Previously presented in abstract form at the American College of Veterinary Internal Medicine (ACVIM) Forum, Montreal, Canada, June 3–6, 2009.
Corresponding author: Dr Ramiro E. Toribio, Department of Veterinary Clinical Sciences, College of Veterinary Medicine, The Ohio State University, 601 Vernon L. Tharp Street, Columbus, OH 43210; e-mail: email@example.com.
Background: Endocrine dysregulation of hormones of energy metabolism is well documented in critically ill humans, but limited information exists in septic foals. The purpose of this study was to provide information on the hormonal response to energy metabolism in critically ill foals, focusing on insulin, glucagon, and leptin.
Hypothesis: Concentrations of insulin, glucagon, leptin, and triglycerides will be higher, whereas glucose concentration will be lower in septic foals than in healthy and sick nonseptic foals. The magnitude of these differences will be associated with severity of disease and nonsurvival.
Animals: Forty-four septic, 62 sick nonseptic, and 19 healthy foals <7 days of age.
Methods: In this prospective multicenter cross-sectional study, blood samples were collected at admission. Foals with positive blood culture or sepsis score ≥12 were considered septic.
Results: Septic foals had lower glucose and insulin and higher triglyceride and glucagon concentrations than did healthy foals. Glucagon concentrations were not different between septic foals that died (n = 14) or survived (n = 30). Higher insulin and lower leptin concentrations were associated with mortality. Quantitative insulin-sensitivity check index was higher in septic foals.
Conclusions and Clinical Importance: Energy metabolism and the endocrine response of related hormones in septic foals are characterized by hypoglycemia, hypertriglyceridemia, low insulin concentration, and high glucagon concentration. Leptin and insulin may have prognostic value for nonsurvival in septic foals. The hormonal response related to energy metabolism in critical illness differs between foals and humans.
Maturation of the endocrine system and the associated energy metabolism in the equine neonate is delayed and continues in the postnatal period.1–4 Neonates are highly dependent on glucose intake, and endocrine glucoregulatory mechanisms are not always fully competent at birth. Thus, episodes of hypoglycemia are common in critically ill foals.2,4,5 For example, newborn and premature foals have a lower insulin response to increased blood glucose concentrations than do older foals and foals born at term.1,6 Postnatal maladaptation and illness are common in foals, making them prone to various diseases, including sepsis. Sepsis is the most common cause of mortality in foals, resulting in major economical losses in the equine industry.7–10
Septic foals often present to intensive care units with anorexia and hypoglycemia, and rapid intervention with energy-containing fluids frequently is indicated because of the inability of these foals to nurse or their intolerance to enteral feeding. A major complication of parenteral nutrition in foals, including administration of dextrose with IV fluids, is the development of hyperglycemia (ie, glucose intolerance).11,12 Hyperglycemia has been associated with increased mortality in critically ill humans13 because of oxidative stress, glucotoxicity, and β-cell dysfunction.14
Endocrinopathies have become a major focus of research in critically ill humans, and more recently in foals.3,15,16 Insulin resistance and hyperglycemia are common manifestations of endocrine dysregulation in people with sepsis or endotoxemia,17–20 and tight glycemic control utilizing insulin therapy has been shown to increase survival in these patients.21 Similarly, insulin frequently is used in equine neonates when hyperglycemia occurs22; however, limited information on the hormonal control of energy metabolism in septic neonatal foals is available.
Insulin is essential for energy regulation. It increases cellular glucose uptake, glycogenesis, fatty acid synthesis, and cellular potassium uptake; decreases proteinolysis, lipolysis, and gluconeogenesis; and, is a vasodilator in the microcirculation. Insulin also has anti-inflammatory properties by decreasing inflammatory cytokines and enhancing anti-inflammatory mediators,18 which may confer some therapeutic benefit in patients in a proinflammatory state (eg, septicemia or endotoxemia). Insulin resistance is a common feature of sepsis and endotoxemia in humans21,23 and also has been reported in horses.24,25
The quantitative insulin-sensitivity check index (QUICKI) is a novel index of insulin sensitivity based on a single fasting blood sample.26 QUICKI provides a reproducible and robust estimate of insulin sensitivity in humans, because it has strong correlations with insulin sensitivity indices calculated from standard glucose clamp or frequently sampled IV glucose tolerance test (FSIVGTT) in adults and children,27 and also appears useful in rodents.28 QUICKI is a reliable index in insulin-resistant human subjects,29 and shows good correlation with other proxies for insulin sensitivity in adult horses.30 The authors did not find any reference on the use of QUICKI in foals.
