This study was performed at the University of Georgia's Veterinary Teaching Hospital, Athens, GA and Hagyard Equine Medical Institute, Lexington, KY. Portions of this work were presented in abstract form at the 14th Annual International Veterinary Emergency and Critical Care Symposium, September 17–21, 2008, Phoenix, AZ and the 5th Dorothy Havemeyer Neonatal Septicemia Workshop, Salem, MA, November 19–22, 2008.
Corresponding author: Dr Kelsey A. Hart, DVM, PhD, DACVIM, Department of Large Animal Medicine, College of Veterinary Medicine, University of Georgia, Athens, GA 30602; e-mail: firstname.lastname@example.org.
Background: Relative cortisol insufficiency occurs in septic foals and impacts survival. Serum free (biologically available) cortisol concentration might be a better indicator of physiologic cortisol status than serum total cortisol concentration in foals.
Hypotheses: In septic foals, (1) low free cortisol concentration correlates with disease severity and survival and (2) predicts disease severity and outcome better than total cortisol concentration.
Methods: In this prospective clinical study, foals meeting criteria for sepsis at admission were enrolled. University-owned animals served as healthy controls. Basal and cosyntropin-stimulated total cortisol concentration and percent free cortisol (% free cortisol) were determined by chemiluminescent immunoassay and ultrafiltration/ligand-binding methods, respectively. Group data were compared by ANOVA, Mann-Whitney U-tests, and receiver operator characteristic curves. Significance was set at P < .05.
Results: Basal % free cortisol was highest in healthy foals at birth (58±8% mean±SD), and was higher (P≤.004) in healthy foals of all ages (33±6 to 58±8%) than in adult horses (7±3%). Cosyntropin-stimulated total and free cortisol concentrations were lower (P≤.03) in foals with shock (total = 6.2±8.1 μg/dL; free = 3.5±4.8 μg/dL versus total = 10.8±6.0 μg/dL; free = 6.9±3.3 μg/dL in foals without shock) and in nonsurvivors (total = 3.8±6.9 μg/dL; free = 1.9±3.9 μg/dL versus total = 9.1±7.7 μg/dL; free = 5.5±4.4 μg/dL in survivors). Free cortisol was no better than total cortisol at predicting disease severity or outcome in septic foals.
Conclusions and Clinical Importance: Serum free cortisol is impacted by age and illness in the horse. There is no advantage to measuring free over total cortisol in septic foals.
low-dose delta free or total cortisol concentration
low-dose delta total cortisol concentration
multiple organ dysfunction syndrome
relative adrenal insufficiency
receiver operator characteristic
% free cortisol
percent free cortisol
A growing body of evidence suggests that transient hypothalamic-pituitary-adrenal (HPA) axis dysfunction is common in critically ill people and animals, particularly those with severe trauma or sepsis.1–7 This syndrome is often termed relative adrenal insufficiency (RAI) or critical illness-related corticosteroid insufficiency (CIRCI), and is best defined as an inadequate cortisol response to the existing degree of illness-related physiologic stress.1,2,4,5,8–11 The occurrence of RAI/CIRCI during sepsis is associated with an increased incidence of shock, multiple organ dysfunction syndrome (MODS), and death in both people1,3,4,11–15 and foals.16–18 The accurate diagnosis of RAI/CIRCI in septic patients by a rapid and inexpensive test will benefit patients by identifying those that might require therapeutic intervention with physiologic doses of corticosteroids.
Diagnostic criteria for RAI/CIRCI are not fully defined in either human or veterinary medicine. Recent work in foals suggests that a blunted total cortisol response to a supraphysiologic dose of ACTH (100 μg) best indentifies septic foals with clinically relevant RAI/CIRCI.17,19 Some studies in critically ill people, though, suggest that basal or ACTH-stimulated total cortisol concentrations can misrepresent physiologic cortisol status in some patients.20–25 The biologically active, and thus more clinically relevant, form of cortisol is the smaller free (ie, non-protein-bound) fraction. Because both cortisol-binding globulin (CBG) and albumin levels decrease variably but significantly in critical illness,22,23 basal and ACTH-stimulated serum total cortisol measurements might not accurately reflect systemic cortisol status in critically ill patients. Thus, in septic patients likely to have altered CBG and albumin concentrations, measurement of serum free cortisol concentration could provide a more accurate method for diagnosis of RAI/CIRCI than basal and ACTH-stimulated total cortisol concentration.23
To the authors' knowledge, basal or ACTH-stimulated serum free cortisol concentrations have not been evaluated in healthy or septic foals. Given recent evidence suggesting the usefulness of free cortisol concentrations in determining adrenal function in humans with sepsis, RAI/CIRCI, or both,20–25 measurement of free cortisol might offer a more accurate method for assessment of HPA-axis function in septic foals. Thus, the primary objectives for this study were (1) to adapt an ultrafiltration assay for measurement of free cortisol in equine serum; (2) to compare free cortisol concentrations between healthy foals and adult horses, and between healthy and septic foals; (3) to determine relationships between free cortisol concentration and indicators of disease severity and outcome in septic foals; and (4) to compare the predictive value of free cortisol and total cortisol parameters for predicting disease severity and outcome in septic foals. We hypothesized that (1) in septic foals, low basal and ACTH-stimulated free cortisol concentration will correlate with disease severity and survival; and (2) serum free cortisol concentrations will more accurately predict disease severity and outcome than serum total cortisol concentrations in septic foals.
