Vasopressin in septic shock: effects on pancreatic, renal, and hepatic blood flow

Low-dose vasopressin has been proposed for treatment of severe hypotension in septic shock that is otherwise unresponsive to high doses of alpha-adrenergic agents [1, 2]. To date, smaller controlled studies of human subjects receiving low-dose vasopressin in septic shock have been rather encouraging, but adverse events, possibly related to the use of vasopressin, have also been reported [3, 4].

Vasopressin can produce intense vasoconstriction that is independent of tissue oxygenation and metabolism [5]. The capacity of vasopressin to decrease mesenteric and portal blood flow has been demonstrated by its efficacy in reducing gastrointestinal bleeding [6], including hemorrhage from blunt liver trauma [7, 8]. The effects of vasopressin were well documented in the 1970s and 1980s in human [9] and animal [10‐12] studies, but this was mostly in non-septic conditions and with doses significantly exceeding what today is considered to be a 'safe' range.

Recently published results from animal studies have confirmed previous findings that high doses of vasopressin (greater than 0.1 units per minute) clearly redistribute regional blood flows and decrease tissue oxygenation [13, 14]. However, reported effects of low-dose vasopressin on regional blood flow and metabolism are more conflicting and range from 'deleterious' [15] to increased mesenteric blood flow and beneficial effects on tissue metabolism [16].

The effects of low-dose vasopressin on other organs, such as the pancreas, are largely unknown. Decreased blood flow in the pancreas was found when high doses of vasopressin were infused under non-septic conditions [12], but the effects of low-dose vasopressin on the pancreas in septic shock have not been studied. The pancreas appears to be particularly vulnerable to low flow as a result of cardiogenic shock [17], hypovolemia [18, 19], and sepsis [20]. Prolonged pancreatic ischemia secondary to hypovolemia may cause secretory dysfunction, edema, and inflammation [18].

Vasopressin has been reported to increase urine output [21, 22] and creatinine clearance [23] in septic subjects. Low-dose vasopressin did not decrease total renal blood flow in endotoxemic pigs. However, it has been found to cause redistribution of intrarenal blood flow, resulting in a reduction of medullary blood flow [24, 25] even with physiologic plasma levels.

We hypothesized that increasing systemic blood pressure by administering vasopressin in fluid-resuscitated experimental septic shock would result in a substantial redistribution of regional blood flow within the splanchnic region and, consequently, in altered microcirculatory blood flow in abdominal organs. Thus, the aim of this study was to compare changes in systemic blood flow with changes in regional splanchnic blood flow and microcirculatory blood flow in the liver, kidney, and pancreas during administration of low-dose vasopressin in fluid-resuscitated septic shock in pigs.

This study was performed according to the National Institutes of Health (Bethesda, MD, USA) guidelines for the care and use of experimental animals. The protocol was approved by the animal ethics committee of Canton Bern, Switzerland.

Thirty-two domestic pigs (weight, 28 to 32 kg) were fasted overnight but were allowed free access to water. The pigs were sedated with intramuscular ketamine (20 mg/kg) and xylazinum (2 mg/kg). After induction of anesthesia with intravenous metomidate (5 mg/kg) and azaperan (2 mg/kg), the pigs were orally intubated and ventilated with oxygen in air (fraction of inspired oxygen [FiO₂] = 0.40). Inhaled concentration of oxygen was continuously monitored with a multi-gas analyzer (S/5™ Critical Care Monitor; Datex-Ohmeda, part of GE Healthcare, Little Chalfont, Buckinghamshire, UK). Anesthesia was maintained with continuous intravenous infusions of midazolam (0.5 mg/kg per hour), fentanyl (20 μg/kg per hour), and pancuronium (0.3 mg/kg per hour) to simulate clinical conditions as closely as possible. The animals were ventilated with a volume-controlled ventilator with a positive end-expiratory pressure (PEEP) of 5 cm H₂O (Servo 900C; Siemens, Dietikon, Switzerland). Tidal volume was kept at 10 to 15 mL/kg and the respiratory rate was adjusted (14 to 16 breaths per minute) to maintain end-tidal carbon dioxide tension (arterial carbon dioxide partial pressure, PaCO₂) at 40 ± 4 mm Hg. The stomach was emptied with an orogastric tube.