Glucagon is secreted by the α-cells of the pancreas and has opposing physiologic roles to insulin because it stimulates gluconeogenesis, glycogenolysis, and lipolysis. Hyperglucagonemia has been reported in septic and endotoxemic humans as well as in endotoxemic dogs and septic rats.31–33 Glucagon is thought to be important to maintain or enhance gluconeogenesis in the catabolic state of critical illness.33
Leptin, an adipocyte-derived hormone (adipokine), is considered the main regulator of satiety and its blood concentrations have been correlated with total body fat in horses, humans, dogs, and other species.34–37 Leptin concentrations decrease in feed-restricted mares38 and increase after a meal in humans,39 and leptin increases insulin sensitivity in humans.35 In addition to energy homeostasis, leptin has immunomodulatory properties.40 Its synthesis is stimulated by inflammatory cytokines,41 and leptin has been described as an acute phase reactant.42,43 Increased leptin concentrations have been associated with sepsis and endotoxemia in humans, rodents, and dogs.43–48 However, an increase in leptin concentration was not identified in septic human neonates,49 or in sheep, cattle,50,51 or pigs52 during experimental endotoxemia. No information is available on leptin in horses (ie, foals) with septicemia or endotoxemia.
The purpose of this study was to investigate the hormones that control energy metabolism in critically ill foals by determining serum or plasma concentrations of insulin, glucagon, and leptin in healthy, sick nonseptic, and septic foals, as well as their association with severity of disease (sepsis score) and outcome (death/survival). To evaluate the relationship between the hormones and metabolic variables, we investigated the correlation with serum glucose and triglyceride concentrations. We hypothesized that blood concentrations of insulin, glucagon, and leptin would be higher in septic foals compared with healthy controls. We also expected an association between the magnitude of these differences with the degree of sepsis and nonsurvival.
Materials and Methods
Foals ≤7 days old of any breed or sex admitted to The Ohio State University Galbreath Equine Center and Hagyard Equine Medicine Institute during the foaling season of 2008 were included. Hospitalized foals were classified into 1 of the 2 groups: sick nonseptic and septic foals. Foals in the septic group had a sepsis score of ≥12, a positive blood culture, or both.53 Foals in the sick nonseptic group were hospitalized for illnesses other than sepsis (eg, meconium impaction, hypoxic ischemic encephalopathy, failure of transfer of passive immunity, flexural deformities) requiring hospitalization. These foals had negative blood cultures and a sepsis score of ≤11. The control group consisted of 18–24-hour-old foals examined on a routine basis at breeding farms in Kentucky. Foals included in the control group were born at the farm and were clinically healthy based on physical examination, a normal CBC, serum biochemistry, a serum immunoglobulin G (IgG) concentration >800 mg/dL, and a sepsis score of ≤4. Foals with a history of having received glucose-containing fluids or corticosteroids before admission to the hospital were excluded from the study.
Any foal that was discharged from the hospital was defined as a survivor. Foals that died or were euthanized because of a grave medical prognosis were defined as nonsurvivors. Individuals euthanized for other reasons such as financial constraints were excluded from the study.
This study was approved by the Ohio State University Veterinary Teaching Hospital executive committee and the Institutional Animal Care and Use Committee, and adhered to the principles of humane treatment of animals in veterinary clinical investigations as stated by the American College of Veterinary Internal Medicine and National Institute of Health guidelines.
History obtained upon admission included expected foaling date, duration of pregnancy, parity, maternal illness, premature lactation, observed or assisted parturition, dystocia, passing and appearance of the fetal membranes, and medications (mare and foal). Clinical data collected included signalment (sex, gestational, and actual age, breed), physical examination findings, CBC, biochemistry profile including serum glucose, fibrinogen, l-lactate, IgG, and triglyceride concentrations. For consistency, the sepsis score was calculated by the 1st author for each foal individually, based on recorded history, physical examination, and laboratory findings.53
Blood samples for hormone assays from foals admitted to both hospitals were obtained on admission by sterile jugular venous catheterization. Blood was placed in plain serum clot tubes and chilled aprotinin-EDTA tubes. Aprotinin was added to preserve sample integrity by preventing potential protease degradation of hormones (500 kU/mL of whole blood). Samples were stored in ice water and centrifuged within 12 hours at 5°C, 2,000 × g for 12 minutes. Serum and plasma then were aliquoted and stored at −80°C until analyzed. A 12-hour delay occurred in few samples, most samples were processed within 2 hours. Based on studies in humans, there is no evidence that sample storage substanitally affects stability of insulin or leptin. For example, human leptin is stable for 1 week at room temperature and for months at 4°C.54 The processing and storage in this study was performed similarly to routine processing of human samples for immunoassays.
