INTRODUCTION

TOP ABSTRACT INTRODUCTION CONCEPTUAL FRAMEWORK METHODS RESULTS DISCUSSION REFERENCES

Septic shock is usually associated with the hemodynamic pattern of systemic hypotension despite normal or increased cardiac output (1). Systemic hypotension is often not responsive to intravascular fluid resuscitation and requires the administration of potent systemic vasopressor agents, such as norepinephrine to reverse it (2, 3). This initial therapeutic approach to counteract systemic hypotension in the septic hyperdynamic circulatory state seems reasonable to sustain cerebral and coronary perfusion pressure. Because norepinephrine induces vasoconstriction via -adrenergic stimulation, it may also decrease visceral organ blood flow if these vascular beds constrict. In such a scenario, intraorgan vascular resistance would increase proportionately more than perfusion pressure. Norepinephrine infusions can decrease splanchnic (4) and renal (5) blood flow under normal circulatory conditions, as well as during essential hypertension and hypovolemic hypotension. If vasopressor infusions induce visceral organ hypoperfusion in the septic patient, then they can potentially induce ischemic organ dysfunction leading to multiple organ dysfunction, loss of gut mucosal integrity (6), a worsening of the intravascular inflammatory state, and death. Thus, concern exists as to the advisability of sustained vasopressor infusions in the hemodynamically unstable septic patient.

It is not clear, however, if the scenario of vasopressor-induced visceral hypoperfusion actually occurs in sepsis. Septic shock is characterized by profound alterations in vascular responsiveness. Downregulation of vascular smooth muscle -adrenergic receptor responsiveness often occurs (7). Furthermore, active vasodilatation has been observed associated with massive nitric oxide production from stimulated vascular endothelium, via expression of inducible nitric oxide synthase (8). Finally, microvascular obstruction by aggregations of platelets and white blood cells, formed by adhesion to the activated vascular endothelium, can disrupt local blood flow distribution independent of -adrenergic tone (9). We hypothesized that norepinephrine infusion would act differently under normal and acute endotoxic conditions, such that we would predict that visceral organ blood flow would increase rather than decrease in endotoxemic conditions.

	CONCEPTUAL FRAMEWORK

TOP ABSTRACT INTRODUCTION CONCEPTUAL FRAMEWORK METHODS RESULTS DISCUSSION REFERENCES

Vascular resistance can be described as the relations between input pressure (P) and blood flow () over a range of pressures and flows, characterizing the vascular P/ relation (10, 11). Using this analysis, vascular resistance can be described as the steady-state P/ ratio. However, this simple ratio inaccurately reflects resistance if the back flow pressure is not zero. To address this concern, analysis of the relation between the change in P and as pressure is rapidly varied is often used to define vascular P/ relation (12). Analysis of the P/ relation over a range of pressures gives two preload independent parameters of vasomotor tone. First, the slope of the P/ relation (changes in P to changes in ) defines the ohmic resistance of the circuit. Increases in ohmic resistance would result in a greater increase in perfusion pressure necessary to increase flow by a constant increment. The second component of the P/ relation is the extrapolated pressure intercept if flow were to cease. For most systemic vascular circuits, the P/ relation decays to an extrapolated pressure at zero flow (Pzf) at a vascular pressure greater than venous pressure (11, 13). This Pzf is felt to reflect the downstream critical closing pressure or vascular waterfall, such that increases in vascular pressures further downstream from this point will not alter flow (11). Importantly, increases in steady-state pressure for a constant flow can occur from an increase in ohmic resistance, Pzf or both (14). Furthermore, ohmic resistance tends to reflect changes in vascular resistance of the small muscular arteries, whereas Pzf tends to reflect vasomotor tone of the very small arterioles, precapillary sphincter, and tissue pressure (10, 12). Thus, one can define the regional P/ relations during basal and acute endotoxemic conditions by both the slope of the P/ relation and its extrapolated Pzf. In summary, the analysis of the P/ relationship during rapid changes in pressure eliminates the confounding effects of changes in preload and allows for the direct measurement of changes in ohmic resistance. This is not possible with conventional techniques such as hemorrhage and resuscitation. In addition, the rapidity of the pressure changes and the short time interval of data collection eliminate the separate contribution of neuroendocrine changes to vasomotor tone. Using this conceptual framework, we hypothesized that norepinephrine infusions would increase organ perfusion pressure under both basal and endotoxemic conditions; however, the changes in the slope of the P/ relation and Pzf during norepinephrine infusion would be different under different conditions. The sum effect of norepinephrine infusions would increase visceral organ blood flow more during endotoxemic conditions than during basal ones. Accordingly, we studied the effects of norepinephrine infusion on the P/ relations of the renal, hepatic, and portal circulations during basal and endotoxic shock states in an acutely anesthetized canine model.

