This study was supported by a starter grant from the Department of Anesthesiology at the Penn State Milton S. Hershey Medical Center (Pennsylvania State University College of Medicine). There was no funding received from any other organization.
Address reprint requests to Dmitri Bezinover, M.D., Ph.D., Department of Anesthesiology, Penn State Milton S. Hershey Medical Center, Pennsylvania State University College of Medicine, 500 University Drive, P.O. Box 850, H187, Hershey, PA 17033-0850. Telephone: 717-531-8433; FAX: 717-531-6221; E-mail: email@example.com
Hemodynamic instability during orthotopic liver transplantation (OLT) is a poorly understood clinical problem. In particular, profound hypotension and significant arrhythmias during liver graft reperfusion are frequent events despite careful volume and electrolyte adjustments. The management of this hemodynamic instability can be especially challenging because of its refractory nature when standard vasoactive agents are used. As a result, only after a comprehensive preoperative evaluation can a medically optimized patient (ie, a patient in stable condition with a favorable short- and long-term cardiac prognosis) be accepted as a transplant candidate.[1, 2] Despite this careful patient selection process, episodes of significant hypotension during reperfusion are still very frequent and, although they are usually of short duration, can significantly affect postoperative outcomes.[3, 4]
The release of nitric oxide (NO) from the vascular endothelium with the subsequent stimulation of cyclic guanosine monophosphate (cGMP) production has been proposed as a possible cause of hemodynamic instability during OLT.
We have hypothesized that cGMP in patients with end-stage liver disease (ESLD) is chronically elevated, and the degree of preoperative cGMP elevation is associated with intraoperative hemodynamic instability during OLT. Potentially, this information could be used to develop an effective strategy for perioperative hemodynamic management.
PATIENTS AND METHODS
This prospective, observational study was approved by the local institutional review board. Fifty consecutive patients undergoing OLT were enrolled after written informed consent was obtained. Two patients were excluded from the data analysis because of errors in cGMP assays. Four patients could not be evaluated because of incomplete data.
Anesthetic and Surgical Procedures
Anesthesia was provided according to a standardized departmental protocol. After initial noninvasive monitoring was established, general anesthesia was induced with the administration of propofol and fentanyl. Cisatracurium was used for muscle relaxation if rapid sequence induction was not required. For patients with significant ascites, a modified rapid sequence induction with rocuronium bromide was performed. After tracheal intubation, anesthesia was maintained with sevoflurane or desflurane, fentanyl, and a continuous infusion of cisatracurium.
Bilateral radial artery lines were placed. A femoral artery line was used if the placement of a radial artery catheter was not technically possible. The right internal jugular vein was cannulated with ultrasound guidance and a 9-Fr introducer. One or 2 rapid infusion catheters (7.5 or 8.5 Fr) were inserted into the cephalic or basilic veins of both arms. If the placement of a rapid infusion catheter was not possible, cannulation of the internal jugular vein on the left side was performed with a 7-Fr double-lumen catheter. For volume resuscitation, a Belmont rapid infusion system (Belmont Instrument Corp., Boston, MA) with double-lumen tubing was connected to the rapid infusion catheter and one of the central lines. If intraoperative continuous venovenous hemofiltration was deemed necessary, a left subclavian 12-Fr Shiley catheter was placed.
Coagulation was monitored with hourly measurements of the partial thromboplastin time, prothrombin time/international normalized ratio, fibrinogen level, and platelet count as well as thromboelastography. Hourly measurements of hemoglobin, hematocrit, lactate, and electrolytes were also routinely performed.
A standard surgical technique with complete cross-clamping of the inferior vena cava (IVC) was used with 30 patients, and a piggyback technique with partial clamping of the IVC was used in 14 patients.
One gram of methylprednisolone was infused over the course of 30 minutes at the beginning of the anhepatic phase.
At the beginning of graft reperfusion, each patient was standardized to the following parameters: a central venous pressure of 8 to 12 cm H2O, a mean arterial pressure greater than 65 mm Hg, a serum potassium level less than 4.5 mEq/L, and an arterial pH greater than 7.35. Coincident with the autologous blood flush of the liver graft (through the portal vein), a bolus of a colloid solution equal to the volume of the blood flush used was administered with the Belmont infuser. Flush blood was collected through a vena cava catheter placed in the donor vena cava at the end of the caval anastomosis. The amount of flush used was measured. The vena cava catheter was then removed, and the caval anastomosis suture was secured.
Samples and Catecholamine Requirement
The blood sampling times and locations were chosen in order to investigate cGMP levels in different locations and follow their distribution in the reperfusion period.
