Work was performed at The Veterinary Clinical Centre, School of Animal and Veterinary Sciences, Charles Sturt University, Locked Bag 588, Wagga Wagga, New South Wales 2678, Australia.
The Summary is available in Chinese – see Supporting information.
Reasons for performing study
Metformin is a potential therapeutic agent for the treatment of insulin resistance (IR). In laboratory animals, orally administered metformin reduces intestinal glucose absorption and may therefore affect insulinaemic responses to oral carbohydrate ingestion.
To determine whether pretreatment with metformin reduces plasma glucose concentration and insulin responses following consumption of dextrose in horses.
Therapeutic cross-over study.
Seven healthy Standardbred and Thoroughbred geldings were subjected to an oral dextrose challenge test on 4 occasions: with and without metformin, before and after induction of IR with dexamethasone. Metformin was administered by nasogastric tube at 30 mg/kg bwt 1 h before administration of dextrose. Glucose and insulin concentrations in plasma/serum were measured at regular intervals during each test. Linear mixed models were specified for each predetermined outcome variable, and for each model the ‘treatment’ was included as a fixed effect with 4 categorical levels (none, metformin, dexamethasone and dexamethasone with metformin) and horse accounted for as a random effect.
In healthy horses, the administration of metformin resulted in a statistically significant reduction in peak glucose concentration (P = 0.002), area under the glucose curve (P<0.001) and insulin concentration 120 min after dextrose administration (P = 0.011). Following the induction of IR, administration of metformin was associated with significant differences in peak glucose concentration (P<0.001), the percentage increase in glucose concentration (P = 0.010), the area under the glucose curve (P<0.001) and insulin concentration at 120 min (P = 0.034) and 150 min after dextrose administration (P = 0.014).
Metformin resulted in reduced glycaemic and insulinaemic responses both in healthy horses and in horses with experimentally induced IR.
Metformin may benefit horses with naturally acquired IR by reducing glycaemic and insulinaemic responses to dietary nonstructural carbohydrates. Further investigations into the mechanisms of action of metformin in horses and controlled clinical trials are warranted.
Equine metabolic syndrome (EMS) describes a collection of risk factors for the development of laminitis, with insulin resistance (IR) and increased adiposity being central features . Robust epidemiological studies of the prevalence of EMS are lacking; however, the prevalence of obesity in horses is recognised to be increasing, and in one study there was an estimated prevalence of 45% . Obesity and EMS have a major impact on equine welfare , and effective management and treatment strategies are required.
Management of diet and exercise regimens is fundamental to effective prevention and treatment of EMS; however, clinical improvements induced by management changes alone are frequently disappointing and may be limited by poor owner compliance . Options for adjunctive pharmacological treatment of EMS have recently been reviewed . Metformin is an affordable medication proposed to be of benefit in EMS, but recent studies have indicated that although serum insulin and glucose concentrations may be reduced by treatment [6, 7], the drug has poor bioavailability in horses [8, 9] and has not been found to improve insulin sensitivity (IS) [10-12].
The relative importance of different mechanisms by which metformin may exert antihyperglycaemic and insulin-sensitising effects is unknown , although in man it is assumed that reduction in hepatic glucose output is the major mechanism and the drug has some additional effects on peripheral glucose utilisation . Metformin is thought to have both insulin-dependent and insulin-independent effects on glucose metabolism ; hence, appraisal of efficacy based on assessments of IS alone may be too limited. Additionally, complex pharmacokinetic and pharmacodynamic properties of metformin may confound the direct relevance of plasma concentrations of the drug . Although previous investigations in horses have indicated that metformin does not increase IS, effects of the drug on intestinal absorption of glucose and subsequent insulin responses have not so far been published . The aim of this study was to investigate whether orally administered metformin reduced glycaemic and insulinaemic responses in healthy horses and in horses with experimentally induced IR.
Materials and methods
The institution's Animal Care and Ethics Committees approved the study. Seven geldings (4 Thoroughbred and 3 Standardbred) from the institution's research herd were used in the study. Ages ranged from 5 to 19 years (median 10 years). All horses had a body condition score between 3/9 and 6/9 and were considered in good health based on the results of clinical, haematological and blood biochemical examinations. Plasma adrenocorticotrophic hormone concentration measured <25 pg/ml in all horses. Horses were acclimatised to housing in dirt yards and a diet of 2% bodyweight (bwt) grass hay (dry matter) daily for 2 weeks. No supplementary concentrate feed was provided. Feeding and management remained consistent throughout the study period, and all hay was sourced from the same batch. An intravenous catheter was placed in one jugular vein of each horse at least 24 h prior to each test, and all samples were collected via the catheters.