Materials and Methods
Three groups of animals were evaluated. The septic foal group included 51 client-owned neonatal foals that were ≤7 days of age at hospital admission and that met criteria for sepsis with either a positive blood culture (23/51 foals) or a sepsis score26≥11 (28/51) at admission. Mean sepsis score in all septic foals was 13.4±5.2 (range 1–25). Septic foals receiving corticosteroids in the 24 hours before admission were excluded from the study population. Eleven healthy neonatal Quarter Horse foals from a herd maintained at the University of Georgia's equine breeding facility were also evaluated at 4 time points during the 1st week of life to provide age-matched comparisons for the septic foal group. Each healthy foal was sampled within 1 hour of birth, at 12–24 hours, at 36–48 hours, and at 5–7 days of age.27 All healthy foals were full-term foals and were born without assistance. Foals in this group were determined to be healthy before and during inclusion in the study by lack of abnormalities identified on physical examination. Adequate transfer of passive immunity was confirmed by a serum immunoglobulin concentration ≥800 mg/dL at 12–24 hours of age in all healthy foals.a Finally, 6 healthy adult horses from the University of Georgia's equine research herd were also evaluated once during the study period, to allow comparisons of total and free cortisol between healthy foals and adult horses.
Study methods were approved by the University's Institutional Animal Care and Use Committee and the College of Veterinary Medicine's Clinical Research Committee, and informed owner consent was obtained before foal enrollment in the septic foal group. Each healthy adult horse and healthy mare/foal pair was stabled with daily paddock turnout during the study period, and was cared for according to the principles and guidelines stated in an Animal Use Protocol determined by the University of Georgia's Department of Animal Resources.
Blood was collected from all 51 septic foals for measurement of basal serum total cortisol concentration (total cortisol concentration) and basal serum free cortisol fraction (free cortisol fraction [FCF]) after study enrollment at hospital admission. A paired low-dose (10 μg)/high-dose (100 μg) cosyntropinb (synthetic ACTH, α1–24 corticotropin) stimulation test17,27 (henceforth referred to as “paired cosyntropin stimulation test,”Fig 1) was also performed within the first 12 hours of hospital admission in 45/51 septic foals. Blood was collected for measurement of total cortisol concentration and FCF before and 30 minutes after IV administration of 10 μg cosyntropin to assess the cortisol response to the low dose of cosyntropin. Ninety minutes after administration of the 10 μg dose, 100 μg cosyntropin was administered IV, and blood was collected 30 and 90 minutes later to assess the cortisol response to the high dose of cosyntropin. Cosyntropin was administered and all blood samples for the paired cosyntropin stimulation test were collected through the foal's indwelling jugular catheter. Detailed total cortisol findings from this group of septic foals have been reported previously.17
Clinical and clinicopathologic data were collected from the medical records of septic foals to determine the incidence of shock, MODS, and nonsurvival in this population. Specific clinical definitions for shock and MODS in neonatal foals have not been developed to date, so criteria for shock and MODS for use in the study herein were adapted from recent definitions used in other species.28–31 Shock was considered present in foals that had at least 2 of the following at admission: (1) mean arterial pressure <60 mmHg as measured indirectly via oscillometric sphygmomanometry with a tail cuff; (2) cold extremities; (3) weak peripheral pulses; (4) capillary refill time >2 seconds; (5) altered mental status (eg, marked depression or inability to stand, nurse, or track the mare); (6) rectal temperature <99.0°F (< 37.2°C); or (7) plasma lactate >5 mmol/L. Any foal meeting the above criteria was considered to have shock, regardless of the specific cause of shock. MODS was defined as at least 2 of the following criteria present at any time during hospitalization: (1) anuria, oliguria, or persistent azotemia (serum creatinine >2.2 mg/dL for ≥72 hours) after initial fluid resuscitation and rehydration; (2) clinical diagnosis of neonatal encephalopathy; (3) respiratory dysfunction (hypoxemia or hypercapnea or both) requiring either nasal insufflation of oxygen or ventilation; (4) persistent ileus necessitating withholding of enteral feeding for >24 hours; (5) >24 hours of vasopressor therapy; or (6) a clinical diagnosis of disseminated intravascular coagulation (as defined previously32,33 by  the presence of clinical signs of hemorrhage or multiple sites of thrombosis and  ≥3 of the following 6 abnormal coagulation test findings: decreased platelet count, decreased plasma fibrinogen concentration, prolonged prothrombin time, prolonged activated partial thromboplastin time, decreased antithombin activity, or increased serum fibrinogen degradation products). Survival was defined as survival to hospital discharge; nonsurvival was defined as death or euthanasia for reasons of worsening disease or poor prognosis during hospitalization.
In the healthy foals, HPA-axis function was assessed at the aforementioned time points during the 1st week of life to provide age-matched comparisons with the hospitalized foal group. Time of birth was determined by use of a Foalertc system placed on each mare before the anticipated foaling date, and defined as the time at which the Foalert alarm was activated. At each age, blood was collected for measurement of basal serum total cortisol concentration and FCF, and a paired cosyntropin stimulation test was performed as outlined above. Detailed total cortisol findings from this serial HPA-axis assessment in these healthy foals have been reported previously.27 In the 6 adult horses, blood was collected for measurement of basal total cortisol concentration and FCF at 1 time point in each horse. Cosyntropin stimulation tests were not performed in adult horses. In healthy foals and adult horses, blood samples were collected and cosyntropin administered (foals only) by direct jugular venipuncture.
Blood for measurement of total cortisol concentration and FCF was collected into a sterile glass tube without additives and allowed to clot at room temperature for 30–60 minutes, and then stored at 4°C until processing. All samples were centrifuged and the serum removed within 3 hours of collection, and stored frozen at −80°C until assays were performed.
Total and Free Cortisol Assays
Total cortisol concentrations were determined on an automated analyzer by a chemiluminescent enzyme immunoassayd validated for use in the horse.34,35 The lower limit of detection of this assay is 0.2 μg/dL.d
FCF was determined by a modification of an ultrafiltration assay described previously in humans and pigs.36,37 Briefly, 0.1 μCi of [3H]-cortisole in ethanol was placed into 2 mL glass tubes and allowed to evaporate to dryness. Then, 400 μL of serum was added to each tube, vortexed, and incubated at 37°C for 30 minutes. Each equilibrated sample was then diluted with 400 μL of assay buffer (1X phosphate-buffered saline [PBS] + 0.1% gelatin). For measurement of total radioactivity, a 50 μL portion of this diluted serum sample was retained and placed into a scintillation vial. Four hundred microliters of the diluted equilibrated serum was then placed into an ultrafiltration devicef that had been preconditioned with assay buffer, and centrifuged at 14,000 ×g for 60 minutes at 25°C. For measurement of ultrafiltrate radioactivity, 50 μL of the ultrafiltrate was retained and placed into a scintillation vial. Radioactivity was measured in 2 mL of scintillation fluidg in a liquid scintillation counter.h Because total and ultrafiltrate radioactivity was measured in identical sample volumes within the same vehicle system, counting efficiency was similar for both measurements, so total counts per minute were not corrected for quench. Further, because undiluted serum samples were equilibrated with [3H]-cortisol, and 1 : 2 dilution of equilibrated serum aliquots in assay buffer was performed immediately before ultrafiltration, dilution would have a minimal effect on cortisol-binding kinetics. Thus, no correction of scintillation counts for sample dilution was necessary.
FCF, percent free cortisol (% free cortisol), and estimated free cortisol concentrations were calculated using the following formulas:
Free Cortisol Assay Optimization
Because measurement of FCF by ultrafiltration methodology has not been described in the horse, a series of preliminary experiments were carried out to optimize the reported methodology36,37 for equine samples. First, filtration efficiency was tested to ensure that significant binding of serum protein to the ultrafiltration device did not occur. Three 400 μL replicates of undiluted equine serum and 1 : 2, 1 : 5, and 1 : 10 dilutions of equine serum with 1X PBS were centrifuged at 14,000 ×g for 30 and 60 minutes. Undiluted and 1 : 2 dilutions of equine serum were also centrifuged for 60 minutes at 10, 15, and 25°C to determine maximal filtration efficiency centrifugation temperature. Sample and ultrafiltrate weights were determined before and after filtration, and the recovery fraction was calculated by subtracting the ultrafiltrate weight from the sample weight. Filtration efficiency was maximized with a 60-minute centrifugation at 25°C and by diluting the serum sample at least 1 : 2 (data not shown).
Next, serial dilutions of 3H-cortisol were prepared in assay buffer to ensure that significant binding of 3H-cortisol to the filter did not occur with the above optimized ultrafiltration protocol. Total radioactivity was recorded in triplicate for each sample as above, and then triplicates of each preparation were placed into the ultrafiltration devices and centrifuged for 60 minutes at 25°C. Ultrafiltrate radioactivity was determined, and the radioactivity recovery was calculated by dividing the ultrafiltrate radioactivity by the total radioactivity. Radioactivity recovery ranged from 87.3 to 93.3%, and was not significantly different between the various dilutions of 3H-cortisol (P= .235, data not shown). Thus, the previously described ultrafiltration protocol for FCF determination36,37 was modified for use in the horse to include a 60-rather than 30-minute centrifugation at 25°C and a 1 : 2 dilution of the original serum sample with assay buffer to maximize filtration efficiency.
The modified protocol was then tested with thawed equine serum samples to determine if prefiltering the serum to remove large molecular aggregates further increased filtration efficiency. Duplicate serum samples were obtained and frozen for at least 96 hours at −80°C and then thawed at room temperature. The ultrafiltration assay was performed as described above, except 1 of the 2 samples was filtered through a sterile 22 μm, 13-mm-membrane filteri before equilibration with 3H-cortisol. The recovery fraction was determined from the sample and ultrafiltrate weights, and the FCF was calculated from the total and ultrafiltrate radioactivity as described above. The recovery fraction and the FCF did not differ significantly between unfiltered and prefiltered samples (P= .199 and .841, respectively, data not shown); thus, prefiltering the serum was not included in the final modified protocol.
Basal total cortisol concentration, FCF, and estimated free cortisol concentration were determined for all foals and adult horses. For clarity and consistency, FCF is henceforth expressed as % free cortisol. Serum cortisol responses to both the 10 μg and the 100 μg dose of cosyntropin were also determined for all healthy and septic foals that had a paired cosyntropin stimulation test performed. The low-dose delta total cortisol concentration (LDdelta total) was defined as the cortisol concentration reached 30 minutes after administration of 10 μg cosyntropin minus the immediate precosyntropin (time 0) cortisol concentration. The high-dose delta cortisol total concentration (HDdelta total) was defined as the peak cortisol concentration reached either 30 or 90 minutes (whichever was higher) after administration of 100 μg cosyntropin minus the time 0 concentration. The low-dose and high-dose estimated delta free cortisol concentrations (LDdelta free, HDdelta free) were defined similarly using basal and cosyntropin-stimulated estimated free cortisol concentrations.
Total and free cortisol parameters were compared between healthy adult horses and healthy foals by the Mann-Whitney U-test. To test for an effect of age on free cortisol parameters in healthy foals, basal % free cortisol and estimated free cortisol concentration were compared among the 4 healthy foal age groups by a repeated measures analysis of variance, with multiple comparisons conducted by Tukey's test.
Total and free cortisol parameters were compared similarly between healthy and septic foals by the Mann-Whitney U-test. Because previous work has shown a significant effect of age on basal total cortisol concentrations and on total cortisol responses to exogenous ACTH in neonatal foals during the 1st week of life,27,38,39 all comparisons between the healthy and septic foal groups were age matched. Based on findings from these previous studies,27,38,39 septic foals were divided into 4 age groups: (A) ≤4 hours; (B) >4 hours to ≤30 hours; (C) >30 hours to < 3 days; and (D) ≥3 days to ≤ 7 days of age; data from age groups A, B, C, and D were compared with healthy foals at birth, 12–24 hours, 36–48 hours, and 5–7 days of age, respectively.
Finally, within the septic foal group, data were further stratified by the Mann-Whitney U-test to compare total and free cortisol parameters between septic foals that did and did not meet criteria for shock, MODS, and nonsurvival. Receiver operator characteristic (ROC) curves were constructed to compare the predictive value of basal total cortisol concentration with basal % free cortisol concentration, and LDdelta and HDdelta total cortisol concentrations with LDdelta and HDdelta free cortisol concentrations for predicting disease severity and outcome in septic foals.