Blood samples were obtained at these time points during surgery:
Sample 1a, a preoperative sample, was obtained from the radial artery just before the surgical incision.
Sample 2 was obtained by the surgeon from the portal vein just before the portal vein anastomosis reconstruction.
Sample 2a was obtained from the radial artery just before the portal vein anastomosis reconstruction.
Sample 3 was obtained from the catheter placed in the IVC for liver graft flushing. The volume of flush blood was measured with a catheter placed in the donor vena cava, which drained into a graduated beaker at the beginning of liver flushing. This was measured immediately after the unclamping of the portal vein.
Sample 4a was obtained from the radial artery 20 minutes after liver graft reperfusion.
Norepinephrine (NE) was the only vasopressor routinely used to maintain hemodynamic stability. The amount of NE necessary to maintain a desired mean arterial pressure greater than 65 to 75 mm Hg was recorded during the period beginning with IVC unclamping and ending 20 minutes after reperfusion. The total amount of NE that was administered was compared to preoperative cGMP levels.
In this study, 16 grafts from extended criteria donors (ECDs) and 28 grafts from standard criteria donors (SCDs) were used. The ECD criteria included the following: donor age > 60 years, hypernatremia > 155 mmol/L, cold ischemia time > 15 hours, steatosis > 30%, aspartate aminotransferase/alanine aminotransferase levels > 100 U/L, intensive care unit (ICU) stay > 7 days, and downtime requiring cardiopulmonary resuscitation. We did not use any grafts from donation after cardiac death donors.
All blood samples were immediately placed on ice, and this was followed by centrifugation (1500 rpm for 15 minutes). The serum supernatant was removed, frozen, and stored at −80°C. Enzyme-linked immunosorbent assays (Cell Biolabs, Inc., San Diego, CA) were performed on each sample in the immunology laboratory to determine cGMP concentrations. These assays were performed with an anti-rabbit immunoglobulin G polyclonal antibody adsorbed onto a microtiter plate. Samples and standards containing cGMP were combined with a peroxidase-labeled cGMP tracer and a rabbit anti-cGMP polyclonal antibody. The peroxidase-labeled cGMP and the sample cGMP coupled to the rabbit anti-cGMP antibody and competed for binding on the microtiter plate. After incubation and washing, any bound peroxidase cGMP tracer was detected with the addition of a substrate solution. The colored product that formed was inversely proportional to the amount of cGMP present in a sample. Substrate reactions were terminated with acid, and the absorbance was measured at 450 nm. The cGMP concentration of the serum samples was determined by comparison to a cGMP standard curve with an effective range of 0.001 to 1000 μM/L. The interassay coefficient of variation was 6.9% to 11.3%, and the intra-assay coefficient of variation was 1.8% to 5.7%.
The cGMP enzyme immunoassay used in this investigation can be performed in clinical practice. The majority of clinical laboratories are equipped for the test. The only clinical limitation to performing this assay is that it takes 4 hours to complete.
Calculations and Statistics
A Pearson correlation coefficient and a regression analysis were used to detect any associations between preoperative cGMP levels and other factors, which included the following:
The amount of NE required to maintain the desired mean arterial pressure during graft reperfusion.
The Model for End-Stage Liver Disease score.
The length of the hospital stay.
The length of the postoperative ICU stay.
The transfusion requirement.
cGMP levels at different time points were compared with the Mann-Whitney rank-sum test. Recursive partitioning analysis was used to divide all participants into 2 groups (high and low preoperative levels of cGMP). The Mann-Whitney rank-sum test and Fisher's exact test were used to identify statistically significant differences between patient groups. Significance was demonstrated with a P value less than 0.05.
Fifty patients were initially enrolled in the study, and the results from 44 patients were evaluated. Six patients were excluded from the evaluation process for reasons cited previously.
A correlation analysis of the preoperative levels of cGMP in the systemic circulation and the amount of NE used during reperfusion revealed a significant Pearson correlation coefficient (r = 0.52). The association between these parameters was confirmed by regression analysis (P < 0.001; Fig. 1). We also demonstrated a significant correlation between the preoperative levels of cGMP and the length of the hospital stay (r = 0.38, P = 0.01) and the length of the postoperative ICU stay (r = 0.44, P = 0.004). There was no correlation between the preoperative levels of cGMP and the Model for End-Stage Liver Disease score (r = 0.04, P = 0.81) or between cGMP levels and transfusion requirements (r = 0.15, P = 0.33).
