivalent to 1 g/kg bwt, with samples collected 2 h later .
Hyperinsulinaemia is caused by increased insulin secretion or delayed insulin clearance, and this problem may be a cause or a consequence of IR (Fig 1). The accepted explanation for hyperinsulinaemia is that insulin secretion increases as a consequence of decreased tissue insulin sensitivity and this is referred to as compensated IR . Decreased hepatic insulin clearance is also associated with IR and can contribute to hyperinsulinaemia because >70% of insulin secreted from β cells is normally cleared from the portal blood before reaching the peripheral circulation in horses . Low connecting (C)-peptide/insulin ratios have been detected in horses with EMS and provide evidence of decreased insulin clearance . Connecting-peptide is a component of the proinsulin prohormone, and is cleaved prior to secretion of the mature insulin peptide from pancreatic β cells. Since C-peptide is released in equal amounts to insulin, concentrations reflect insulin secretion. Unlike insulin, which has a plasma half-life of only 5–8 min in humans, C-peptide does not undergo rapid clearance from the blood by the liver, so blood concentrations reflect pancreatic function and can be used to evaluate hepatic insulin clearance.
Although the cause-and-effect relationship between IR and hyperinsulinaemia appears straightforward, the authors have encountered horses and ponies with postprandial hyperinsulinaemia that have normal glucose and insulin responses during the combined glucose-insulin test (CGIT), which is a test for insulin sensitivity. Other causes of postprandial hyperinsulinaemia following ingestion of sugars should therefore be considered, including increased stimulation by incretin hormones, including glucagon-like peptide 1 and glucose-dependent insulinotropic peptide . These incretin hormones are secreted from L and K cells, respectively, within the small intestine in response to ingestion of sugars, amino acids and fats, and the activity of both hormones is regulated through degradation by the enzyme dipeptidyl peptidase 4. Both hormones have been identified in equine plasma after oral administration of glucose [23, 24]. Incretins stimulate pancreatic β cell insulin secretion and slow gastric emptying, thereby minimising postprandial hyperglycaemia. Postprandial hyperinsulinaemia in horses with EMS might therefore reflect increased incretin hormone secretion or decreased degradation by dipeptidyl peptidase 4. Furthermore, incretin hormones increase pancreatic islet mass in rodents by stimulating β cell proliferation, induction of islet neogenesis, and inhibition of β cell apoptosis . It is conceivable that chronic incretin hormone stimulation contributes to β cell hyperplasia and hyperinsulinaemia in horses with EMS. In humans, fasting glucagon-like peptide 1 concentrations increase in association with greater clustering of metabolic syndrome components .
If the first manifestation of insulin dysregulation in genetically predisposed horses is postprandial hyperinsulinaemia, then IR would raise insulin concentrations even further. These relationships are illustrated in Figure 2. Obesity, systemic inflammation and concurrent endocrinopathies such as pituitary pars intermedia dysfunction (PPID) are potential causes of IR that can exacerbate hyperinsulinaemia by slowing insulin clearance. Hyperinsulinaemia might even induce IR itself through the process of homologous desensitisation, a process by which tissue sensitivity to insulin decreases as blood hormone concentrations increase . Acute increases in insulin concentrations can still stimulate glucose uptake, but chronic hyperinsulinaemia results in downregulation of the insulin receptor and downstream signalling. This is the mechanism by which insulinomas induce IR and is referred to as the Somogyi effect, when higher doses of exogenous insulin have a diminished effect on blood glucose concentrations . Evidence for hyperinsulinaemia inducing IR was provided by a recent study in mice : a high-fat diet induced obesity and IR, and streptozotocin was administered to one group of mice to lower insulin concentrations. Insulin resistance was only detected in mice that developed hyperinsulinaemia.
Obesity has been a component of EMS since the syndrome was first proposed in 2002 , but hyperinsulinaemia is also detected in lean horses and ponies  and some obese animals have normal insulin concentrations [11, 12]. These inconsistencies can be explained by considering obesity as modifying factor, with the genetics of the individual animal determining the magnitude of insulin dysregulation. Hyperinsulinaemia itself might even promote obesity through the anabolic effects of insulin on lipid metabolism  and contribute to the ‘easy keeper’ or ‘good doer’ phenotype. Horses and ponies with this phenotype readily become obese when overfed, and obesity can develop when grass is the only source of energy. A recent study performed with knockout mice confirmed that hyperinsulinaemia alters lipid metabolism and promotes obesity . Genetic manipulations affecting the brain and pancreas generated mice that were incapable of diet-induced hyperinsulinaemia, and these animals were also protected from obesity and hepatic lipid infiltration. Hyperinsulinaemia, obesity and hepatic IR might therefore represent a vicious cycle in equids if lipid infiltration of the liver slows insulin clearance and higher fatty acid concentrations induce fasting hyperinsulinaemia.