	METHODS

TOP ABSTRACT INTRODUCTION CONCEPTUAL FRAMEWORK METHODS RESULTS DISCUSSION REFERENCES

The study protocol was approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh and conforms to the guiding principles of the American Physiologic Society and the position of the American Heart Association on research animal use. Nine male mongrel dogs (weight range 18.2 to 21.6 kg) were studied. They were anesthetized with intravenous pentobarbital sodium (30 mg/ kg) after taking nothing by mouth (NPO) for at least 12 h. A 7-French polyvinyl chloride catheter was inserted into the right femoral vein for the continuous infusion of pentobarbital at 2 to 4 mg/kg/h. Supplemental doses of 50 to 75 mg pentobarbital were subsequently given intravenously as necessary to prevent spontaneous movement. The trachea was intubated with a Hi-Lo cuffed endotracheal tube with an interior diameter of 9.0 mm (National Catheter, Argyle, NY). Ventilation was accomplished by a constant-volume ventilator (Servo Ventilator 900C; Siemens-Elena, Sweden) at a tidal volume of 10 to 15 ml/ kg. Enriched inspired O₂ (fraction of inspired oxygen [FI_O2] 28%) was given, and arterial blood gases were periodically monitored (ABL-30; Radiometer, Copenhagen, Denmark). Changes in ventilatory frequency or supplemental intravenous doses of sodium bicarbonate were given as necessary to maintain arterial PCO₂ between 38 and 42 mm Hg, arterial pH between 7.37 and 7.42, and PO₂ > 90 mm Hg. A standard lead II electrocardiogram was monitored for heart rate. A saline-filled polyethylene catheter with multiple end-side holes and a flow-directed balloon-tipped pulmonary artery catheter with a 15-cm proximal port (model 93A-095; American Edwards, Irvine, CA) were advanced into the right atrium and pulmonary artery, respectively, from peripheral cutdown sites. Placement of the proximal port in the right atrium was verified by waveform analysis of the pressure tracing, and placement of the distal tip in a pulmonary artery was verified by fluoroscopy. A 5-French saline-filled polyethylene catheter with multiple side holes was advanced in the midabdominal aorta from a femoral artery for the measurement of arterial pressure.

To minimize intravascular volume and hematocrit shifts due to splenic contraction during the experimental protocol, a splenectomy was performed after maximal splenic contraction to 0.5 ml of topical epinephrine (1:10,000). Ultrasonic flow probes (Transonic Systems, Ithaca, NY) were placed around the portal vein and hepatic artery. The left renal artery was isolated, and an ultrasonic flow probe was placed around it. All flow probes were cemented into place with an agar gel mixture that minimized movement artifact and improved acoustic signal. The infrahepatic vena cava was isolated, and a hydraulic vascular occluder (In-Vivo Metric, Healdsburg, CA) was placed around it. Following splenectomy, venous catheters were inserted into the left hepatic and left renal veins (via the right external jugular vein), the portal vein (via the splenic vein). These catheters as well as another ultrasonic flow probe around the left femoral artery were placed for the purpose of a concomitant study.