The level of cGMP in the systemic circulation that was obtained from the radial artery during the anhepatic phase (sample 2a) was increased in comparison with preoperative levels (sample 1a; P = 0.06). The concentration of cGMP in flush blood decreased at the beginning of reperfusion (sample 3) and then rose slightly in the systemic circulation after 20 minutes (sample 4a). These changes were not statistically significant (P = 0.16-0.41).
The concentration of cGMP in the portal vein (sample 2) was elevated in comparison with preoperative levels (sample 1a; P < 0.001). The level of cGMP in the portal vein was also significantly higher than the level in the radial artery during the anhepatic phase (sample 2a; P = 0.003) in samples collected at the same time. The cGMP level in the portal vein was also significantly elevated in comparison with the levels in flush blood (sample 3; P < 0.001) and in arterial blood 20 minutes after graft reperfusion (sample 4a; P < 0.001; Fig. 2).
The patients were divided into 2 subgroups on the basis of their preoperative plasma cGMP levels. This separation was determined by a statistical analysis of our data with recursive partitioning. This type of analysis was used to build a hierarchical, binary classification of the analyzed variables (the baseline cGMP level and the NE requirement) to select a dichotomization value of the independent variable (the cGMP level) that maximally discriminated the dependent variable (the NE requirement). This binary split provided a prediction rule that was used to classify subjects according to the probability of being a patient with a high or low NE requirement. According to the results of recursive partitioning, a preoperative cGMP level of 0.05 μmol/L was defined as a cutoff (P < 0.008). The patients were then retrospectively allocated into 2 groups: group 1 included 19 patients with high levels of preoperative cGMP (≥0.05 μmol/L), and group 2 included 25 patients with low levels (<0.05 μmol/L). We found statistical significance in preoperative cGMP levels (P < 0.001), NE requirements (P < 0.001), and postoperative ICU stays (P = 0.02). Comparisons of group data and demographics are presented in Table 1.
Table 1. Patient Demographics: A Comparison of the Groups
Group 1 (n = 19)
Group 2 (n = 25)
NOTE: Fifty patients were initially involved in the study. Six patients were excluded from the data collection.
aRanges are shown within parentheses.
bStatistically significant difference.
cThe standard procedure involved complete IVC cross-clamping; the piggyback procedure involved partial IVC cross-clamping.
A statistical analysis did not demonstrate any correlation between the duration of cold ischemia and NE requirements (P = 0.33). We also did not find any difference in NE requirements between recipients of SCD grafts and recipients of ECD grafts (P = 0.38).
Two surgical techniques [standard (n = 30) and piggyback (n = 14)] were used in the study. A comparison of preoperative cGMP levels and NE requirements with these 2 techniques failed to demonstrate a statistically significant difference (P = 0.90 for both techniques).
The results of our study suggest that NO release with the subsequent activation of cGMP might be one of the major causes of hemodynamic instability during OLT. Severe hypotension during liver graft reperfusion is a frequent event and was initially described as postreperfusion syndrome (PRS). PRS has been defined as a decrease in the systolic pressure greater than 30% in the first 5 minutes after graft reperfusion that lasts at least 1 minute.[3, 12, 13] Most recently, the incidence of PRS has been reported to be approximately 50%. One of the proposed causes of PRS is an increased production of NO. Other potential causes include the response to the release of inflammatory cytokines, particularly tumor necrosis factor α (TNF-α), from the liver graft, which is known to activate the production of cGMP.[16, 17]
Understanding NO synthesis dysregulation is an important component of identifying the cause of hemodynamic instability during OLT. cGMP is a significant mediator responsible for systemic vasodilatation and hypotension despite the compensatory increase in cardiac output. Vallance and Moncada developed a hypothesis with an animal liver cirrhosis model and suggested that the release of NO resulted in increased cGMP production. The formation of NO from L-arginine is catalyzed by a family of nitric oxide synthase (NOS) enzymes, including inducible, constitutive, and other isoforms. An increase in NOS availability can be produced by 2 mechanisms:
An increase in the rate of NOS gene translation and transcription activated by shearing forces in the portal system.[18, 19]
The activation of protein kinase B, which directly phosphorylates NOS. This is caused by portal shearing forces as well as ischemia. This leads to the activation of NOS with a subsequent increase in NO production. The importance of this mechanism has been demonstrated in the early stages of portal hypertension.
NO interacts with soluble guanylate cyclase (GC), and this leads to the subsequent conversion of guanosine triphosphate into cGMP. Although cGMP serves primarily as an intracellular mediator, it also diffuses through cellular membranes into the systemic circulation, where it can be measured.
