Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Nitric oxide (NO) has been proposed as the predominant mediator of sepsis-induced shock (Radomski et al., 1990; Evans et al., 1993). In a healthy state and in early stages of sepsis, NO produced by a constitutive, calcium-dependent NO synthase in endothelial cells (eNOS) regulates vascular tone (Vallance et al., 1989). In a later stage of sepsis, inflammatory mediators induce a calcium-independent isoform of NO synthase (iNOS) that has been closely associated with the hypotension and the catecholamine hyporesponsiveness of septic shock in animals (Radomski et al., 1990; Szabo et al., 1993) and humans (Evans et al., 1993). This hypothesis, linking NO and the pathogenesis of septic shock, has led to an attempt to develop NO synthase inhibitors for treating this syndrome (Petros et al., 1994).

Commonly used cyclooxygenase inhibitors, such as ibuprofen, may represent an alternative approach for controlling the NO pathway in sepsis, as well as in other conditions where NO may play a pathogenic role, such as inflammatory arthritides (Sakurai et al., 1995), neurodegenerative diseases (Smith et al., 1997), and atherosclerosis (Wever et al., 1998). Cyclooxygenase inhibition has been shown in vitro to down-regulate iNOS expression and NO production (Aeberhard et al., 1995; Amin et al., 1995; Farivar et al., 1996; Kepka-Lenhart et al., 1996). However, others have reported iNOS up-regulation (Tetsuka et al., 1994; Xu et al., 1995), and one study with human endothelial cells found eNOS inhibition, but iNOS activation, at clinically achievable concentrations of ibuprofen (Menzel and Kolarz, 1997). Furthermore, ibuprofen has failed to increase blood pressure in septic patients (Bernard et al., 1997) or in volunteers challenged with endotoxin (Martich et al., 1992). These divergent results suggest two possibilities. Either ibuprofen does not decrease NO production in vivo, or it decreases NO production, but other vasoregulatory pathways compensate for this loss and blood pressure is unaffected.

The purpose of this study was to determine whether cyclooxygenase inhibition alters NO production in humans. Blood pressure, NO pathway activity, and prostaglandin synthesis were measured in the presence or absence of ibuprofen in normal and in endotoxin-challenged volunteers. This model produces a hyperdynamic, vasodilated state that develops over 2 to 3 h and persists for at least 8 h (Suffredini et al., 1989). The accompanying inflammatory response is well characterized, is self-limited (Martich et al., 1991), and spares the pulmonary airways (Boujoukos et al., 1993).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subjects. This investigation was approved by the Institutional Review Board on Human Experimentation of the National Institute of Allergy and Infectious Diseases. Written, informed consent was obtained from all of the participants. Twenty-nine healthy subjects (22 men, 7 women), ranging in age from 20 to 37 years old (mean, 28 years), were studied. For some of these subjects, data on fever, heart rate variability, and circulating levels of leptin and macrophage inflammatory protein-1 have been published elsewhere (Godin et al., 1996; Bornstein et al., 1998; O'Grady et al., 1999). After an overnight fast, subjects had i.v. and radial artery catheters placed (Arrow International, Reading, PA), and their electrocardiogram and blood pressure were monitored continuously. Heart rate and mean arterial pressure (MAP) were recorded every hour starting before the administration of sodium nitroprusside (SNP) or endotoxin (baseline, time = 0).

Exhaled NO, Exhaled Carbon Dioxide (CO₂), and Airway Flow Measurements. Subjects wore nose clips and breathed NO-free air (Roberts Oxygen, Rockville, MD) through a mouthpiece. A NO standard curve with a sensitivity of 1 part per billion was generated each day of the study. Single-breath vital capacity maneuvers were analyzed for NO (270B Chemiluminescence Analyzer; Sievers Instruments, Inc., Boulder, CO), CO₂ (Capnogard ETCO₂ Monitor 1265; Novametrix Medical Systems, Inc., Wallingford, CT), and airway flow (CP-100 Pulmonary Monitor; Bicore Monitoring Systems, Inc., Irvine, CA). CO₂ measurements defined the quality of breath-holding and the portion of exhaled air that represented alveolar gases. Labview 3.1 software (National Instruments, Austin, TX) was used to determine peak and alveolar NO concentration and exhaled NO flow rates. Each data point represents the average of five single-breath measurements.