The abdomen was loosely closed with interrupted sutures. All animals were fluid resuscitated with 0.9% saline as necessary to maintain the right atrial pressure between 2 and 5 mm Hg. The conditions of the animals were allowed to stabilize. Stability was defined as constant heart rate, arterial pressure, end-tidal CO₂, and organ flow signals for at least 30 min. The vascular catheters were connected to low-displacement transducers (Micron MP-50; Gould, Cleveland, OH). All vascular pressures were referenced to the midcardiac plane. At the conclusion of the experiment, the animals were killed by intravenous injection of KCl. To determine their respective zero hydrostatic pressure, all intrathoracic catheters were exposed to atmospheric pressure while in situ during a necropsy by excising surrounding tissues. The pressure signals were continuously recorded on an eight-channel recorder (Model 2800; Gould) and on a computer using customized hardware and software, digitized at a sampling rate of 150 Hz per channel. These data were stored on a computer disc for subsequent analysis (Model DN3550; Apollo Computer Inc., Chelmsford, MA). The acquisition system (Model RTS-132; Significat, Hudson, MA) consisted of an 80186-coprocessor card and a separate acquisition hardware chassis.

Experimental Protocol

The study entailed examining the hemodynamic responses to five sequential conditions: an initial baseline state (Control 1), a continual infusion of norepinephrine during control (NE), a repeat baseline to control for time-dependent factors (Control 2), following induction of acute endotoxemia (Endo), and Endo plus NE. Control 1 was defined as the initial hemodynamically stable state at least 30 min after completion of the surgical preparation. NE was created as a continuous intravenous infusion of norepinephrine at 0.3 µg/kg/min. This infusion was continued until a new hemodynamic steady state was achieved, as defined by a constant heart rate, arterial pressure, end-tidal CO₂, and renal blood flow (RBF) for a period of 5 min. Following NE, the infusion of norepinephrine was stopped and the animal observed until a new steady state was achieved. In practice, this new hemodynamic steady state was achieved within 10 to 20 min after discontinuing the norepinephrine infusion. Control 2 studies were then done. Endo was then induced by the infusion of 1 mg/kg of Escherichia coli endotoxin (L-2880 lipopolysaccharide; Sigma, St. Louis, MO) over 5 min via the right atrial port. Because acute endotoxemia induced immediate hemodynamic instability the hemodynamic response of each animal was closely monitored. Acute resuscitative efforts were avoided unless strictly needed. If necessary, however, a bolus of 200 ml of 0.9% saline was administered for a  < 65 mm Hg and respiratory rate increased if end-tidal CO₂ > 45 mm Hg (four animals). No animal received sodium bicarbonate during this resuscitative interval. All dogs were in a hemodynamic steady state 30 to 45 min after endotoxin administration. After Endo data collection, each animal was again started on an intravenous infusion of 0.3 µg/kg/min of norepinephrine. Saline was also administered intravenously in seven animals during the study in order to achieve and maintain a right atrial pressure of 1 to 3 mm Hg.

During each of the five steps data were collected during two sequential stages. First, after 10 s of apnea, apneic steady-state pressures and flow data were collected over the next 5-s interval. These data were averaged to define the mean pressure and flow data for vascular beds. Immediately after this interval a rapid and transient occlusion of the inferior vena cava was performed and the hemodynamic response followed over the next 5 s. The beat-to-beat data pressure and flow data pairs were used to construct hepatic and renal arterial P/ relations. Example of these analogue data are shown in Figure 1. After these measurements, each animal was returned to mechanical ventilatory support and allowed to stabilize and return to pre-inferior vena caval occlusion hemodynamic values. Then, cardiac output was measured by thermodilution technique using the average of five 5-ml iced dextrose 5% in water (D5W) injections done at random times with respect to the ventilatory cycle.

View larger version (115K): [in this window] [in a new window]   Figure 1.   Strip chart recording of the changes in, in descending order, Pa, RBF, hepatic artery blood flow, and portal venous flow immediately after inferior vena caval occlusion.

Measurements, Calculations, and Statistical Analysis

Ultrasonic flow data were obtained from ultrasonic flowmeters calibrated ex vivo using a standard perfusion circuit as recommended by the manufacturer (Transonic system). Paired arterial P/ data during transient inferior vena caval occlusion were used to generate beat-to-beat P/ relations for each vascular bed. Abdominal aortic pressure was coupled to both renal arterial flow and hepatic arterial flow. Due to the nature of the portal circulation, no measure of portal circulatory P/ changes was made during the inferior vena caval occlusion maneuver interval.

This experimental design allowed us to test for differences of mean flow between the five states (Control 1, NE, Control 2, Endo, and Endo-NE) while controlling for input pressure. This is because during the inferior vena caval occlusion maneuver input pressure decreased under all conditions, becoming similar across all conditions along some common pressure range for each individual. However, the effect of each experimental condition (NE, Endo, or both) on flow will be determined by its direct effects on the organ being studied and by its remote effects on other vascular beds. This is because a change in flow to one vascular bed will result in a reciprocal change in flow to other beds. In this way, changes in blood flow to an individual vascular bed are produced by the net effects of all other vascular beds. Because agents such as NE and endotoxin are known to change blood flow distribution, it is important to separate the local and systemic effects of these agents. One limitation of the standard modeling techniques used to address this question is that each observation is required to be independent of the other. The generalized linear model used in this report (15) does not require that observations be independent of each other. The model estimates the intraclass correlation (a measure of homogeneity of the sample) and adjusts for this variable in estimating each effect. Therefore, multiple observations from a given test subject can be included in the analysis. Using this analysis, changes in the slope of the P/ relationship were inferred to reflect changes in ohmic vascular resistance, whereas parallel shifts in P/ relations were inferred to reflect changes in the critical closing pressure of the vascular bed usually referred to as the Pzf (10, 11). Once overall statistically significant difference of means was detected between the five groups, post hoc comparisons were made between treatments using the same generalized linear model and controlling for pressure. A p < 0.05 was considered statistically significant.

Comparisons between apneic steady-state hemodynamic values across the five different conditions in the nine experimental animals were performed using nonparametric analysis of variance (ANOVA) (Friedman's test) corrected for multiple comparisons. Correlations between systemic hemodynamic variables and static preocclusion blood flows in the hepatic arterial, renal, and portal circulation were tested for with Spearman's test. Data are presented as mean ± SD unless otherwise stated and differences corresponding to a p value of < 0.05 are considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION CONCEPTUAL FRAMEWORK METHODS RESULTS DISCUSSION REFERENCES

The mean steady-state hemodynamic data for all the measured variables for the five conditions are summarized in Table 1. The most striking hemodynamic effects involved changes in Ppa among the conditions. Arterial pressure during Control 1 and Control 2 were not different. Endo was associated with a decrease in Ppa and in both control and endotoxemic conditions, norepinephrine infusion increased Ppa as compared with the pre-NE condition (p < 0.005) (Figure 2). Steady-state RBF showed very similar values for the two control states and for NE. RBF tended to decrease during endotoxemia and increase when NE was added in the endotoxemic condition. However, these changes were not statistically significant. Calculated renal vascular resistance (RVR) was the same in both control conditions and only slightly higher with NE and lower with Endo (p = NS). RVR remained low during endotoxemia even with the addition of NE (Endo-NE condition) and was significantly lower than NE (p = 0.015). No relationship between Ppa and hepatic arterial flow could be established under any study condition. Neither could apneic hepatic arterial blood flow or portal venous flow be correlated with cardiac output and . Thus, all subsequent analysis pertains only to the renal P/ relations.

View larger version (27K): [in this window] [in a new window]   Figure 2.   Graph illustrating the changes in Pa in each dog in the five different physiologic conditions under study. Pa increases from the first control period (Control 1) in all animals as they receive norepinephrine (NE), returns to baseline values when the infusion is ceased (Control 2), falls markedly with endotoxemia (Endo), and returns toward control value after the reintroduction of norepinephrine (Endo-NE) (p < 0.0001).

Using a generalized linear model we observed that, after controlling for the effect of Pa, mean RBF was significantly different across the five conditions (p < 0.001) (Table 2). Pair-wise comparisons revealed that, after controlling for the effect of blood pressure, mean RBF increased with NE compared with Control 1 (74.9 ± 22.9 ml/min versus 83.4 ± 29.9 ml/min, p < 0.001), decreasing in Control 2 to values similar to Control 1 (76.2 ± 20.8 ml/min, p < 0.019). During Endo, RBF decreased from Control 2 (to 60.47 ± 41.8 ml/min, p < 0.001) returning again to near-control values with addition of NE (73.2 ± 35.9 ml/min, p < 0.001 compared with Endo). The effects of treatment (Endo, NE, or none), independent of the effect of changes in , on RBF obtained from the parameter estimates of the general linear model are summarized in Figure 3.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Histograms summarizing the relative size and direction of the "treatment" effects with endotoxin or NE or both on median RBF using the parameter estimates of the general linear model. The height of each box represents the median RBF under each condition. The shaded portions represent the change from the preceding condition. The lightly shaded portions are the effects attributable to blood pressure alone while the dark shaded portions represent the size of the treatment effect not attributable to blood pressure. Thus, the decline in flow during Endo was due in small part to the fall in pressure but much more significantly affected by other effects of endotoxin. Similarly, this decline in flow was reversed by NE in part by the increase in blood pressure and in part by other effects (e.g., increase in nonvisceral vascular resistance). *Significant differences from control (p < 0.01); **difference from Endo (p < 0.01).

The P/ relationship for the renal vasculature was altered by the administration of NE both before and after the administration of endotoxin (Figure 4). However, the changes in this relationship were different depending on whether norepinephrine was infused under basal or acute endotoxemic conditions. The infusion of norepinephrine during control conditions induced two independent effects. First, the slope of the P/ relation decreased indicating decreased ohmic resistance. Second, the renal P/ relation was shifted to the right (pressure on the x-axis) such that Pzf increased (increased renal vascular critical closing pressure) and for a given pressure, flow was less compared with control. By contrast endotoxin induced a leftward shift in the renal P/ relation such that Pzf decreased and RBF was higher at any given pressure relative to control while the slope of the P/ relation still decreased (decreased ohmic resistance). Interestingly, in the endotoxemic condition, the addition of NE produced a further decrease in the slope of the P/ relation (decreased ohmic resistance) and the renal P/ relation was shifted further to the left such that Pzf decreased and less pressure was required to maintain RBF (Figure 4). All these effects were significant relative to the preceding condition, p < 0.001. 

View larger version (12K): [in this window] [in a new window]   Figure 4.   Graph illustrating the pressure/flow relationship for the left renal vasculature during the averaged control states (empty circles-C ), the infusion of norepinephrine (stars-N ), after endotoxin (squares-E ), and after norepinephrine during endotoxemia (circle with cross-N&E ). The critical closing pressures (Pzf) are also displayed on the x-axis using the same abbreviations. Endotoxemia causes decreased vascular resistance (decreased P/ slope) and decreased outflow pressure at the vascular waterfall (left shift of Pzf) compared with controls. Norepinephrine infusion during endotoxemia tends to further decrease resistance (decreased P/ slope) and decreases Pzf while increasing P/ such that for a given P, is greater compared with Endo.

	DISCUSSION

TOP ABSTRACT INTRODUCTION CONCEPTUAL FRAMEWORK METHODS RESULTS DISCUSSION REFERENCES

The findings of our investigation demonstrate that norepinephrine infusion, at clinically relevant dosages, affects RBF differentially during basal and acute endotoxemic conditions. When normal circulatory controls exist in the otherwise unstressed circulation, norepinephrine infusion fails to proportionally increase dynamic RBF despite increasing arterial pressure. By contrast, once the systemic circulation has been perturbed by the insult of acute endotoxemia, identical dosages of norepinephrine increase both dynamic RBF and perfusion pressure. Importantly, the methodology used allowed us to isolate the effect of the intravenous infusion of norepinephrine on the determinants of steady-state RBF independent of perfusion pressure. Under normal conditions, norepinephrine, infused intravenously at a rate capable of increasing by approximately 15 mm Hg, induces a decrease in renal vascular ohmic resistance but an increase in vascular critical closing pressure, such that in the aggregate, these combined renal vasoactive effects serve to reduce RBF for a constant perfusion pressure. However, during acute endotoxemic conditions, the initial state of the renal vasculature is altered, reflecting the profound effects that endotoxemia has on vascular smooth muscle tone and vascular responsiveness. Under these conditions the addition of norepinephrine infusion further decreases renal vascular ohmic resistance but also decreases the vascular critical closing pressure, such that in the aggregate, these combined renal vascular effects serve to increase RBF for a constant perfusion pressure. Surprisingly, no consistent effect of norepinephrine or endotoxin could be demonstrated on the hepatic arterial bed.