A number of different phosphodiesterases (PDEs) are responsible for cGMP degradation by hydrolyzing cGMP into 5′-GMP. cGMP deactivation takes 10 to 40 seconds, with the time depending on the PDE type. The concentration of cGMP is maintained by the continuous stimulation of NOS production during the progression of ESLD. Different tissues contain a variety of PDE isomers. They belong to different families of cGMP-specific PDEs with overlapping activity. This is the reason that an increase in NOS production (and not an insufficiency of PDE release) is likely the main source of elevated cGMP. The rate of gene expression regulates PDE levels. This can be affected by multiple factors, including inflammatory stimulating factors, TNF-α, epinephrine, angiotensin II, isoproterenol, and others. It is important to note, however, that there is very little known about the regulation of PDE protein levels.[23, 25]
Possible causes of increased cGMP levels in patients with ESLD include the following:
The production of inducible NOS can be increased by macrophages, neutrophils, hepatocytes, and endothelial and vascular smooth muscle cells in response to endotoxin stimulation. In fact, high levels of endotoxins are frequently detected in patients with cirrhosis, even in the absence of an infection. Possible causes for this phenomenon are as follows:
Shunting of blood between the portal and systemic circulations, which bypasses the normal clearance of endotoxins by the reticuloendothelial system.
Reduced reticuloendothelial system activity in patients with cirrhosis.[26, 27]
Endotoxins also directly activate the production of various cytokines such as interleukin-1β, TNF-α, and interferon-γ. These cytokines are able to initiate iNOS production in target cells. Galley et al. demonstrated a positive correlation between the concentration of NOS and the stage of liver cirrhosis. At the same time, increases in NOS were associated with decreased systemic blood pressure. A similar trend was shown in multiple prior investigations.[6, 28]
Chronically increased portal pressure in patients with liver cirrhosis can also lead to the initiation of NO production. Several studies have demonstrated a correlation between elevated portal pressure and the release of cGMP.[29, 30] Patients with cirrhosis have a significant imbalance between increased portal pressure and low systemic hydrostatic pressure (decreased colloid osmotic pressure and a massive loss of fluid to ascites). This pressure difference leads to dramatic changes in hydrodynamic forces at the endothelium with subsequent activation of shearing force receptors. The results of our investigation support this theory. The concentration of portal vein cGMP in our study was significantly elevated in comparison with the concentrations at all other locations. Because of obstructed blood flow toward the liver, shearing forces are higher in the portal field versus other parts of the vascular system. This leads to the continuous activation of NOS production[20, 31, 32] and results in NO release from the vascular endothelium with subsequently increased cGMP levels.[9, 10]
The elevation of cGMP in the portal vein can also be considered a compensatory mechanism. This compensation is not very effective because of the ongoing obstruction of blood flow in the portal system. It is not clear what degree of portal pressure elevation is necessary to activate this mechanism.
It is unlikely that cross-clamping the portal vein itself with subsequent splanchnic ischemia was the primary cause of the increased portal cGMP levels. The time between portal vein clamping and blood sampling was 25 to 30 minutes. This time was insufficient to allow the increase in NOS synthesis that would have been necessary for cGMP production. Cross-clamping the portal vein, however, may have partially contributed to increases in cGMP in the portal vein system.
In our investigation, we were able to demonstrate considerable fluctuations in cGMP levels during different stages of surgery. The observed tendency toward increases in systemic levels of cGMP after the surgical incision might be related to the fact that cGMP is involved in several physiological responses related to the inflammation cascade activated at the beginning of surgery. Decreases in cGMP after graft reperfusion have been demonstrated in previous investigations. Bzeizi et al. found a significant decrease in cGMP after graft reperfusion. cGMP levels in their investigation were measured in the right atrium and correlated with increases in pulmonary vascular resistance as well as pulmonary arterial pressure. However, the reduction in cGMP was not associated with hemodynamic changes in the systemic circulation. Similar results were demonstrated in our investigation: a tendency toward a decreased cGMP concentration immediately after graft reperfusion without associated hemodynamic changes. The cause of reduced cGMP levels at this stage of surgery is not completely understood. It might be related to the release of free radicals from the liver graft into the systemic circulation.[34, 35] When free radicals interact with NO, superoxide anions are formed. Superoxide anions have an inhibitory effect on GC and can also directly induce vasoconstriction. This may explain the reported frequent arterial blood pressure increases during the 20 to 30 minutes following reperfusion.
In our study, we also demonstrated significant decreases in cGMP between portal and caval blood after graft reperfusion. This might be explained by partial sequestration of cGMP in the liver during graft irrigation.