Plasma and Urine Assays. Blood samples were collected in 5-ml vacuum tubes with or without 7.4 mM EDTA (Becton Dickinson Vacutainer Systems, Rutherford, NJ). The phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (0.1 mM; Sigma Chemical Co., St. Louis, MO) was added to the tubes for measurement of cyclic GMP (cGMP). Tubes for S-nitrosothiol determinations were wrapped with aluminum foil. Blood was centrifuged for 10 min at 1300g at 4°C. Supernatants were frozen and stored at 70°C until assayed. A 12-h urine collection into 2-propanol (J.T. Baker, Phillipburg, NJ) was obtained before, during, and after the study from each subject. Urine volumes were recorded and samples were frozen at 70°C until assayed.

1

SNP Administration (Methods Validation Study). To evaluate the methods used to measure NO, patients were given an infusion of SNP (Elkins-Sinn Inc., Cherry Hill, NJ), a NO donor, diluted in 5% dextrose in water (final concentration, 200 µg/ml). Before the initiation of SNP or placebo (vehicle alone) infusion, all of the subjects received 30 ml/kg of 0.9% saline i.v. over 30 min. A maintenance infusion of 0.9% saline at a rate of 1 ml/kg/h was then administered for the remainder of the study.

Endotoxin and Ibuprofen Administration. Twelve subjects were given endotoxin (4 ng/kg i.v.; U.S. Standard Reference Endotoxin, Lot EC-5, Escherichia coli 0:113; Bureau of Biologics, U.S. Food and Drug Administration) or 0.9% saline (vehicle) separated by 1 week in a single-blinded, randomized order. Endotoxin in 2 ml of 0.9% saline (vehicle) or vehicle alone was administered over 1 min followed by a 10-ml flush of 0.9% saline. Subjects were randomized further to receive either 800 mg of ibuprofen (Upjohn, Kalamazoo, MI; n = 6) or a lactose placebo (n = 6) p.o. at 1.5 h before, at the time of, and at 3 h after the administration of endotoxin or 0.9% saline.

Statistical Analysis. Data from the SNP infusion study were analyzed by first calculating a summary statistic for each individual (e.g., maximum change). Summary statistics were analyzed for group differences with a nonparametric Kruskal-Wallis test.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

SNP Effects (Methods Validation Study). SNP increased heart rate in the high-dose group (Fig. 1A). In both the low- and high-dose groups, MAP was decreased (Fig. 1B) and the exhaled total NO flow rates were increased (Fig. 1C) compared with the placebo group (see Fig. 1 for P values). Other measures of exhaled NO, including the peak NO concentration and the alveolar NO flow rate, were also increased in the SNP groups (P < .05 for all; data not shown). Urinary NOx excretion increased only in the high-dose group during the poststudy urine collection (Fig. 1D). In contrast, compared with the placebo group, low- or high-dose SNP had no effect on the urinary excretion of cGMP (P  .2; data not shown) or the plasma concentrations of NOx and cGMP (P  .16 for both; data not shown). Changes from baseline in S-nitrosothiol, considered a more specific measure of endogenous NO production, were decreased by SNP in both the high- and low-dose groups compared with the placebo group (0.3 ± 0.2 and 0.6 ± 0.8 µM, respectively, versus 2.6 ± 1.3 µM; P  .05). Therefore, the exhaled NO flow rates and the urinary excretion of NOx were the most sensitive indicators of systemic NO release.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Changes from baseline in heart rate (A), MAP (B), exhaled total NO flow rate (C), and poststudy urinary NOx excretion for subjects in the placebo, SNP low, and SNP high groups (D) .

Ibuprofen Effects Independent of Endotoxin. Ibuprofen concentrations were similar at 0, 3, and 6 h, with an overall mean of 46.6 ± 4.3 µg/ml (P = .001). Ibuprofen had no significant effects on heart rate (P  .5; Fig. 2A). In contrast, ibuprofen blunted the small fall in MAP associated with bed rest, both in the early period (<3 h) and the late period (3 h) after administration of either saline (normal) or endotoxin (P = .005; Fig. 2B). Similarly, ibuprofen decreased temperature (P = .004; Fig. 3A). Furthermore, measures of NO were decreased by ibuprofen in normal and endotoxemic subjects, both in the early and the late period. These measures included the alveolar NO concentration (P = .05; Table 1), the alveolar NO flow rate (P = .03; Table 1), the total NO flow rate (P = .06; Fig. 3B), and the urinary NOx (P = .02; Fig. 3C). The plasma NOx concentrations were decreased by ibuprofen but only in the late period (P = .04; Table 2).