Relative hypoinsulinaemia and diabetes mellitus
The term hypoinsulinaemia is rarely used because lower limits are not provided for insulin reference ranges. However, the concept of relative hypoinsulinaemia is a key component of diabetes mellitus. Type 1 diabetes mellitus is most commonly caused by immune-mediated destruction of pancreatic β cells, resulting in inadequate insulin secretion, and a rarer idiopathic form is also recognised in humans. In contrast, type 2 diabetes mellitus develops when insulin production by β cells decreases following a prolonged period of hyperinsulinaemia. The term pancreatic insufficiency or pancreatic exhaustion are used to describe this process of decompensation, and Treiber et al.  adopted the term uncompensated insulin resistance to describe horses and ponies with high blood glucose concentrations that are inadequately controlled by insulin. In diabetes mellitus, pancreatic insulin secretion decreases as a result of direct damage to the pancreas and loss of β cell mass, whereas inhibition by hormones such as somatostatin and catecholamines can suppress insulin secretion during stress or illness . In veterinary medicine, type 1 diabetes mellitus is most commonly observed in dogs, whereas type 2 diabetes mellitus is more often detected in cats. Transient Type 1 diabetes mellitus has recently been reported in a foal , and there are reports of type 2 diabetes mellitus in mature horses in association with pancreatitis , PPID [33, 34], granulosa cell tumour , and presumed autoimmune polyendocrine syndrome .
Increased understanding of glucose and insulin dysregulation in horses has led to more frequent recognition of diabetes mellitus, particularly in horses with PPID . Horses with advanced PPID can develop transient diabetes mellitus when stress and systemic disease are encountered, so it is important to monitor blood glucose and triglyceride concentrations in hospitalised patients. Hyperglycaemia is marked (>11 mmol/l [2.0 mg/l]) in some horses and accompanied by glucosuria. Severe hypertriglyceridaemia is also a concern in horses with diabetes mellitus, and insulin therapy is required to manage hyperglycaemia and suppress hormone-sensitive lipase. This condition appears to involve β cell insufficiency because horses are very sensitive to exogenous insulin and return to adequate glycaemic control when the primary problem resolves, or when pergolide treatment is initiated or increased. Hyperglycaemia alters the amino acid lysine and this leads to the formation of fructosamine, which alters other amino acids and creates advanced glycosylation end-products that damage tissues. Higher fructosamine concentrations have been detected in laminitic horses  and are likely to reflect the accompanying stress- and pain-associated hyperglycaemia.
Insulin resistance refers to the failure of insulin-sensitive tissues to respond to insulin, and skeletal muscle, adipose, and liver are primarily affected. Consequences of IR include impaired glucose uptake into tissues, increased glucose synthesis in the liver via gluconeogenesis, and increased lipolysis resulting in higher circulating free fatty acid concentrations. Insulin resistance is associated with glucose transporter 4 dysfunction within skeletal muscle and adipose tissues [38, 39].
Insulin resistance and glucose
Resting glucose concentrations are often towards the higher end of reference range in horses and ponies with EMS , and it is reasonable to assume that both hyperglycaemia and hyperinsulinaemia are encountered when these animals are grazing on pasture (Fig 3). Glucotoxicity is a consequence of persistent hyperglycaemia in human diabetes mellitus patients, and hyperglycaemia can result in glycosylation of amino acids within tissues and the formation of advanced glycosylation end-products. Tissues become damaged as a result of glycosylation and this process contributes to the development of diabetic angiopathy . Endothelial dysfunction is also induced as hyperglycaemia stimulates diacylglycerol synthesis and activation of protein kinase C. Hyperglycaemia might therefore contribute to the development of laminitis in horses, and this hypothesis has been tested in 2 studies. Hoof lamella changes suggestive of laminitis were induced in healthy Standardbred horses by infusing glucose i.v. for 48 h . However, in a related study using the hyperinsulinaemia model  advanced glycosylation end-products were not detected in hoof tissues collected after laminitis was induced. Glucose might therefore play a role in the development of acute laminitis by inducing endogenous insulin secretion or altering endothelial function, but not through glycosylation of amino acids.