Blood samples for CBC,a serum biochemistry, and IgGb were processed immediately. Samples from healthy control foals were obtained during routine newborn foal examinations at the farm and processed the same day.
Blood concentrations of insulin (serum) and glucagon (plasma) were determined by human radioimmunoassayc,d (lowest level of detection: 1.2 μIU/mL and 13 pg/mL, respectively), whereas leptin (serum) was measured with a multispecies leptin radioimmunoassaye (lowest level of detection: 1.0 ng/mL). All assays have been validated previously for the equine species.34,38,55,56
Insulin sensitivity was assessed by the QUICKI, calculated by following formula:
The Shapiro-Wilk statistic was used to assess data normality. Only glucose concentrations were normally distributed. The remainder of the data was not normally distributed. Therefore, median and interquartile ranges were calculated for continuous variables. Nonparametric comparisons among the groups were computed with the Kruskal-Wallis statistic and a Dunn's posttest to compare each group individually, by a statistical software program.f The Mann-Whitney U-test was applied to compare survivors with nonsurvivors. Significance was set at P < .05. The Spearman rank order (rs) was used to define correlations between variables.g Continuous variables were categorized by cutoff values based on distribution within a group, and analyzed by logistic regression (procgenmodh) for binomial distribution. Clinical data analyzed included sepsis score, survival, age, breed, sex, temperature, heart rate, respiratory rate, mucous membrane color, capillary refill time, cold extremities, pulse quality, diarrhea, hematocrit, total white blood cell count, segmented neutrophil count, band neutrophil count, lymphocyte count, monocyte count, platelet count, fibrinogen, total protein, albumin, l-lactate, IgG, sodium, potassium, chloride, anion gap, glucose, blood urea nitrogen, creatinine, total calcium, phosphorus, total bilirubin, and triglycerides. Crude odds ratios (OR) and 95% confidence intervals (CI) were determined based on categories. The dependent variable was survival/nonsurvival. All variables were screened and any variables with a P value < .25 were tested in a forward and backward stepwise multivariate logistic regression to determine a final model. The Hosmer and Lemeshow Goodness-of-Fit was determined with proc logistic.h,57 Variables that resulted in a P value < .05 were retained in the model.
A total of 125 neonatal foals were included, of which 106/125 were hospitalized and 19/125 were healthy foals. Forty-four/106 (41%) were classified as septic, and 62/106 (58%) as sick nonseptic. Of the 44 septic foals, 30 foals (68%) survived to discharge from the hospital and 32 had positive blood cultures (73%). The median age of all hospitalized foals at admission was 12 hours (range: sick nonseptic, 1–168 hours; septic, 1–192 hours). Healthy controls all were between 18 and 24 hours old.
All healthy controls were Thoroughbreds (n = 19). Breeds representing the group of hospitalized foals included Thoroughbred (71), Quarter Horse (11), Standardbred (9), Appaloosa (4), Warmblood (3), Friesian (2), American Paint Horse (2), Arabian (1), Gypsy Vanner (1), Percheron (1), and 1 mixed breed. Of the hospitalized foals, 50 were fillies and 56 were colts, but 15/19 healthy controls were fillies.
Serum Glucose and Triglycerides
Serum glucose concentrations were significantly lower in septic foals compared with sick nonseptic and healthy foals (P < .001). Septic and sick nonseptic foals had significantly higher serum triglyceride concentrations than did the healthy controls (P < .001, Table 1). Glucose and triglyceride concentrations were inversely correlated in all sick/hospitalized foals of the study population as well as in septic foals only (Tables 3a and 5a), but not in healthy foals.
Table 1. Serum glucose, triglyceride, hormone concentrations, and QUICKI in neonatal foals at admission (n = 125).
Septic foals had significantly lower insulin and higher glucagon concentrations than did the healthy foals (P < .001). Glucagon concentrations were higher and insulin concentrations were lower in sick nonseptic than in healthy foals (P < .05 and P < .001, respectively). Glucagon concentrations also were higher in septic than in sick nonseptic foals (P < .01). Insulin was not different between septic and sick nonseptic foals. There was no difference in leptin concentrations among groups (Table 1).