Systemic, regional, and local parameters recorded during the development of septic shock and during fluid resuscitation but before t = 300 minutes are presented in Appendix 1. Data recorded after t = 300 minutes until end of the study at t = 480 minutes are presented below and in Tables 1, 2, 3 and Figures 1 and 2. Three series of LDF measurements from the liver (one each in groups V, S, and SV) and two series from the kidney (one from group C and another from group S) had to be excluded because of excessive motion artifacts and loss of optical coupling to the tissue.

All animals in groups S and SV first developed signs of hypodynamic septic shock, with low MAP, low CI, and decreased microcirculatory blood flow, followed by signs of normo/hyperdynamic sepsis after fluid administration (Appendix 1). Fluid resuscitation increased CI. It restored blood flow in the portal vein, the celiac trunk, and the hepatic and renal arteries. Furthermore, it restored microcirculatory blood flow in the renal cortex. In contrast, fluid administration did not restore microcirculatory blood flow in the liver (down by 15% to 27%) or the pancreas (down by 27% to 32%).

Substantial effects of vasopressin on the systemic and regional circulation were observed within a few minutes after starting the vasopressin infusion (in groups V and SV). The peak effect on most systemic and regional parameters was measured between 30 and 60 minutes after starting vasopressin (Tables 1, 2, 3; Figures 1 and 2). Administration of vasopressin to septic animals (group SV) increased MAP and decreased CI and HR.

Administration of vasopressin resulted further in significant redistribution of splanchnic blood flow (Figure 1; Table 2): 60 minutes after the start of vasopressin infusion, blood flow in the portal vein had decreased by 58% in septic animals receiving vasopressin (group SV) but by 19% in septic controls (group S; p < 0.01). Blood flow in the celiac trunk increased by 20% in group SV and by 30% in group V but decreased by 15% in group S (Figure 1; Table 2). The hepatic artery blood flow remained virtually unchanged or increased in some animals (Figure 1). Thus, similar to portal flow, total liver blood flow decreased (p < 0.01) more in group SV (by 32%) than in group S (by 15%; Table 2). Microcirculatory blood flow in the liver remained unchanged in both septic groups. Administration of vasopressin in group SV decreased microcirculatory blood flow in the pancreas further to 36% ± 14% (p < 0.01) of baseline, whereas virtually no change occurred in group S.

Renal artery blood flow remained unchanged in septic controls (group S) as well as in septic animals receiving vasopressin (group SV; Table 2). In group SV, microcirculatory blood flow in the renal cortex decreased by 16% ± 20% (Figure 2; p < 0.01), but urine output increased (Table 1). Microcirculatory blood flow in group S remained unchanged. Systemic, regional, and microcirculatory flow parameters (Table 3) remained stable in control animals not receiving vasopressin (group C) and in vasopressin control animals (group V) during the first 300 minutes.

Administration of vasopressin to non-septic animals (group V) resulted in systemic, regional, and local changes similar to those seen in septic animals (Tables 1, 2, 3). However, the effects of vasopressin on some systemic (pulmonary artery occlusion pressure and mixed venous oxygen saturation) and regional (total liver blood flow, portal blood flow, portal oxygen delivery [DO₂I PV], and renal oxygen delivery [DO₂I kidney]) parameters appeared to be stronger in non-septic than in septic animals.

Two septic and another two non-septic groups were studied in a factorial design with the aim of comparing changes in systemic blood flow with changes in regional splanchnic blood flow and microcirculatory blood flow in multiple abdominal organs during administration of low-dose vasopressin in septic shock. Therefore, the results of the non-septic groups are presented for reference only and are not discussed in detail.

