INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The use of proinflammatory cytokines (e.g., tumor necrosis factor (TNF) and interleukin-2 [IL-2]) to produce a tumoricidal effect in cancer patients (2, 26) and to boost CD4⁺ cell counts in patients with acquired immune deficiency sundrome (AIDS) (13) is hindered by dose-dependent cardiovascular toxicities (10, 21). Infusions of these proinflammatory cytokines simulate the cardiovascular abnormalities seen in bacterial shock (8, 10, 17, 18, 21). Sepsis is thought to stimulate the release of endogenous cytokines, which induce production of nitric oxide (NO), a potent vasodilator. This release of NO is then immediately responsible for cardiac and peripheral vascular dysfunction, and hypotension (4, 12, 19). This paradigm has led to the investigational use of N-monomethyl-L-arginine (L-NMMA), an inhibitor of nitric oxide synthase (NOS), an enzyme that produces NO, in cancer patients receiving cytokine therapy (11) and in patients with septic shock (24). L-NMMA has been shown to increase blood pressure in these patients (11, 24), but the effects of L-NMMA in vivo on cytokine-induced or sepsis-induced myocardial depression have not been fully investigated.

There are in vitro data supporting a role for NO in cytokine-induced cardiac injury. TNF decreases the contractility of isolated murine myocytes (1, 14, 31) and isolated hamster papillary muscles (9). Inflammatory cytokines or medium from stimulated macrophages, added to isolated murine myocytes, induces transcription and expression of NOS, (20, 23, 31) and increases the formation of nitrite and nitrate, which are stable end products of NO metabolism (1, 31). Agents that suppress NOS transcription or activity prevent these cytokine-induced increases in nitrites and nitrates (25, 30). Importantly, NOS inhibitors block the negative inotropic effects of TNF (9), endotoxin (3), activated macrophage medium (1), and serum from septic humans (14) on hamster papillary muscles (9), guinea pig myocytes (3), and adult (1) and newborn rat myocytes (14), respectively. Thus, a number of in vitro studies suggest that NO released in response to proinflammatory cytokines may cause myocardial depression.

In dogs, bacterial challenges produce a pattern of cardiovascular injury similar to that in human septic shock (17), activate inducible forms of NOS, and increase nitrate, nitrite, and TNF levels (4, 35). TNF alone in dogs produces reversible decreases in left-ventricular (LV) contractility and hypotension similar to those in human septic shock (4, 8, 12, 16, 18, 22, 32). We examined whether cytokine-induced increases in NO production could explain cardiovascular dysfunction associated with TNF.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design and Experimental Subjects

In order to investigate effects of TNF in the setting of blocked NO synthesis, 36 2-yr-old, 10 to 12-kg, purpose-bred beagles were randomly assigned to receive TNF, L-NMMA, both, or neither (0.9% sodium chloride as a control). One half of the animals were studied with a crossover experimental design, and were given either TNF (n = 10) or L-NMMA (n = 8) on two occasions (Figure 1). The other 18 animals were not part of the crossover design, and were given TNF or L-NMMA only once because of the increased stress produced by the higher doses of TNF used.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Study design for 36 animals.

Seven days before TNF or L-NMMA challenges (baseline), we obtained blood for laboratory studies and performed hemodynamic measurements as previously described (Figure 2) (5, 8, 17, 18). On Day 0, animals received an intravenous bolus of either L-NMMA or normal saline (time = 0), followed by a 6-h infusion of either L-NMMA or normal saline. At 10 min after the bolus, animals were given an intravenous bolus of TNF or normal saline over a 20-min period.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Time course of treatments, hemodynamic studies, and blood work.

Each animal received a fluid bolus at 168 h (7 d), 6 h, and 24 h, and on Day 0, from 1 to 6 h, received an infusion of Ringer's solution (Figure 2). At 27 h, a subset of animals received an intravenous infusion of L-arginine over a period of 1 h. Hemodynamic evaluations were done at the times shown in Figure 2. For 7 d before baseline studies and during the experiment, animals were fed a diet with a nitrite and nitrate concentration of approximately 1 ppm (Harlan Teklad, Madison, WI).