Glucagon was positively correlated with leptin in hospitalized foals and in septic foals only (Tables 3b and 5b). There were no correlations between individual hormones in healthy or sick nonseptic foals (Tables 2b, 4b).
Table 3b. Correlation (rs) between the hormones (all sick/hospitalized foals, n = 106).
Table 2b. Correlation (rs) between the hormones (healthy foals only, n = 19).
Table 4b. Correlation (rs) between the hormones (sick non-septic foals, n = 62).
Association of Hormone Concentrations with Glucose and Triglycerides
Insulin concentrations were positively correlated with glucose concentrations in all groups. However, glucagon was inversely correlated with glucose only in the healthy controls. When data of all hospitalized foals (sick nonseptic and septic) were analyzed, all hormones were correlated with glucose (Table 3a), but when sick nonseptic foals and septic foals were analyzed as individual groups, there was no association between glucagon or leptin and glucose (Tables 4a and 5a). Glucagon was positively correlated with triglycerides in all groups of sick/hospitalized foals (Tables 3a, 4a, and 5a), but not in healthy foals. Furthermore, there was no correlation between insulin or leptin and triglycerides in any of the groups.
Table 4a. Correlation (rs) between blood glucose and triglyceride concentrations, QUICKI, and hormones (sick non-septic foals, n = 62).
Association of Glucose and Triglyceride Concentrations with Survival
Of all septic foals, 68% survived (30/44). Glucose concentrations were significantly lower in nonsurvivors than in septic foals that survived. However, we found no difference in triglyceride concentrations between septic nonsurvivors and septic survivors. Serum triglyceride concentration was one variable that was retained in the final logistic regression model. In the entire foal study population (healthy, sick nonseptic, and septic foals), multivariate logistic regression showed that overall nonsurvival was more likely with serum triglyceride concentrations <60 mg/dL (P < .01).
The 8 septic foals with the highest glucose concentrations (147–329 mg/dL) survived. Only 2 of these 8 individuals had a higher insulin response than normo- or hypoglycemic septic foals (Fig 1).
Association of Hormone Concentrations with Survival
Among septic foals, we found significantly higher insulin concentrations in nonsurvivors than in survivors (P < .001). Septic foals that had insulin concentrations >4 μIU/mL were more likely to die than those that had low insulin on admission (OR, 6.0; 95% CI, 1.2–36.4).
Glucagon was not different among the groups. Leptin concentrations were lower in septic foals that died (P < .05, Table 5). Foals with leptin concentrations <1.1 ng/mL were more likely to die (OR, 9.8; 95% CI, 1.4–199.5). Likewise, when compared with the whole study population, the likelihood of nonsurvival was higher in foals with low leptin concentrations (OR, 3.98; 95% CI, 1.1–18.7).
Insulin Sensitivity (QUICKI)
QUICKI was higher in sick nonseptic and septic foals than in healthy foals (Table 1), but was not different between septic survivors and septic nonsurvivors (Table 6). QUICKI was negatively correlated with leptin in all groups (Tables 2a, 3a, 4a, 5a). In all hospitalized foals (sick nonseptic and septic), QUICKI also was negatively correlated with insulin concentrations (Tables 3a, 4a, 5a). Furthermore, QUICKI was positively correlated with glucagon and triglycerides in septic foals as well as in all hospitalized foals (Tables 3a and 5a).
Table 6. Comparison of serum glucose, triglyceride, hormone concentrations, and QUICKI between surviving and nonsurviving septic neonatal foals at admission.
The sepsis score correlated significantly with insulin, glucagon, triglycerides, and QUICKI, but it was not associated with leptin concentrations (Table 7). Sepsis score was the second variable retained in the final logistic regression model. In this study population, nonsurviving foals had higher sepsis scores (P < .01).
Table 7. Correlation (rs) between sepsis score and hormones/laboratory parameters.
The final model is presented in Table 8. The model includes 3 variables: triglyceride concentrations, cold extremities (yes/no), and sepsis score (<12/≥12). The Hosmer and Lemeshow Goodness-of-Fit test indicated that the data fit the model well (P= .98).
Table 8. Results for the final model of the multivariate logistic regression for risk factors associated with nonsurvival in neonatal foals.
OR, odds ratio, 95% CI, 95% confidence interval, Reference, reference group for comparison, N/A, not applicable; QUICKI, quantitative insulin-sensitivity check index.
In the current study, we documented that changes in energy metabolism in critically ill foals are characterized by hypoglycemia, hypertriglyceridemia, and the endocrine response consists of low insulin and high glucagon concentrations. Mortality in septic foals was associated with low leptin and high insulin concentrations.
Septic foals had significantly lower blood glucose concentrations than did sick nonseptic and healthy foals, which was not unexpected, because hypoglycemia is a common finding in critically ill foals.10,12,53 Hypoglycemia commonly occurs with illness because of decreased suckle, sepsis, and endotoxemia. Among septic foals, nonsurvivors had lower glucose concentrations than did survivors, which is in agreement with a recent study in critically ill foals.12 We found hyperglycemia in few critically ill foals of this study, perhaps caused by increases in cortisol and catecholamine concentrations arising from the stress of transportation or illness. Hyperglycemia also has been associated with mortality and normoglycemia with survival in critically ill foals,12,16 but septic foals in our study that were hyperglycemic on admission (glucose >130 mg/dL) survived. The effect of prolonged hyperglycemia was not evaluated in this study.
Serum triglyceride concentrations were increased in hospitalized foals, further supporting a metabolic response to decreased energy intake. Because carbohydrate stores in foals are limited, energy demand during anorexia depletes carbohydrate stores, leading to mobilization of fat depots. Ultimately, when the liver cannot maintain glucose production from fatty acids, blood triglyceride concentrations increase. Similarly, hypertriglyceridemia is the main feature of altered fat metabolism in critically ill humans.23 In calves, glucose concentrations decrease and triglyceride concentrations increase in response to administration of endotoxin or TNF-α.58 High TNF-α concentrations have been measured in septic foals59 and additionally may have contributed to the hypoglycemia and hypertriglyceridemia found in the septic foals of this study.
We hypothesized that insulin would be higher in critically ill foals, based on what has been observed in adult horses and calves with experimental endotoxemia.25,58 Sepsis and endotoxemia induce insulin resistance in humans,17 horses,24 and mice.60 However, in our study, insulin was significantly lower in septic and sick nonseptic foals as compared with healthy controls, and insulin and glucose were positively correlated in all groups of the study population. This, as well as the inverse relationship between glucose and glucagon, indicates an appropriate physiological response of insulin and glucagon to blood glucose concentrations and stands in contrast to the aforementioned studies.
To our knowledge, ours is the first study to document the role of glucagon in the endocrine response to sepsis in foals. Higher glucagon concentrations in the septic foals could be interpreted as a physiological response, because glucagon is a catabolic hormone that stimulates gluconeogenesis. Increased glucagon concentrations also have been documented in septic rats and dogs,32,33 and in endotoxemic humans.31
Based on multivariate logistic regression analysis of all foals of the study population, low serum triglyceride concentrations were associated with nonsurvival, which likely represents decompensation in fat metabolism in severely affected individuals, in which fat stores may either be depleted, or fat is not mobilized for energy. The wide range of triglyceride concentrations in nonsurvivors also could have accounted for the lack of statistical difference in triglyceride concentrations between survivors and nonsurvivors.
In septic foals of our study, insulin concentrations did not exceed those of the healthy controls. However, lower glucose but higher insulin concentrations in nonsurvivors possibly occurred because of decreased insulin clearance from impaired kidney or liver function in the sickest foals.61 Both increased and decreased insulin clearance has been found during sepsis.62,63 Although most publications show that insulin secretion is suppressed by inflammatory cytokines such as IL-1, IL-6, and TNF-α,64 there is also evidence that systemic inflammation can increase insulin concentrations in horses.24 Therefore, the conflicting results in our study cannot be fully explained without further investigation, including dynamic glucose studies and assessment of systemic inflammation.
To the authors' knowledge, insulin resistance has only been assessed in healthy neonatal foals, and appears to be associated with maternal diet during late gestation.65 Insulin resistance as a response to systemic inflammation has not been evaluated in critically ill foals by glucose clamp or FSIVGTT, and it was not feasible to do so in this study. To alternatively determine insulin sensitivity based on a single sample, we calculated QUICKI. QUICKI has not been validated as an indicator of insulin sensitivity in foals, and the fact that QUICKI was higher in septic foals compared with the other groups and positively associated with sepsis score was unanticipated. We expected evidence of insulin resistance in septic foals, because insulin resistance is a common feature of sepsis and endotoxemia in humans21,23 and has been reported in endotoxemic horses.24,25 Assuming that QUICKI predicts insulin sensitivity correctly, increased insulin sensitivity in septic foals may be a protective mechanism during times of energy deprivation to provide peripheral tissues and vital organs with sufficient energy by increased glucose uptake. This may occur despite low blood glucose concentrations or as a response to low glucose and low insulin concentrations. However, one must be cautious with this interpretation because hypoglycemia is frequent in septic foals and it overestimates insulin sensitivity by affecting QUICKI. Contrary to what is observed in sick foals, hypoglycemia rarely occurs in adult horses, whether healthy or insulin resistant (in horse populations for which QUICKI has been validated). Thus, without dynamic studies (eg, glucose clamp or FSIVGTT), we cannot rule out insulin resistance in septic foals.
The finding that surviving septic foals had higher leptin concentrations is in agreement with a study by Arnalich et al,44 in which high leptin concentrations in septic human patients were associated with survival. In contrast, leptin concentrations were higher in septic children that did not survive66 as well as septic and endotoxemic rodents,41 dogs,47,48 and adult humans.44 However, other studies found no association between leptin concentrations and septicemia, shock, multiorgan failure, or severity of disease in human neonates, children, or endotoxemic sheep and cows.44–47,49–51,66 The lower blood glucose and leptin concentrations in nonsurviving septic foals indicate a physiological response of leptin to increased caloric demand, as has been shown in feed-restricted mares.38 A similar relationship between leptin and glucose has been demonstrated in horses that had an increase in first insulin and then leptin after feeding, indicating that insulin drives an increase in leptin.67 Conversely, we did not find a correlation between insulin and leptin concentrations. Although there is evidence that leptin increases insulin sensitivity,35 leptin concentrations were negatively correlated with QUICKI in all groups. The leptin response to systemic inflammation or sepsis appears to be variable among species, age groups, and possibly individuals. Thus, it is difficult to make strong conclusions on leptin and its prognostic value in septic foals with the current data, and additional research using foals with various levels of sepsis or a controlled study will be necessary to further address this question.
One limitation of this study was the sex bias in the control group (15/19 foals were fillies). It is unclear how the hormones of energy metabolism are influenced by sex in the neonatal period. In sexual maturity, leptin has been shown to be influenced by sex hormones68; however, in the neonatal period it is unlikely that sex hormones play an important role in hormonal regulation. Another limitation is the use of the sepsis score to define sepsis in this study. Hypoglycemia is one of the variables used to calculate the sepsis score and this has been an accepted approach in the current literature. With regard to hypoglycemia, this may have caused bias in the analysis. Of the foals defined as septic in this study, 73% had positive blood cultures, and 27% of the foals were classified as septic by sepsis score only. This approach may have confounded the results.
In conclusion, most of the endocrine responses documented in this study appeared to be appropriate physiological responses to metabolic changes in the face of negative energy metabolism. The value of leptin and insulin as prognostic indicators for survival in septic foals and the association with systemic inflammation should be evaluated in future studies.
a Cell-Dyn 3500R analyzer, Abbott Laboratories, Abbott Park, IL
b Boehringer Mannheim/Hitachi 911 system, Boehringer Mannheim Corporation, Indianapolis, IN
c Coat-A-Count human insulin radioimmunoassay, Siemens Healthcare Diagnostics, Los Angeles, CA
d Coat-A-Count human glucagon radioimmunoassay, Siemens Healthcare Diagnostics
e Multispecies leptin radioimmunoassay, Linco, Millipore, St Charles, MO
f Prism, version 4.0a, GraphPad Software Inc, San Diego, CA
g SigmaStat 3.5, Systat, Chicago, IL
h SAS version 9.1, SAS Institute Inc, Cary, NC
We thank all of the technical staff and veterinarians at Hagyard Equine Medical Institute and Galbreath Equine Center, The Ohio State University, for their cooperation and assistance with this project. We acknowledge the help of Drs Michele L. Frazer, Katherine MacGillivray, Peggy S. Marsh, and Rhonda A. Rathgeber for their expertise and sample collection on hospitalized and healthy foals as well as Tina Elam for sample processing. Special thanks go to Holly Brown and Samantha Coe for their assistance with laboratory techniques.