Administration of low-dose vasopressin in this porcine model of volume-resuscitated septic shock increased arterial blood pressure, decreased systemic blood flow and oxygen delivery, and resulted in a marked redistribution of blood flow in the splanchnic region. Portal venous flow decreased almost by half in the group receiving vasopressin. In contrast, hepatic arterial blood flow either remained unchanged or increased. This finding suggests different effects of vasopressin on the arterial versus the portal venous blood supply in the liver. In fact, it has been shown in non-septic rats that effects of vasopressin on the liver are heterogenous and more pronounced on the portal venous side than on the arterial side, due to receptor density, which favors the portal zone [29]. In non-septic low-flow states, liver blood flow is known to be regulated by the hepatic arterial buffer response, in which a decrease in portal flow leads to increased hepatic arterial blood flow due to vasodilatation, which is mediated locally by the accumulation of adenosine [30]. In the present study, hepatic arterial buffer response did not fully compensate for decreased portal flow, except perhaps in one animal out of eight (Figure 1). Our results in septic pigs are also in accordance with a study by Schiffer and colleagues [31] on endotoxic sheep showing that capacity of the hepatic arterial buffer response is diminished during endotoxemia [32].

Although total liver blood flow decreased during administration of vasopressin, average microcirculatory blood flow measured on the surface of the liver remained unchanged. This finding must be interpreted with caution. One question that has to be addressed is whether microcirculatory flow measured on the surface of the liver is representative of the entire organ. In rats, microcirculatory blood flow measured on the hepatic surface using LDF has been reported to reflect changes in total liver blood flow [33]. Similar findings were found in a porcine model [34]. However, the authors of the latter study also reported an increased sensitivity of LDF to changes in arterial blood flow [34].

Microcirculatory blood flow in the pancreas decreased markedly during the development of septic shock. Although intravenous fluids appeared to have effectively restored systemic and regional blood flows, microcirculatory flow in the pancreas remained approximately 30% below baseline after fluid administration. Administration of vasopressin further decreased pancreatic blood flow by approximately 50% despite the fact that blood flow in the supplying regional artery (celiac trunk) increased (Table 2). Why the hepatic artery was apparently getting a larger share of flow in the celiac trunk than the pancreas cannot be answered from the present data. It is possible that the V1 receptors (V1Rs) are more dense in the pancreatic vascular bed than in the hepatic artery or that, even if the hepatic arterial buffer response could not increase arterial hepatic flow enough to maintain liver blood flow unchanged, it may have limited the reduction in liver flow by reducing the resistance in the hepatic artery and thereby favoring distribution of flow in the celiac trunk to the liver.

Previous studies demonstrate that the pancreas is very vulnerable to deterioration of systemic and splanchnic blood flow caused by hemorrhage [19], sepsis [20, 35], and administration of vasoconstrictors such as vasopressin under non-septic conditions [11, 36, 37]. We are not aware of any other study that has investigated the effects of vasopressin on the pancreas in septic shock. Hypoperfusion of the pancreas may be a clinically relevant problem; the pancreas has been suggested to be a source of toxic mediators after ischemia and reperfusion injury [38], and impaired pancreatic function has been found after prolonged hypoperfusion [18].

Blood flow in the renal artery decreased moderately after the vasopressin infusion began but recovered to baseline with time. Microcirculatory blood flow in the renal cortex also decreased but remained low. Despite decreased regional and microcirculatory blood flow in the kidney, urine output increased. Vasopressin produces vasoconstriction via the V1Rs, whereas osmoregulation, antidiuretic effects, and nitric-oxide-dependent vasodilatation are mediated via the V2 receptors (V2Rs) [39]. In the present study, we used the vasopressin analogue ornithin-8-vasopressin, which has effects very similar to those of arginine vasopressin but a slightly higher affinity for V1R. However, it can still bind to the V2R once V1Rs are saturated. There is experimental evidence that, in the kidney, vasopressin preferentially constricts efferent arterioles [40]. Thus, increased diuresis was related to increased filtration pressure rather than to renal blood flow. Increased diuresis during administration of vasopressin has also been reported in patients in septic shock [21] and with hepatorenal syndrome [41].

