The effect of indomethacin on systemic and renal hemodynamics in neonatal piglets during experimental endotoxemia

Abstract

Systemic and renal hemodynamics are affected by prostaglandin production during endotoxemia. To study indomethacin effects on endotoxinemia in a neonatal piglet model, sixteen 7–10 day old piglets were anesthetized, ventilated, and catheterized. Mean arterial pressure (MAP), heart rate (HR), and urine output were continuously monitored. Endotoxin (0.06 mcg/kg) was injected after baseline measurements. We studied two groups with either endotoxinemia alone (n = 7) or an additional indomethacin infusion (0.2 mg/kg per h, n = 9). HR, MAP, renal blood flow (RBF), systemic and renal vascular resistance (SVR, RVR), cardiac index (CI), and glomerular filtration rate (GFR), were obtained at baseline, at 1, 2 and 3 h. We observed a drop in CI and an increase in SVR and HR within 3 h of endotoxinemia, while MAP remained unchanged. These effects were prevented by indomethacin. RVR was not altered significantly. Endotoxinemia triggered a drop of RBF in both control (P < 0.01) and intervention group (P < 0.05). In the intervention group, drop of GFR, urine volume, and paraaminohippuric acid clearance were apparent signs of nephrotoxicity (P < 0.01, <0.05, and <0.01). In conclusion, indomethacin maintains hemodynamic parameters during endotoxinemia at the expense of nephrotoxicity. We speculate that indomethacin counteracts the renoprotective effect of prostaglandins.

Introduction

Sepsis caused by gram-negative bacilli is a substantial cause of morbidity and mortality in neonatal intensive care units [1]. Gram negative sepsis is one of the most important causes of renal failure [2], which is often the first organ system to fail in neonatal sepsis and contributing to its high mortality [3]. The devastating effects of gram negative sepsis are mainly triggered by endotoxin released from gram negative bacteria during cellular division or lysis triggering endothelial damage, loss of vascular tone, coagulopathy and multiple-system organ failure, often resulting in shock and death [4]. These effects are, to a major part, mediated via pro-inflammatory cytokines such as tumor necrosis factor (TNF)-α and interleukin-1 (IL-1)-β, among others [5]. Indomethacin, a nonselective inhibitor of cyclooxygenase, has been suggested to alleviate sequelae of uncontrolled cytokine release in septic patients [6]. On the other hand, Azab et al. showed that while indomethacin was able to impair prostaglandin E2 production and hypothermia, it failed to reduce the cardiovascular alterations from endotoxemia in a rat mode [7]. We therefore tested a null hypothesis of assuming indomethacin not to prevent hemodynamic deterioration from endotoxinemia against an alternative hypothesis that it does have protective effects in a septic shock animal model. In addition, our objective was to analyze potential adverse effects on renal hemodynamic and function parameters as well as blood cell counts.

Methods

This investigation was approved by the University of Illinois at Chicago’s Animal Care Committee, and was in compliance with the 1996 Guide for the Care and Use of Laboratory Animals and with the Animal Welfare Act.

Animals

Sixteen 7–10 day old neonatal farm piglets weighing 2.2–4.0 kg (mean 2.8 ± 0.5) were sedated with ketamine (25 mg/kg), xylazine (2 mg/kg), and atropine (0.05 mg/kg). They were then anesthetized and paralyzed with fentanyl (50 μg/kg per h) and tubocurare (0.5 mg/kg). Ventilatory support was achieved with a Respiration Pump, model 607A, (Harvard Apparatus, Dover, MA). Ventilator settings were adjusted to keep arterial PO₂ at 80–100 mmHg and PCO₂ at 35–45 mmHg. Subsequently, catheters were placed into the jugular veins, as well as the femoral artery and left ventricle for blood pressure monitoring and blood sampling, medication, as well as fluid infusion. Urine collections were conducted via suprapubic catheterizations. Body temperature was maintained between 37 and 39°C using a heating blanket. After the last line was placed, a 0.9% sodium chloride solution was given as a bolus (10 ml/kg). The animals were then allowed 30 min of recovery to normalize heart rate (HR), mean arterial pressure (MAP) and arterial blood gases before baseline values were obtained.

Intervention and measurements

The animals were randomized into two groups, i.e. a control group with endotoxin alone (n = 7), as well as an intervention group with endotoxin and indomethacin (n = 9). Following randomization, 0.06 μg/kg of endotoxin, derived from Escherichia coli serotype 0111:B4 (Sigma Chemical Co., St. Louis, MO), was administered to the piglets. Subsequently, the animals of the intervention group received a continuous infusion of indomethacin 0.2 mg/kg over the entire experiment. Thereafter, measurements of cardiac index (CI), renal blood flow (RBF), systemic and renal vascular resistance (SVR, RVR), arterial blood gases, white blood cell and platelet counts were conducted at 1, 2 and 3 h post-endotoxin exposure.

