Materials and Methods

Animals and groups—Mature horses evaluated at the Ontario Veterinary College Teaching Hospital between March 2007 and September 2007 were used for this study. Horses were assigned to a SIRS¹¹ group, an NSIRS group, or a healthy control group on the basis of results of clinical evaluation. Information and data that were available for each horse from both colic groups (SIRS and NSIRS) included that obtained from a complete physical examination including palpation per rectum, nasogastric intubation, and abdominal ultrasononography; venous blood analyses for sodium, potassium, chloride and TP concentrations, PCV, and blood gases; and for most horses, a CBC. In horses in which a CBC was performed at a later time point than on admission, CBC data were not used.

Inclusion criteria for the SIRS group were that horses had colic signs, an abnormal finding on palpation per rectum, heart rate > 40 beats/min, discolored mucous membranes (red to purple), and capillary refill time > 2 seconds. Inclusion criteria for the NSIRS group were that horses had colic signs, an abnormal finding on palpation per rectum, and normal mucous membrane color and capillary refill time. The control group horses were from the teaching herd; were bright, alert, and responsive; had normal findings on physical examination; and had CBC values within reference ranges.

Blood endotoxin activity measurements—Twenty milliliters of venous blood was withdrawn from all horses after completion of the physical examination and placed into 2 EDTA blood collection tubes. In all except 4 horses, 1 blood sample was used for an initial CBC. For all horses, the other blood sample was maintained at room temperature (21°C) and endotoxin activity was assayed within 60 minutes of collection. An endotoxin activity assay^a was used with slight modifications of the manufacturer's recommended procedure to measure endotoxin activity in equine blood. Briefly, blood spiked with a maximal LPS concentration of 9,200 pg/mL was used for the positive control (on the basis of results from a pilot study in healthy horses), the assay time was extended from 20 to 40 minutes, and integrals were calculated over that time period.

Forty microliters of blood was added to a reagent mixture containing saturating concentrations of a murine IgM monoclonal antibody directed against lipid A of Escherichia coli J5, luminol, and zymosan (Saccharomyces cerevisiae). The resulting respiratory burst activity of neutrophils was detected as light released from oxidized luminol in a chemiluminometer.^b Each assay was run in duplicate and consisted of 3 measurements: basal activity (chemiluminescence in the absence of anti-endotoxin antibody to measure nonspecific oxidative burst activity of sample neutrophils), test sample activity (chemiluminescence of the sample measured by addition of specific anti-endotoxin antibody to stimulate oxidative respiratory burst activity of neutrophils in proportion to the concentration of endotoxin), and maximal activity (chemiluminescence of sample after addition of anti-endotoxin antibody and excess exogenous endotoxin [9,200 pg of LPS/mL]). The endotoxin activity of the test sample was calculated by normalizing chemiluminescence in the test sample against the maximum chemiluminescence and correcting both measurements for basal activity chemiluminescence.¹²

Overlap performance—Recognizing the lack of a gold standard to measure endotoxin activity in blood from horses, diagnostic overlap performance was assessed indirectly.¹⁰ Endotoxin activity was measured in 20 horses in the SIRS group, 8 horses in the NSIRS group, and 20 healthy horses. In all horses, data recorded included heart rate, mucous membrane color, PCV, TP concentration, base excess from blood gas analysis (SIRS and NSIRS horses only), WBC count, presence of band neutrophils (yes or no), presence of so-called toxic neutrophils (yes or no), and outcome (survival or euthanasia).

Statistical analysis—One-way ANOVA was performed on endotoxin activity values to determine whether there was a significant difference among the 3 groups. A Shapiro-Wilk test and examination of residuals were used to evaluate whether the data were normally distributed. A Tukey comparison of means was performed if the F test from the ANOVA was significant. Intra-assay accuracy was assessed by use of concordance correlation. Determination of closeness of comparisons was evaluated on the basis of the kappa coefficient.¹³ Exact conditional logistic regression was used to determine whether high endotoxin activity values increased the likelihood of euthanasia. The Pearson correlation coefficient was calculated to assess the relationship between WBC count and endotoxin activity. Significance was set at values of P ≤ 0.05, and all analyses were performed with a statistical software package.^c

Results

In the SIRS group, 10 horses had an obstruction of the small or large intestine, 5 had enteritis, and 5 had intestinal perforation with septic peritonitis. The 8 horses in the NSIRS group either had a right, left, or cranial displacement of the colon; in 1 horse, mild colitis was also diagnosed. All horses in the NSIRS group had WBC count values within reference ranges and survived with either medical or surgical treatment.