Factors potentially affecting hemodynamic stability during liver transplantation have been investigated in previous studies. Kang et al. failed to demonstrate an association between the preoperative cardiac index and hemodynamic instability during graft reperfusion. Paugam-Burtz et al. found a correlation between the occurrence of PRS and the cold ischemia time in patients with cirrhosis. The type of surgical technique (eg, the absence of a portocaval shunt) has also been associated with an increased incidence of PRS. It has also been demonstrated that the quality of liver grafts can have an effect on hemodynamic stability.
In the present study, we have been able to demonstrate that preoperative levels of cGMP correlate with the amount of NE used to maintain the mean arterial pressure during liver graft reperfusion. This information potentially could be used to predict intraoperative hemodynamic instability.
Increased cGMP was the only identified factor with a significant effect on a patient's hemodynamic stability during reperfusion of the liver graft. Other factors such as the surgical technique, quality of the liver graft, and duration of cold ischemia were not associated with profound hypotension. Several intra-operative events such as the cardiodepressive action of anesthetic agents, acute volume shifts, partial or complete IVC cross-clamping, and release of inflammatory cytokines may result in acute hemodynamic decompensation. The degree of decompensation depends not only on these intraoperative events but also on preoperative physiological changes such as vascular predilatation. These changes in vascular tone may be related to increased baseline cGMP levels due to the progression of ESLD. The current literature strongly supports the hypothesis that the increase in cGMP production in patients with ESLD might reflect the progression of liver failure.
According to the preoperative levels of cGMP, patients in group 1 (preoperative cGMP level ≥ 0.05 μmol/L) were at risk for developing profound hypotension during the reperfusion stage. The NE requirements in group 1 were significantly increased in comparison with group 2 (preoperative cGMP level < 0.05 μmol/L). The concentration of cGMP in group 1 had a tendency to be elevated not only preoperatively but also during and after reperfusion.
Even though hypotension during liver graft reperfusion is a frequent occurrence and usually responds well to therapy, it can have a major impact on a patient's postoperative course and can affect survival. The significantly longer ICU stay for patients in group 1 may be related in part to the downstream effect of increased cGMP. The total hospital stay was also prolonged in this group. A correlation analysis also confirmed the trend that patients with increased preoperative cGMP levels had prolonged ICU and hospital stays. The exact mechanism by which cGMP affects a patient's postoperative course is unknown. It is possible that the physiological changes related to the more advanced systemic vasodilatation seen in group 1 prolong recovery.
The NO/cGMP signaling system participates in the function of the N-methyl-D-aspartate (NMDA) receptor. This is a reason that anesthesia medications can potentially affect the final cGMP concentration. It has been demonstrated that volatile and intravenously administered anesthetics can inhibit the action of the NO/cGMP signaling system. The exact mechanism of action is not completely understood, but tissue-specific interactions of anesthetics with NOS and GC along with the subsequent inhibition of synaptic transmission and suppression of NMDA receptor function have been discussed in the literature. It has also been demonstrated that both volatile and intravenous anesthetics can diminish vasodilatation because of a reduction in NO synthesis and decreased neuronal NOS activity. The actions of different anesthetics are, however, not absolutely equal and are tissue-dependent. Isoflurane has been demonstrated to directly inhibit NO production and suppress NMDA receptors. Halothane has been shown to have an indirect activating function on NMDA receptors. It has also been demonstrated that cGMP-mediated pulmonary vasodilatation is preserved during phenobarbital anesthesia but not during halothane anesthesia. The interaction of anesthetics with NO/cGMP signaling systems can affect cGMP levels in the body. Kant et al. demonstrated that under 2% halothane anesthesia, the concentration of cGMP in different parts of the brain was reduced by 10% to 60%. It is unclear whether all general anesthetics have this effect on the systemic concentration of cGMP. We cannot completely rule out an anesthesia medication effect on the final results of our investigation.
That the cGMP pathway plays an important role in hemodynamic stability during liver transplantation has potential therapeutic implications. It has been shown that methylene blue (MB) can be used for the treatment of cGMP-mediated vasodilatation. MB is able to inhibit iNOS with subsequent suppression of NO release. MB also binds soluble GC and interrupts the cGMP activation pathway.[48, 49] The effectiveness of MB for the treatment of refractory hypotension during liver transplantation has been demonstrated.
In summary, the mechanism of hemodynamic instability during OLT is most certainly multifactorial, but dysregulation of NO production most likely plays an important role. Our study confirms that cGMP is one of the factors contributing to intraoperative hypotension during OLT. We have also been able to demonstrate that preoperative cGMP levels might have a possible prognostic value. If further prospective investigations confirm the positive predictive value of preoperative levels of cGMP with respect to intraoperative hemodynamic instability, a patient-specific management plan can be established.