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Changes from baseline in heart rate (A) and MAP (B) for subjects in Saline/Placebo (), Saline/Ibuprofen (), Endotoxin/Placebo (), and Endotoxin/Ibuprofen () groups during the early (<3 h) and the late (3-8 h) periods after administration.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Changes from baseline in temperature (A), total NO flow rates (B), and urinary excretion of NOx (C) for subjects in Saline/Placebo (), Saline/Ibuprofen (), Endotoxin/Placebo (), and Endotoxin/Ibuprofen () groups during the early (<3 h) and the late (3-8 h) periods after administration.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Correlation coefficients of MAP changes versus exhaled total NO flow rate changes for each individual subject (A) and as group means (B). Data shown are without () or with () ibuprofen after either saline or endotoxin challenge. In the absence of ibuprofen, the mean correlation coefficient (r) for saline was 0.37 ± 0.16 and for endotoxin was 0.34 ± 0.15 (combined r = 0.36 ± 0.10, P = .009). In the presence of ibuprofen, the mean correlation coefficient (r) for saline was 0.14 ± 0.20 and for endotoxin was 0.40 ± 0.14 (combined r =.27 ± 0.12, P = not significant).

Endotoxin Effects. Endotoxin increased heart rate (Fig. 2A) both in the early period (P = .04) and the late period (P = .003) after its administration, but it decreased MAP (Fig. 2B) only in the late period, 3 to 8 h after its administration (P = .002). These cardiovascular effects of endotoxin were unaltered by ibuprofen. Endotoxin also significantly increased temperature both in the early and the late periods after its administration (P = .002 for both; Fig. 3A), and increased the total NO flow rates (P = .04; Fig. 3B) and the alveolar NO flow rates in the late period (P = .04; Table 2), but only in subjects not treated with ibuprofen. Urinary NOx excretion during the poststudy collection period showed a similar pattern (P = .07; Fig. 3C). Endotoxin did not alter the urinary cGMP excretion or the plasma NOx, S-nitrosothiol, and cGMP concentrations (P > .5 for all). Endotoxin-induced increases in urinary levels of 2,3-dinor 6-keto-prostaglandin F₁, a metabolite of prostacyclin, and levels of 11-dehydro-thromboxane B₂, a metabolite of thromboxane A₂, were blocked by the administration of ibuprofen (Table 2; P  .001; see "Statistical Methods" in Materials and Methods).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study demonstrates that ibuprofen, a cyclooxygenase inhibitor, reduces NO production and dissociates it from MAP in humans. This observation confirms in vivo a potentially important mechanism of action for ibuprofen and possibly other nonsteroidal anti-inflammatory drugs. Although ibuprofen blocked increases in NO production and blunted bed rest-associated reductions in blood pressure, this agent failed to alter the hemodynamic response to endotoxin, suggesting that vasoregulatory mechanisms other than NO determine vasomotor tone under these conditions.

The in vivo effects of ibuprofen on the NO pathway previously have not been reported in humans and its effects in vitro have been variable. Some in vitro studies have found that cyclooxygenase inhibitors down-regulate NOS expression and function (Aeberhard et al., 1995; Amin et al., 1995; Farivar et al., 1996; Kepka-Lenhart et al., 1996), but others have demonstrated up-regulation (Tetsuka et al., 1994; Xu et al., 1995). A study with human endothelial cells concluded that the administration of high doses of ibuprofen to humans would likely increase net NO generation (Menzel and Kolarz, 1997). Ibuprofen does have some in vivo effects that might tend to increase rather than decrease NO responses. For example, in human endotoxin-challenged volunteers, ibuprofen increases the release of proinflammatory cytokines (Martich et al., 1991; O'Grady et al., 1999), an effect which might increase iNOS induction. Although our study demonstrates that ibuprofen down-regulates the NO pathway in vivo, it does not address whether this represents the net effect of several competing mechanisms.