Mean ± SE endotoxin activity was 0.16 ± 0.05 for control horses, 0.18 ± 0.07 for horses in the NSIRS group, and 0.53 ± 0.05 for horses in the SIRS group. A significant (P < 0.001) difference among NSIRS, SIRS, and control groups was present. Mean endotoxin activity was significantly (P < 0.001) higher in the SIRS group than in the NSIRS and control groups. No significant difference between the control and NSIRS groups was present. Intra-assay accuracy between duplicate measurements was excellent (r = 0.92).

Twelve of 20 horses in the SIRS group were euthanized because of a poor prognosis as determined on the basis of the attending clinician's judgment. The clinician was unaware of the endotoxin activity value. The higher the measured endotoxin activity value, the more likely it was for the horses to be euthanized (P = 0.008; odds ratio, 90.18 [95% confidence interval, 2.77 to 999]). The cumulative probability for horses to be euthanized on the basis of blood endotoxin activity was determined (Figure 1). A significant (P = 0.04) though weak negative correlation (r = −0.34) was noted between endotoxin activity and WBC count.

View Full Size

Cumulative probability of euthanasia as a function of endotoxin activity in horses. Diamonds indicate predicted probability; squares and triangles indicate upper and lower limits, respectively, of the 95% confidence interval.

Citation: American Journal of Veterinary Research 70, 10; 10.2460/ajvr.70.10.1183

• Download Figure
• Download figure as PowerPoint slide

Discussion

On the basis of results from this study, the chemiluminescent endotoxin assay appeared useful for measuring endotoxin activity in equine blood. Unlike the results of previous studies^5,14 of horses in which endotoxin activity was measured with the LAL assay, endotoxin activity in this study seemed to be more consistently increased in horses with clinical signs of endotoxemia. In humans, an endotoxin activity of 0.4 is used as a cutoff value to determine that there is a risk of severe endotoxemia and a need for treatment with polymyxin B.¹⁵ A similar cutoff value was reported for dogs.¹⁰ However, because of test kit modifications required to achieve meaningful chemiluminescent values in horse samples, endotoxin activity values derived in the present study are not directly comparable to those identified in dogs and humans. In the study reported here, 7 horses of the SIRS group had values < 0.4, which likely means that different cutoff values would need to be determined from the modified endotoxin activity assay for horses, relative to humans and dogs. It is also possible that the 7 horses from the SIRS group did not have a large amount of endotoxin circulating through their blood. On the basis of findings from the present study, horses with endotoxin activity values > 0.25 may be considered to require intensive intervention; however, a greater number of horses with colic and controlled interventions should be assessed to provide precise guidelines. Outcome prediction based on endotoxin activity values is presently also limited by the modest number of horses assessed, as indicated by wide confidence intervals for the odds ratio estimates in the present study. Low endotoxin activity in horses with signs of septicemia was consistent with previous reports^5,14 and may relate to the ability of horses to quickly clear endotoxin from blood. Other possibilities for the low endotoxin activity detected in horses with endotoxemia are less than optimal interaction of the murine antibody with LPS in the presence of equine blood immunoglobulins; relatively less abundant respiratory burst activity of equine neutrophils, compared with neutrophils of other species; limited specificity of the lipid A antibody for the gram-negative organisms typically causing endotoxemia in horses; and variable neutrophil concentrations. In humans, the assay was reliable over neutrophil concentrations ranging from 500 to 20,000 cells/μL, and horses in the present study had neutrophil concentrations within this range; therefore, limited neutrophil numbers were considered an unlikely cause of the low values. It was considered more likely that priming of neutrophils by endotoxin-antibody complexes was less efficient, possibly because of differences in endotoxin binding proteins such as CD14 or because of less than optimal concentrations of endotoxin and antibody.¹² However, it was encouraging that all healthy horses and horses from the NSIRS group had endotoxin activity values < 0.25, suggesting that further validation of the assay is warranted. Interestingly, all horses in the SIRS group with endotoxin activity ≥ 0.7 died. In humans, endotoxin activity > 0.6 is significantly associated with development of severe endotoxemia and a longer stay in an intensive care unit.¹⁵ Evaluating endotoxin activity in a larger number of horses will be required to establish meaningful cutoff values for horses that might be used to predict outcome than present methods. The chemiluminescent endotoxin assay evaluated in this study was easy to use, required a short time (40 minutes) to perform, could be completed at the patient's side, and with the indicated modifications, may be a useful component in the clinical assessment and prognostication of horses with colic.