Insulin resistance and obesity
Obesity is a pathological condition associated with altered adipokine production and IR, and inflammation plays an important role in obesity in other species [43-46]. As adipocytes reach their maximal storage capacity, events leading to energy failure, inflammation and cellular stress are initiated . Hypoxia is a factor in adipocyte stress because oxygen only diffuses approximately 100 μm from capillary beds, and adipocyte diameter expands beyond this limit in obesity [47, 48]. The endothelium of vessels within adipose tissues can also become unresponsive to nitric oxide-induced vasodilation, and this further contributes to hypoxia [47, 49]. Hypoxia leads to a reduction in mitochondrial function and stimulation of hypoxia-induced genes that promote inflammation. Stressed adipocytes, and those that have become necrotic as a result of hypoxia, release cytokines such as macrophage chemoattractant proteins that recruit new macrophages to adipose tissues [47, 50-52]. Resident tissue macrophages within adipose tissue tend to be anti-inflammatory in profile, whereas those recruited by stressed cells have a more proinflammatory profile . Resident macrophages can also convert to a proinflammatory profile under conditions of adipose tissue hypoxia or when influenced by recruited proinflammatory macrophages [53, 54].
In addition to suffering from hypoxia, adipocytes that are laden with triglyceride lose their ability to buffer lipid fluxes within the body [47, 55]. Ectopic lipid deposition and lipotoxicity ensue, contributing to the development of IR in adipose and other tissues. Lipid accumulation in liver, muscle, pancreas and other tissues has a negative impact on insulin metabolism throughout the body [55-58]. Insulin normally induces nitric oxide-dependent vasodilation in certain capillary beds, so loss of insulin sensitivity further impedes postprandial capillary recruitment and the delivery of lipids to appropriate storage depots, promoting ectopic deposition . Accumulation of intermediate lipid metabolites such as diacylglycerol, long-chain fatty acyl CoA, and ceramide exacerbates cell stress and incites inflammation, and this leads to IR [59-61].
Inflammatory mediators secreted from adipose tissue act locally and at distant locations, which may be relevant to equine laminitis because the vascular endothelium is targeted [62, 63]. In obese human individuals and rodents, increased adipose tissue secretion of inflammatory mediators and adipokines is well described , but studies on associations between obesity and inflammatory cytokines have yielded conflicting results in horses. Vick et al.  detected positive correlations among body condition score, blood mRNA expression of tumour necrosis factor α (TNFα) and interleukin (IL)-1, and blood TNFα protein concentrations in light breed mares. Regional adiposity, assessed using the cresty neck score , has also been associated with increased plasma TNFα protein concentrations in ponies . Previously laminitic ponies had significantly higher plasma TNFα protein concentrations than nonlaminitic ponies, and the same animals had higher neck scores and insulin concentrations. Detection of hyperinsulinaemia in horses and ponies with regional adiposity and laminitis raises the question of whether insulin affects inflammatory cytokine concentrations in horses. This has been demonstrated in horses using an experimental model, with significantly higher plasma TNFα and IL-6 protein concentrations detected during i.v. infusion of insulin for 6 h . Average TNFα concentrations were 16% and 20% higher than baseline at 4 and 6 h, respectively, in the insulin group.
Not all studies have supported the development of a basal proinflammatory state in obese and hyperinsulinaemic horses. Holbrook et al.  demonstrated lower endogenous gene expression of IL-1 and IL-6 in peripheral blood mononuclear cells from hyperinsulinaemic obese horses than in controls, and comparable cytokine responses in both groups following peripheral blood mononuclear cell stimulation. However, neutrophil oxidative burst activity was significantly higher in obese horses, suggesting that immune function is altered by obesity and hyperinsulinaemia. Suagee et al.  were unable to associate obesity with plasma concentrations of TNFα, IL-1β or IL-6 protein in a larger population of 110 horses. Of the variables examined, only serum amyloid A concentrations were positively correlated with insulin concentrations and body condition score. When insulin-resistant and insulin-sensitive horses were compared by Burns et al. , no differences in TNFα, IL-1β, IL-6 or macrophage chemoattractant protein-1 mRNA expression were detected within adipose tissues collected from multiple sites. This study revealed, however, that expression of IL-1β and IL-6 was significantly higher in nuchal ligament adipose tissue, than in other depots, suggesting that regional adiposity contributes to the increased risk of laminitis in insulin-resistant horses. It therefore remains challenging to separate the effects of adiposity, hyperinsulinaemia and IR on inflammatory cytokine profiles in equids.
Hepatic insulin resistance
Hepatic lipid infiltration accompanies obesity in man and can lead to the development of nonalcoholic fatty liver disease [68-70]. Increased plasma gamma glutamyl transferase activity and post mortem evidence of hepatic lipidosis have also been detected in obese horses (N. Frank, unpublished data). Enhanced peripheral lipolysis secondary to IR represents the major cause of lipid accumulation in the steatotic liver, with a smaller contribution by de novo lipid synthesis within hepatocytes . Intermediates of fatty acid metabolism are agonists for Kupffer cell pattern recognition receptors and therefore trigger an inflammatory cascade [56, 72], which induces hepatic IR and contributes to systemic inflammation . Recent studies suggest that IL-6 plays an important role in hepatic IR by suppressing insulin receptor autophosphorylation and tyrosine phosphorylation of insulin receptor substrate 1 .
Hepatic IR may also lead to the development of hypertriglyceridaemia in horses and ponies. Although plasma triglyceride and very low-density lipoprotein (VLDL) concentrations are only mildly increased in horses and ponies with obesity and hyperinsulinaemia [2, 10], they are potential indicators of hepatic IR. Insulin secreted from the pancreas arrives at the liver via the portal blood and normally suppresses VLDL secretion after eating, so that triglycerides are stored at times of positive energy balance. Pathways involved in insulin-mediated suppression of VLDL secretion are affected by hepatic lipidosis and IR, and lipoprotein secretion increases as a result . Increased ferritin concentrations are also detected in human nonalcoholic fatty liver disease patients, and serve as a predictor of liver damage . Ferritin concentrations were recently measured in obese horses and a positive correlation existed between ferritin concentrations and insulin area under the curve values after oral dextrose administration and corn ingestion .
Insulin resistance and adipokines
Leptin is an adipocyte-derived hormone that is produced almost exclusively by white adipose tissue, and concentrations increase in obese and insulin-resistant horses as a result of increased fat mass and leptin resistance [2, 10, 78-80]. Leptin acts as an adiposity signal to keep other systems in the body apprised of the adipose depot size, and interaction between leptin and its long-form receptor in the hypothalamus modulates satiety [81, 82]. Intact leptin signalling results in appetite suppression  and might also enhance energy expenditure and reduce hepatic fatty acid synthesis . In humans, leptin resistance is implicated in overeating and obesity . Leptin concentrations are positively correlated with body condition score in horses, ponies and donkeys [78, 85], although hyperleptinaemia is also detected in equids with normal body condition scores. Hyperinsulinaemia and tissue IR are both associated with hyperleptinaemia in mares . Leptin might therefore serve as an indirect measure of adipocyte pathology and insulin dysfunction, and leptin concentrations are included on a diagnostic panel for EMS, with a cut-off value of 4 μg/l for hyperleptinaemia.
Adiponectin is an insulin-sensitising hormone with pleotropic effects that include enhancement of fatty acid oxidation and improvement of glucose tolerance . This adipokine also possesses anti-inflammatory properties and counteracts IR caused by cytokines such as TNFα . Adiponectin circulates in the form of trimers, hexamers and high molecular weight multimers , and lower concentrations of total  and high molecular weight  adiponectin have been detected in obese horses. High molecular weight adiponectin concentrations were negatively correlated with insulin concentrations .
Leptin and adiponectin are important to the discussion of EMS and laminitis because both adipokines have direct effects on the vascular endothelium, as well as indirect effects via modulation of insulin sensitivity and inflammation [91-93]. Adiponectin has anti-inflammatory actions and can therefore counteract the effects of cytokines such as TNFα on the endothelium [82, 94-97]. In contrast, leptin is generally proinflammatory and exacerbates endothelial activation and the generation of free radicals that damage the endothelium [98, 99]. Obese people with adiponectin deficiency have a higher risk of mortality during sepsis, and loss of adiponectin's immunomodulatory effects and protective actions on the endothelium is believed to be a contributing factor .