Notably, ibuprofen prevented endotoxin from augmenting NO production, but the fall in blood pressure was unaffected. However, ibuprofen did have vascular effects, as it decreased the fall in blood pressure that occurred with the institution of bed rest. The inability of ibuprofen to alter inflammation-associated abnormalities of vascular tone has been reported before in septic patients (Bernard et al., 1997) and in volunteers challenged with endotoxin (Martich et al., 1992). In contrast, cyclooxygenase inhibitors have been shown to raise blood pressure in septic pigs (Weitzberg et al., 1995), dogs (Jacobs et al., 1982), and baboons (Fletcher et al., 1976). NO responses vary across species and do not always correlate well with changes in vascular tone (Evans et al., 1993; Jacob et al., 1993; van den Berg et al., 1994; Cobb et al., 1995). Therefore, the administration of low doses of endotoxin to humans may provide insights into blood pressure regulation that are not identical with those obtained from animal models (Suffredini et al., 1989; Martich et al., 1992).

In our subjects, mechanisms other than NO probably contribute to endotoxin-induced vasodilatory responses. In addition to preventing increases in NO production, ibuprofen inhibited synthesis of prostacyclin and thromboxane A₂. Blockade of both vasodilatory and vasoconstrictive prostaglandins could cause either increases or decreases in blood pressure, depending on their net effect on vascular tone. Furthermore, in endotoxin-challenged volunteers, ibuprofen decreases the release of vasoconstrictive catecholamines (Revhaug et al., 1988), an effect that favors vasodilation. Alternatively, a fall in blood pressure in the absence of increased NO production could be caused by increased NO effectiveness. Nonsteroidal anti-inflammatory drugs block production of superoxide by cyclooxygenases (Wolin, 1996). Because superoxide inactivates NO (Gaboury et al., 1993), a decrease in superoxide could increase NO bioavailability. Regardless of the mechanism, our results demonstrate that in the presence of ibuprofen, endotoxin-induced vasodilation occurs without a measurable increase in NO production, possibly because of reciprocal changes in other vasoregulatory pathways.

Ibuprofen was found to reduce NO production in both normal and endotoxin-challenged volunteers, although this effect was more pronounced after endotoxin. This suggests that ibuprofen may also interfere with homeostatic NO production, which primarily arises from constitutive isoforms of NOS. An in vitro study with human endothelial cells (Menzel and Kolarz, 1997) and an in vivo study conducted in swine (Dahm et al., 1997) both support the possibility that cyclooxygenase inhibitors decrease the activity of eNOS, a constitutive isoform. In our study, ibuprofen may have suppressed NO production in normal volunteers by blunting eNOS responses, by decreasing low basal levels of iNOS expression, or both.

Among methodologies used as markers for NO production, we found that the exhaled NO flow rates and the urinary NOx excretion were the most sensitive. Exhaled NO has a number of sources, but increases caused by SNP and endotoxin in this study likely arose from NO release within the pulmonary vasculature (Husain et al., 1994; Persson et al., 1994; Stitt et al., 1997). Although airway inflammation increases exhaled NO (Kharitonov et al., 1994), the administration of low doses of endotoxin to humans produces a self-limited inflammatory response that spares the airways (Boujoukos et al., 1993). Importantly, increases in exhaled NO paralleled increases in urinary NOx excretion, a measure that reflects global changes in NO production. Other tests used in this study, plasma NOx, S-nitrosothiol, and cGMP or urinary cGMP, were relatively insensitive to changes in NO production. Notably, plasma S-nitrosothiol, a more specific measure of endogenous NO production than NOx (Stamler et al., 1992), decreased in response to SNP.

In summary, the administration of ibuprofen to humans blocked NO production but did not prevent an endotoxin-induced fall in blood pressure. These results suggest that mediators other than NO may contribute to alterations in vascular tone during endotoxemia, particularly in the presence of cyclooxygenase inhibition. In support of this concept, iNOS-deficient "knockout" mice have been shown to have NO-independent mechanisms of endotoxin-induced hypotension and death (Macmicking et al., 1995). Thus, the vasodilatory response to endotoxin is complex and may be regulated by a network of overlapping mediators. The ability of ibuprofen to decrease NO production is a mechanism of action that has possible experimental and therapeutic applications.