The analysis of the dynamic P/ relationships was not meant to replace the steady-state analysis but rather to complement it by isolating the intrinsic organ P/ relationship from the systemic hemodynamic state. Analysis of the steady-state data also reveals certain information. First, RBF tends to decrease with endotoxemia and increases once blood pressure is restored with NE. Second, RVR decreases with endotoxemia and remains low even when NE is added such that the RVR is significantly lower in the Endo-NE condition compared with NE without endotoxemia. These findings are consistent with the view that, in the normal animal, NE increases RVR and arterial pressure to a similar degree such that in the aggregate there is little change in steady-state RBF. However, during endotoxemia, hypotension and a decrease in RVR occurs. NE added to this condition restores arterial pressure without increasing RVR and as such, RBF tends to increase. However, these steady-state findings do not provide a complete picture by themselves. Our preparation induced significant changes in systemic hemodynamics making it impossible to separate the individual effects of NE and endotoxin. For these reasons we chose to analyze the dynamic P/ relationships induced by rapid inferior vena caval occlusion (to control for systemic hemodynamics) and used a general linear model (to control for the effects of arterial pressure).

Our findings agree with the results of previous animal studies and extend these observations to include analysis of regional vascular characteristics. We have previously shown that acute endotoxemia decreases systemic Pzf (12). In the present study, we saw similar decreases in regional Pzf only in the renal and not the hepatic vascular beds. Furthermore, the finding of norepinephrine-induced intense vasoconstriction has only been seen to occur with the infusion of the drug directly into the renal artery, not via the systemic route (16). In addition, the dose of drug used in such models of norepinephrine-induced acute renal failure was twice to three times that used in our study and well beyond the mean dose usually administered in clinical practice. Schaer and coworkers have also reported the renal effects of norepinephrine infusion at different doses with or without the addition of dopamine (17). They measured RBF with the technique of regional thermodilution and found that, although RVR appeared to increase from baseline (there was no placebo arm in their study), total RBF progressively increased with increasing doses of intravenous NE up to 1.6 µg/kg/min. In their study these adverse effects of norepinephrine infusion on RVR were seen in animals with a baseline of 151 mm Hg. Furthermore, norepinephrine infusion increased approximately 30% to 200 mm Hg. The relevance of these data to clinical practice is unclear. Instead, our findings are in keeping with a study by Anderson and coworkers (18). These investigators infused norepinephrine intravenously at 0.2 to 0.4 µg/kg/min in conscious dogs and, using an electromagnetic flow probe, studied RBF, RVR, and glomerular filtration rate. They, like us, found that RBF increased and RVR decreased in response to norepinephrine infusions. Such renal vasodilatation was unaffected by pretreatment with indomethacin, propranolol, or angiotensin-converting enzyme (ACE) inhibition. However, efferent autonomic sympathetic nerve blockade with pentolinium prior to norepinephrine infusion, completely abrogated norepinephrine-induced renal vasodilatation. These investigators concluded that, in keeping with previous experimental data (19), most of the renal vasodilating effect of intravenous norepinephrine could be attributed to the increase in systemic blood pressure which in turn decreases renal sympathetic tone. The effect of norepinephrine infusion on regional blood flow has also been recently assessed by Zhang and coworkers (20). These investigators have also demonstrated that, following the administration of endotoxin, norepinephrine did not induce any decrease in renal or hepatic blood flow. Our data agree with their results and extend them to assess the renal vascular effects of norepinephrine apart from its effect on blood pressure.