Hypothesis tests were 2-tailed and statistical significance was set at P < .05 for all analyses. Statistical analyses were performed by commercial statistical software.j,k
Body weight was not significantly different between the septic and healthy foal groups (P= .498) (Table 1). Septic foals were divided into the previously described age groups for comparison with healthy foals (Table 1).27 Eleven septic foals fit age group A (≤4 hours of age), 12 foals fit age group B (>4 to <30 hours of age), 23 foals fit age group C (≥30 hours to <3 days of age), and 5 foals fit age group D (≥3 hours to ≤7 days of age).
Table 1. Animal group characteristics
Age (Horses) or Gestational Age (Foals)
Reason for Admission (Note: Some foals had >1 complaint)
305 days (n = 1) 315–320 days (n = 3) 321–329 days (n = 4) 330–345 days (n = 32) 346–359 days (n = 2) 360–370 days (n = 3) Unknown (n = 6)
Recumbency (n = 12) Failure to nurse well (n = 10) Weakness/lethargy (n = 12) Colic (n = 13) Diarrhea (n = 12) Postdystocia or cesarean section (n = 6) Failure of transfer of passive immunity (n = 5) Premature parturition (≤320 days, n = 3) Joint swelling (n = 1) Premature placental separation (n = 2) Orphans (n = 2) Agalactia in the dam (n = 2) Seizures (n = 1) Fever (n = 1) Inguinal hernia (n = 1) Surviving twin (n = 1) Severe extensor tendon contracture (n = 1)
47.6 ± 13.1(11.8–79.5)
Healthy foals (n = 11)
6 male 5 female
Quarter Horse (n = 11)
≥330 days (n = 11)
50.4 ± 7.5 (38.7–61.4)
Healthy adult horses (n = 6)
4 male 2 female
Quarter Horse (n = 6)
There was a significant effect of age on basal total cortisol concentration, as well as on % free cortisol and estimated free cortisol concentration, was found in healthy foals (P < .001 for all analyses), with the highest values for all parameters seen in foals at birth (Table 2). Total cortisol concentration was not significantly different between healthy foals and adult horses (foals 12–24 hours of age: P= .313; foals 36–48 hours of age: P= .428; foals 5–7 days of age: P= .159) except in foals at birth, in which it was significantly higher (P < .001). However, healthy foals had significantly higher % free cortisol and estimated free cortisol concentration than adult horses at all ages (P < .001 to P= .004).
Table 2. Basal total cortisol concentration, basal % free cortisol, and basal free cortisol concentration in healthy foals (n = 11) at 4 ages during the 1st week of life and in healthy adult horses (n = 6)
Basal Total Cortisol (μg/dL)
Basal % Free Cortisol (%)
Basal Free Cortisol (μg/dL)
% free cortisol, percent free cortisol.
Within a column, values with different letter superscripts are significantly different between individual foal age groups (P < .05).
Numbers in parentheses represent the data range.
Data are reported as the mean ± SD.
Significant differences between foals and adult horses (P < .05).
Basal total cortisol concentration was significantly higher in septic foals than in healthy foals at all ages except in the youngest foals (≤4 hours of age) (Table 3). Basal estimated free cortisol concentration was significantly higher in septic foals in age groups B (>4 to ≤30 hours) and C (>30 hours to <3 days) but was not significantly different in the youngest (≤4 hours) or oldest (≥3 to ≤7 days) septic foals. LDdelta and HDdelta total cortisol concentrations and LDdelta and HDdelta estimated free cortisol concentrations were not significantly different between healthy and septic foals in any age groups.
Table 3. Basal and cosyntropin-stimulated total and free cortisol concentrations (mg/dL) in age-matched healthy and septic foals
HDdelta free, high-dose delta free cortisol concentration; HDdelta total, high-dose delta total cortisol concentration; LDdelta free, low-dose delta free or total cortisol concentration; LDdelta total, low-dose delta total cortisol concentration.
Cosyntropin-stimulated values (LDdelta and HDdelta parameters) were determined in response to a paired low- (10 μg) and high- (100 μg) dose cosyntropin stimulation test.27 Foals were divided into 4 age groups for age-matched comparisons: (A) ≤4 hours of age; (B) >4 to ≤30 hours of age; (C) >30 hours to <3 days of age; and (D) ≥3 to ≤7 days of age.
Data are shown as mean ± SD (range).
Significant difference (P < .05) between age-matched healthy and septic foals.
Septic foals that met criteria for shock and for nonsurvival had significantly higher basal total and free cortisol concentrations, and significantly lower HDdelta total and free cortisol concentrations (Table 4). Septic foals that met criteria for MODS had significantly lower LDdelta total and free cortisol concentrations, but no significant differences in any other total or free cortisol parameters.