Blood flow

Blood flow measurements were conducted via the left ventricular and the femoral artery catheters. The left ventricular catheter was connected to a pressure transducer (403 Ivy Biomedical Systems Inc., Branford, CN), and the femoral artery catheter was connected to a 1 ml syringe placed in a syringe pump (Harvard Apparatus 944, South Natick, MA). For each blood flow determination, blood withdrawal from the femoral artery catheter was initiated at a rate of 0.25 ml/min. At the same time arterial blood pressure was recorded from the left ventricular catheter. HR was derived from the pulsatile arterial pressure output of the blood pressure analyzer. After 30 s of blood withdrawal, the left ventricular catheter was disconnected from the pressure transducer, and approximately 4 × l0⁴ microspheres of 15 ±5 μm labeled with ¹²⁵I iothalamate and suspended in Tween 80 to a concentration of 0.12 μCi/ml. Before injection the radioactivity was determined in a gamma well scintillation counter using a multichannel analyzer [8]. Three measurements, at 1 h intervals, were made by injecting three different microsphere labels. After completion of the third measurement point the animals were killed by exsanguination. The heart, lungs, and kidneys were removed and weighed. The larger organs were cut into several pieces and placed in several scintillation vials to increase the geometric efficiency of counting. To check for adequate mixing and streaming of the microspheres each kidney was counted separately. The fraction of cardiac index to each organ was calculated from the ratio of radioactivity of each organ to total injected radioactivity (where total injected radioactivity was obtained by subtracting the residual radioactivity from the radioactivity in the Silastic tubing before injection). Absolute organ arterial flow (in ml/min) could therefore be calculated by multiplying the fraction of CI in each organ by the total CI. Vascular resistance for that organ was obtained by dividing mean arterial pressure by that organ blood flow.

Statistical analysis

All data assembled in this study were analyzed using SPSS 15.0 (Statistical Package for the Social Sciences, SPSS Inc., Chicago, IL, USA). Data were tested for normal distribution using the Kolmogorov–Smirnov test. All normally distributed data are displayed as mean ± SD, the non-normally distributed data as median with 1st and 3rd quartile. We analyzed our hypothesis question by generating three null hypotheses. The first null hypothesis is that there is no difference between the two groups of animals with the tested values averaged over time (H₀ group), the second is that the parameters do not differ between time points (H₀ time), and the third that there are no difference across time points between both groups (H₀ group × time). If the data were normally distributed, measurements of the intervention points were compared to the baseline values of each individual animal using Student’s t test, and, if otherwise, the Mann–Withney U test was applied (H₀ time and H₀ group). For (H₀ group × time), we conducted a mixed model ANOVA analysis with control and experimental group as between subject and measurement time points (baseline, 1, 2, 3 h) as within subject factor. A P value <0.05 was considered significant, P < 0.01 as highly significant.

Results

Baseline values of all categories assessed were non-significant between both control and intervention group. All data followed normal distribution.

Systemic circulation

In the group receiving endotoxin alone, HR increased throughout the experiment (P < 0.01 at 2 and 3 h). In addition, CI dropped by 75% from 407 ± 85 to 98 ± 7 (P < 0.01). While MAP changes were not significant within the 3-h interval of our observation, SVR increased significantly, with P < 0.05 and P < 0.01 at 1 and 3 h, respectively. In the intervention group with continuous indomethacin infusion, the effects observed were attenuated, compared to the internal control (baseline values of both the endotoxin plus indomethacin groups prior to randomization), only HR increase was significant from 157.44 ± 20.26 at baseline to 207 ± 38.09 at 3 h (P < 0.01), while CI, SVR and MAP remained stable throughout the experiment. Comparing cardiovascular parameters between both groups at the equivalent time, we could identify a trend to less cardiovascular effects, but findings did not reach significance (Fig. 1). The mixed model ANOVA analysis showed HR to be significantly higher in the control group (Fig. 2). For the mixed model ANOVA on SVR, a trend was noted towards lower values in the control group, but significance not reached. Therefore, we maintain H₀ group, reject H₀ time, and reject H₀ group × time.

Fig. 1

Cardiovascular (upper panel) and renal function (lower panel) before and 3 h after endotoxin administration. Intervention group with continuous infusion of indomethacin 0.2 mg/kg per h. The top, bottom, and line through the middle of the box correspond to the 75th percentile, 25th percentile, and 50th percentile, respectively. The whiskers on the bottom extend from the 10th percentile to top 90th percentile. Outliers marked by circle. Significance shown as *P < 0.05 and **P < 0.01. HR heart rate, MAP mean arterial pressure, SVR systemic vascular resistance, RVR renal vascular resistance, RBF renal blood flow, GFR glomerular filtration rate, CPAH paraaminohippuric acid clearance, gKW gram kidney weight

Fig. 2

Mixed model ANOVA analysis showing estimated marginal means over time for HR heart rate and SVR systemic vascular resistance

Renal perfusion and function

In the control group, endotoxin exposure increased the mean renal vascular resistance but the changes were not significant. In contrast, renal blood flow was reduced by 47.28% at 3 h after endotoxin injection (P < 0.01). The glomerular filtration rate was initially unchanged, but fell along with the renal blood flow at 3 h (P < 0.05). Similarly, urine volume and para-aminohippuric acid clearance (C _PAH) decreased at 2 h (P < 0.01), and 3 h of endotoxin exposure (P < 0.05, respectively). In the intervention group, renal blood flow was reduced by 18.7% at 3 h into the experiment (P < 0.05). Mean renal vascular resistance increased by 50% from 2.24 ± 1.11 to 3.39 ± 2.3, but the change was not significant. In addition, urine volume, GFR, and CPAH were all significantly decreased at 3 h into the experiment (P < 0.01, Fig. 1). Notably, serum potassium levels were low throughout the experiment in the indomethacin exposure group (P < 0.01).