The effects of norepinephrine infusion on RBF may not be unique to this vasopressor, but representative of the group of potent vasoconstrictor agents. Bersten and coworkers have recently studied the renal effects of epinephrine, another vasopressor agent with a strong -adrenergic effect. These investigators administered epinephrine by continuous infusion at clinically relevant doses in the normal and septic sheep (21, 22). They demonstrated that, in normal animals, after a short-lived (minutes) small decrease in RBF at the highest doses tested (0.4 to 0.8 µg/kg/min), RBF progressively increased and remained elevated for up to 6 h of epinephrine infusion. They found a similar increase in RBF in septic animals. This increase in RBF occurred in association with a nonsignificant trend toward greater RVR that was offset by greater renal perfusion pressure. Thus, -adrenergic agents may increase, rather than decrease, RBF. If this is so in the experimental situation, similar systemic and local effects ought to occur in humans and may translate into increased urine output and/or increased creatinine clearance. Indeed, although studies that directly measure RBF and RVR in humans are not available, many clinical reports now support the notion that the continuous infusion of norepinephrine may increase urine output and improve creatinine clearance in patients with hyperdynamic septic shock (3, 23, 24).

We were unable to demonstrate a predictable P/ relationship for the hepatic arterial circulation during inferior vena caval occlusion. This is not surprising given the complex nature of arterial blood flow autoregulation within the liver which actively increases with either increases in hepatic metabolism or decreases in portal blood flow (hepatic buffer response) (25, 26), rather than to transient variation in cardiac output or mean arterial pressure. In addition, hepatic arterial and portal venous flows may have been more powerfully affected by changes in hepatic venous pressure induced by the inferior vena caval occlusion itself. Such changes may induce an increase in liver blood flow despite a transient reduction in preload (26). Our model did not, therefore, allow meaningful conclusions to be drawn concerning the hepatic arterial response to endotoxin and norepinephrine infusion. It is worthy of note that recent work performed in humans with septic shock also receiving norepinephrine infusion demonstrated higher splanchnic blood flows during norepinephrine-treated septic shock than septic patients not receiving norepinephrine (27).

Limitations. First, our animal model was a highly invasive preparation studied under light general anesthesia requiring laparotomy, the insertion of multiple perivascular flow probes, the cannulation of several vessels, the insertion of a vascular occluder around the inferior vena cava, and a splenectomy. This may have affected our findings but is unlikely to have biased them since we compared our data among conditions in which all animals had the same treatments. Nonetheless, following suturing of the abdomen, all animals were hemodynamically and metabolically stable for a period of 30 min prior to initiation of our study protocol. Similarly, in order to diminish the potential bias associated with stress-induced, time-dependent physiologic changes, two baseline measurements were introduced, one before (Control 1) and one after (Control 2) the first administration of norepinephrine. Furthermore, prior studies by our group using this model have shown that this preparation remains hemodynamically stable over at least 4 h, a time interval in excess of our entire observation period. Similarly, the use of the transient vascular occlusion technique (13, 14, 28) eliminated many of the effects of systemic stress and of variability in cardiac output and preload otherwise induced by such hemodynamic changes during each step. For instance, in the absence of inferior vena caval occlusion, it would have been impossible to determine if the changes in RVR seen during Endo or NE were in response to steady-state changes in systemic pressures and flows. The infusion of endotoxin in dogs appears to reasonably reproduce several hemodynamic aspects of human septic shock and has been widely used for this purpose (20, 29). In addition, endotoxin itself may play an important role in the pathogenesis of septic shock-associated renal failure (30), causes hypotension (as was the case in our experiment), and has been shown to induce renal afferent arteriolar vasoconstriction in mammals (31, 32). We, therefore, believe that our animal model of septic shock was also clinically relevant. In addition, the selection of the dose of norepinephrine to be infused was aimed at maintaining clinical relevance and reproduced the mean dose range reported in the management of human septic shock (33). Although we did not use fluid administration to assess the effects of NE and endotoxemia as some previous studies have, we did maintain a constant right atrial pressure by basal fluid administration. Thus, our model was that of resuscitated endotoxemia and quite comparable to clinical conditions.

In summary, we have demonstrated that norepinephrine infused at clinically relevant doses increases and induces a decrease in ohmic resistance but an increase in Pzf in the renal artery of the dog. Furthermore, we have shown that norepinephrine infusion in acute endotoxemia at similar doses reverses systemic hypotension and improves RBF independent of perfusion pressure. Therefore, norepinephrine appears to have two beneficial effects on RBF during endotoxic shock, one related to its effects on RVR and the other related to its effect on renal perfusion pressure. These findings in association with other literature cited provide a physiologic basis for the administration of norepinephrine during septic shock.