Preparation and Administration of Experimental Agents

L-NMMA. Before use, L-NMMA monoacetate (Calbiochem-Novabiochem Corporation, San Diego, CA) was reconstituted with sterile 0.9% sodium chloride to a concentration of 40 mg · ml¹. Animals assigned to receive L-NMMA were given an intravenous bolus of 40 mg · kg¹ over a period of 10 min, followed by a 6-h infusion (Infu-Med 300 pump; Medfusion, Duluth, GA) of 40 mg · kg¹ · h¹.

TNF. Recombinant human TNF (lot DOE 270   /  91) was provided by Bayer Corporation (Berkeley, CA). TNF was expressed in a yeast system with a purity of > 95.3% as demonstrated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The protein concentration, as determined by the Bradford protein assay, was 6.6 mg · ml¹. Individual doses of TNF (20, 45, or 60 µg · kg¹) were diluted in 100 ml of 0.9% sodium chloride in sterile bags coated with 5 ml of canine serum. At 10 min after the L-NMMA bolus, animals received TNF intravenously at 20 µg · kg¹ or 45 µg · kg¹ over a period of 20 min.

L-arginine. Ten-percent arginine hydrochloride, USP, was obtained from Kabi Pharmacia, Inc. (Clayton, NC) in a concentration of 0.1 g · ml¹. At 27 h, 1.5 g · kg¹ of L-arginine was infused intravenously over a period of 1 h.

Hemodynamic Measurements

All hemodynamic measurements were made on awake animals, using previously described techniques (5, 8, 17, 18). At 168 h (7 d), 2 h, and 24 h, femoral and pulmonary artery catheters were inserted under local anesthesia (1% lidocaine). Heart rate (HR, min¹), mean arterial pressure (MAP, mm Hg), central venous pressure (CVP, mm Hg), mean pulmonary artery pressure (MPAP, mm Hg), pulmonary capillary wedge pressure (PCWP, mm Hg), and cardiac output (ml · min¹) were measured. In order to determine left-ventricular ejection fraction (LVEF), we performed radionuclide gated blood-pool scans, using described techniques (8, 17, 18, 21). At the end of each study day, intravascular catheters were removed. Hemodynamic data were indexed to body weight in kilograms. Cardiac index (CI, ml · min¹ · kg¹), stroke volume index (SVI, ml · kg¹), left-ventricular stroke-work index (LVSWI, g · ml¹ · kg¹), systemic vascular resistance index (SVRI, dyn · s · cm⁵ · kg¹), oxygen delivery index (DO₂I, ml · min¹ · kg¹), end-diastolic-volume index (EDVI, ml · kg¹), end-systolic-volume index (ESVI, ml · kg¹), and alveolar-arterial oxygen gradient ([A-a]O₂, mm Hg) were calculated with standard formulas (5, 8, 17, 18).

Laboratory Measurements

Routine chemistries and complete blood counts were obtained at times shown in Figure 2, as previously described (5, 8, 17, 18). Arterial and mixed venous blood-gas analysis (pH, Pa_O2 [mm Hg], Pa_CO2 [mm Hg], and arterial base excess [mmol · L¹]) and lactate (mmol · L¹), were measured with each hemodynamic evaluation. Serum nitrite/nitrate levels (µmol · L¹) were measured by converting serum nitrate to nitrite and using a colorimetric assay based on the Griess reaction (33). Serum nitrosylated protein (µmol · L¹) was determined with the Saville nitrosothiol assay as previously described (27).

Animal Care

This experimental protocol used in the study was approved by the Animal Care and Use Committee of the Clinical Center of the National Institutes of Health. Throughout the studies, all efforts were taken to minimize animal pain and suffering, and to limit the number of animals needed for the study.

Statistical Methods

Analysis of variance (ANOVA) was performed to determine the effects of L-NMMA and TNF, and to determine whether TNF effects were altered by L-NMMA. For all analyses, a group-time interaction was included in the model along with main effects (33). All interactions, including dog, were pooled to form the error term. To identify specific differences in the time course of the groups, contrasts were constructed using group-time interactions. To control for multiple comparisons, nonindependent contrasts were corrected with Bonferroni's procedure (33). The effect of TNF dose was examined, and in all instances the doses of TNF were found to have produced nonsignificant interactions and were pooled. In addition, for the crossover studies, a prior study did not significantly affect subsequent studies. The analyses of Frank-Starling ventricular function plots, as well as of a class of cardiac-function parameters, were done with a multivariate analysis of variance (MANOVA) procedure, which incorporates information associated with the correlation among the measurements. One-SE elliptical confidence regions were drawn for the Frank-Starling curves, using a previously described procedure (17).