Arterial blood gases

In the control group, mean pH dropped significantly at 1 and 3 h (P < 0.01). When compared to control pCO₂ and pO₂ levels were significantly increased and decreased, respectively (P < 0.01). In the intervention group, pH dropped at 1 and 2 h (P < 0.01), and pCO₂ was found increased at 1 h (P < 0.05). However, neither of these changes was persistent and the levels were recovered to baseline following adjustment of ventilator settings, HCO₃ ⁻ infusions.

Hematological parameters

In the control group, WBC dropped rapidly within the first hour of observation from a mean of 4,800 down to 1,700/μl (P < 0.01). The WBC drop improved yet stayed significantly lower compared to control (P < 0.01). In contrast, the platelet count was more stable, with significant decrease in number at 1 and 3 h (P < 0.05, Fig. 2). In the intervention group, the pattern of WBC and platelet drop was very similar, with significantly more reduction in platelet count compared to the same time post-endotoxin in the control group at 2 and 3 h (P < 0.01, Fig. 3).

Fig. 3

White blood cell (WBC) and platelet counts (right panel) before and 3 h after endotoxin administration. Control group (a) endotoxin only, intervention group (b) with endotoxin and additional continuous infusion of indomethacin 0.2 mg/kg per h. Pattern of WBC and platelet drop was similar between both groups

Discussion

Intervention against the events following endotoxinemia has been attempted on numerous levels. Neutralization of circulating endotoxin via antibodies is not feasible due to the short time needed to initiate an irreversible chain of events leading to inflammatory cytokine release [9]. TNF-α and IL-1β antibodies did not translate into clinical applications, mainly due to being given to late to stop the changes initiated and the risk of impaired host defense [5, 10]. Therefore, the focus of interest in sepsis treatment has been shifted towards more selective treatment. Peevy et al. investigated prostaglandin synthesis inhibition via indomethacin in a rabbit model of group B streptococcal sepsis, and found that indomethacin administration could alleviate cardiovascular effects of gram positive sepsis, however, did not alter neutropenia and thrombocytopenia [11]. Our data suggest that the therapeutic potential lies mainly within stabilization of hemodynamic function, which is demonstrated by the mixed model ANOVA analysis on HR, and suggested by observed trends failing significance on group comparisons that are averaged over time. Azab et al. showed indomethacin not to prevent endotoxin-related decrease of left ventricular function as documented by echocardiography [7]. Aside from multiple differences in their model, our study did not assess left ventricular function directly but rather hemodynamic parameters affected by compensatory mechanisms. Cyxlooxygenase derivates can, based on their local predominance, promote either vasoconstriction or vasodilation. Stoclet et al. found that vasodilatory cyclooxygenase products mediated a nitric oxide synthase induction leading to an increased vasoconstriction threshold to norepinephrine [12]. In contrast, cycloxygenase-derived vasoconstrictors such as thromboxane A2 can inhibit inducible nitric oxide synthase activity [13]. Thus, blocking cyclooxygenase can help attenuating extreme fluctuations of blood pressure in either direction. However, side effects are substantial. Aside from renal, gastrointestinal, hematological and cerebral damage, indomethacin treatment induces lymphocyte apoptosis, which can lead to impaired host defense critically important in sepsis [14]. Our own data underline pronounced impairment of renal function, which is particularly concerning in consideration of the vital importance of this organ system during septic shock [3]. Endotoxin initially attenuates renal medullary vasodilation during hypotension, exerts transient cortical vasodilation, and following repeated exposure induces profound renal vasoconstriction [2]. The prostaglandins E2 and I2 cause afferent arteriole vasodilatation, increasing the glomerular filtration rate, sodium chloride excretion, water excretion, and stimulating renin secretion, whereas thromboxanes lead to constriction [15, 16]. Maintenance of adequate renal function is dependent on a balance of substances causing mesangial relaxation and contraction. Our results indicate that this balance is shifted towards vasoconstriction, which is reflected by increased RVR and decreased RBF. Wang et al. documented indomethacin to lower GFR significantly in a low dose endotoxinemia murine model, where the endotoxinemia alone has not even altered GFR significantly, which underlines its potential of renal toxicity [17]. Considering the challenges endotoxinemia already poses on the renal function and other systems by itself, it seems unlikely to have a sustained role in anti-inflammatory early sepsis intervention.