Insulin resistance and PPID
It is often assumed that PPID predictably causes IR, but this assumption must be challenged on the basis of clinical experience and the detection of hyperinsulinaemia in only 32% of horses with PPID . There are a number of mechanisms that might link PPID with hyperinsulinaemia and IR in horses. Cortisol antagonises the actions of insulin and causes IR, with accompanying hyperinsulinaemia, as is readily demonstrated in the rat by administering dexamethasone . Administration of dexamethasone to healthy horses also increases plasma insulin concentrations and lowers insulin sensitivity [102, 103]. Hyperinsulinaemia and decreased insulin sensitivity are recognised consequences of hyperadrenocorticism in dogs, and diabetes mellitus is commonly detected . Higher C-peptide concentrations are also detected following glucagon infusion in dogs with hyperadrenocorticism and provide evidence of increased insulin secretion . Corticotropin-like intermediate peptide is a derivative of pro-opiomelanocortin and adrenocorticotrophic hormone (amino acids 18–39), and concentrations increase in horses with PPID [106, 107]. This hormone acts as an insulin secretagogue when added to the media of pancreatic explants , and might therefore contribute to the development of hyperinsulinaemia in equine PPID. Alpha melanocyte-stimulating hormone (αMSH) concentrations are also increased with PPID  and this hormone, along with leptin, plays an important role in energy metabolism. Under normal conditions, αMSH stimulates neural melanocortin receptors 3 and 4 within the hypothalamus, opposes the action of leptin, and reduces body fat mass . However, downregulation of this system as a result of increased αMSH concentrations might have the opposite effect and raise insulin concentrations through increased visceral adiposity and hepatic IR.
One explanation for the variability in insulin concentrations detected in horses with PPID is that the condition itself is less predictable than other forms of hyperadrenocorticism. Adrenal hyperplasia is a common manifestation of hyperadrenocorticism caused by pars distalis tumours in dogs, whereas only approximately 20% of horses with PPID develop this problem . Polyuria and polydipsia are also common features of canine hyperadrenocorticism, but are only reported in 30% of horses with PPID .
Convergence of PPID and EMS
An alternative explanation for the variability in insulin concentrations detected in equids with PPID is that this endocrinopathy acts only as a modifying factor (Fig 2). If hyperinsulinaemia is a genetically determined trait in equids, then PPID exacerbates the problem by stimulating insulin secretion via increased corticotrophin-like intermediate peptide concentrations or by inducing IR though hyperadrenocorticism. When viewed from this perspective, it is not surprising that in the study by van der Kolk et al.  hyperinsulinaemia had 92% sensitivity as a diagnostic test for PPID because ponies were over-represented. Hyperinsulinaemia occurs independently of PPID in ponies  and is detected at all ages [2, 9]. If PPID is an exacerbating factor for hyperinsulinaemia, even as pituitary dysfunction is first developing, then early diagnosis of this endocrine disorder is essential. Detection of higher insulin concentrations at noon, rather than 08.00 h, in horses with PPID fed hay and chaff the morning of testing also suggests that postprandial hyperinsulinaemia is exacerbated by PPID , and this could increase the risk of hyperinsulinaemia-induced laminitis.
Systemic inflammation and the stress response to illness
Insulin resistance can develop as a result of mediators released during the systemic inflammatory response (Fig 2). A number of the inflammatory mediators upregulated by nuclear factor κB signalling, as well as the counter-regulatory hormones cortisol, epinephrine, glucagon and growth hormone, can induce IR [117, 118]. Transient IR has been induced in healthy horses by administering lipopolysaccharide i.v., and this is presumed to occur through increased production of proinflammatory cytokines [119, 120]. Cross-talk between the insulin signalling cascade and inflammatory signalling cascades occurs on many levels, and IR results from inappropriate serine phosphorylation of insulin receptor substrates and the insulin receptor itself . Cortisol downregulates the expression of proteins involved in the phosphoinositide 3-kinase signalling cascade [122, 123], including the p85 regulatory subunit . Glucocorticoids also reduce cellular insulin receptor substrate-1 concentrations in adipocytes [122, 123] and inhibit translocation of glucose transporter 4 to the plasma membrane .
The overall goal of these counter-regulatory mechanisms is to ensure the provision of adequate energy in the form of carbohydrates, proteins and lipids to meet the body's requirements while responding to infection . Inducing a state of peripheral and hepatic IR tips the balance in favour of glucose liberation from tissue stores and glucose synthesis via gluconeogenesis in the liver . However, stress-associated inhibition of insulin secretion by catecholamines and relative hypoinsulinaemia might also exacerbate hypertriglyceridaemia in hospitalised horses . Hyperadrenocorticism increases lipolysis and ketogenesis in horses , although exogenous corticosteroid administration did not exacerbate hypertriglyceridaemia in one study of ponies . Although the effects of catecholamines on fat metabolism have not been extensively studied in equids, in other species plasma triglyceride concentrations increase in response to administration or endogenous release of catecholamines [131, 132].