Following acclimatisation to the housing conditions and testing stalls, each horse was subjected to an oral glucose challenge test (OGCT) on 4 occasions: with and without pretreatment with metformin, both before and after induction of IR with dexamethasone. The order of testing was not randomised because concurrent studies necessitated the administration of metformin at the end of each phase of testing to avoid washout periods. On Day 1, all 7 horses had an OGCT performed as described below (GC). On Day 3 the OGCT was repeated, 1 h after the administration of 30 mg/kg bwt metformin by nasogastric tube (GC-MET). All horses were then given 0.08 mg/kg bwt dexamethasone i.v. every second day from Day 3 to Day 19 to induce IR . An OGCT was performed on Day 18 without metformin (GC-DEX) and then repeated on Day 20, 1 h after the administration of 30 mg/kg bwt metformin by nasogastric tube (GC-DEX-MET). Metformin tablets (Metforbella) were powdered using a pestle and mortar and suspended in 1 l of water for administration by nasogastric tube. The bucket was rinsed with a further 1 l of water that was also administered through the nasogastric tube, to ensure that each horse received the full dose and no residue remained in the bucket or tube.
The OGCT was performed according to a standardised protocol. Horses were fed 1% bwt grass hay at 16.00 h on the day preceding the test and received no further feed overnight. At 06.00–07.00 h the following morning, horses were moved to individual open stalls adjacent to their yards. Baseline blood samples were collected into plain and oxalate fluoride Vacutainer tubes for measurement of serum insulin and plasma glucose concentrations, respectively. Following baseline sampling, all horses received 0.5 g/kg bwt dextrose powder mixed into a feed of 2 g/kg bwt rice bran pellets (digestible energy 11 MJ/kg, crude protein 12%, crude fat 6%; Coprice Cool Conditionerb), with a small amount of water. One horse would not eat the dextrose powder and the dextrose was administered in water by nasogastric tube on all occasions. The horse was accustomed to frequent intubation for teaching purposes and the procedure was not considered stressful. Blood samples were collected and plasma glucose concentrations measured every 30 min for 4 h or until plasma glucose concentrations returned to baseline levels, whichever was sooner. In addition, serum insulin concentrations were measured 90, 120 and 150 min after dextrose was administered. Two hours after feeding the dextrose, horses were provided with 1% bwt grass hay.
Plasma glucose concentrations were measured using a point-of-care glucometer (Accutrend plus)c validated for use with equine plasma. Plasma insulin concentrations were measured using a chemiluminescent assay (Immulite)d validated for use in horses .
Data were collated in an Excel databasee. Statistical analysis was performed using IBM SPSS softwaref. Linear mixed models were specified for each outcome variable (baseline glucose concentration [Gbase], peak glucose concentration [Gpeak], the percentage increase in glucose concentration above baseline [G%inc], area under the glucose curve [AUCg], insulin concentration prior to dextrose challenge [Ibase] and insulin concentration 90, 120 and 150 min after dextrose challenge [I90, I120 and I150]). For each model, ‘treatment’ was included as a fixed effect with 4 categorical levels (none, metformin, dexamethasone and dexamethasone with metformin) and horse accounted for as a random effect. All main effects were compared. Adjustment for multiple comparisons was made in each analysis, residuals were checked for normality and significance was set at P<0.05.
Median values and interquartile ranges for predetermined outcome measures in both healthy and IR horses before and after administration of metformin and the results of statistical analyses performed are shown in Table 1. For each intervention, glucose concentrations at each time point are displayed in Figure 1 and highlight the reduced peaks of the glucose curves in both normal and IR horses following the administration of metformin. Insulin concentrations for each intervention at 90, 120 and 150 min are shown in Figure 2.
Table 1. Median values (interquartile ranges) of parameters of glucose and insulin responses for 7 horses in which oral dextrose was administered with and without metformin, before and after induction of insulin resistance with dexamethasone
Abbreviations: AUCg = area under the glucose curve; DEX = treatment with dexamethasone; GC = glucose challenge; and MET = administration of metformin. Within each row, statistically significant differences occurred between those values with the same superscript. Glucose concentrations are expressed as millimoles per litre and insulin concentrations as international units per litre.