Analysis of serum nitrite/nitrate was confounded, since the food supplied to some dogs was inadvertently changed from a low-nitrate/ nitrite formulation to a standard formulation for a number of days prior to the study. This event increased nitrate/nitrite values for some dogs at baseline, irrespective of study group. However, by 6 h no diet effects were observed (i.e., mean nitrate/nitrite levels were similar for both diets in all treatment groups). This observation was verified with two independent measures of assessing NO production: nitrite/nitrate levels and nitrosylated protein levels, which were highly significantly correlated. Thus, for analyses of nitrate/nitrite levels changes from baseline to 6 h represent dogs whose diet was not changed, and changes from 6 h to 24 h represent all dogs.

Survival curves were tested for statistical significance with a Kruskal-Wallis procedure. Statistical significance is declared for values of p < 0.05, although values of p in the range from 0.05 to 0.10 are reported to reflect potential type II errors.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Survival

Baseline studies for all treatment groups showed similar mean values for all hemodynamic and laboratory parameters measured (p = NS, data not shown). There was no significant difference in mortality rates when the TNF-challenged and TNF-plus-L-NMMA-challenged study groups were compared. One of 10 (10%) TNF (20 µg · kg¹)-challenged animals, one of nine (11%) TNF (20 µg · kg¹)-plus-L-NMMA-challenged animals, two of four (50%) TNF (45 µg · kg¹)-challenged, and three of five (60%) TNF (45 µg · kg¹)-plus-L-NMMA-challenged animals died. The one animal given TNF at 60 µg · kg¹ died. Comparison of all animals that received high-dose TNF (45 and 60 µg/kg, n = 10) with those that received low-dose TNF (20 µg/kg, n = 18), irrespective of whether L-NMMA was given or not, showed a significant increase in lethality rates (p = 0.02).

Effects of L-NMMA on Intravascular Pressures, Cardiac, Metabolic, and Pulmonary Parameters (1 h to 27 h)

Persistent effects of L-NMMA (variables with L-NMMA effects from after bolus to 27 h). (Figure 3I): Both immediately after the L-NMMA bolus at 0 h and during the 6-h infusion, there were increases in mean MAP (p = 0.0001), SVRI (p = 0.0001), PCWP (p = 0.001), and CVP (p = 0.005), and decreases in mean CI (0.0001), DO₂I (p = 0.0001) and HR (p = 0.0004), as compared with controls (Figure 3I). From 6 h to 24 h the differences in mean MAP (p = 0.05), SVRI (p = 0.0001), PCWP (p = 0.03), CI (p = 0.009), and DO₂I (p = 0.008) became smaller. However, at 24 h and 27 h, mean MAP (p = 0.06), SVRI (p = 0.004), PCWP (p = 0.04), and CVP (p = 0.01) were still increased, and mean CI (p = 0.0002), DO₂I (p = 0.002), and HR (p = 0.0001) were still decreased. In the L-NMMA-treated animals at 27 h, L-arginine decreased mean MAP (p = 0.005), PCWP (p = 0.02), CVP (p = 0.0001), and SVRI (p = 0.003), and increased mean HR (p = 0.001), CI (p = 0.001), and DO₂I (p = 0.003), as compared with controls. Furthermore, after L-arginine infusion, mean values for all these variables returned to control values (all: p = NS).

View larger version (23K): [in this window] [in a new window]   Figure 3.   Serial mean changes (± SEM) produced by L-NMMA (closed circles connected by solid lines) versus controls (open circles connected by dashed lines). Open rectangles, at the bottom left of each graph, indicate time of L-NMMA bolus and infusion. Thick open arrows, at the upper right of each graph, indicate time of L-arginine bolus. Thin open arrows, originating from the last open and closed circles in each graph, illustrate mean changes from before (arrow origin) to after (arrow tip) L-arginine. Variables denoted as persistent (I, A to G) include L-NMMA effects that occurred immediately after the bolus and persisted but were fully reversed to control values by L-arginine at 27 h. Transient variables (II, A and B) indicate those showing L-NMMA effects at 6 h that were fully reversed to control values with fluid administration (see Tables 2 and 3), and at 27 h were similar to control values. Late variables (III, A and B) included L-NMMA effects that became apparent after 6 h, rather than acutely, and did not reverse with L-arginine. See RESULTS for statistical values.

Transient effects of L-NMMA (variables with L-NMMA effects at 6 h that at 27 h were similar to control values). (Figure 3II): The L-NMMA bolus did not, but the 6-h infusion of L-NMMA did, produce decreases in mean SVI (p = 0.01) compared with controls. Immediately after the L-NMMA bolus at 0 h, and during the 6-h infusion, there was a decrease in mean LVEF (p = 0.0001) as compared with controls. From 6 h to 24 h, mean SVI (p = 0.0005) and LVEF (p = 0.09) increased in the L-NMMA-treated animals to the point that, at 24 h and 27 h, mean SVI and LVEF were no longer different than in controls (both: p = NS). In L-NMMA-treated animals, L-arginine had no significant effects on mean SVI (p = NS), but it increased mean LVEF (p = 0.05). In addition, after L-arginine, mean LVEF and SVI in L-NMMA-treated animals were similar to control values (p = NS).

Late effects of L-NMMA (variables with L-NMMA effects that occurred from 6 h to 27 h, not acutely). (Figure 3III): The L-NMMA bolus at 0 h had no significant effects on mean arterial pH or (A-a)O₂ compared with controls (both: p = NS) (Figure 3III). However, in L-NMMA-treated animals, there was a decrease in mean arterial pH (p = 0.02) at 6 h, and at 24 h and 27 h, a decrease in mean arterial pH (p = 0.0001) and an increase in (A-a)O₂ (p = 0.09) compared with controls (Figure 3III). In L-NMMA-treated animals at 27 h, L-arginine had no significant effect on pH or (A-a)O₂ (both: p = NS). After L-arginine infusion, mean arterial pH (p = 0.0001) was still decreased and (A-a)O₂ (p = 0.002) was still increased as compared with controls.

Effects of L-NMMA on Blood Chemistries

In L-NMMA-treated animals at 6 h, there were significant decreases in mean total protein (p = 0.04), albumin (p = 0.05), or cholesterol (p = 0.05), from control values. At 24 h, there were decreases in L-NMMA-treated animals in mean triglycerides (p = 0.05), glucose (p = 0.01), blood urea nitrogen (BUN)/creatinine ratio (p = 0.02), and BUN (p = 0.04), and increases in alanine aminotransferase and aspartate aminotransferase (both: p = 0.059) as compared with controls. Further, at both 6 h and 24 h, there were decreases in mean globulin (p = 0.06) and serum calcium (p = 0.03) as compared with controls. For all other serial blood chemistry parameters measured, including hemoglobin and hematocrit, mean values in L-NMMA-treated animals were similar to those in controls (p = NS) (data not shown).

Effects of Fluids at 6 h and 24 h on L-NMMA

In response to fluid challenge, L-NMMA-treated animals at 6 h had greater increases in mean PCWP, SVI, and LVEF, and a greater decrease in mean SVRI than did controls (Table 1). In L-NMMA-challenged animals at 6 h after fluid administration, absolute values for mean MAP, SVRI, PCWP, and CVP were still increased, and absolute values for mean CI, DO₂I, HR, and pH were still decreased as compared with controls (all p < 0.005; Table 2). However, in L-NMMA challenged animals at 6 h after fluid administration, SVI and LVEF effects disappeared (both: p = NS; Table 2). In response to fluid challenge at 24 h, L-NMMA-treated animals had significantly smaller increases in mean CI, and at 6 h and 24 h had significantly smaller increases in mean HR than did controls.

TNF Effects Altered by L-NMMA

Effects of TNF altered by L-NMMA at 6 h (CI, SVI, DO₂I, and Frank-Starling Plots of PCWP versus LVSWI). During the first 6 h after infusion, TNF produced decreases in mean CI (p = 0.0004), SVI (p = 0.001), and DO₂I (p = 0.0002) as compared with controls (Figure 3, controls; Figure 4, TNF alone). During that same time period, L-NMMA also produced significant decreases in mean values of these 3 parameters (Figure 3). However, in animals treated with the combination of TNF and L-NMMA, the changes in mean CI (p = 0.036), SVI (p = 0.11), and DO₂I (p = 0.02) were less than the sums of the corresponding changes caused individually by TNF and L-NMMA (Figure 5). For these three measures of cardiac function at 6 h, the overall p value for a TNF interaction with L-NMMA was 0.003 (MANOVA, see METHODS).

View larger version (22K): [in this window] [in a new window]   Figure 4.   Serial mean changes (± SEM) in TNF-plus-L-NMMA-treated animals (closed circles connected by straight lines) versus those given TNF alone (open circles connected by dashed lines). The large open rectangle at the lower left of each graph indicates the time of L-NMMA bolus and infusion, and the small open rectangle indicates the time of TNF infusion.

View larger version (19K): [in this window] [in a new window]   Figure 5.   Graphs illustrating TNF-L-NMMA interactions (i.e., whether TNF effects were different with and without L-NMMA). The following analysis was used: effects of L-NMMA alone were subtracted from the serial changes produced by TNF-plus-L-NMMA (closed circles connected by solid lines), and these differences were then compared with the changes produced by TNF alone, subtracting the effect of saline (open circles connected by dashed lines). If an interaction was present ([TNF-plus-L-NMMA] minus L-NMMA  TNF minus control, p < 0.05) at 6 h (I, graphs A to C) or 24 h (II, graphs A and B), we examined the effects of a fluid bolus (arrows) on this relationship. The arrow tips denote mean values after fluids. The SE bar in the bottom left-hand corner of each graph represents the pooled source of variability used for comparisons.

Reflecting another method of analyzing cardiac function at 6 h, Figure 6 displays plots of LV filling pressures (PCWP) versus performance (LVSWI) before (Figure 6A) and after (Figure 6B) a fluid challenge. TNF produced decreases in mean LVSWI (p = 0.0008), but no change in mean PCWP (p = NS) as compared with controls. During that same time period, L-NMMA produced a decrease in mean LVSWI (p = 0.02) and an increase in mean PCWP (p = 0.0001). However, in animals treated with the combination of TNF and L-NMMA, the decrease in LV performance (PCWP versus LVSWI) was smaller (p = 0.047) than the sum of the individual changes produced by TNF and L-NMMA. Thus, in animals treated with the combination of TNF and L-NMMA, plots of mean LVSWI versus PCWP were shifted upward and to the left of those corresponding to the calculated sum of the changes induced in PCWP versus LVSWI by TNF alone and by L-NMMA alone.

View larger version (26K): [in this window] [in a new window]   Figure 6.   Frank-Starling LV function plots at 6 h of PCWP versus LVSWI before (A) and after (B) fluids, in controls (circle), L-NMMA-treated (triangle), TNF-treated (square), and TNF-plus-L-NMMA-treated (diamond ) animals. The ellipses around each symbol represent one SD. In B, the geometric symbols indicate the same groups and values as before fluids in A; the black arrows indicate the shifts in ventricular performance from before (arrow origin) to after (arrow tip) the fluid challenge. In A and B, the position of the L-NMMA group is where one would expect the TNF-plus-L-NMMA group to be found if TNF effects were completely reversed by NO inhibition. In A, the TNF-plus-L-NMMA group is found between the TNF-alone and L-NMMA-alone groups, suggesting a partial beneficial effect. However, after fluids, the TNF-plus-L-NMMA group is shifted downward and to the right of the TNF group, and below the L-NMMA group. This position is not indicative of a beneficial effect of NO inhibition on cardiac performance (LVSWI) at comparable preloads (PCWP).

For all other cardiac, pressure, metabolic, and pulmonary parameters measured at 6 h, including LVEF (p = 0.30, for interaction), the changes in animals treated with the combination of TNF and L-NMMA were equivalent to the sum of the individual effects of TNF and L-NMMA, (i.e., the effects were additive).

Effects of fluids at 6 h on TNF and L-NMMA interactions. Fluids altered (p = 0.03) the TNF and L-NMMA interactions at 6 h for CI, SVI, and DO₂I (Figure 5). In response to fluids at 6 h, the TNF-treated group had greater increases in mean CI, DO₂I, and SVI than did the TNF-plus-L-NMMA-treated group (Table 3). Consequently, after fluids, these TNF and L-NMMA interactions were no longer statistically significant (CI: p = 0.70; SVI: p = 0.30; DO₂I: p = 0.35), (i.e., the effects of TNF alone and L-NMMA alone, as compared with those of TNF and L-NMMA combined, were additive. In addition, the 6-h fluid challenge significantly altered (p = 0.02) the TNF and L-NMMA interactions demonstrated by ventricular performance plots (Figure 6B). The plot of mean LVSWI versus PCWP for TNF-plus-L-NMMA treated animals after fluids was shifted (p = 0.03) downward and to the left (no longer upward, as seen before fluids), as compared with the plot corresponding to the calculated sum of the changes produced after fluids by TNF alone and L-NMMA alone.

TNF effects altered by L-NMMA at 24 h. From baseline to 24 h, TNF produced decreases in mean MAP (p = 0.0002) and arterial pH (p = 0.0002), as compared with controls (Figure 3, controls; Figure 4, TNF alone). During this same time period, L-NMMA produced increases in mean MAP and decreases in arterial pH, as compared with controls (Figure 3). However, in animals treated with the combination of TNF and L-NMMA, the net changes in these two parameters were significantly different than the sum of the individual effects produced by TNF alone and by L-NMMA alone (Figure 5). For example, after the L-NMMA infusion, TNF was associated with less of a decrease in MAP (p = 0.001). Also, at 24 h, with a prior L-NMMA infusion, TNF was associated with less of a decrease in arterial pH (p = 0.004). For other hemodynamic parameters measured at 24 h, the changes in TNF-plus-L-NMMA-treated animals were not statistically different than the sums of the individual effects of TNF and of L-NMMA.

Effects of fluids at 24 h on TNF and L-NMMA interactions. The 24-h fluid challenge altered TNF interactions with L-NMMA for MAP (p = 0.06) such that after the challenge these interactions were no longer significant (p = NS). However, at 24 h, fluids did not alter (p = NS) TNF interactions with L-NMMA for pH. At 24 h, both before and after fluid administration, there was a significant TNF interaction with L-NMMA for arterial pH (p = 0.04).

Serum Nitrite/Nitrate Levels

From baseline to 6 h, mean serum nitrite/nitrate levels (and nitrosothiol levels; data not shown) similarly and significantly (p < 0.0001) decreased in all four treatment groups. However, from 6 h to 24 h, animals not given L-NMMA (both those given TNF alone and controls) had similar and significant increases (p = 0.007) in mean serum nitrite/nitrate levels over those of animals given L-NMMA (L-NMMA-alone- and TNF-plus-L-NMMA-challenged animals) (Figure 7).

View larger version (18K): [in this window] [in a new window]   Figure 7.   Changes (mean ± SEM) in nitrate/nitrite serum levels from baseline to 6 h (A) and from 6 to 24 h (B) in controls (open bar), TNF- (hatched bar), L-NMMA- (gray bar), and TNF-plus-L-NMMA-treated animals (solid bar).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

At the doses used in this study, TNF alone produced progressive decreases in cardiac performance, which by 24 h returned to near normal. L-NMMA alone increased afterload and decreased cardiac performance. In animals given both TNF and L-NMMA, the decreases in cardiac performance were significantly smaller than the sum of those produced by TNF alone and L-NMMA alone. After fluids were given to correct abnormalities in preload, decreases in cardiac performance in animals given both TNF and L-NMMA became equal to the sum of those produced by TNF and L-NMMA given separately. Among potential explanations for this lack of additivity before giving fluids are that: (1) TNF offsets the afterload effects of L-NMMA, improving cardiac performance only before fluids; (2) L-NMMA, through its vasoconstrictive properties, improves cardiac performance by increasing preload; or (3) L-NMMA inhibits NOS and excess NO production, a proposed mechanism through which TNF decreases contractility and cardiac performance.

TNF-induced decreases in afterload cannot explain the lack of additivity of TNF with L-NMMA. In L-NMMA-treated animals from 0 to 6 h, TNF actually increased SVRI and produced only small decreases in MAP (10 to 15 mm Hg). An alternative explanation is that if the beneficial effects were caused in part by L-NMMA-induced increases in preload, fluid challenges might be expected to interfere with this effect. This is exactly what was observed; fluid challenges at 6 h abolished the beneficial effect of L-NMMA on TNF-induced depression of cardiac performance. This suggests that the beneficial effect of L-NMMA was related to its ability to maintain preload and cardiac performance through a constrictive effect on capacitance vessels. Additionally, this beneficial effect of L-NMMA was unlikely to be related to NO-induced impairment of LV performance, because TNF was not shown to increase end products of NO production in this model, and such an effect should not be abolished by fluid loading.

Studies suggesting a role for NO in TNF-induced cardiovascular dysfunction have largely been done in vitro. Finkel and colleagues (9) showed that inhibition of NOS prevented TNF-induced decreases in contractility of isolated hamster papillary muscles. In contrast, Yokoyama and associates, using feline hearts and myocytes, found that negative inotropic effects of TNF were not blocked by inhibition of NOS (34). Further, Decking and coworkers, examining hearts from endotoxin-challenged guinea pigs, also concluded that contractile dysfunction is not mediated through NO (7). Our previous in vivo investigations in dogs challenged with endotoxin or bacteria found that L-NMMA did not prevent cardiac depression (4, 5, 35). Therefore, our in vivo findings in endotoxin-, bacteria-, and TNF-challenged dogs are most consistent with Yokoyama and colleagues' in vitro data (34) and Decking and coworkers' ex vivo data (7).

There are a number of potential limitations in our ability to interpret the data in this study. We assumed that if the cardiac-depressant effects of L-NMMA and TNF occurred through different mechanisms of action, they would be additive. Although the simplest approach, this model has not been proven to fit the data. Other measures of cardiac performance that were independent of preload and afterload, such as Emax, may have been more sensitive to small improvements in cardiac function produced by inhibition of NO (32). However, we were able to find significant beneficial effects of NO inhibition on cardiac performance that were significantly altered by fluid loading. Furthermore, larger doses of L-NMMA, given earlier, might have been necessary to fully block NO production in the heart. However, we used a total L-NMMA dose that was three times greater than that known to block acetylcholine-induced responses in dog hearts (6).

At 24 h, TNF-induced cardiac dysfunction had resolved, but there were still significant decreases in MAP and pH. These TNF-associated effects were ameliorated to a greater extent than could be explained by the effect of L-NMMA in normal dogs. Among potential explanations for these augmented beneficial effects of L-NMMA during TNF challenges are that: (1) L-NMMA, through its vasoconstrictive properties, reduced TNF-induced abnormalities in preload, and thereby improved cardiac performance, systemic pressures, and tissue perfusion; or (2) L-NMMA inhibited NOS and excess NO production, preventing hypotension and the direct toxicity of NO on tissues.

The vasoconstrictive properties of L-NMMA in healthy, euvolemic animals were associated over time with a progressive metabolic acidosis. Paradoxically, after TNF challenge, the L-NMMA-induced vasoconstriction was actually associated with an improvement in TNF-induced metabolic acidosis. In this setting, vasoconstriction probably becomes beneficial because it reverses TNF-induced defects in preload and perfusion.

The inducible isoform of NOS, activated by cytokines and associated with disease, requires time to be expressed (4, 20). This might explain why NO played a central role only in the late TNF-induced abnormalities described here. A difficulty with this paradigm for our study is that although L-NMMA did decrease nitrate/nitrites, animals given TNF alone did not, relative to controls, have increases in nitrate/nitrites. It is possible that the low doses of TNF used in our study did not increase NOS activity enough to make such increases measurable.

Our methodology could be used to define which physiologic responses are intrinsically dependent on NOS, even though this was not the aim of our study. We reasoned post hoc that if a given parameter was directly related to NOS activity, it would be immediately affected by L-NMMA and not fundamentally altered by fluid challenges, and that the consequences of L-NMMA administration would be reversed by L-arginine. In agreement with others, we found that intravascular pressures (MAP, SVRI), intracardiac pressures (PCWP, CVP), and heart rate were directly regulated by NOS activity (4, 5, 15, 29). By our definition, other, more direct measures of cardiac performance, such as LV ejection fraction and SVI, were not directly regulated by NOS activity; nor were pulmonary pressures, measures of pulmonary function, or metabolic parameters. In fact, L-NMMA had no significant effect on MPAP. The low vascular resistance of the pulmonary circulation under homeostatic conditions appears to be independent of NO. The late onset and irreversibility with L-arginine of the nonlactic metabolic acidosis and increases in (A-a)O₂ produced by L-NMMA suggest that these effects are not directly related to inhibition of NOS. Prolonged vasoconstriction, such as that produced by NOS inhibition, may in normal animals over time result in fluid shifts, perfusion abnormalities, and organ dysfunction.

In summary, we could not prove a direct role for NO in acute cytokine-induced cardiovascular dysfunction. The predominant cardiovascular effect of L-NMMA appears to be vasoconstriction. If potential adverse consequences can be minimized, this property may be useful in certain clinical settings for treating forms of shock characterized by inadequate preload, hypotension, and metabolic acidosis.