Alterations in insulin metabolism in the foal
The endocrine system of the foal undergoes many adaptations in the perinatal period and this subject was recently reviewed by Fowden et al. . Insulin sensitivity decreases in foals during the first 24 h of life and this is attributed to increased circulating cortisol and catecholamine concentrations. Clearance of exogenous glucose is slower during the first 12–24 h after birth, despite appropriate insulin responses . This may be part of the adaptation to intermittent nursing, rather than continuous delivery of glucose via the placenta. Relative hypoinsulinaemia may be observed in normal foals during the immediate post partum period, and glucose-stimulated insulin secretion from pancreatic β cells is significantly lower in healthy foals 2 h after birth, compared to one week of age [135, 136]. It has been suggested that perinatal alterations in the abundance of insulin receptors and intracellular signalling proteins affect insulin sensitivity in foals, as has been noted in other species .
Survival of premature and dysmature foals is influenced by the degree of maturity of several key endocrine systems [133, 137]. In particular, immaturity of the hypothalamic-pituitary-adrenal axis is associated with impaired stress responses and increased mortality in the face of sepsis [138-140]. Hypoglycaemia and hyperglycaemia can both occur in critically ill foals [141, 142], as in human neonates [143, 144]. Premature human infants and those born small for gestational age because of impaired placental function are at risk for neonatal hyperglycaemia . Both peripheral and hepatic IR are demonstrated in preterm human infants [143, 145], and IR can be further exacerbated by sepsis and other inflammatory conditions [144, 146]. Relative hypoinsulinaemia also contributes to the development of hyperglycaemia because pancreatic β cells are immature in premature infants and fail to efficiently process proinsulin to biologically active insulin . As in humans, normal physiological processes, including transient reduction in insulin sensitivity within 24 h of birth and relative insulin deficiency caused by inefficient β cell secretory function, render neonatal foals susceptible to hyperglycaemia and other metabolic derangements in the face of prematurity or critical illness.
Insulin concentrations are low after birth in premature foals delivered via induced parturition before 320 days, and pancreatic β cell responses to exogenous glucose at age 2 h are also blunted in comparison to full-term foals . These findings are attributed to the inhibitory effects of increased catecholamine concentrations in premature foals. In contrast, increased β cell responsiveness to exogenous glucose, without an accompanying change in glucose clearance, has been observed in foals born 24–48 h premature after induced parturition . It is assumed that pancreatic compensation is occurring in response to decreased peripheral insulin sensitivity as a result of hypercortisolaemia. Studies examining insulin sensitivity in healthy, full-term human infants born small for gestational age have yielded mixed results, with both increased and decreased insulin sensitivity documented over the first 2–3 days of life [148, 149]. Relative insulin deficiency resulting from reduced pancreatic β cell mass has also been suggested as a contributing factor to hyperglycaemia , which is associated with increased morbidity and mortality in human neonates [150, 151].
Sepsis is a leading cause of death in neonatal foals [152-156]. Septic foals have lower glucose and insulin concentrations and higher triglyceride and glucagon concentrations than healthy foals, and higher insulin and lower leptin concentrations are associated with mortality . Hyperglycaemia is associated with increased mortality in critically ill foals [138, 141] and is likely to result from increased production of inflammatory cytokines and stress hormones [116, 157]. Sepsis-associated hyperglycaemia develops as stress hormones such as epinephrine and cortisol induce gluconeogenesis and IR, and toxins including lipopolysaccharide bind to Toll-like receptor 4 and impair insulin signalling. Although some septic foals experience relative adrenal insufficiency, circulating concentrations of cortisol and other stress hormones such as catecholamines can still be markedly increased and contribute to IR [138, 139].
In addition to pathological causes of IR, such as obesity, systemic inflammation, hyperadrenocorticism and acromegaly, there are physiological causes such as pregnancy and stress. Pregnancy is associated with IR in people and horses (Fig 2), and represents a normal physiological adaptation . In women, insulin action is 50–70% lower than normal during late pregnancy, and both basal and 24 h mean insulin concentrations show a 2-fold increase . Postprandial glucose concentration, basal endogenous hepatic glucose production and total gluconeogenesis are also increased during late pregnancy. These alterations are attributed to rising concentrations of prolactin, cortisol and glucagon, and ensure continuous delivery of nutrients to the growing fetus . Hyperinsulinaemia, enhanced β cell sensitivity to endogenous and exogenous glucose, increased degradation of insulin, and decreased insulin sensitivity have all been demonstrated in pregnant mares . George et al.  also used the frequently sampled i.v. glucose tolerance test (FSIGTT) and minimal model analysis to demonstrate that pregnant Thoroughbred mares had slower glucose clearance and greater insulin secretion at 28 weeks of gestation than nonpregnant mares. These mares developed higher glucose and insulin concentrations after meals, consistent with lower insulin sensitivity and increased insulin secretion.
Assessment of insulin sensitivity
Tissue insulin sensitivity is most accurately assessed by performing dynamic tests such as the hyperinsulinaemic-euglycaemic clamp (HEC) procedure and FSIGTT with minimal model analysis . The HEC procedure involves i.v. infusion of soluble (regular) insulin and concurrent infusion of dextrose solution to maintain glucose concentrations at the preinfusion euglycaemic level. Because blood glucose concentrations are held constant during the insulin infusion, the glucose infusion rate represents insulin sensitivity in muscle and adipose tissues. Vick et al.  evaluated a population of 60 mixed light horse breed mares on pasture using this test and found that body condition scores were significantly and positively correlated with resting insulin concentrations (Spearman's rho = 0.53), and negatively correlated with insulin sensitivity (Spearman's rho = -0.57) at the time of testing. To perform the FSIGTT procedure, an i.v. bolus of dextrose is administered first, followed several minutes later by an i.v. insulin bolus. While these tests provide the most information about glucose and insulin metabolism, the time and equipment required to perform them largely restrict their utility to the research setting.
An insulin tolerance test (ITT) can also be used to measure tissue insulin sensitivity, and this procedure involves i.v. injection of soluble insulin as a bolus in amounts ranging from 20 to 125 μiu/kg bwt, followed by serial blood sampling over a 3 h period . Area under the glucose curve is calculated, or the dose of insulin required to cause a 50% decline in glucose concentration can be recorded. The ITT has been simplified to a 2-step insulin-response test consisting of an i.v. insulin injection of 100 μiu/kg bwt soluble insulin followed by blood glucose determination at 0 and 30 min after injection . Insulin-resistant horses fail to undergo a 50% reduction in blood glucose concentration within 30 min. When 6 insulin-resistant horses were compared with 6 controls, the 2-step ITT performed as well as the conventional test that required 10 blood samples, and is therefore practical to use in the clinical setting.
A CGIT can also be performed to assess insulin sensitivity, and involves i.v. infusion of dextrose solution, followed immediately by soluble insulin [10, 163]. Blood glucose concentrations initially increase and then decrease in response to exogenous (and endogenous) insulin. The time required for blood glucose concentrations to return to baseline is recorded and the insulin concentration is measured in blood collected at 45 min. The rate of decrease in glucose concentrations caused by exogenous insulin reflects tissue insulin sensitivity, whereas the insulin concentration at 45 min is affected by insulin clearance rate and pancreatic insulin output. Insulin resistance is diagnosed when the blood glucose concentration fails to return to baseline by 45 min. However, IR may also be present if normal glucose results are accompanied by an increased (>100 μiu/ml) insulin concentration at 45 min. In these cases, compensated IR or β cell hyperplasia are suspected, with compensatory increases in insulin secretion remaining sufficient to maintain normal glucose homeostasis. Uncompensated IR, in contrast, results in higher glucose concentrations after dextrose infusion with lower peak insulin concentrations.
Care must be taken when interpreting results of earlier studies that assessed insulin sensitivity on the basis of resting insulin concentrations. Factors such as stress or duration of fasting can affect resting insulin concentrations and introduce variability. Since proxy measures of insulin sensitivity, including the reciprocal of the square root of insulin (RISQI), are calculated using resting insulin concentrations , hyperinsulinaemia caused by factors other than IR could also affect these results. The same concern can be raised with acute insulin response to glucose values calculated from FSIGTT results , because enhanced insulin secretion caused by increased β cell function or hyperplasia might also raise acute insulin response to glucose values. Proxy measures of insulin sensitivity remain useful screening tests for detecting insulin dysregulation in larger populations, but dynamic tests involving the administration of exogenous insulin are preferred for the diagnosis of IR. Further studies are required to determine the degree of agreement among FSIGTT, HEC, CGIT and ITT results in the same horses.
Management of hyperinsulinaemia and insulin resistance
Potential exacerbating factors for hyperinsulinaemia, including obesity, high-sugar diets, inadequate exercise and concurrent PPID, must be considered before establishing a management plan. If obesity is present, the initial focus of management should be weight loss, and when PPID is diagnosed, pergolide should be prescribed. A complete diet history should be collected and assessed, and the diet adjusted to reduce the amount of sugars available if hyperinsulinaemia is a concern. Horses should be regularly exercised, as this is an integral part of weight management plans when obesity is a factor.
Body fat mass can be reduced in obese horses and ponies by lowering energy intake and increasing exercise . Some horses with EMS are being overfed by their owners, and removal of grain from the diet is sufficient to induce weight loss. Other obese horses and ponies are resistant to weight loss and require greater reductions in energy intake . Pasture access should be restricted in obese animals because obesity often persists until grass intake is limited, and sugars and amino acids contained in pasture grass contribute to hyperinsulinaemia. Recommended strategies for limiting grass consumption include placing the obese horse and a companion in a small grass paddock that is 0.13–0.2 ha in size (one-third to one half an acre), or the application of a grazing muzzle during turnout. Responses to management should then be reassessed and housing conditions adjusted over time. If removing grain from the diet and limiting pasture access does not result in weight loss, the amount of hay fed should be progressively reduced. Hay should initially be fed in amounts equivalent to 1.5% of current bodyweight and then lowered to 1.5% of ideal bodyweight. An obese horse or pony that is not losing body mass after 4 weeks on this diet can be reduced to 1.25%, and then down to the minimum level of 1.0% bodyweight after another month if there has been no progress. A balanced vitamin/mineral supplement or ration balancer should be provided to horses on a hay-only diet to maintain adequate micronutrient and protein intake.
Attention must also be paid to the sugar content of the diet, and the importance of this component of the management plan depends on the severity of hyperinsulinaemia. Hay containing large amounts of simple sugars and hydrolysable starches is a significant concern for severely affected horses. Analysis of the hay is recommended for these animals to ensure that the nonstructural carbohydrate content of the forage is <10% on a dry-matter basis . Hay can also be soaked in cold water for one hour to lower the nonstructural carbohydrate content, although results vary according to the type of hay used . A supplement containing additional protein, vitamins and minerals should also be provided. Hyperinsulinaemic horses with normal body mass are more challenging to manage. The first question to be addressed is whether the animal is concurrently affected by PPID, so horses older than 10 years should undergo testing for both PPID and IR [112, 167]. Previously obese horses that remain hyperinsulinaemic after weight loss must also be managed carefully. If weight gain is required, feeds providing more energy in the form of fat and fibre should be selected.
Two medical treatments are also used to manage horses with EMS and both have specific indications.
Levothyroxine sodium can be administered to accelerate weight loss when severe insulin dysregulation is present in an obese horse that cannot be exercised because of laminitis, or when obesity persists despite intensive diet and exercise management. Levothyroxine sodium lowers bodyweight and improves insulin sensitivity in horses , and can be administered by mouth or in the feed at a dosage of 0.1 mg/kg bwt once daily. A higher dosage of 0.15 mg/kg bwt can be selected if obesity persists after 3 months of treatment. Levothyroxine is administered until the horse reaches an ideal body condition score or until the end of a 6-month period. When treatment is discontinued, the dosage is lowered to 0.05 mg/kg bwt/day for 2 weeks and then 0.025 mg/kg bwt/day for 2 weeks. Complications have not been associated with the administration of levothyroxine to healthy horses for 12 months , although some horses exhibit increased activity and mild hyperexcitability when treated. Pasture access must be restricted when horses are treated with levothyroxine to prevent the horse from increasing its own feed intake. Levothyroxine is reasonably priced in the USA, but expensive in the UK.
Metformin hydrochloride is commonly administered to treat diabetes mellitus in man and has also been reported to improve resting insulin concentrations and proxy measures of insulin sensitivity in horses and ponies . However, studies evaluating metformin use in equids have yielded conflicting results, and insulin sensitivity did not improve in nonobese insulin-resistant ponies when i.v. glucose tolerance tests were performed to assess glucose and insulin dynamics . This discrepancy can be explained by the low oral bioavailability of metformin in horses  and recent results indicating that metformin acts at the level of the intestine to blunt blood glucose and insulin concentrations after feeding . Metformin therefore appears to be better suited to controlling postprandial hyperinsulinaemia than to treating IR in horses. The current recommendation for metformin is 30 mg/kg bwt per os q. 8–12 h, given one hour before feeding or turnout.
Looking forward, our understanding of insulin dysregulation in mature horses will be further enhanced when results of studies examining the genetic basis of EMS are completed and incretin biology is further investigated in the horse. Insulin dysregulation in foals also requires additional research so that glucose concentrations can be better managed in premature foals and those with systemic illness. Relationships between EMS and PPID also require further study, and evidence must be gathered to establish links between insulin dysregulation and the development of pituitary disease.
Authors' declaration of interests
Dr Frank consults for Boehringer Ingelheim on research study design.
Source of funding
Dr Frank has received research funding in the past from Boehringer Ingelheim Vetmedica.
Manuscript prepared by Dr Frank with sections contributed by Dr Tadros.