Glucose increase (%)
Insulin 90 min
Insulin 120 min
Insulin 150 min
Administration of dexamethasone had significant effects (identifying a significant association between treatment and outcome) on Gbase (P<0.001), Gpeak (P<0.001), G%inc (P = 0.012), AUCg (P<0.001) and I120 (P = 0.011), indicating that reductions in tissue sensitivity to insulin had been induced.
When assessing the main effects, it was found that in the healthy horses the administration of metformin resulted in a statistically significant reduction in the Gpeak (P = 0.002), AUCg (P<0.001) and I120 (P = 0.011).
Following the induction of IR, administration of metformin was associated with significant differences in Gpeak (P<0.001), G%inc (P = 0.010), AUCg (P<0.001), I120 (P = 0.034) and I150 (P = 0.014).
The results of this study indicate that the administration of metformin prior to the administration of oral dextrose is associated with a reduction in the resultant glycaemic and insulinaemic responses. This effect was observed both in horses that were considered to demonstrate normal glucose–insulin dynamics and following the experimental induction of IR in the same horses.
Recent evidence from human and laboratory animal studies indicates that the effect of metformin on the intestine may be of considerable importance and is independent of plasma metformin concentrations [16, 18]. Following oral administration in man and laboratory animals, metformin accumulates in the intestine [18, 19] and may achieve concentrations in the human jejunum that are 30–300 times greater than those in plasma . In rats, metformin delays absorption of glucose by the intestine  and increases glucose utilisation by the intestinal tissue . Metformin reduces intestinal glucose absorption  by altering regulation of the 2 major intestinal glucose transporters  and has a pronounced first-pass pharmacodynamic effect [13, 16]. The findings of the present study indicate that metformin may have similar inhibitory effects on glucose absorption in the equine intestine. This may account for the disparity between previous studies, in which reduced insulin concentrations were identified following the administration of metformin [6, 7], yet bioavailability is poor and insulin sensitivity has not been increased [7, 10-12]. Clearly, further investigations are necessary to determine whether metformin accumulates and exerts similar inhibitory effects on glucose absorption in equine intestine. In mice, concentrations of metformin in gastrointestinal tissues decline rapidly after peaking at 2 h, and within 24 h reduce to <2% of peak values . Whether the effects of metformin on the equine intestine are of sufficient duration to result in clinical benefit remains to be determined.
Recent experimental models have demonstrated that laminitis may be induced in horses by hyperinsulinaemia in the absence of IR, suggesting that insulin may be intrinsically harmful to the lamellae [23, 24]. In the lamellae and the digital vascular endothelium glucose uptake is predominantly, if not entirely, insulin independent [25, 26], and therefore IR per se may not be directly relevant in these tissues. Furthermore, IR may represent a protective mechanism against the harmful effects of hyperinsulinaemia . The insulin-lowering effect of metformin demonstrated in the present study could be of benefit in animals prone to laminitis by moderating diet-induced hyperinsulinaemia even in the absence of demonstrable effects on IS. In addition, reduction in glucose absorption may be of benefit in lowering caloric intake and thereby aiding weight loss, an effect reported in man . However, the potential benefit of these effects in horses fed appropriate forage that is low in nonstructural carbohydrates versus more glycaemic diets, such as grass, forage with a high nonstructural carbohydrate concentration and cereals, may be questionable.
The horses used in the present study were not typical of the breeds most often diagnosed with EMS. Standardbreds and Thoroughbreds were used based upon availability and to limit variability in the study population. The method used to induce IR in the present study has been demonstrated previously to result in a significant decrease in IS with a compensatory increase in pancreatic B cell response . Although the induction of IR according to an established experimental method  had the advantage of being consistent between individuals, it is unknown how well this model reflects naturally acquired IR in animals with EMS. Numbers of animals were small, and the absence of a statistically significant difference in insulin concentration at 90 min may be due to the low numbers of horses and the small magnitude of the insulin response when horses did not receive metformin.
In conclusion, the present study indicates that orally administered metformin reduces the glycaemic and insulinaemic responses to orally administered dextrose in horses. Further work is required to determine whether these effects may translate into clinical benefits in horses with EMS and hyperinsulinaemia.
Authors' declaration of interests
None of the authors have personal or commercial affiliations that might prejudice the results.
Ethical animal research
The study was approved by the Animal Care and Ethics Committee of Charles Sturt University.
Sources of funding
The study was funded by an internal grant from the School of Animal and Veterinary Sciences, Charles Sturt University and by The Liphook Equine Hospital Laboratory.
The authors are grateful to staff and students at Charles Sturt University for their assistance.
All authors were involved in the study design, data collection or analysis and preparation of the manuscript.