Table 4. Total and free cortisol parameters in septic foals that did or did not meet criteria for shock, MODS, or nonsurvival
Basal Total Cortisol (μg/dL)
Basal Free Cortisol (μg/dL)
LDdelta Total Cortisol (μg/dL)
LDdelta Free Cortisol (μg/dL)
HDdelta Total Cortisol (μg/dL)
HDdelta Free Cortisol (μg/dL)
HDdelta free, high-dose delta free cortisol concentration; HDdelta total, high-dose delta total cortisol concentration; LDdelta free, low-dose delta free or total cortisol concentration; LDdelta total, low-dose delta total cortisol concentration; MODS, multiple organ dysfunction syndrome.
Basal total and free cortisol data were available from 51 foals, and cosyntropin-stimulation test results (LDdelta and HDdelta total and free cortisol) were available from 45 foals.
Significant difference (P< .05) in the specified parameter in septic foals that met specific criteria for shock, MODS, or nonsurvival, as compared with septic foals not meeting that specific criterion.
ROC curves and corresponding areas under the curves (AUCs) were used to compare the predictive value of basal and cosyntropin-stimulated total and free cortisol concentrations for predicting disease severity and non-survival in septic foals (Fig 2 and Table 5). No significant differences in the AUCs were found between any total and free cortisol parameters; thus, there was no meaningful predictive advantage of measuring basal or cosyntropin-stimulated free cortisol concentration over total cortisol concentration for predicting shock, MODS, or nonsurvival in septic foals.
Table 5. Mean (±SE) areas under the curve (AUCs) for receiver operator characteristic curves comparing the predictive value of basal and cosyntropin-stimulated total versus free cortisol concentration for predicting shock, MODS, and nonsurvival in septic foals (n = 51)
Shock (AUC ± SE)
MODS (AUC ± SE)
Nonsurvival (AUC ± SE)
HDdelta free, high-dose delta free cortisol concentration; HDdelta total, high-dose delta total cortisol concentration; LDdelta free, low-dose delta free or total cortisol concentration; LDdelta total, low-dose delta total cortisol concentration; MODS, multiple organ dysfunction syndrome.
No significant differences (P< .05) in the AUCs between total and free cortisol parameters were found.
Basal total cortisol (μg/dL)
0.838 ± 0.056
0.425 ± 0.088
0.759 ± 0.072
Basal free cortisol (μg/dL)
0.857 ± 0.057
0.441 ± 0.088
0.755 ± 0.071
LDdelta total cortisol (μg/dL)
0.631 ± 0.084
0.696 ± 0.093
0.621 ± 0.103
LDdelta free cortisol (μg/dL)
0.636 ± 0.083
0.719 ± 0.091
0.599 ± 0.107
HDdelta total cortisol (μg/dL)
0.707 ± 0.079
0.652 ± 0.096
0.725 ± 0.096
HDdelta free cortisol (μg/dL)
0.727 ± 0.076
0.626 ± 0.096
0.753 ± 0.084
The above findings illustrate unique plasma cortisol-binding dynamics in foals. While % free cortisol in adult horses, at approximately 5–10%, was consistent with previous reports40 and findings in other species,20,21,36,37,41 the proportion of cortisol present in the free fraction in healthy neonatal foals throughout the 1st week of life was 3–6-fold higher than in adult horses, ranging from approximately 60% at birth to approximately 30% by 5–7 days of age. Total cortisol concentrations were significantly higher in foals at birth than in adult horses, but no difference in total cortisol concentration was found between adult horses and foals 12 hours to 7 days of age. Thus, the substantial increase in % free cortisol and estimated free cortisol concentration observed in healthy foals is because of a dramatic reduction in plasma cortisol-binding capacity in foals rather than an overall increase in total cortisol concentration.
As described previously, plasma cortisol is primarily bound to CBG, with a small percentage (<5%) bound to albumin.20,21 Given that only a small fraction of cortisol is bound to albumin, the reduction in plasma cortisol-binding capacity found in foals in this study is most likely because of decreased levels of serum CBG in foals as compared with adult horses. Unfortunately, to the authors' knowledge, an assay for equine CBG is not currently available to confirm this theory. Reductions in CBG concentrations and concomitant increases in plasma FCF from adult levels have also been documented in premature and full-term infants.42,43 However, in full-term infants, % free cortisol ranged from 32% at birth to 19% by 3 months of age,42 levels approximately half of those found in foals during the 1st week of life in the study herein. Thus, while a decrease in plasma cortisol-binding capacity associated with low plasma CBG concentrations may be common to both human and equine neonates, the plasma cortisol-binding capacity in healthy neonatal foals is still substantially less than that of human infants.
This study demonstrates that septic foals >4 hours of age have significantly higher basal total cortisol concentrations and estimated free cortisol concentrations than healthy age-matched foals. Such an increase in total cortisol concentrations is appropriate and expected with HPA-axis activation by the physiologic stresses of sepsis, and is consistent with previous reports.16–18 Because of the neonatal foal's limited plasma cortisol-binding capacity as described above, this stress-induced increase in cortisol secretion most likely results in saturation of limited plasma protein cortisol-binding sites and thus also leads to an increase in estimated free cortisol concentration. Concurrent suppression of CBG production in septic foals because of negative acute phase effects22,23 may also contribute to reduced plasma cortisol-binding capacity in septic foals.
Despite the above differences in basal total and free cortisol between healthy and septic foals, total and free cortisol responses to cosyntropin were similar between healthy and septic foals. Further stratification of the data in septic foals by disease severity and outcome illustrated higher basal total cortisol and estimated free cortisol concentrations, and lower total and free delta cortisol responses to cosyntropin stimulation in septic foals with shock and nonsurviving foals as compared with septic foals that did not have shock and survivors. Thus, low basal free cortisol concentration was not correlated with more severe disease or decreased survival as initially hypothesized. These findings for free cortisol are consistent with total cortisol findings described previously,17 and these consistencies observed between basal and cosyntropin-stimulated total and free cortisol are likely because of the fact that such a large proportion of total cortisol is present in the free form in foals.
As discussed above, the increases in basal total and free cortisol in the most ill and nonsurviving foals are consistent with stress-induced HPA-axis activation, and are not unexpected. The lower delta cortisols in the most ill and nonsurviving foals might indicate that adrenocortical cortisol synthesis is already at or near maximal capacity and further trophic hormone stimulation cannot increase cortisol concentrations any further (loss of adrenal reserve). Alternatively, adrenal sensitivity to ACTH might be reduced in critical illness, resulting in this limited adrenocortical cortisol response. Feedback inhibition of cortisol synthesis by the high basal cortisol concentrations observed in severe illness is unlikely to cause this decreased responsiveness to exogenous ACTH in the most ill foals, as studies in humans have demonstrated that cortisol responses to ACTH stimulation testing are independent of short duration hypercortisolemia.44,45
The results herein also do not support the second hypothesis, that free cortisol concentration better predicts disease severity and death in septic foals than total cortisol concentration. This was unexpected, and differs from clinical studies in septic adult humans and children showing better correlations between basal and cosyntropin-stimulated free cortisol concentrations and disease severity and outcome than with total cortisol.20–25 Again, this difference between foals and humans is most likely because of the dramatically higher FCF in foals. In adult humans, with ≤10% free cortisol, a small increase in the FCF is likely to have substantial physiologic effects as more biologically active cortisol is available. In foals, though, with such a large FCF even in health, a small increase in this free fraction could have minimal to no effect on the biologic availability of cortisol to enter cells and interact with its receptors. Thus, this substantially larger FCF in foals as compared with people explains the lack of predictive advantage of measuring free cortisol versus total cortisol in septic foals as compared with septic adult humans. While human neonates do have higher % free cortisol, at approximately 20–30%,42,43 than adult humans, % free cortisol in foals in the study herein is still markedly greater at approximately 30–60%. To the authors' knowledge, the predictive value of free versus total cortisol parameters in septic human neonates has not been investigated.
One important limitation of this study lies in the potential for misdiagnosis of sepsis in the study foals with the criteria applied in this study. Given the time delay to obtain results and the substantial potential for a false-negative result when blood cultures are used as the sole means for sepsis diagnosis, a combination of blood culture results and sepsis scoring was used in an attempt to maximize the diagnostic accuracy for sepsis in the study population. However, the use of the sepsis scoring system is controversial in foals, as is the cut-off value for sepsis diagnosis (score ≥11). It is possible that some foals in the study with a low positive sepsis score were not septic, which could have contributed to the absence of substantial differences in cortisol responses to cosyntropin identified between healthy and septic foals. However, significantly higher total cortisol and estimated free cortisol concentrations in septic foals than in age-matched healthy foals were identified, consistent with substantial HPA-axis activation associated with the stress of critical illness in the septic group. Additional study in a larger group of foals, including both septic and sick but nonseptic foals, might be helpful to further elucidate key impacts of sepsis on HPA-axis function in foals.
The group of septic foals studied was a heterogenous group, consisting of foals of various breeds comprising a wide weight range. Smaller foals received a larger dose of cosyntropin on a body weight basis, which might have impacted the results. However, the majority of the foals in the septic group were Thoroughbreds or Quarter Horses, and all but 2 were horse breeds. If an important impact of weight on cortisol responses to cosyntropin was present, one would expect greater cortisol responses in the smallest foals. However, inspection of the data from the 2 smallest foals (an 11.8 kg donkey and an 13.5 kg Shetland pony) revealed cortisol responses to cosyntropin comparable to responses in larger foals, suggesting weight had minimal to no impact on the cosyntropin stimulation test responses reported herein.
Finally, given the large FCF in foals, one could wonder if the occurrence of CIRCI is then even possible in septic foals, because so much cortisol is theoretically biologically available to cross cell membranes and interact with the cytoplasmic cortisol receptor to induce relevant physiologic effects in response to stress. However, while it initially might seem counterintuitive, the substantially increased FCF and reduced cortisol-binding capacity in the neonatal foal have several important consequences that could make foals at even greater risk for CIRCI. Circulating free cortisol is preferentially metabolized and eliminated via renal excretion and hepatic conjugation over bound cortisol46; therefore, bound cortisol essentially provides a reservoir of cortisol in the plasma, and decreased plasma cortisol-binding capacity can result in more rapid cortisol clearance and relative cortisol insufficiency during illness.47 In addition, CBG appears to play a vital role in the mediation of some of cortisol's physiologic effects, by promoting delivery and release of cortisol to specific steroid-responsive target tissues and sites of inflammation and by direct interactions with cell surface receptors that bind CBG-cortisol complexes.48–50 Recent work also documents hyporesponsiveness to glucocorticoids and an aggravated response to septic shock in mice that are genetically deficient for CBG.51 Thus, the substantially decreased plasma cortisol-binding capacity observed in septic foals might limit rather than enhance the biologic activity of cortisol in stress responses by increasing cortisol clearance and impairing CBG-cortisol complex mediated cellular and physiologic responses. Ultimately, this could impair cortisol-mediated physiologic responses to sepsis and contribute to the development of CIRCI in septic neonatal foals.
This study documents substantially greater plasma FCF in neonatal foals as compared with adult horses and both adult and neonatal humans. This increased FCF is the result of reduced plasma cortisol-binding capacity in foals as compared with adult horses, presumably because of decreased CBG concentrations although further study is needed to confirm this supposition. Septic foals with shock or MODS had decreased total and free cortisol responses to cosyntropin, consistent with impaired adrenocortical sensitivity to ACTH or exhaustion of adrenocortical cortisol synthetic capacity. However, basal and cosyntropin-stimulated free cortisol concentrations were no better than total cortisol concentrations for predicting disease severity and outcome. Thus, measurement of basal or cosyntropin-stimulated free cortisol appears to offer no advantage over measurement of cosyntropin-stimulated total cortisol for diagnosis of CIRCI or prognostication in septic foals. Further study of cortisol concentrations and cosyntropin stimulation test responses over time in critically ill foals, and documentation of CBG concentrations and cortisol clearance rates in healthy and septic neonatal foals, is necessary to better understand the consequences of these unique cortisol-binding dynamics on the pathogenesis of CIRCI in septic foals.
aSNAP Foal IgG Test, IDEXX Laboratories Inc, Westbrook, ME
bCortrosyn, Amphastar Pharmaceuticals Inc, Rancho Cucamonga, CA
cFoal Alert Inc, Atlanta, GA
dImmulite, Diagnostics Product Corporation, Los Angeles, CA
eAmersham Radiochemicals, GE Healthcare Life Sciences, Pittsburgh, PA
fMillipore, Billerica, MA
gScintiverse BD Cocktail, Fisher Scientific, Pittsburgh, PA
hLS 6500 Multipurpose Scintillation Counter, Beckman Coulter, Brea, CA
iAcrodisc 0.2 μm 12 mm Supor Membrane Syringe Filter, Pall, Port Washington, NY
jGraphPad Prism Statistical Software (Version 4), GraphPad Software Inc, San Diego, CA
kStata (Version 11.0), StataCorp LP, College Station, TX
The authors acknowledge the clinicians, technicians, and students at the University of Georgia and Hagyard Equine Medical Associates for assistance in sample collection and processing.
This study was funded by the University of Georgia College of Veterinary Medicine's Clinical Research Fund and the American College of Veterinary Internal Medicine Foundation.