The aim of this study was to measure the effects of vasopressin on regional and microcirculatory blood flow in abdominal organs during septic shock. Severe, irreversible microcirculatory disturbances have been associated with poor outcome in patients with septic shock [42]. In patients dying from septic shock, these disturbances have been shown to persist even after correction of systemic variables [26, 43]. Nevertheless, treatment of circulatory shock is mostly guided by systemic variables alone because direct measurements of regional and local splanchnic blood flow in patients are invasive, time-consuming, and require special skills and instruments that are not readily available at the bedside.

We intended to simulate clinical conditions in critically ill patients as closely as possible. The pig model appeared suitable because of the pig's anatomical and physiologic similarity to humans with respect to the cardiovascular system and the digestive tract [44, 45]. Fecal peritonitis is a frequent cause of septic shock in humans, and clinical conditions in a critical care unit were imitated as closely as possible in the laboratory (sedation, mechanical ventilation, monitoring, and drug administration). Still, the results of this study are not based on human data, and that is the study's main limitation. Furthermore, due to the small number of animals per group, some biologically relevant effects may have been missed. The full factorial design used in this study, comprising three different control groups, was intended to minimize this risk. Another limitation of this study may be the fact that we measured only organ blood flow, but not metabolism. However, a recent study performed in our laboratory suggested that signs of anaerobic metabolism in tissues may be detected only relatively late and only when blood flow is substantially reduced (that is, by 60% or more) and that an even greater reduction of blood flow may be required in order to detect these signs in regional venous blood [46]. The vasopressin dose used in the present study (0.06 U/kg per hour, which is approximately 0.06 U/minute in a 70-kg human) was determined in pilot studies as the dose required to raise MAP by 20 to 25 mm Hg in septic pigs. This perhaps may be considered 'high' by some investigators since it is higher than the 0.04 U/minute proposed by several authors, including Malay and colleagues [14]. However, the dose used in the present study was lower than 0.10 U/minute, which Martikainen and colleagues [15] advised as the upper limit since higher doses caused reduction in systemic and splanchnic blood flow in endotoxemic animals. On the other hand, Klinzing and colleagues [47] administered vasopressin as a single vasopressor in septic patients in doses that were up to 50 times higher than those used in the present study (average, 27 U/minute).

Another matter that deserves mentioning here is the fact that most clinicians who use vasopressin in septic shock use it as a supplementary vasopressor to norepinephrine or in combination with dobutamine. However, in the present study, we used vasopressin alone because our purpose was to study the effects of vasopressin without possible interactions of other vasoactive agents.

Administration of low-dose vasopressin in septic shock resulted in increased arterial blood pressure and decreased systemic blood flow. Splanchnic regional blood flow was substantially redistributed. It decreased markedly in the portal vein, but remained unchanged or increased in the hepatic artery, and increased in the celiac trunk. This resulted in significantly decreased total liver blood flow. Microcirculatory blood flow remained unchanged in the liver but decreased markedly in the pancreas. Initially, blood flow in the renal artery decreased, but it returned to baseline levels after 3 hours, whereas microcirculatory flow in the renal cortex remained decreased. This study also showed that increased urine output does not necessarily reflect increased renal blood flow. Considering these disturbances in blood flow and the fact that the safety of vasopressin in septic shock has not yet been demonstrated in humans, vasopressin should be used with great caution for treatment of hypotension in septic shock.

Administration of low-dose vasopressin in porcine septic shock resulted in: